Legal Tech and The New Sharing Economy

Perspectives in Law, Business and Innovation
Marcelo Corrales Compagnucci

Nikolaus Forgó
Toshiyuki Kono
Shinto Teramoto
Erik P. M. Vermeulen Editors
Legal Tech and

the New Sharing
Economy
Series Editor
Toshiyuki Kono, Faculty of Law, Kyushu University, Fukuoka, Japan
Over the last three decades, interconnected processes of globalization and rapid technological
change—particularly, the emergence of networked technologies—have profoundly disrupted
traditional models of business organization. This economic transformation has created
multiple new opportunities for the emergence of alternate business forms, and disruptive
innovation has become one of the major driving forces in the contemporary economy.
Moreover, in the context of globalization, the innovation space increasingly takes on a global
character. The main stakeholders—innovators, entrepreneurs and investors—now have an
unprecedented degree of mobility in pursuing economic opportunities wherever they arise.
As such, frictionless movement of goods, workers, services, and capital is becoming the
“new normal”.
This new economic and social reality has created multiple regulatory challenges for
policymakers as they struggle to come to terms with the rapid pace of these social and
economic changes. Moreover, these challenges impact across multiple fields of both public
and private law. Nevertheless, existing approaches within legal science often struggle to deal
with innovation and its effects.
Paralleling this shift in the economy, we can, therefore, see a similar process of disruption
occurring within contemporary academia, as traditional approaches and disciplinary
boundaries—both within and between disciplines—are being re-configured. Conventional
notions of legal science are becoming increasingly obsolete or, at least, there is a need to
develop alternative perspectives on the various regulatory challenges that are currently being
created by the new innovation-driven global economy.
The aim of this series is to provide a forum for the publication of cutting-edge research in
the fields of innovation and the law from a Japanese and Asian perspective. The series will
cut across the traditional sub-disciplines of legal studies but will be tied together by a focus
on contemporary developments in an innovation-driven economy and will deepen our
understanding of the various regulatory responses to these economic and social changes.
The series editor and editorial board carefully assess each book proposal and sample
chapters in terms of their relevance to law, business, and innovative technological change.
Each proposal is evaluated on the basis of its academic value and distinctive contribution to
the fast-moving debate in these fields.
Series Editor
Toshiyuki Kono (Professor, Faculty of Law, Kyushu University, Fukuoka, Japan)
Editorial Board
Erik P. M. Vermeulen (Professor of Business & Financial Law, Tilburg University & Philips
Lighting, The Netherlands)
Claire Hill (James L. Krusemark Chair in Law, University of Minnesota Law School, USA)
Wulf A. Kaal (Associate Professor & Director of the Private Investment Institute, University
St. Thomas, USA)
Ylber A. Dauti (Founding Partner The Dauti Law Firm, PC, USA)
Pedro de Miguel Asensio (Professor, Complutense University of Madrid, Spain)
Nikolaus Forgó (Professor, University of Vienna, Austria)
Shinto Teramoto (Professor, Kyushu University, Japan)
Urs Gasser (Executive Director, Berkman Klein Center for Internet & Society at Harvard
University; Professor of Practice, Harvard Law School, USA)
More information about this series at http://www.springer.com/series/15440

Marcelo Corrales Compagnucci•
Nikolaus Forgó Toshiyuki Kono

• •
Shinto Teramoto Erik P. M. Vermeulen

•
Editors
Legal Tech and the New

Sharing Economy
123
Editors
Marcelo Corrales Compagnucci Nikolaus Forgó
Center for Advanced Studies in Biomedical Department of Innovation
Innovation Law (CeBIL) and Digitalisation in Law
Faculty of Law, University of Copenhagen University of Vienna
Copenhagen, Denmark Vienna, Austria
Toshiyuki Kono Shinto Teramoto

Faculty of Law Faculty of Law
Kyushu University Kyushu University
Fukuoka, Japan Fukuoka, Japan
Erik P. M. Vermeulen
Department of Business Law
Tilburg University
Tilburg, The Netherlands
ISSN 2520-1875 ISSN 2520-1883 (electronic)

ISBN 978-981-15-1349-7 ISBN 978-981-15-1350-3 (eBook)
https://doi.org/10.1007/978-981-15-1350-3
© Springer Nature Singapore Pte Ltd. 2020
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface
This volume is part of the book series: Perspectives in Law, Business and
Innovation. The aim of this series is to provide a forum for the publication of
cutting-edge research in the fields of innovation and law from a global perspective.
The series aims to cut across the traditional sub-disciplines of legal studies, but is
tied together by a focus on deepening our understanding of the various responses to
technological, economic, and social change.
This volume constitutes the result of a joint cooperative effort drawing on the
extensive global network of four academic institutions: The Centre for Advanced
Studies in Biomedical Innovation Law (CeBIL), Faculty of Law, University of
Copenhagen (Copenhagen, Denmark); the Department of Innovation and
Digitalization in Law, University of Vienna (Vienna, Austria); the Faculty of Law,
Kyushu University (Fukuoka, Japan); and the Department of Business Law, Tilburg
University (Tilburg, The Netherlands).
Contributors to this book—including business, software engineering, legal
scholars, and practitioners from Europe, East Asia, and the Americas—attempt to
provide some of the latest thinking and assessment of current challenges and
opportunities with regard to Legal Tech and the new sharing economies.
This book is designed for anyone interested in an up-to-date legal analysis of
current trends related to new technologies and the emerging sharing economies. In
particular, legal scholars, students, practitioners, software developers, managers, and
experts looking to deepen their understanding of this “game-changing phenomena.”
The editors would like to thank the authors and co-authors of each chapter for
their hard work, patience, and cooperation throughout the whole process from
initial conception to the final manuscript. Finally, the editors are grateful to the
Springer staff for their support and efforts in ensuring the final publication.
Copenhagen, Denmark Marcelo Corrales Compagnucci

Vienna, Austria Nikolaus Forgó
Fukuoka, Japan Toshiyuki Kono
Fukuoka, Japan Shinto Teramoto
Tilburg, The Netherlands Erik P. M. Vermeulen
v
Contents
Legal Aspects of Decentralized and Platform-Driven Economies . . . . . . 1

Marcelo Corrales Compagnucci, Toshiyuki Kono and Shinto Teramoto
Sharing Economy and Platforms

Building Platforms for Collaboration: A New Comparative Legal
Challenge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Annelise Riles
Sharing City Seoul and the Future of City Governance . . . . . . . . . . . . . 21
Benjamen Franklen Gussen
Digital Age and Personal Data

The Digital Person—A New Legal Entity? On the Role of Law
in an AI-Based Society . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Cecilia Magnusson Sjöberg
Nudging Consent and the New Opt-Out System to the Processing
of Health Data in England . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Janos Meszaros, Chih-hsing Ho and Marcelo Corrales Compagnucci
National Electronic Health Record Systems and Consent
to Processing of Health Data in the European Union
and Australia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Danuta Mendelson
Blockchain and Code

Legal Education in a Digital Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Mark Fenwick, Wulf A. Kaal and Erik P. M. Vermeulen
Reputation Protocol for the Internet of Trust . . . . . . . . . . . . . . . . . . . . 123
Craig Calcaterra and Wulf A. Kaal
vii
viii Contents
Legal Issues Related to Blockchain Technology—Examples

from Korea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Gyooho Lee
Autonomous Systems and Future Challenges

Lawyers’ Views on Autonomous Driving . . . . . . . . . . . . . . . . . . . . . . . . 169
Shinto Teramoto
Adversarial Machine Learning: A Blow to the Transportation
Sharing Economy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Steven Van Uytsel and Danilo Vasconcellos Vargas
The New EU Directive on the Protection of Trade Secrets
and Its Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Thomas Hoeren
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Editors and Contributors
About the Editors
Marcelo Corrales Compagnucci is attorney-at-law specializing in intellectual

property (IP), information technology (IT), and corporate law. His research interests
are the legal issues involved in disruptive innovation technologies. He is currently a
Postdoctoral Researcher at the Center for Advanced Studies in Biomedical
Innovation Law (CeBIL), Faculty of Law, University of Copenhagen in Denmark.
His past activities have included working as a consultant and lawyer for law firms
and IT companies. He was also a research associate with the Institute for Legal
Informatics (IRI) at Leibniz Universität Hannover in Germany, and a visiting
research fellow in various research centers around the world, including the Max
Planck Institute for Comparative and International Private Law (Hamburg), the Max
Planck Institute for Innovation and Competition (Munich), the Shepherd and
Wedderburn Centre for Research in Intellectual Property and Technology
(SCRIPT) within the University of Edinburgh in Scotland, and the Academia Sinica
in Taiwan. He has a Doctor of Laws (LL.D.) degree from Kyushu University in
Japan. He also holds a Master of Laws (LL.M.) in international economics and
business law from Kyushu University, and an LL.M. in law and information
technology and an LL.M. in European intellectual property law both from the
University of Stockholm in Sweden. He has several publications in the field of
IT-Law and IP-Law. His most recent publications include New Technology, Big
Data and the Law (Springer, 2017); Robotics, AI and the Future of Law (Springer,
2018); and Legal Tech, Smart Contracts and Blockchain (Springer, 2019).
Nikolaus Forgó born 1968, is Professor of IT-Law and IP-Law and head of the
department of innovation and digitalization in law at the University of Vienna Law
School. He studied law in Vienna and Paris with a Ph.D. in legal theory. From 1990
to 2000, he served as assistant professor and delegate of Information Technology at
University of Vienna Law School. From 2000 to 2017, he was full professor for
Legal Informatics and IT-Law at Leibniz Universität Hannover (Germany),
ix
x Editors and Contributors
between 2010 and 2017 head of department, and between 2013 and 2017, he also
served as data protection officer and chief information officer of this university. He
regularly teaches and consults in all fields of IT-Law, legal informatics, civil law,
and legal theory and has been responsible for more than 50 research projects
including more than 20 EU-funded research projects with overall budgets
exceeding 100 Mio €. He has specific interest and expertise in IT-related trans-
formation of law, in particular, in data protection, privacy, governance, and intel-
lectual property.
Toshiyuki Kono is a distinguished professor at Kyushu University (Fukuoka,

Japan). He currently is a titular member of the International Academy of
Comparative Law and serves as the president for the Academy of Commercial and
Consumer Law, as well as the chair of the Committee for Intellectual Property and
Private International Law at the International Law Association. In 2019, he received
the Reimar-Lust Award from the Alexander von Humboldt Foundation for his
academic excellence and contribution to the bilateral relationship between Germany
and Japan. In the field of private international law, he gave special lectures on
“Efficiency in Private International Law” at the Hague Academy of International
Law in 2013. He was selected as one of three lecturers who are invited to publish
lectures as a monograph in its Pocket Book series in 2014. He also serves as the
president of the International Council of Monuments and Sites (ICOMOS), an
advisory body of UNESCO, since 2017, after he served as a vice president of
ICOMOS from 2014 through 2017. He has been active in UNESCO as an inde-
pendent expert as well. For example, he served in 2010 as the chairperson of the 3rd
General Assembly of the State Parties of the UNESCO Convention for the
Safeguarding of Intangible Cultural Heritage and chairperson of the Legal
Committee of the 34th UNESCO General Conference in 2007. His recent publi-
cations in the field of international heritage law include “Authenticity, notions and
principles,” Change over Time, Fall 2014, Vol. 4, No. 2, pp. 436–460.
Shinto Teramoto born 1963 and having over 25 year experience as a practicing
attorney, is devoting himself to promote the social implementation of the products
of engineering, medicine, and other disciplines, while is very active in educating
young legal professionals and candidates for lawyers. He passed the bar exam in
1984 while he was an undergraduate student and represented and gave legal advices
to a number of venture capitalists, investment fund managers, and technology
ventures as their attorney. Also, he is a board member of two commercial com-
panies, as well as multiple non-profit organizations and foundations. His profound
experience contributes much to make the courses provided by him or jointly by him
and his colleagues very practical and effective, including those provided for the
students of International Programs in Law as described below, as well as intel-
lectual property courses provided for Japanese Law School students, business law
courses provided for the Business School students, and legal training courses for
undergraduate students who are preparing themselves to enter law schools.
Although his study and contribution to society, as well as educations provided to
Editors and Contributors xi
students, cover very wide scope from intellectual property, diffusion of medical and
healthcare information with better quality, to sharing of medical records among
patients and multiple medical teams, always the essence thereof is the question that
how we can promote the wider and quicker diffusion of knowledge and innovation.
In order to respond to this question, he is combining the perspective from social
network with empirical knowledge derived from legal practices.
Erik P. M. Vermeulen is a Professor of Business and Financial Law at Tilburg

University in The Netherlands, Senior Legal Counsel at Signify (formerly known as
Philips Lighting) and an Innovation Advisor. He can best be described as an
innovator. He has a particular interest in how emerging technologies, such as
artificial intelligence, sensors, and blockchain, are creating opportunities across all
areas of the economy and society. In this fast-changing world, he revisits long-held
assumptions, theories, and business models and seeks to develop new models more
appropriate to the new world. His thought-provoking and innovative views have
attracted international attention. He regularly serves as an expert advisor to inter-
national organizations, such as the European Commission, the Organization for
Economic Co-operation and Development, the United Nations, the World Bank,
and national and local governments around the world. He is an Innovation Advisor
at a law firm (Pels Rijcken & Drooglever Fortuijn) in the Hague, and a
board/advisory member of several companies/organizations, including a healthcare
provider in The Netherlands and several start-ups. He teaches regularly at univer-
sities in Europe, the USA, Colombia, and Japan and has appeared at numerous
conferences as a featured or keynote speaker. He has a blog at medium.com/
@erikpmvermeulen.
Contributors
Craig Calcaterra Department of Mathematics, Metropolitan State University,

Minneapolis, USA
Marcelo Corrales Compagnucci Center for Advanced Studies on Biomedical
Innovation Law (CeBIL), Faculty of Law, University of Copenhagen, Copenhagen,
Denmark
Mark Fenwick Faculty of Law, Kyushu University, Fukuoka, Japan
Benjamen Franklen Gussen The Swinburne School of Law, Swinburne
University of Technology, Victoria, Australia
Chih-hsing Ho Institute of European and American Studies, Academia Sinica,
Taipei, Taiwan
Thomas Hoeren Institute for Information, Telecommunication and Media Law,
Faculty of Law, University of Münster, Münster, Germany
xii Editors and Contributors
Wulf A. Kaal University of St. Thomas School of Law, Minneapolis, USA

Toshiyuki Kono Faculty of Law, Kyushu University, Fukuoka, Japan
Gyooho Lee Institute for Culture, Media, and Entertainment Laws, School of Law,
Chung-Ang University, Seoul, Republic of Korea
Cecilia Magnusson Sjöberg Faculty of Law, Stockholm University, Stockholm,
Sweden
Danuta Mendelson Deakin Law School, Deakin University, Melbourne, VIC,
Australia
Janos Meszaros Institute of European and American Studies, Academia Sinica,
Taipei, Taiwan
Annelise Riles Buffett Institute for Global Affairs and Pritzker School of Law,
Northwestern University, Chicago, Illinois, USA
Shinto Teramoto Faculty of Law, Kyushu University, Fukuoka, Japan
Steven Van Uytsel Faculty of Law, Kyushu University, Fukuoka, Japan
Danilo Vasconcellos Vargas Faculty of Information Science and Electrical
Engineering, Kyushu University, Fukuoka, Japan
Erik P. M. Vermeulen Department of Business Law, Tilburg University, Tilburg,
The Netherlands;
Legal Department, Signify (formerly Philips Lighting), Amsterdam, The Netherlands
Acronyms
ADHA Australian Digital Health Agency

AI Artificial Intelligence
AVs Autonomous Vehicles
B2P Business-to-Peer
BCE Before the Common Era
BetrVG Works Constitution Act (Germany)
BGB German Civil Code (Germany)
BGH Federal Supreme Court (Germany)
BIT Behavioral Insights Team
BMJV Federal Ministry of Justice (Germany)
CCBG Citizens Coalition for Better Government
CCEJ Citizens’ Coalition for Economic Justice
CEO Chief Executive Officer
CLDC Common Law Duty of Confidentiality
CMD Civic Movement for Decentralization
CPU Central Processing Unit
CTC Centralized Train Control
DAG Directed Acyclic Graph
DAOs Decentralized Autonomous Organizations
DL Deep Learning
DLT Distributed Ledger Technology
DNN Deep Neural Network
ELABs Ethical and Legal Advisory Boards
EPC European Patent Convention
EU European Union
EU GDPR European Union General Data Protection Regulation
G2P Government-to-Peer
GP General Practitioner
GPS Geographical Positioning System
HIV Acquired Immune Deficiency Syndrome
xiii
xiv Acronyms
ICCPR United Nations International Covenant on Civil and Political

Rights
ICESCR International Convenant on Economic, Social and Cultural
Rights
ICO Information Commissioner’s Office
ICT Information and Communication Technology
IoT Internet of Things
IP Internet Protocols
ISO International Organization for Standardization
IT Information Technology
JHABA Japanese Housing Accommodation Business Act
(Act No. 65 of 2017)
LAA Local Autonomy Act 1947 (Korea)
LIDAR Light Detection and Ranging
M2M Machine-to-Machine
MBA Master of Business Administration
MHLW Ministry of Health, Labor and Welfare (Japan)
MHR My Health Record
ML Machine Learning
MOPAS Minister of Public Administration and Safety
NDG National Data Guardian
ND opt-out National Data Opt-out System
NEHR National Electronic Record Schemes
NeHTA National E-Health Transition Authority
NGOs Non-Governmental Organizations
NHS National Health Service
NIO National Infrastructure Operator
OECD Organization for Economic Cooperation and Development
P2P Peer-to-Peer
PatG Patent Law (Germany)
PCD Presidential Commission for Decentralization
PCEHR Personally Controlled Electronic Health Record
PCGID Presidential Committee on Government Innovation and
Decentralization
PCs Personal Computers
PoR Proof of Reputation
PoS Proof of Stake
PoW Proof of Work
PSPD People’s Solidarity for Participatory Democracy
SMG Seoul Metropolitan Government
SPoS SEMADA Proof of Stake Protocol
SRI Semada Research Institute
StGB Criminal Code (Germany)
TAM Total Available Market
TCP Transmission Control Protocol
Acronyms xv
TRIPS Agreement Agreement on Trade-Related Aspects of Intellectual Property

Rights
UK United Kingdom
UrhG Copyright Law (Germany)
US United States
UWG Unfair Competition Act (Germany)
WHO World Health Organization
WoT Web of Trust
Legal Aspects of Decentralized
and Platform-Driven Economies
Marcelo Corrales Compagnucci, Toshiyuki Kono and Shinto Teramoto
Abstract The sharing economy is sprawling across almost every sector and activity
around the world. About a decade ago, there were only a handful of platform-driven
companies operating on the market. Zipcar, BlaBlaCar and Couchsurfing among
them. Then Airbnb and Uber revolutionized the transportation and hospitality indus-
tries with a presence in virtually every major city. “Access over ownership” is the
paradigm shift from the traditional business model that grants individuals the use of
products or services without the necessity of buying them. Digital platforms, data and
algorithm-driven companies as well as decentralized blockchain technologies have
tremendous potential. But they are also changing the “rules of the game.” One of
such technologies challenging the legal system are AI systems that will also reshape
the current legal framework concerning the liability of operators, users and manu-
facturers. Therefore, this introductory chapter deals with explaining and describing
the legal issues of some of these disruptive technologies. The chapter argues for a
more forward-thinking and flexible regulatory structure.
Keywords Sharing economy · Platforms · AI · Blockchain · Data protection ·

Autonomous vehicles
M. Corrales Compagnucci (B)

Center for Advanced Studies on Biomedical Innovation Law (CeBIL), Faculty of Law, University
of Copenhagen, Copenhagen, Denmark
e-mail: marcelo.corrales13@gmail.com
T. Kono · S. Teramoto
Faculty of Law, Kyushu University, Fukuoka, Japan
© Springer Nature Singapore Pte Ltd. 2020 1

M. Corrales Compagnucci et al. (eds.), Legal Tech and the New Sharing Economy,
Perspectives in Law, Business and Innovation,
https://doi.org/10.1007/978-981-15-1350-3_1
2 M. Corrales Compagnucci et al.
1 Introduction
The sharing economy is a new model of organizing economic activity that may
substitute traditional corporations and capitalism around the world. This activity
is based on acquiring, providing or sharing access to goods and services based on
underutilized assets.1 The sharing economy is facilitated by a community based on
digital platforms that enable people who have never met before to share resources
and trust each other. Information and innovation technologies are used in order to
match individuals who possess such superfluous resources with existing demand in
the market.2
Decentralized platforms are sometimes associated with “peer-to-peer” (P2P) tech-
nologies (such as Napster in the late 1990s). However, it is characterized by two very
different business models: (i) Sometimes individuals own and get to share their
assets with each other, like extra rooms (Airbnb3 and Homestay),4 cars (Uber,5 Lyft
and Turo),6 parking spaces (Just Park),7 skills (Taskrabitt)8 and even their own pets
(Rover)9 ; (ii) In other cases, companies own and lend out the assets, such as cars,
bicycles and motorbikes. Prime examples are Zipcar10 and Car2Go11 now owned by
Avis and Daimler-Benz, respectively.12
It is indisputable that the sharing economy produces an enormous amount of
wealth. In 2015, PriceWaterhouseCoopers projected growth from $15 billion in
global revenue to $335 billion in 2025.13 Faster than ever, we have barely begun
to scratch the surface of the possibilities of innovation and dynamic capability
behind this new economic model.14 However, the extraordinary growth of the shar-
ing economy creates unprecedented legal problems. This book anthology takes up
various disruptive technologies that are currently transforming the legal system
around the world. Such technologies include: cloud computing, Big Data, Internet of
Things (IoT), artificial intelligence (AI), machine learning (ML), deep learning (DL),
blockchain, algorithms and other related autonomous systems—such as self-driving
vehicles.
1 Sundararajan (2016).
2 See, e.g., Malik and Wahaj (2019), p. 249 et seq.
3 Maurer (2016), p. 6.
4 See Homestay. Available at: https://www.homestay.com. Accessed 10 June 2019.
5 See Uber. Available at: https://www.uber.com. Accessed 10 June 2019.
6 See Turo. Available at: http://www.turo.com. Accessed 10 June 2019.
7 See Justpark. Available at: https://www.justpark.com. Accessed 10 June 2019.
8 See Taskrabbit. Available at: https://www.taskrabbit.com. Accessed 10 June 2019.
9 See Rover. Available at: https://www.rover.com. Accessed 10 June 2019.
10 See Zipcar. Available at: https://www.zipcar.com. Accessed 10 June 2019.
11 See Car2go. Available at: https://www.car2go.com/US/en/. Accessed 10 June 2019.
12 Cusumano (2018), pp. 26–28.
13 PriceWaterhouseCooper (2015), pp. 1–30.
14 Gazzola (2017), pp. 75–94.
Legal Aspects of Decentralized and Platform-Driven Economies 3
There is no doubt that the expansion of the sharing economy is changing our
world. It has also triggered the emergence of new products and services. As a result,
companies are becoming increasingly more data and algorithm-driven, making use
of so-called “decentralized platforms.” New transaction and/or payment methods
such as Bitcoin, Ethereum, etc., which are based on trust building systems using
blockchain, smart contracts and other distributed ledger technology (DLT) also con-
stitute an essential part of such a new economic model and central to the analysis of
this work.
Most of these digital platforms within the sharing economy rely on cloud-based
infrastructures to operate at the upper level. This paradigm shift would not be pos-
sible without the adoption of cloud computing deployment models and services.
Individuals and companies in general, are fast gearing up for the “on-demand” and
“pay-as-you-go” culture, which constitute the building blocks of cloud computing
transactions.15
All these new technological breakthroughs have brought complex ways of pro-
cessing and analyzing information at a larger scale.16 From a legal perspective, the
uncertainties triggered by the emergence of a new digital reality are particularly
urgent. How should these tendencies be reflected in legal systems in each jurisdic-
tion? This collection brings together a series of contributions by leading scholars in
the emerging field of the sharing economy. The aim of this book is to enrich legal
debates on the social, economic and political meaning of this new economic model
along with these cutting-edge technologies.
It is indisputable that this paradigm shift is changing the scope in which law is
designed, interpreted and applied in a constantly evolving environment. There is,
therefore, an increasing awareness that the traditional concepts and approaches of
the law must be more flexible and expanded to encompass new areas associated
to this new economic model. Based on this new reality, this work aims to provide
insights on some of the key legal topics that will affect the future of our daily lives.
The chapters presented in this edition attempt to answer some of these questions
from the perspective of different legal backgrounds. The aim is to answer some of
these questions from an inter-disciplinary and integrated point of view taking into
account a variety of legal systems.
2 Parts
Addressing the many challenges created by the sharing economy requires going
beyond one single disciplinary perspective or frame of reference. As such, after this
introductory chapter, the book is divided into four parts comprising 12 chapters as fol-
lows: Part I—“Sharing Economy and Platforms;” Part II—“Digital Age and Personal
Data;” Part III—“Blockchain and Code;” and, Part IV—“Autonomous Systems and
15 Tandon (2018).
16 See, e.g., Chen et al. (2014), p. 12 et seq.
Future Challenges.” Each part focuses on one particular area of the sharing economy
by adopting different approaches and methods.
Part I—“Sharing Economy and Platforms”—focuses on the impact of the sharing
economy as an emerging economic model. The chapters in this section cut across
different aspects of digital platforms, including platforms for building collaboration
and the governance of cities in real world settings using the use case examples of
sharing cities in Seoul and the study of the Japanese housing accommodation legal
system.
Part II—“Digital Age and Personal Data”—has the digital person and the pro-
tection of personal data as its overarching subject. The starting point is that there
is the need to discuss the concept of “digital person” as a new legal entity in light
of the development and application of AI technology. This takes us directly to the
necessity of changing the legal education system and the role of future lawyers as
this will affect the design of the new global architecture. This part also focuses on the
protection of the personal data of individuals in the context of health systems. It does
this by comparing different approaches of consent under the EU General Data Pro-
tection Regulation (GDPR)17 and the specific situations in Australia and the United
Kingdom (UK). The GDPR has been generally well received for strengthening some
of the rules in the previous EU Data Protection Directive18 by granting individuals
more control over their data when using electronic health records. However, it has
also generated hot debate around the world regarding its practicability and flexibility
within modern processing technologies.
Part III—“Blockchain and Code”—discuss blockchain technologies and how
learning the legal issues embedded in the “code” of computer software can help
legal professionals to reinvent themselves. Some think that lawyers or even bankers
and notaries might soon have to offer new services and change the way they do
business otherwise they might become redundant and obsolete.19 The blockchain
is one of the most hyped terms of this new century and it has been said that it can
revolutionize the world.20 Overall, the chapters of this part discuss some of the main
features of blockchain technology to increase trust and transparency in decentralized
networks.
Part IV—“Autonomous Systems and Future Challenges”—looks into some of the
legal implications of autonomous vehicles. Autonomous driving represents a crucial
part of the mobility of the future. However, there are still vexed legal issues. What
17 Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on
the protection of natural persons with regard to the processing of personal data and on the free
movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation).
While the Regulation entered into force on 24 May 2016, it applies to all EU Member States from
25 May 2018. See European Commission, Reform of EU Data Protection Rules https://eugdpr.org.
Accessed 10 June 2019.
18 Directive 95/46/EC of the European Parliament and of the Council of 24 October 1995 on the
protection of individuals with regard to the processing of personal data and on the free movement
of such data.
19 Vermeulen (2017).
20 Corrales et al. (2019), p. 2 et seq.
happens for example when there is a car accident? Who should be liable for it? Is it
the AI or automated system software developer? Or, is it the auto manufacturer who
assembled the car pieces together? How shall we handle the insurance companies?
In sum, how can we reduce and mitigate these legal risks?21 These are just some of
the main questions that come to our minds immediately when we are talking about
driverless cars. The chapters in this section explore some of the legal problems and
suggest that this will affect the sharing economy, which might not be ready yet to face
these problems. Finally, the remaining chapter of this part discusses the protection
of trade secrets in light of the new EU Directive on the protection of trade secrets
and outlines the options for implementation.
3 Chapters
After this introduction, the book comprises twelve substantive chapters. Part I—
Sharing Economy and Platforms—consists of three contributions.
Annelise Riles starts by explaining the collaborative genius of today’s intercon-
nected world. The image of the lonely genius working by himself in his office is
largely over. The pervasive and dynamic nature of our current society presents dif-
ficult technical and legal questions. Transactions frequently take part in complex
relationships where several actors are involved across different jurisdictions. There-
fore, the genius of our time is a collaborative genius. From the business world to the
academy, and from the leading financial centers to grassroots development projects
around the world, collaboration is increasingly perceived as a necessity. This chapter
deals with some of these challenges, and also some of the possibilities that are inher-
ent in collaboration, taking into account the example of a recent experiment with
Meridian 180, a global engagement platform for policy experimentation founded in
2011.
Benjamen Franklen Gussen retains a focus on the impact of the sharing econ-
omy on the governance of cities using the lessons learnt from the Sharing City
Seoul Project. The chapter starts by explaining the historical grounds of collabora-
tion among societies. The analysis suggests that collaboration was always present in
different forms from the beginning of civilization. The nature of collaboration, how-
ever, shifted to a hierarchical organizational model as a result of increasing population
density. This led to the emergence of new thriving cities and markets. The arrival of
new waves of technological innovation in the 21st century, however, brought with it a
new change to a network organization and collaboration. This innovation allows for
a return to a network organization at a scale (population-density) never seen before.
Seoul is a very good example of this paradigm shift as it is one of the leading cities
when it comes to innovation in the sharing economy. Its experience has influenced
already other major cities across the world. The chapter explains the policy inter-
ventions that took place in Seoul to inform future approaches to governing cities in
21 https://www.daimler.com/innovation/case/autonomous/legal-framework.html.
other countries and looks at the role of law in enabling cities as a supple network of
digital technologies that stimulate innovation within the sharing economy.
Yuichiro Watanabe focuses on a study of the Japanese Housing Accommodation
Business Act (Act No. 65 of 2017), which is the first national law in the world legit-
imizing home-sharing. The starting point of this chapter is to explain the making
process of the new Japanese Housing Accommodation Business Act, which came
into effect in 2018 after 3 years of negotiations. The chapter also explains the pre-
vious regulatory framework which had remained almost untouched since 1948. The
chapter identifies and reviews three legal issues: (a) extraterritorial application out-
side of Japan; (b) lacking the consistency between the other existing Japanese laws
in terms of regulating digital platforms; and (c) illegal local ordinance beyond the
Act, preventing the sharing economy by imposing additional restrictions. Finally, the
author suggests that the Act might not be the best model to be implemented in other
jurisdictions.
Part II—Digital Age and Personal Data—comprises three chapters.
Cecilia Magnusson Sjöberg discusses the concept of “digital person” as a new legal
entity taking into consideration the recent development and application of AI tech-
nology. The purpose is not to just add another term to those of the natural person
and the legal person, but to introduce a concept that could eventually, under certain
circumstances, be implemented in the legal order. This new legal figure could also,
tentatively, be granted legal capacity, with rights and responsibilities. This new legal
person i.e., the “digital person” would as such be possible to describe as a constella-
tion of algorithms consisting of a basic algorithmic identity, which could be profiled
and specified with reference to various purposes. The risk is otherwise an emerging
dysfunctional legal society where there is no legal entity, which can take the role of
the subject. Issues that arise concern e.g., self-driving car liability, pricing algorithms
on the competitive market and data protection when profiling consumers.
In their chapter, Janos Mészáros, Chih-Hsing Ho and Marcelo Corrales Compag-
nucci examine the challenges of the revised opt-out system and the secondary use
of health data in the UK. The secondary use of data refers to the processing of data
collected during direct care for new purposes, such as research and policy planning.
The analysis of this data could be very valuable for science and medical treatment as
well as the discovery of new drugs. For this reason, the UK government established
the “care.data program” in 2013. The aim of the project was to build a central nation-
wide database for research and policy planning. However, the processing of personal
data has been planned without proper public engagement, and the central database
was aimed to be used by public and private third-parties, including IT companies.
The care.data program established a double opt-out system which turned out to be
very controversial due to scandals such as the Google DeepMind deal with the U.K.’s
National Health Service (NHS).22 Google’s artificial intelligence firm was allowed
to access health data from over 1.6 million patients to develop an app monitoring
kidney disease called “Streams.” Public concerns and corroborative research studies
22 See National Health Service (NHS). Available at: https://www.nhs.uk. Accessed 10 June 2019.
suggested that the Google DeepMind deal had access to other kinds of sensitive data
and failed to comply with data protection law.
For this reason, since May 2018, the UK government launched the “national data
opt-out” (ND opt-out) system23 with the hope to regain the public trust. Nevertheless,
there is no evidence of significant changes in the ND opt-out. Neither in the use
of secondary data, nor in the choices that patients can make. The only notorious
difference seems to be in the way that these options are communicated and framed to
the patients. Most importantly, according to the new ND opt-out, the type-1 opt-out
option—which is the only choice that truly stops data from being shared outside direct
care—will be removed in 2020. According to the Behavioral Law and Economics
literature (Nudge Theory),24 default rules—such as the revised opt-out system in
the UK—are very powerful, because people tend to stick to the default choice made
readily available to them. The crucial question analyzed in this chapter is whether
it is desirable for the UK government to stop promoting the type-1 opt-outs, and
whether this could be seen as a kind of “hard paternalism.”
Danuta Mendelson focuses on the National Electronic Health Record System and
Consent to the Processing of Health Data in the European Union (EU) and Australia.
She does this by comparing the legal framework in the EU within the scope of the
GDPR in the context of the EU national electronic record (NEHR) schemes with the
approach of the Australian national health record scheme called My Health Record
(MHR).25 The main difference being the different levels of developments of consent
between the two approaches. In the EU, the GDPR proclaims that individuals (data
subjects) should give their consent for the processing of their personal health data
or have the right to refuse the processing of their personal health data not just in
general, but in specific instances. Whereas in Australia, legislators did not predicate
the lawfulness of personal health data processing on the individual data-subject’s
valid and informed consent. For this reason, the creators of the MHR system did
not embed in its architecture the right of patients to give free, informed, specific and
unambiguous indication that they agree to the processing of their personal health
data in every (reasonable) instance. This might be one of reasons why Australia is
not seeking certification for attainment of adequate level of data protection from
the European Commission. The chapter concludes that under the MHR scheme in
Australia, data subjects—in particular those vulnerable patients such as the very
young and the elderly—will remain disempowered data subjects.
Part III of the book—Blockchain and Code—contains three contributions.
In their chapter, Mark Fenwick, Wulf A. Kaal and Erik Vermeulen discuss the impor-
tance for lawyers and law students to learn how to code in this new digital age. The
authors explain the benefits of introducing a Coding for Lawyers course in the legal
23 See “national data opt-out” (ND opt-out) system Available at: https://digital.nhs.uk/national-
dataopt-out. Accessed 10 June 2019.
24 See, e.g., Thaler and Sunstein (2009).
25 See My Health Record. Available at: https://www.myhealthrecord.gov.au. Accessed 10 June 2019.
curriculum and they share their initial experiences with the course. The main argu-
ment is that Legal Technology (LegalTech) is profoundly disrupting the legal profes-
sion and the development of cutting-edge technologies—such as blockchain, AI, Big
Data, smart contracts, etc.—have triggered the emergence of new business models.
The authors conclude that since all these technologies are code-based, lawyers—as
active “transaction engineers”—need to be able to understand and communicate in
and about code to participate in the design of such technologies. The computer code
is ubiquitous by nature, yet hidden. This is affecting the legal framework, particularly
in terms of the on-going legal profession. Therefore, lawyers of the future will be
transaction engineers managing the responsible deployment of new technologies and
the design of a new global architecture, and that to perform this function effectively,
legal professionals need to develop a number of new skills and capacities, including
an understanding of the basic concepts and power of coding.
Craig Calcaterra and Wulf A. Kaal explain the reputation protocol for the Internet
of Trust. Trust is very important and became one of the most important drivers in
our global economy. Internet-based platform business outcompetes traditional busi-
ness and DLT shows a lot of promise in multiple business vertical. Studies suggest,
however, that they have not reached their full potential due to the decreasing global
trust in the Internet and under-developed trust in decentralized technology solutions.
Semada and the Semada Research Institute (SRI)26 believe that a decentralized rep-
utation network can reverse that trend to increase trust in the Internet and increase
decentralized technology adoption. The authors call this solution the Semada Inter-
net of Trust—a network that uniquely captures real world information, context, and
value in cryptographic transactions generating transparently validated consensus on
truth. Through the creation of unconscious and conscious trust in decentralized net-
work adoption becomes a desirable outcome and increases. The Semada Internet of
Trust creates a framework of trust through reputation and incentive optimization that
enables market conditions for unprecedented business models and market capital-
ization while reducing adverse selection and moral hazards for transacting parties.
The network will provide information symmetry while reducing transaction costs
for market participants. Businesses will leverage the platform to achieve efficiencies
across verticals.
Gyooho Lee discusses two issues raised by the Korean legal community in terms
of blockchain technology. One is intellectual property-related issues of open source
software on which blockchain technology is based. When it comes to open source
software, one pivotal court case needs to be explored. It is related to the conflict
between the possessor of a trade secret and copyright owner of open source software.
The other issue concerns how to guarantee the authenticity of e-Apostilles by using
blockchain technology. The lack of authenticity of e-Apostilles is a big legal hurdle
which prevents e-Apostilles from gaining popularity in many countries. Blockchain
computing is a good solution for guaranteeing the authenticity of e-Apostilles. The
chapter explores an IP-related issue which inherently concerns blockchain technol-
ogy and proposes a method for ensuring the authenticity of e-Apostilles by using
26 See Semada Research Institute (SRI). Available at: https://semada.io. Accessed 10 June 2019.
blockchain technology. In sum, the chapter illustrates an inherent legal issue of

blockchain technology and blockchain technology as a method to solve a current
legal problem.
Part IV of the book—Autonomous Systems and Future Challenges—comprises three

chapters.
Shinto Teramoto explains the issues of autonomous driving from a lawyer’s perspec-
tive. The development of driverless cars is still in its early stages. At this point, the
focus of attention should be on the safety and user-friendliness of autonomous driv-
ing on public roads. Therefore, the current discussion must be who can effectively
prevent traffic incidents and accidents involving autonomous driving by investing
their own financial, human, and technological resources. Safe autonomous driving is
a product of a well-organized network system. Contending hastily that only specific
classes of nodes should be liable does not make sense. The importance of telecom-
munication predicts that telecom and network service industries will play major
roles in realizing safe road traffic involving autonomous driving. If lawyers discuss
the allocation of cost caused by traffic incidents or accidents involving autonomous
driving without considering the involvement of telecom and network service indus-
tries, it simply shows the carelessness of lawyers. Autonomous driving is the way
to realize the sharing of surplus resources, which have been unused or wasted, by
means of aggressive involvement of information and communication technology
(ICT). The chapter concludes with a suggestion that lawyers should go beyond the
current debate of liability issues caused by traffic accidents and incidents involving
autonomous driving. Yet, the discussion should involve collaborative ways of devel-
oping interfaces and standards for every road traffic participant to effectively and
efficiently communicate with each other.
Steven van Uytsel and Danilo Vasconcellos Vargas discuss some of the legal
implications of adversarial machine learning. Research revealed that perturbations
to a picture—even in small size—may disable a deep neural network from correctly
qualifying the content of a picture. This research has been transplanted to traffic signs.
The test results were disastrous. For example, a perturbated stop sign was recognized
as a speeding sign. Because visualization technology is not able to overcome this
problem yet, the question arises who should be liable for accidents caused by this
technology. Manufacturers are being pointed at and for that reason it has been claimed
that the commercialization of autonomous vehicles may stall. Without autonomous
vehicles, the sharing economy may not fully develop either. This chapter shows that
there are alternatives for the unpredictable financial burden on the car manufacturers
for accidents with autonomous cars. This chapter refers to operator liability, but
argues that for reasons of fairness, this is not a viable choice. A more viable choice
is a no-fault liability on the manufacturer, as this kind of scheme forces the car
manufacturer to be careful but keeps the financial risk predicable. Another option is
to be found outside law. Engineers could build infrastructure enabling automation.
Such infrastructure may overcome the problems of the visualization technology, but
could potentially create a complex web of product and service providers. Legislators
should prevent that the victims of an accident, if it were still to occur, would face years
in court with the various actors of this complex web in order to receive compensation.
The final chapter, by Thomas Hoeren explores some of the provisions enshrined in
the new EU Directive on the protection of trade secrets (Directive (EU) 2016/943).27
The Directive was adopted in June 2016 and is now to be transposed into national law
by June 2018. This led to discussion at national levels whether the protection of trade
secrets should be included in a comprehensive set of rules of intellectual property
rights, or at least regulated by a special law. The chapter focuses on the producer’s
liability in accordance with Article 4 (5) of the Directive and outlines the options for
implementation. The chapter concludes that the Directive will permanently change
European secrecy law. Compared to current German regulations, important differ-
ences can be observed, concerning for example the definition of a trade secret or
the legitimacy of reverse engineering. Therefore, the final implementation of the
Directive can be suspenseful. In the meantime, companies are summoned to conduct
concrete confidentiality measures and to adjust to the changed field of secrets. In this
context, extended nondisclosure agreements and a strategy to deal with the freedom
of reverse engineering are most important. A new culture of secrecy in companies
and with suppliers must be added.
References
Chen M et al (2014) Big data: related technologies, Challenges and future prospects. Springer,
Cham
Corrales M, Fenwick M, Haapio H (2019) Digital technologies, legal design and the future of
legal profession. In: Corrales M, Fenwick M, Haapio H (eds) Legal tech, smart contracts and
blockchain. Springer, Singapore
Cusumano M (2018) The sharing economy meets reality. Commun ACM 61(1):26–28
Gazzola P (2017) Behind the sharing economy: innovation and dynamic capability. In: Vătămănescu
EM, Pînzaru F (eds) Knowledge management in the sharing economy. Springer, Cham
Malik F, Wahaj Z (2019) Sharing economy digital platforms and social inclusion/exclusion: a
research study of uber and careem in Pakistan. In: Nielsen P, Kimaro H (eds) Information and
communication technologies for development: strengthening southern-driven cooperation as a
catalyst for ICT 4D, Part 1. Springer, Cham
Maurer J (2016) Sharing economy: regulatory approaches for combating Airbnb’s controversy
regarding taxation and regulation. GRIN Verlag, Open Publishing GmbH, s.l., Munich
PriceWaterhouseCooper (2015) The sharing economy. Consumer intelligence series https://www.
pwc.fr/fr/assets/files/pdf/2015/05/pwc_etude_sharing_economy.pdf. Accessed 10 Jun 2019
Sundararajan A (2016) The sharing economy: the end of employment and the rise of crowd-based
capitalism. The MIT Press, Cambridge
Tandon A (2018) Cloud computing and birth of the sharing economy. What are the dominat-
ing trends?https://altimetrik.blog/2018/06/27/cloud-computing-birth-of-the-sharing-economy/.
Accessed 10 Jun 2019
27 Directive (EU) 2016/943 of the European Parliament and of the Council of 8 June 2016 on the
protection of undisclosed know-how and business information (trade secrets) against their unlawful
acquisition, use and disclosure.
Thaler R, Sunstein C (2009) Nudge: improving decisions about health, wealth, and happiness.
Penguin Books Ltd., New York
Vermeulen E (2017) There is no escape from blockchains and artificial intelligence…. Lawyers bet-
ter be prepared! https://medium.com/@erikpmvermeulen/there-is-no-escape-from-blockchains-
and-artificial-intelligence-lawyers-better-be-prepared-2d7a8221c627. Accessed 10 Jun 2019
Sharing Economy and Platforms
Building Platforms for Collaboration:
A New Comparative Legal Challenge
Annelise Riles
Abstract Collaboration has emerged as a panacea for the ills facing societies around
the world, and also as a methodology for comparative legal scholars who seek to
understand the same. Yet the rise of collaboration as a political and scholarly method
masks the substantive and practical challenges of creating productive and meaningful
transnational and transcultural relationships. This chapter considers some of these
challenges, and also some of the possibilities that are inherent in collaboration,
through the example of a recent experiment with Meridian 180, a global engagement
platform for policy experimentation founded in 2011.
Keywords Collaboration · Platforms · Comparative legal studies · Expertise
1 Introduction
There is an image of how the epoch-making ideas in any field—business, policy, the
arts, or the academy—are made: the singular genius alone working alone in an office.
The days of singular geniuses of this kind are largely over. The problems confronting
society are too complex, too interconnected across different fields and across juris-
dictional and cultural boundaries for any one person to be able to define them clearly,
let alone resolve them. Rather, the genius of our time is a collaborative genius. From
the business world to the academy, and from the leading financial centers to grass-
roots development projects around the world, collaboration is increasingly perceived
as a necessity. The innovations in policy that garner the most attention, likewise, turn
to collaboration of various kinds—from public-private partnerships to harnessing
the power of crowdsourcing, to “peer review” as a norm generating tool and an
alternative to legal enforcement.1
1 See, e.g., Sabel and Zeitlin (2008), Wilkinson (2010), Green and Boehm (2011), Riles (2013a, b).
A. Riles (B)
Buffett Institute for Global Affairs and Pritzker School of Law, Northwestern University, Chicago,
Illinois, USA
e-mail: annelise.riles@northwestern.edu

https://doi.org/10.1007/978-981-15-1350-3_2
16 A. Riles
But there is also a darker side to our newfound obsession with collaboration—
a story of our collective loss of faith in expertise. When government regulators
no longer believe in the power of institutions of their own making to coordinate
social action, they look to collaborative private-public partnerships as a way to avoid
sharing responsibility for planning and rulemaking with the private sector. When
corporations do not trust market mechanisms to sustain a global production chain
held together by long-term labor relations they turn to crowdsourcing as an “anti-
crisis solution.”2 When the producers of consumer products, or of advertisements
for those products, lose confidence in their ability to predict consumers’ desires they
turn to aggressive data collection and surveys to crowdsource market predictions and
nudge the aggregate consumer toward certain products.
Hence it is no surprise that the turn to collaboration has also spawned a number of
quasi-utopian but ultimately quite limited and conventional templates and buzzwords
emanating from the corporate world to the world of NGOs and the international
human rights field where web-based exchanges promise to overcome cultural, polit-
ical, and economic differences, leaving only “win-win scenarios.” As two observers
recently put it in the Harvard Business Review blog, collaboration “is at risk of
simply becoming a new form of ‘green washing.’”3
In comparative legal studies, likewise, comparison is increasingly supplanted by
a new watchword for research and practice: collaboration.4 Collaboration is at once
a necessary legal skill to be taught at law schools, a novel research methodology,
and more generally, a malleable new answer to old legal problems. Law schools
everywhere are endeavoring to produce lawyers who collaborate effectively in their
practice. From the point of view of some university administrators, collaborative
opportunities also obviate the need for comparative scholarship: Who needs to read a
scholarly comparison of legal institutions in Pakistan and Canada, when the Pakistani
legal thinker can be brought into one’s project collaboratively?5 It is no longer enough
to signal cosmopolitanism in law school programming; success is about building
collaborative relationships that obviate the need for cosmopolitanism altogether. In
a world in which everyone is already an expert, solutions are produced through
crowdsourcing rather than knowledge work and there is no need for fine-grained
comparative descriptions, the legal academy is left without an overt role. This view
that the crowd is ultimately wiser than the expert may help to explain the increasing
premium placed on popular writing such as blog posts in the legal academy.6 The
implication is that what is needed in such a condition is not comparative analyses
but attention from potential collaborators and methods for collaborating per se.
2 Oihab and Salvator (2011).

3 Ellingstad and Love (2013).
4 Riles (2015), p. 147.
5 See, e.g., generally, Anderson (2014), Munir (2012).
6 See, e.g., Liptak (2013), Wagner (2011), Schwartz (2014).
Building Platforms for Collaboration: A New Comparative Legal … 17
2 Collaboration in Practice
Collaboration is in practice painstaking, riddled with glitches, laced with potential

for exploitation and politically fragile. One has only to consider recent episodes of
scandal and failure in the collaborative economy, from Uber to Facebook to AirBnB
to see how easily collaboration fails, or worse yet, empowers the worst of human
instincts and enables people to turn inward, to avoid rather than embrace what is
challenging or unknown. Transnational collaborations in particular—among teams
building the latest gadget or consumers sharing reviews of those gadgets, or aca-
demics or artists creating the newest work—face tremendous challenges, technical
and political.
Why is this the case? Collaboration by definition means working with someone
different from oneself—someone in a different institution, or with differing exper-
tise from one’s own, or perhaps with a person in another location, another market,
or another culture altogether. In practice, there are almost insurmountable barriers
to doing this. Institutions such as corporations and universities are not designed or
organized to share resources and ideas with outsiders. Individuals have trouble trust-
ing people who are different from themselves. Barriers of language and culture lead
to a lack of interest in, or patience for, the very different points of view that need to
be brought together in a successful collaboration.
Too often, the law seems to assume that collaboration happens sui generis. All
the state has to do, it is assumed, is get out of the way and people will naturally
collaborate. Yet research in the social sciences demonstrates that if we truly wish to
produce more collaboration we will need to encourage, incentivize and educate both
experts and lay people alike to step outside their comfort zones to collaborate.7 How
do we include a wider, global community of stakeholders in building the frameworks
and categories for idea generation? This is a question with political, economic, and
ethical consequences. It is a technical question but also a scholarly question. It is
a question of ethical imagination that comparative lawyers are uniquely poised to
answer.
If we turn the issue around, away from the denigration of expertise, to focus
instead on collaboration, therefore, we see that comparative lawyers, in partnership
(collaboration!) with other experts such as systems designers, are uniquely posi-
tioned to play a new and vital role. A new kind of engagement platform, capable of
transcending national borders and cultural divides between experts and non-experts,
government officials and civil society is needed in this moment. This platform needs
to be able to address normative and ethical questions while also developing solutions
to specialized, technical problems. It must encompass diverse points of view and
incorporate agendas driven from the bottom up, all the while remaining closely man-
aged and results-driven. It must help connect linguistic, social, economic, scholarly,
and political vantage points to usefully input into the process.
7 See Riles (2013a, b), Hansen (2009).

18 A. Riles
3 Experimenting with Platforms: Meridian 180
Beginning in 2011, I began experimenting with a prototype of this kind of platform as

a project of interactive and institutionalized comparative law. Meridian 180, “a mul-
tilingual platform for policy experimentation + innovation,” brings together more
than 1000 policy-makers, industry representatives, academics and representatives
of civil society from 39 countries. The platform blends the structure of an inter-
national membership organization—in which policy-makers and representatives of
civil society participate as private individuals and then feed ideas and solutions to
their own institutions, communities and networks—with that of a university feder-
ation—in which resources, links to national policy-makers, university presses, and
intellectual talent pools are shared across national boundaries—and a digital plat-
form. Crucially, all conversation takes place in four languages, and hence questions
of language and translation so critical to comparative law, are always at the forefront
of our discussions.
A platform is not a traditional category of legal analysis, let alone comparative
legal practice. It is part regulation, part technology and part social organization. It
incorporates elements of both states and markets and welcomes different kinds of
stakeholders. It comes with certain theoretical assumptions built in, but these are not
unpacked or elaborated at every moment. But the science, meaning and practice of
collaboration deserves now to take a central place in comparative legal studies, and
hence it behooves us to begin thinking in sophisticated ways both theoretical and
practical about the nature of platforms. Towards that end, I offer the following initial
lessons from the Meridian 180 experiment.
The first is the value of collaboration: it produces insights that none can produce
alone. It produces prospective knowledge, not just retrospective knowledge: Working
together, a group of thought leaders are often able to anticipate problems and address
them before they become crystallized in public consciousness in ways they could
not do singularly.
The second is that collaboration does not just happen sui generis. It is frustrat-
ing, difficult, and outside of everyone’s comfort zone by definition. It is logistically
complex, and requires a methodology, theoretical grounding, and extensive institu-
tional support. The successes or failures of these can be studied systematically, like
any social, legal or technical process. Some collaborations are more effective, more
worthwhile, and more ethical than others.
The third, is that like cultural sensitivity, collaboration is a skillset to be learned.
People are not simply natural collaborators, although some are perhaps more pre-
disposed or more prepared than others. One has to create opportunities for practice,
and even for failure, as well as for people to get a taste of what the benefits (and even
pleasures) of collaboration might be in their own work.
The fourth is that collaboration is also a set of normative commitments, commit-
ments that I believe we as comparative lawyers should champion professionally. In
particular, collaboration requires commitment to personal and institutional empathy
and curiosity about the world, tethered to a hope that there are undiscovered new
Building Platforms for Collaboration: A New Comparative Legal … 19
ideas to be gained from struggling to understand points of view foreign to one’s

own. These basic insights about the nature and value of collaboration in turn have
implications for the architecture of platforms. Again, here are some insights from
our experiment:
First, the platform must be deeply inclusive and diverse. Representatives of various
national, linguistic, social, economic, scholarly, professional and political vantage
points must have a seat at the table. At the idea incubation stage, the unfiltered
participation of individuals is critical.
Second, space for risk-taking and free play must be preserved in order to cre-
ate fertile ground for new ideas and social ties to emerge. Thought leadership must
emerge from the collective rather than the top, a process that can be encouraged by
new information technologies. Unnecessary barriers to participation can be lessened
through online meeting technologies, artificial intelligence and crowd-sourced trans-
lation technologies. New technological innovations can enable ideas to be bundled,
weighted according to their degree of support, and edited collectively.
The methodology must also facilitate discussion of policy details without losing
sophistication. The challenge is to create a discursive structure that enables serious
consideration of key issues while allowing the agenda to emerge organically. This
is crucial for idea generation, but even more so for the legitimacy of the process.
Meridian 180 stages multiple rounds of discussions, with shifting groups of partic-
ipants depending on the scope or stage of the discussion, in order to keep open a
channel of ideas emerging from the bottom up.
Finally, we must design the pipeline from idea incubation to production. We
must maintain porosity between deliberation and implementation. The people who
create the ideas become the most appropriate champions for those ideas. One way
of thinking about this question is in terms of the interface between platforms. The
task of the platform is not just to incubate but to interface, and this too must be a
transnational, transcultural endeavor.
4 Conclusion
All of these are criteria that we can reflect on as we consider what kind of regulation
is appropriate for existing platforms, as is the theme of this volume. Yet beyond this,
I believe we can also think of regulation itself as enabling platform-like interaction.
Platforms in this sense are emerging as an alternative to nation states and nationally
bound regulatory cultures on the one hand and markets on the other, as sites of
collaboration, innovation and value creation. One goal of law reform should be
to facilitate the flourishing of such platforms, and one project of comparative law
might then become to study platform successes and failures in light of the incentives
created by differing local, institutional, national and transnational regulatory systems.
I believe we can understand the enabling of the emergence of new ideas, of the new
20 A. Riles
genius, if you will, as a critical and intellectually challenging task for comparative
law, one that draws upon our full faculties as scholars and citizens of the world, and
our full tradition of comparative study and analysis.
References
Anderson K (2014) Through our glass darkly: does comparative law counsel the use of Foreign
Law in U.S. Constitutional Adjudication. Duquesne Law Rev. 52:115
Ellingstad P, Love C (2013) Is Collaboration the New Greenwashing? HBR Blog Network. http://
blogs.hbr.org/2013/03/is-collaboration-the-new-green-1/. Accessed 3 Mar 2019
Green EF, Boehm JL (2011) The limits of “name-and-shame” in international financial regulation.
Cornell Law Rev 97:1083
Hansen M (2009) Collaboration: how leaders avoid the traps, build common ground, and reap big
results. Harvard Business Press, Cambridge
Liptak A (2013) The Lackluster reviews that lawyers love to hate, New York Times http://www.
nytimes.com/2013/10/22/us/law-scholarships-lackluster-reviews.html?_r=1&. Accessed 3 Mar
2019
Munir MA (2012) South Asian chapter of CJEI—a blueprint for establishing a regional judicial
education body in South Asia, law and justice commission of Pakistan/National Judicial Policy
Making Committee. Supreme Court of Pakistan, Islamabad, p 1
Oihab AC, Salvator M (2011) Collaboration as an anti-crisis solution: the role of the procurement
function. Int J Phys Distrib Logistics Manage 41:860
Riles A (2013a) Is new governance the ideal architecture for global financial regulation? Monetary
Econ Stud 65
Riles A (2013b) Market collaboration: finance, culture, and ethnography after neoliberalism. Am
Anthropol 115:555
Riles A (2015) From comparison to collaboration: experiments with a new scholarly and political
Form. Law Contemp Probl 78:147
Sabel CF, Zeitlin J (2008) Learning from difference: the new architecture of experimentalist
governance in the EU. Eur Law J 14:271
Schwartz J (2014) This is law school? Socrates takes a back seat to business and tech,
New York Times http://www.nytimes.com/2014/08/03/education/edlife/socrates-takes-a-back-
seat-to-business-andtech.html. Accessed 5 Mar 2019
Wilkinson M (2010) Three conceptions of law: towards a jurisprudence of democratic experimen-
talism. Wisconsin Law Rev 2010:673
Wagner A (2011) Making law accessible to the public, The Guardian
http://www.theguardian.com/law/2011/jul/26/tort-law-access-legal-aid. Accessed 5 Mar 2019
Sharing City Seoul and the Future
of City Governance
Benjamen Franklen Gussen
Abstract This chapter looks at the impact of the sharing economy on the gover-
nance of cities. The Sharing City Seoul Project is used as a case study to ascertain this
impact. Seoul is one of the leading cities when it comes to innovation in the sharing
economy. Its experience has already informed other global cities around the world,
including New York City and Amsterdam. The chapter unpacks the policy interven-
tions that took place in Seoul to inform future approaches to governing cities in other
countries. The short-term micro effects of the sharing economy on the governance of
cities are already underway. These effects suggest that regulatory schemes are being
adjusted to enable innovation in the sharing economy. The long-term macro effects
seem to usher in wider governance implications. There is a constitutional imperative
where cities are emerging as independent actors on the international stage, with their
own sui generis legal personalities. Cities need to adjust their regulatory approaches
to secure a first mover advantage on reaping the benefits from these effects. The ver-
tical relationship between federal, state and local government (in particular, urban
or city-region governments) needs to be reassessed in light of these effects.
Keywords Sharing economy · Seoul · Cities · Governance ·

Cooley-Eaton-McQuillin thesis
1 Introduction
This chapter looks at the governance of sharing cities. More specifically, the chapter
argues two effects of the sharing economy on the governance of cities. First, there is
AN earlier version of this chapter appeared in Benjamen Franklen Gussen, ‘On the Governance
of Sharing Cities: Lessons from the Sharing Economy in Seoul’ (2018) 12 Northeast Asian Law
Review 9.
This paper was made possible by a 2018 grant from the Australia Korea Foundation.
B. F. Gussen (B)
The Swinburne School of Law, Swinburne University of Technology, Victoria, Australia
e-mail: bgussen@swin.edu.au

https://doi.org/10.1007/978-981-15-1350-3_3
22 B. F. Gussen
a micro-level, short-term, effect where governance (qua legal rules) is already under-
going a change in emphasis that is modifying the current legal framework given the
new combinations enabled by the sharing economy. This effect is illustrating how the
regulatory scheme is adjusting to facilitate collaboration between citizens or between
citizens and industries. The current regulatory scheme puts emphasis on internaliz-
ing externalities (also known as external diseconomies), namely costs affecting third
parties (to the economic transaction).1 The sharing economy is changing the calculus
of these diseconomies through ecologically-sensitive, Internet-driven, technologies
of what is known as the “Internet of Things”—the interoperation of physical devices
through embedded digital connectivity to (automated) control systems across exit-
ing infrastructure.2 The issue is to understand how these technological advances are
changing the production and distribution of negative externalities.
The second effect flowing from the title of this chapter is a macro-level, long-term
(positive feedback) effect where the sharing economy (through decentralization and
equitable collaboration) is changing how cities are governed (in a constitutional
sense). City power is on the ascendancy thanks to the bottom-up, self-regulation
mechanisms enabled by the sharing economy. My prediction is that innovation in
the sharing economy will allow cities to play a much larger role in an emerging new
world order, where cities become the main actors on the “international” stage. Cities
will displace the Westphalian nation-state as the main body politic for organizing eco-
nomic and social activity. There will be a shift in emphasis towards acknowledging
a sovereign role for cities as direct actors in the world economy, without the current
mediation of nation-states. The process is asymptotically a Tier 1 civilization on the
Kardashev scale3 : a planetary civilization. Technological advances will eventually
result in one civilization, where there are thousands of (non-contiguous) sovereign
cities forming a network of loose (continental) confederations under a global consti-
tutional framework. The next phase towards this Kardashev I civilization is driven
by the Internet of Things, and the Cooley-Eaton-McQuillin thesis, which recognizes
cities as the building blocks in a new world order.4
The analysis starts by recasting the sharing economy in the context of human
civilization (Sect. 2). The objective is to gauge the novelty of the sharing economy,
given the historical stages of development of human settlement. My proposition
is that a feedback loop between shared economic activity and the emergence of
cities is creating a reiterative process to resolve the problem of scale created by
increases in population density.5 Through this lens, the sharing economy is seen as a
staccato process that started over ten thousand years ago—as bouts of technological
innovation that allow for optimizing this density intra- and inter- city.6 What is
1 Buchanan and Stubblebine (1962).

2 Vermesan and Friess (2013).
3 Kardashev (1964), pp. 217, 219.
4 Frug (1999). See also Gussen (2016a, b, 2019).
5 Gussen (2013).
6 Innovation is understood in the Schumpeterian sense, namely as “new combinations” leading to new
energy resources, new products, new methods of production, new markets, or new organization of
Sharing City Seoul and the Future of City Governance 23
new in the 21st century is the speed with which available technology is allowing for
increasing population density in existing cities, and the much faster connectivity with
new ones.7 The technological innovation, nevertheless, is always aiming at the same
result: the modulation of efficiency (in the transformation of energy) to maximize
human connectivity intra- and inter-cities. Increases in this efficiency allow for a
restoration of the conviviality function of human interactions.
To understand how the sharing economy puts different demands on the regulation
of human behavior, the chapter revisits the nexus between efficiency and the law
through the lens of the Coase theorem (Sect. 3). I elaborate on how the sharing
economy opens new options for self-organizing bargaining processes, and by doing
so, necessitates a revision of the role of government in regulating human behavior
(Sect. 4). To be clear, my assertion is that governance for effective coordination
will always be necessary, although its objective will now be to underwrite self-
organization (rather than a large-scale government-enforced system of rules).8
To illustrate the interaction between cities, the sharing economy, and legal sys-
tems, the project called Sharing City Seoul is presented as a case study (Sect. 4). The
project choice is motivated by the status of Seoul as a role model in sharing econ-
omy innovation, which has catapulted Seoul into becoming an avant-garde model
for others to emulate.9 The project illustrates how innovation in the sharing economy
requires a new thinking as to the role of local governments, and to the role of cities
in a new world order (what can be referred to as continentalization).10
The chapter ends with some remarks to motivate further research on the challenges
and opportunities that the 21st century economy is presenting to legal systems. The
role of (transaction) automation in particular is flagged as likely to see a radical
change in the profession and its institutions.
2 The Historical Context of Sharing Cities
The sharing economy is not a new phenomenon. The role of the sharing economy can
be traced back to the dawn of civilization. Historically, the sharing of coproduction
resulted in the creation of cities.11 To motivate this assertion, however, I need to
explain what is meant by the sharing economy.
economic activity (such as networks as opposed to markets or hierarchies). See Schumpeter (1939),
pp. 84–85. See Schumpeter (1934), p. 66.
7 Population density is interpreted in a network sense, namely, as “the degree to which the members
of any network are interconnected.” See Thompson (2003), p. 34.

8 Governance refers to: (1) the regulation of agents in a given organizational model (such as hierar-
chies, markets or networks), (2) the effectiveness of their reproduction, and (3) their alignment and
coordination. See Thompson (2003), p. 34.
9 Guerini (2014).
10 Rifkin (2011), pp. 161–192.
11 Agyeman and McLaren (2017).
24 B. F. Gussen
There is no one universal definition of what we refer to as the “sharing economy.”

Usually, the sharing economy is said to include four categories: “recirculation of
goods, increased utilization of durable assets, exchange of services, and sharing of
productive assets [including space].”12 These categories are delivered using either
a peer-to-peer (P2P) or a business-to-peer (B2P) model; and using both, not-for-
profit (public-benefit) and for-profit (private-benefit) platforms.13 The P2P model
generates commissions and hence depends on the number of trades. This model is
frequently democratically-organized and complements the “public good” function
of the government-to-peer (G2P) model. The B2P model is closer to traditional
business models in that it maximizes revenue per transaction and has historically
led to monopolies where platforms are backed by large corporations (for example,
Google or Amazon). Both types of platforms allow for exchanges to occur in a
market setting, although the profit objective affects the way sharing takes place and
how revenues are distributed to owners and management.
Other definitions focus on the role of the sharing economy as a wider paradigm
shift, away from the hereto dominant models of economic organization, namely, cap-
italism and socialism. This definition sees the sharing economy as part of a 21st cen-
tury Third Industrial Revolution.14 The sharing economy, referred to in this discourse
as the collaborative economy, is about organizing economic activity to achieve sus-
tainability, partly through the use of distributed renewable energies that are generated
at local intelligent mini-grids, and then shared over intelligent electricity networks
connecting continents (for example from Canada to Chile).15 Jeremy Rifkin explains
the collaborative economy in these terms:
The Third Industrial Revolution is the last stage of the great industrial saga and the first stage
of the emerging collaborative era rolled together. It represents an interregnum between two
periods of economic history—the first characterized by industrious behavior and the second
by collaborative behavior.16
This model is driven by the Internet of Things and presumes an access-neutral

network platform. The economic (big) data generated from the Internet enables using
analytics to discover ways to increase aggregate efficiency. This increase in efficiency
will in turn increase productivity and hence reduce the ecological footprint of eco-
nomic activity. The end result is future reduction of (marginal) costs. “The Internet
of Things is the first general purpose technology in history that can potentially take
large parts of the economy to near zero marginal cost.”17 The collaborative economy
transforms civic public spaces into urban commons, and by doing so, the sharing
economy introduces a third actor into city governance: the urban commons (net-
work) mediates the existing state (hierarchy) and private (market) institutions. This
12 Schor (2016).
13 Schor (2016), p. 11.
14 Rifkin (2011), pp. 161–192.
15 Rifkin (2011), p. 115. Rifkin presents the European Union as the first continental union. Rifkin
(2011), p. 165.
16 Rifkin (2011), p. 259.
17 Rifkin (2014), p. 138.
environment of open-source access and peer-to-peer collaboration breeds prosumers

(consumers who are their own producers) who have different ideals of freedom,
power, and community. Geopolitics will make way to a biosphere consciousness
where governance institutions move to a decentralized model. More specifically,
this technological innovation will allow for a return to network organization and the
original emphasis on the conviviality function (see below).
Klaus Schwab, on the other hand, recognizes the sharing economy as a deep
shift in a Fourth Industrial Revolution.18 Schwab identifies the shift 10,000 years
ago from foraging to farming as the first revolution. The First Industrial Revolution
followed in the 18th century. The Second Industrial Revolution started in the 19th
century. The Third Industrial (computer or digital) Revolution began in the 1960s.
Today the Fourth Industrial Revolution is bringing a more ubiquitous and mobile
Internet, and artificial intelligence (AI) and machine learning.19 In essence, Rifkin’s
Third Revolution is Schwab’s Fourth. What is clear is that both identify the sharing
economy as a pillar of both revolutions. Schwab explains the sharing economy (in
the context of car sharing) in the following terms:
The common understanding of this phenomenon is the usually technology-enabled abil-
ity for entities (individuals or organizations) to share the use of a physical good/asset, or
share/provide a service, at a level that was not nearly as efficient or perhaps even possible
before. This sharing of goods or services is commonly possible through online marketplaces,
mobile apps/location services or other technology-enabled platforms. These have reduced
the transaction costs and friction in the system to a point where it is an economic gain for
all involved, divided in much finer increments.20
This understanding of the sharing economy identifies an adjustment of taxation

and regulation from ownership- and sale-based models to use-based ones.21 There
is also a shift in power from state to non-state actors, where established institutions
are replaced by loose networks.22 Micro powers such as the city will now be able to
constrain national governments.23 Schwab adds the following:
Governments…will be forced to change as their central role of conducting policy increasingly
diminishes due to the growing levels of competition and the redistribution and decentraliza-
tion of power that new technologies will make possible. Increasingly, governments will be
seen as public-service centers that are evaluated on their abilities to deliver the expanded
service in the most efficient and individualized ways.
…governments will be forced to change their approach when it comes to the creation, revision
and enforcement of regulation. In the “old world”, decision-makers had enough time to study
a specific issue and then create the necessary response or appropriate regulatory framework.
The whole process tended to be linear and mechanistic, following a strict top-down approach.
For a variety of reasons, this is no longer possible.24
18 Schwab (2016).
19 Schwab (2016), p. 7.
20 Schwab (2016), p. 157.
21 Schwab (2016), p. 158.
22 Schwab (2016), p. 67.
23 Schwab (2016), p. 68.
24 Schwab (2016), pp. 68–69.
26 B. F. Gussen
The general role of regulation in this Fourth Revolution is to facilitate technical

innovation in the sharing economy.25
As to cities, Schwab adds:
Cities have been the engines of economic growth prosperity and social progress through-
out history, and will be essential to the future of competitiveness of nations and regions.
Today, more than half of the world’s population lives in urban areas, ranging from mid-sized
cities to megacities, and the number of city dwellers worldwide keeps rising. Many factors
that affect the competitiveness of countries and regions—from innovation and education to
infrastructure and public administration—are under the purview of cities.26
Hence, some define the sharing economy by anchoring it in civilization, more

specifically, in civic society, or (literally) human settlement in cities. In the same
direction, Julian Agyeman and Duncan McLaren suggest moving beyond the
bounded concept of the sharing economy to that of sharing cities: a concept that
imagines cities’ social capital as enhancing novel forms of sociocultural sharing.27
Unprecedented levels of high population density, and the increased connectivity
enabled by digital technologies are identified as reinventing the city. The city is
reimagined as a shared entity: a bundle of shared infrastructure, resources and spaces.
Smart (Internet-enabled automation) technologies are geared towards an agenda of
sharing, which itself is based on trust and solidarity.28 In this sense, government
intervention is not as much as to help the poor, at least in the welfare sense, but
to decentralize economic power through the sharing economy. Such government
intervention, however, has to be sensitive to the fast-paced innovation underlying
the sharing economy. To help deliver the required agility, hypotaxis (also known as
subsidiarity or shared sovereignty) has to be enlivened. Any government intervention
to help communities utilize (idle) built-in capacity will have to be at the city scale.
The purpose of this intervention is to enable more self-regulation within the sharing
economy. Such outcomes are only transitory towards a “steady state” (or equilib-
rium) beyond the sharing economy, where matching is actioned by (self-regulating)
automated platforms.
Agyeman and Duncan give the example of Seoul. The city championed a project,
Sharing City Seoul that aims at expanding its physical and digital sharing infrastruc-
ture. Initiatives under this project include incubation for sharing start-ups and the
sharing of public resources.
The sharing city goes beyond the classical “tragedy of the commons” to what came
to be known as the “comedy of the commons.”29 The “comedy” refers to the self-
regulation of the sharing economy by its users. The governance of the sharing city is
25 Schwab (2016), p. 74.

26 Schwab (2016), p. 76.
27 McLaren and Agyeman (2015).
28 Solidarity refers to commitment to the common good. Note that trade-offs between hypotaxis and
solidarity become binding only under extreme conditions. The existence of idle capacity negates
the need for such trade-offs.
29 Rose (1986).
therefore inspired by the scholarship of Elinor Ostrom in relation to the design prin-
ciples for self-organized governance systems.30 This is also what Friedrich Hayek
referred to as catallaxy: a self-organized system of voluntary cooperation.31 The
commons are the city resources accessible by citizens for collective benefit. Man-
agement of such resources is based on use value and maintenance costs rather than
a market-value. The governance of such resources is driven by citizens rather than
national government or the private sector. The sharing city is enabling new forms of
political participation that are invented and controlled by the people. In a sharing city,
local government takes a leading role in governance. Examples of such governance
include citizen participation in budgeting.32
The above definitions imply that the sharing economy is intertwined with tech-
nological advances—especially at the city scale. To be clear, the definition of a city
is itself a function of technology. That is to say, the city is a human settlement that
employs technology (that maximizes the efficiency of energy transformations) to
the end of optimizing population density (in other words, allowing more people to
enjoy the conviviality function from living in a city). Today, for example, a city can
be distinguished from a town by populations of over 100,000. That is to say, at that
population level, technology (from the Second Industrial Revolution) is employed
to maintain the conviviality function. This becomes more difficult as we reach mil-
lions of inhabitants. The Third Industrial Revolution, however, will allow attaining
conviviality at higher population levels intra-cities and inter-cities. This is done by
allowing sharing to occur on a continental scale – eventually on a global (Kardashev)
scale (see above). It follows that the definition of a city (in terms of its population)
changes over time, depending on available technology. The beginning of this story
can be found in the very first of human settlements. These settlements were able to
amass thousands of inhabitants, given the available technology at the time (while
still allowing the conviviality function to dominate the economic and political func-
tions—see below). With more technological advances, the numbers climbed up to
hundreds of thousands, to millions, and today to tens of millions. A 21st century city
is one that pushes the boundary of its population density to the optimal level that can
be achieved using the Internet of Things. Other cities continue to function at a ceiling
imposed by earlier technologies, but through the Internet of Things, can partake in
continental networks of cities.
The function of first human settlements was strictly convivial, driven by the desire
for companionship.33 Simultaneously, human settlements began to assume an eco-
nomic role. Having to “chase his own food, cook his own meals, build his own hut,
mend his own shoes, weave his own cloth, till his own fields,” meant that there was no
time for convivial purposes.34 This required increasing productivity through special-
ization and then exchanging the surplus with others. To allow for this specialization
30 Poteete and Ostrom (2010).

31 Hayek (1976), pp. 108–109.
32 Ebdon and Franklin (2006), p. 437.
33 Kohr (1977).
34 Kohr (1977), pp. 14–16.
28 B. F. Gussen
to free up enough time (through exchange with others) required more settlers. This
optimum city size was also reached through loose confederations with other (geo-
graphically proximate) settlements. But the increase in the number of settlers added
complexities both intra- and inter-settlements. The original objective of conviviality
now required a political function—peace, justice and defense.
The optimal size of cities depends on three factors: technology, education and
organization.35 In this context, history is a footnote to advances in these fields, the
result of which was a continuing increase of population (density) intra-cities and
increasing inter-city connectivity. The end result was an optimal level of population
density (given these three factors).
Given our discussion above, in its historical context, the sharing economy can be
best illustrated by the rise in the number of cities as well as the increase in the popu-
lation of these cities. The first cities appeared between 6000 and 5000 BC.36 These
were possible by technological advances that moved from manual energy, oral com-
munication and “on foot” logistics, to one with smelting, writing and domesticated
animals. This allowed for a larger city footprint and hence an optimal population
density.37 The size of cities now accommodated a ceiling of 40,000 inhabitants. By
the Medieval period, the use of renewable energy (wind and water mills) saw the
advent of the printing press and the introduction of roadways and marine navigation.
The ceiling was now pushed to over 100,000 inhabitants. The First Industrial Rev-
olution (1760–1840), brought the use of coal, the steam printing press and the tele-
graph, and railroads and steamships powered by steam engines. These advances saw
London reach one million inhabitants.38 By the time of the Second Industrial Revo-
lution, cities were reaching a ceiling of tens of millions of inhabitants. Smaller cities
continued to spring up throughout history, limited only by the available technolo-
gies that could bridge the distance between existing urban networks (as kingdoms,
nation-states, and today as regional alliances).
In line with these technological advances, cities continued to spread across the
world. From the eastern corner of North Africa (including Arabia and the Levant),
to southern Asia, Europe, and today, to all continents.39 The number of cities grew
from less than 50 cities in 1950 BCE to over a hundred by 1000 CE, and to over 4000
today (with over 100,000 in population).40
While seen by many as a positive dynamic, some question whether the sharing
economy is able to improve on equitable distribution of wealth,41 due to an internal
35 Kohr (1977), p. 20.

36 Davis (1955), pp. 429–430.
37 Population density was managed mostly by larger (geographical) footprints. Over time, however,
density continues to climb back to its earlier levels. The sharing economy of the 21st century is
allowing for a reduction of density while holding the footprint constant.
38 Davis (1955).
39 Galka (2016).
40 The 4037 Cities (2015).
41 Schor (2016).
tension between the efficiency of allocation and the equity of distribution. This ten-
sion maps onto the B2P and P2P models respectively. Resolving this tension imports
the principle of hypotaxis with its dual origins from economics and ethics.42 To illus-
trate the rationale of hypotaxis, the sharing economy is defined as a self-organizing
system of voluntary (online and offline) cooperation. It follows then that the sharing
economy is more than a (Hayekian) catallaxy. The current iteration of the sharing
economy illustrates its emphasis on social justice in the form of trust and solidarity,
and a higher level of social capital,43 making it closer to a hypotaxy.44 Social capital
itself was identified since the 1990s as leading to enhanced economic activity.45 In
Italy, for example, the horizontal networks of northern-central Italy were contrasted
with the vertical hierarchies of southern Italy. The former networks were found to
be responsible for the rise of a civic society that in turn was responsible for creating
wealth. Hence, even when holding technology constant, a history of strong cities and
continuing trust and solidarity leads to networks of governance, as exemplified, for
example, in the economy in Emilia Romagna.
3 The Role of Legal Systems
To set the stage for the arguments in this section, I analyze how negative externalities
should be regulated with the help of a classic example from law-and-economics. The
setup is shown in Fig. 1.
At this point, it is helpful to clarify what is meant by “externality.” An externality
is present where46 :
u A = u A (X 1 , X 2 , . . . , X m , Y1 ) (1)
The utility of an individual A, u A , depends on a set of activities: (X 1 , X 2 , . . . , X m ).

These activities are under the control of A. Activity Y1 on the other hand, is under the
control of B—a member of the same human settlement as A. Individual A maximizes
his utility by modifying the activities under his control to account for the activity
outside his control. A negative marginal externality exists when,
∂u A
≡ δ− < 0 (2)
∂Y1
In our example, there is a farm and a ranch. They have a shared boundary with
no fence. In this example, Y1 is the cattle raising activity by the rancher (B), which
the farmer (A) has no control over. The cattle wander onto the farmer’s land and
42 Gussen (2018).
43 Rifkin(2011), pp. 161–192.
44 Gussen (2014), p. 123; Gussen (2016b), p. 383.
45 Putnam et al. (1994).
46 Buchanan and Stubblebine (1962).
30 B. F. Gussen
Fig. 1 The basic

Rancher-farmer example
(Cooter and Ulen 2014,
p. 68.)
damage the tilled area. Of course, there would be no crop damage without the cattle.
Nor would there be crop damage without the crops.
The traditional (English common law) legal approach is to internalize the negative
externality. That is to say, the cost of rectifying ∂u
A
∂Y1
is assigned to B. Given that the
damage runs from the rancher to the farmer, under English common law, the rancher
must pay for the damage. The US approach, however, assigns the property right to
the rancher (open range approach). The rancher would not be liable for any damages
to the crops unless the farmer did in fact have a fence, and the cattle broke through
the fence). We therefore have two different approaches to assigning property rights,
one where ranchers have responsibility to control the cattle, and hence assigns the
rights to the farmer. The other approach assigns the rights to rancher and allows the
cattle to roam free. The issue is to decide which approach is more efficient.
In 1960, Ronald Coase published an analysis of the problem. The analysis is based
on efficiency: on optimizing farming activities and ranching activities.47 According
to Coase: (1) in the absence of transaction costs, and (2) if property rights are well
defined and (3) tradable, then voluntary negotiations will lead to an efficient allocation
of rights (but not to efficient distribution).
Let’s use some numbers to illustrated Coase’s approach.48 Let’s assume that the
damage caused by the cattle costs the farmer $25 annually. Putting up and maintaining
a fence around the farmer’s tilled area costs $50 annually. In contrast, given the
longer border between the ranch and the farm, installing the fence at the border
costs the rancher $75 per year. Since building the fence by the farmer costs less,
it would be more efficient for the farmer to build that fence around the tilled area.
The efficient outcome is the exact opposite of what would be expected under the
47 Coase (1960), p. 1.
48 Coase (1960), p. 69.
traditional approach (of internalizing the externality). Given that the quantum of the
damage caused by the cattle is lower than building the fences (either at the border
or around the tilled area), if the approach is closed range (traditional approach), the
rancher will simply pay for the damages rather than build the fence. If the legal
approach is open range, the farmer will simply live with the damage.
Let’s now assume that the damage caused by the cattle costs the farmer $100
annually. Again, the traditional approach assigns rights to the farmer and requires
the rancher to bear the damage of $100 per year. To the rancher, the option is to either
pay this fine, or to install and maintain the fence for $75 per year. Given that the cost
of the latter is less than the fine, the rancher would choose to pay for the fence. If on
the other hand, the legal rule assigns the rights to the rancher (to damage the farmers
tilled area), the farmer will have to build and maintain the fence at a cost of $50 per
year. Coase’s insight was that if the rancher and the farmer can bargain successfully,
it does not matter what rights the law assigns. Coase argued that the law needs to be
structured so as to remove impediments to private agreement. This is in contrast with
the traditional approach where the law is structured to minimize the harm caused to
others by failure in private agreements—what is known as the Normative Hobbes
Theorem, which dominated during hierarchical periods of organization.49
Coase’s theorem is generally formulated in relation to transaction costs (i.e., costs
that represent resource losses due to lack of information),50 and predicated on well-
defined property rights that can be traded. These costs arise from exchange under
a market organization. They are a necessary condition for the existence of negative
externalities.51 When these costs are zero, externalities are internalized, and the legal
assignment of property rights is irrelevant. An efficient use of recourses will result
from private bargaining. Hence, Coase makes negative externalities a function of
both transaction costs and property rights:

δ − = f Tc , R p < 0 (3)
Negative externalities (δ − ) are hence related to transaction costs (Tc ) and to prop-
erty rights (R p ). The higher the transaction costs, the higher are the negative external-
ities. On the other hand, the less well-defined property rights are, for example where
there are missing property rights, the higher are the negative externalities. Hence,
holding Tc constant, negative externalities can be reduced by making property rights
(R p ) complete. Alternatively, holding property rights constant, we can reduce these
externalities by reducing transaction costs.
To elaborate on how this insight applies in the sharing economy, I first need to
further clarify the meaning of property rights and transaction costs.52 Let’s start with
the former.
49 Cooter (1982), p. 1. The Hobbesian theorem says that the law should be structured so as to
minimize the harm caused by failures in private agreements. See Cooter and Ulen (2014), p. 68.
50 Dahlman (1979), pp. 141–148.
51 Dahlman (1979), p. 142.
52 Williamson (1981), pp. 548–559.
32 B. F. Gussen
Property rights function as incentives to internalize externalities.53 An externality

exists where the cost of transaction in the rights between the parties (i.e., internaliza-
tion) exceeds the gains from internalization. Hence, were property rights are assigned
to the farmer, the rancher has gains from internalization. If the transaction costs are
zero, there will be an incentive to internalize the externality by paying the farmer
to build the fence. However, where the damages are low ($25 annually versus a
cost of $50 to build the fence), the outcome is still not efficient (i.e., the fence will
not be built). The efficient outcome obtains, however, where there is a collaborative
approach to the problem. Hence, if the farmer and rancher want to maximize their
joint profits, they would choose to share the cost of installing the fence around the
tilled area, given the lower cost compared to building the longer fence at the border.
Regardless of which rule is enforced (farmer rights versus rancher rights), in a col-
laborative economy, the outcome will always be the efficient one, namely building
the lower cost fence. In this collaborative economy property rights are irrelevant.
What matters is the level of transaction cost. Where these costs are zero, there is no
gain from private property. We can hence rewrite Eq. (3) above as follows:
δ − = f (Tc ) < 0 (4)
Let’s therefore look at how these transaction costs work. Imagine an entity A that is
faced with the decision of buying or making components of its production activities.54
In essence, the entity has to choose between a market or hierarchy organization (see
below). The former is favored where components are standardized, while the latter
where components are specific. Hence, where asset specificity is high, production cost
difference (C) approaches zero, while governance cost difference (G) goes below
zero reflecting the benefit of command over adjudication in resolving disputes.55
Markets will be favored over hierarchies as long as:
C + G > 0 (5)
This condition is satisfied where C > G. Where the opposite holds, hier-
archies are favored. A network organization is favored where the transaction costs
(C + G) are close to zero (C ≈ G ≈ 0). Institutions such as legal rules can
help achieve this equilibrium. The key point is that under a sharing economy, the
role of law is not to assign rights, but to facilitate the bargaining process between
the parties. The law is also responsible for protecting third parties through a similar
facilitation of bargaining.
53 Demsetz (1967), pp. 347–348.

54 Hierarchies are rule-driven processes that rely on “administrative” functions for coordination and
on “management” for direction.

55 Note that both cost differences are also a function of the level of production. The optimal value
of asset specificity depends on demand effects and absolute cost effects. This discussion looks at a
fixed level of output with demand effects.
Fig. 2 Effect of technology

on transaction costs and
population density
Here it is helpful to compare and contrast markets and networks. Both are mech-
anisms for coordination and governance (i.e., effective coordination).56 The former,
however, is a spontaneous distributed process of gathering and disseminating infor-
mation based on price signals. Under the neoclassical model, the interaction between
supply and demand leads to an equilibrium price. Under the Austrian approach, there
is no static equilibrium. The market is interpreted with emphasis on production and
innovation, and hence is seen as an evolving process (of creative destruction) in a
continuous disequilibrium. Networks on the other hand go beyond the price mecha-
nism. They organize, either spontaneously or deliberately, through social structures,
such as friendships, institutional structures, semi-public organizations and techni-
cal networks. Networks, sometimes seen as a hybrid form of organization between
hierarchies and markets, are informal processes of coordination that rely on direct
personal contact, and hence tend to be localized. The Internet, however, allowed for
these networks to extend across continents. The coordination in networks is driven by
trust rather than prices (instrumental rationality) or administrative orders (procedural
rationality). They therefore aim at a collective common purpose, which results in a
flat architecture that emphasizes equality between the participants. Networks reduce
transaction costs by nurturing trust and solidarity (i.e. common experience57 ). The
lower the population density is, the higher trust, and the lower the transaction costs.
In other words:
Tc = f (D) (6)
Transaction costs (Tc ) are a positive but declining function of (population) density
(D), as seen in Fig. 2.58
The Third Industrial Revolution is shifting the transaction costs curve downwards
(see Fig. 2). This is resulting in lower costs for any given density. Effectively, a new
56 Thompson (2003), p. 34.

57 Thompson (2003), p. 40.
58 Thompson (2003), p. 40.
34 B. F. Gussen
Fig. 3 The conceptual chain leading to negative externalities
technology lowers the effective population density on the old technology curve. The
Internet of Things is enabling this reduction in transaction costs through increasing
trust. Trust increases with embeddedness, or the degree of connectivity between
network participants, which itself is a function of the number of participants and the
technology used for connectivity.
Unlike the governance of hierarchies, which depends on ex ante coordination,
and unlike governance of markets, which is ex post coordination, the governance
of networks is a hybrid between active governance and “non-governance.”59 Net-
works reduce the long-run average transaction costs where there is low economic
system complexity.60 In the network-based sharing economy, complexity is reduced
by adopting a distributed architecture (with lower network densities). The arguments
so far can be summarized as shown in Fig. 3.
The fact that networks are based on high levels of trust and solidarity reduces the
occurrence of negative externalities. Where these externalities still occur, they are
resolved through cooperative approaches. Hence, in the rancher-farmer example, the
approach to damages caused by the cattle is one where the interests of the farmer and
of the rancher are seen as one (aligned) interest. The objective is to maximize their
combined profits from farming and ranching. The outcome is efficient regardless of
the rule of law.
To facilitate the bargaining process, the law needs to reduce transaction costs,
which encompasses all impediments to bargaining. These costs are endogenous to the
legal system in that legal rules can be redesigned to reduce these costs.61 The emphasis
shifts from internalizing negative externalities (ex post) to reducing transaction costs
(ex ante). When these costs approach zero, the use of resources will be efficient. These
costs can be divided into three main types of costs: (1) search costs, (2) bargaining
costs, and (3) enforcement costs.62 The Internet of Things and blockchains helps
drive all three types to zero by reducing population densities with larger mobility
(including that of information). The communication Internet makes the cost of search
for unique goods and services closer to that of standardized ones. An access-neutral
network for the Internet of Things ensures that the threat values and the cooperative
solution in negotiations is public, which in turn reduces bargaining costs. The threat
values are defined by parties’ legal rights. Blockchains help define these rights in
a way analogous to how public registration systems work. Similarly, the instant
fulfillment of agreements on the Internet reduces enforcement costs to zero. The
59 Thompson (2003), p. 48.

60 Thompson (2003), p. 50.
61 Cooter and Ulen (2014), p. 68.
62 Coase (1960), p. 1.
Internet of Things makes it easier to monitor implementation and cheaper to punish

any breach.
Incentivizing innovation in the sharing economy requires shifting emphasis
towards the reduction of transaction costs. When the surplus (benefit) from agree-
ments exceeds transaction costs, the net benefit (surplus less transaction costs) from
private exchange is positive. The law is required to ensure that this benefit is positive
by reducing transaction costs. This is done not by defining clear property rights,
but by reducing barriers to accessing the sharing economy and by ensuring that the
platform is network-neutral. The law can also reduce transaction costs by increas-
ing mobility. This encapsulates the mobility of all forms of capital: natural, social,
human, financial and manufactured capital.63 This mobility helps reduce population
density at existing cities by creating new ones and ensuring connectivity between
these cities. The drive is to transform the city into a collaborative ecosystem that
enables “collective action for the commons.”64 This drive requires a shift towards
“co-owned, co-managed, and co-produced institutions,”65 including the legal system.
Local services, including legal ones, “must be conceptualized as commons.”66 The
proposition is to provide universal access even where commons are in private hands.
Given the wide scope of economic activity that the law would be hoping to facilitate,
it follows that the general approach can only prescribe a few general principles and
leave much of the detail of how bargaining is to be facilitated to the agreement of the
parties. In turn, we would expect local variety to emerge. Different cities will have
different approaches driven by their unique economic complexities.
Historical precedents suggest that the envisaged shift in emphasis will take time to
materialize. The shift will lag the Internet of Things and blockchains but will eventu-
ally arrive via democratic processes. Inertia from vested interests and sunset remnants
from the Second Industrial Revolution are likely to continue to be operational well
into the 21st century.
In summary, the first effect that the sharing economy is having on cities has to
do with reducing transaction cost, which in turn is shifting the regulatory emphasis
from assigning negative externalities to enabling trust and embeddedness.
4 Lessons from Sharing City Seoul
This section looks at how Seoul was able to shift its regulatory emphasis to promote
the sharing economy. The analysis below is intended to explain how Seoul effected a
transition from a market economy to one based on networks. The approach illustrates
emphasis on creating social capital and on enhancing trust and solidarity in the city.
63 The Firve Capitals (2019).

64 Iaione (2016), pp. 415–415.
65 Iaione (2016), p. 416.
66 Iaione (2016), p. 417.
36 B. F. Gussen
In 2007, Seoul was ranked sixth in the world in terms of population density. The
city had a density of around 17,000 people per square kilometer.67 In comparison,
the city with the highest density in the world in that year, Mumbai, had a density
of around 30,000 km2 .68 In 2018, even though Seoul is ranked as the fifth largest
city in terms of population, its (physical) population density has now dropped to
below 10,000.69 In terms of its density, Seoul is today ranked 242 out of all cities
with a population of 500,000 or more.70 This drop in density was accomplished
through huge investments in primary infrastructure (such as roads) that increased
the urban area, and by doing so reduced the density. The discussion of the sharing
economy in Seoul picks up from this point and shows how the effective population
density (rather than the physical one) can be reduced through a flow-through effect
of lower transaction costs and lower negative externalities. The transition is possible
only through technological innovation, including innovation in relation to secondary
infrastructure. This is true even with an enlarged urban area. Part of the story of
this transition can be told through the lens of implementing the sharing economy
in Seoul. The driver behind Sharing City Seoul is the development of the urban
commons (the second infrastructure). Policies and regulations are revised to enable
the collaborative economy.
In 2012, the Seoul Metropolitan Government (SMG) proclaimed the Sharing City
Seoul Project.71 The project was intended to stimulate innovation in the sharing econ-
omy as a means to resolve urban problems (economic, social, and environmental).
To this end, SMG policies went beyond the creating of primary infrastructures (such
as roads and schools). The uniqueness of Seoul’s approach is in how it “blends reg-
ulations with support for new sharing enterprises.”72 The city was able to shift its
regulatory emphasis from internalizing externalities to reducing transaction costs.
An example of this approach is city investment in dozens of sharing organizations
since 2013. The SMG also engaged in producing secondary infrastructures to uti-
lize idle capacities in resources ranging from spaces, to objects, and even talents.
This approach required changing regulations to allow for more effective forms of
sharing. This regulatory change started with the Seoul Metropolitan Government
Act for Promoting Sharing. The Act defines “sharing” as “the shared use of space,
objects, or information to enhance their social, economic, or environmental values
and to enhance the citizens’ benefits or conveniences.”73 Five key policies were put in
67 The Largest Cities (2007).

68 The Largest Cities (2007).
69 Demographia (2018).
70 Tokyo, the most populous city in the world today (around 40 million), however, is ranked 646
with a density of 4500 km2 .

71 Heinrichs (2013), p. 228.
72 Greene and McGinty (2016).
73 Seoul Metropolitan Government Act No 5396. December 31, 2012.
place74 : The Sharing Promotion Ordinance, support for sharing enterprises, improve-
ment of laws and institutions, autonomous Gu (a borough or a district within Seoul)
incentive system and opening of public facilities and administrative information.
The Metropolitan Government explain the project as:
…social innovation…designed to create new economic opportunities, to restore reliable
relationships, and to reduce the wasting of resources with a view to resolving urban economic,
social, and environmental problems all together. Notably, while the existing city policies
focus on the construction of primary infrastructures such as roads, parking areas, schools,
and libraries, the city’s future policies focus on the construction of secondary infrastructures
such as spaces, objects, talents, and other unused resources in order to boost the utilization
thereof. Furthermore, Seoul is set to implement other policies that will respect and promote
private-sector capabilities, as well as policies that will require the public sector to open public
resources that are to be shared with citizens.75
The Sharing City Seoul Project has three prongs:76 expanding the sharing infras-
tructure, supporting sharing start-ups, and increasing the utilization of idle public
resources. Being able to achieve these objectives rests on enhancing the social cap-
ital, which in turn requires a new cultural paradigm that emphasizes trust and col-
laboration. The project, therefore, recognized the critical role of social trust and the
principle of hypotaxis in reinventing the city. This social justice prerequisite necessi-
tates an equitable approach to the sharing economy that prioritizes the development
of low socioeconomic neighborhoods of the city.
Seoul illustrates the shift of emphasis in legal rules from internalizing externali-
ties to reducing transaction costs. Sharing City Seoul is creating new social capital
in the form of a sharing culture, which suggests a move from a market organiza-
tion to a network one. The project identified the sharing assets owned by the city,
identified sharing opportunities in existing policies and operations, and invested in
the co-production of sharing services. The project invested in the non-commercial
sharing ecosystem and acted as the hub for the sharing economy. The Metropolitan
Government explains the social capital effects as follows:
Sharing can contribute to the recovery of the disappearing sense of community, increasing
interpersonal exchanges and restoring broken relations since sharing promotes a trust-based,
reciprocal economy.77
Seoul’s experience explains that in order for cities to expand their sharing infras-
tructure, their governance models will have to emphasize building social trust. This
requires enlivening two complementary concepts: hypotaxis and solidarity.78 The
former refers to axioms of assistance and non-interference.79 Hypotaxis invites the
states to help (rather than replace) actors in the sharing economy. Moreover, hypotaxis
74 Bernardi (2016).
75 Seoul Metropolitan Government (2012).
76 Johnson (2013).
78 Booth (2014).
79 Hypotaxis (subsidiarity) however should not be confused with (top-down) delegation. See Gussen
(2016b), p. 383.
38 B. F. Gussen
puts emphasis on transferring competencies to actors in the sharing economy so that

they would acquire spontaneity and require less and less help from government.
While hypotaxis has an efficiency logic to it, it also has a moral anchor. It envisages
association through free choices and acts.
Sharing City Seoul enacted sharing promotion rules intended to reduce transaction
costs, rather than focus on assignment of externalities. These rules “stipulate the
principle for sharing public resources, designates the organizations and enterprises
for sharing resources, provides the administrative and financial support for them, and
lays out the guidelines for the formation of the sharing promotion committee.”80 The
latter is a public-private partnership for city governance. The committee is made up
of “personnel from academia, legal circles, the press, businesses, non-profit private
organizations, and research institutes, as well as of the directors and general officials
who are responsible for economic, welfare, transportation, and innovation affairs.”81
The committee takes up the role of advising on how law and systems can be improved
to the end of promoting the sharing economy.
5 Prediction: Beyond Decentralization
In the long-term, we will see an even more disruptive effect on the current gover-
nance model of the nation-state. The Third Industrial Revolution is resurrecting the
legal personality of cities.82 Under the Cooley-Eaton-McQuillin thesis, cities are the
dominant body politic.83 It was 20th century liberalism that brought about the cur-
rent powerlessness of cities. Liberalism, with its model of the world as dualities, saw
participatory democracy on a small scale like the city-region as unworkable. Cities
with real power were an intermediate structure between the state and the individual.
Through legal doctrine, liberalism proceeded to eliminate this intermediate structure
and replace it with instruments of the state.84

82 Paton (1973), p. 393.
83 Frug (1980), pp. 1057–1113. The thesis was a response to John Dillon’s 1872 treatise on the
authority of states over cities. The latter was based on a rationale for protecting private property.
The first part of the thesis was developed by Judge Thomas Cooley who denied absolute state
supremacy over cities. Cooley argued that local government was a liberty of exceptional importance
in American colonial history. The second component of this thesis came from Amasa Eaton, a
member of the Rhode Island House of Representatives (1865–66 and 1872–74). He argued that the
right to local self-government preceded the incorporation of the states and hence was not subject
to state restriction. The third part of the thesis was proposed by Eugene McQuillin, who was an
American lawyer and a member of the National Conference of Commissioners on Uniform State
Laws. McQuillin canvased the right to local self-government in the history of municipal corporations
in support of the proposition that cities were not created by the states.
84 Frug (1980), p. 1074.
Today we can discern a move towards empowering cities on two other fronts:
one is domestic where there is constitutional recognition, under cooperative mod-
els of federalism, of the local governments of city-regions as co-equal to federal
and state governments, and the development of what is known as the “doctrine of
usurpation of jurisdiction.”85 The approach does not emphasize political autonomy
but rather the idea of hypotaxis where general competence powers are extended to
city-regions.86 The second front is international, where there is an emerging field
of law that acknowledges city-regions as independent international actors. Interna-
tional law has long had an indirect impact on cities, but now there is an emerging
trend where cities are becoming distinct international actors almost co-equal to their
nation-states.
In 2016, Seoul’s mayor, Park Won-soon, signed a joint declaration with other
Korean cities to develop a network of sharing cities. By 2017, Seoul played a major
role in launching the Sharing City Alliance, during the second Sharing City sum-
mit in New York. As of today, in addition to Seoul, the Alliance has over thirteen
city members, including Amsterdam, Barcelona (the venue for the 2018 Summit),
Copenhagen, Singapore, Tel Aviv, and Toronto. My prediction is that these cities
will form the nucleus of loose confederations on a continental scale. The traditional
approach where local government is simply an administrative division of the state is
now being replaced by an approach, driven largely by Rifkin’s Third (or Schwab’s
Fourth) Industrial Revolution, where international institutions redefine the scope of
domestic frameworks. This has come to be known as “International Local Govern-
ment Law,” an interdisciplinary field that draws on comparative urban governance as
well as on the ‘world cities’ hypothesis.87 Unlike these other disciplines, however,
international local government law emphasizes the dual legal nature of city-regions
(i.e., cities and their hinterland) as both sub-national governments and as independent
international actors. This emerging field will determine both, who should regulate the
legal framework for city-regions, and the nature of that framework. Examples of this
approach include decisions by international arbitration tribunals regulating cities’
land use. This international framework envisages (at least tentatively) empowering
cities “principally as a mechanism for promoting private economic development.”88
International instruments such as the United Nations International Covenant on
Civil and Political Rights (ICCPR) and the International Convention on Economic,
Social and Cultural Rights (ICESCR), inter alia, are altering the relationship between
85 Humby (2012), p. 628.

86 Caulfield and Larsen (2002).
87 Frug and Barron (2006) (eds), pp. 1–2. Although Frug and Barron suggest that international local
government law “should not be evaluated in terms of whether the world is enhancing or limiting
local power. The focus instead should be on the kinds of cities that international local government
law is trying to create.” See Frug and Barron (2006) (eds), p. 60. I argue that international local
government law enables city-regions to become independent international actors, which would
make sense only if we empower them to act in such capacity. I suggest that international local
government law does in fact empower city-regions rather than simply making cities creatures of
international statutes rather than domestic ones.
88 Frug and Barron (2006) (eds), p. 4.
40 B. F. Gussen
cities and nation states. City-regions are becoming “nodal points for radially distinct
governance projects that have their common goal to transform cities from mere
subdivisions of sovereign states into legally empowered entities, able to advance
goals and values that are different from their states.”89
What is envisaged then is a new world order where the sharing economy will
enable “world cities”90 (privileged cities that have a global economic status driven by
innovation) to steer political and economic processes on a global scale. Network-like
loose continental confederations will replace the nation-state and regional models
established by the First and Second Industrial Revolutions.
To evaluate this governance effect of the sharing economy, i.e., the rise of city
autonomy, we need to look at the history of local government in South Korea,
including the role of the Seoul Metropolitan Government. While a local government
system was first implemented in South Korea in 1952,91 the authoritarian regime
(1961–1987) abolished the system and replaced it with a command-and-control sys-
tem of intergovernmental relations.92 It was only under the Fifth Republic in the
1980s that an extensive local autonomy was constitutionally enshrined.93 However,
it was the rapid economic development of the 1990s that brought democratic local
autonomy reform to the political agenda.94 The reintroduction of this autonomy
was largely a deal among the political elite at the center.95 The adopted local gov-
ernment system was an elected mayor-council two-tier system. The first tier was the
regional one, which included provinces (Do 道) and metropolitan cities (Kwangyeok-
si 廣域市). The second level was the municipal level, which included rural counties
(Gun 郡), cities (Si 市), and districts (Gu 區).96 A consolidation phase followed in
the mid-nineties. The Seoul Metropolitan City is one of the regional (tier-one) gov-
ernments. All local governments are autonomous bodies with elected councils and
administrative executives. A local government can change its status by meeting the
conditions stipulated for in the Local Autonomy Act 1947 (Korea) (LAA).97 Becom-
ing a metropolitan city requires a necessary condition of a population of one million
or more, and other conditions from the LAA, including financial capacity. However,
differences in size (area and population) continue to exist even among similar local
government types.98 Roughly one-third of the government functions in Korea are
carried out by local government.99
89 Blank (2005–2006), pp. 875–899. See also, Gussen (2018).

90 Friedmann (1986), p. 69.
91 Choi et al. (2013).
92 Seong (2000), pp. 127–148, 130–31.
93 Article 118 of the Constitution of the Fifth Republic.
94 Oh (1999), pp. 89–97.
95 Lee (1996), pp. 60–64.
96 Choi et al. (2013), p. 24.
97 The LAA was repealed and replaced in 2007.
98 Choi et al. (2013), p. 27.
99 Choi et al. (2013), p. 30.
New efforts for decentralization coincided with the advent of the Internet. In
1999 the national government established the Special Committee on the Devolu-
tion of Government Affairs (later renamed the Presidential Commission for Decen-
tralization or PCD).100 This effort resulted in the Special Act on the Promotion of
Decentralization in 2004.101 The Act entrusted the PCD with promoting a systematic
allocation of functions between the central and local governments. Nevertheless, cen-
tral government continues to exert a high level of control over local governments.102
For example, Article 155 of the LAA authorizes central government intervention in
the daily operation of local governments. Article 156 states that tier-one governments
fall under the supervision of the central government. Article 158 stipulates for audits
of local government functions by the Minister of Public Administration and Safety
(MOPAS). While the MOPAS has also a veto on the decisions of local governments,
the local council can override the veto by a two-third majority vote, with over half
of the members present. This would be followed by a lawsuit in the Supreme Court
to resolve the matter. It should be noted, however, that MOPAS has a strong fiscal
control on local governments through grants and tax sharing schemes.103
These changes were prompted by a global political and economic environment that
favored decentralization.104 The decentralization efforts were also influenced by civil
society and non-government organizations (NGOs).105 For example, the Citizen’s
Coalition for Economic Justice (CCEJ), the People’s Solidarity for Participatory
Democracy (PSPD), the Citizen’s Coalition for Better Government (CCBG) and the
Civic Movement for Decentralization (CMD) played a decisive role in promoting
decentralization.106 These organizations worked with the predecessor to the PCD, the
Presidential Committee on Government Innovation and Decentralization (PCGID)
to produce a 2003 roadmap for decentralization.107
Given the new drive towards the sharing economy, and the concurrent shift from
globalization to continentalization, my prediction is that the decentralization efforts
in Korea will potentially see Seoul become the first truly autonomous global city in
the 21st century.
100 Choi et al. (2013), p. 31.

101 This Special Act has since been repealed by the Special Act on Decentralization and Restruc-
turing of Local Government Entities 2013 (Korea). The latest version is the Special Act on Local
Autonomy and Decentralization, and Restructuring of Local Administrative Systems 2018 (Korea).
102 Choi et al. (2013), p. 61.
103 Choi et al. (2013), p. 63.
104 Kim (2003), p. 55.
105 Lee and Arrington (2008), pp. 75–81.
106 Bae (2009), pp. 471–72.
107 Bae (2009), p. 470.
42 B. F. Gussen
6 Conclusion
This chapter looks at the role of the law in enabling cities as a network of combi-
nations of digital technologies that drive the sharing economy. The chapter traces
the sharing (or collaborative) economy in its historical context to ascertain its nov-
elty. The analysis suggests that collaboration in economic activity has been taking
place since the very beginning of civilization. The nature of the collaboration, how-
ever, changed as a result of increasing population densities in human settlements.
This change was moderated by technological innovation to allow continued sharing
intra-cities and inter-cities, but technological innovation lagged population growth.
This intermission in new combinations necessitated changes in the nature of col-
laboration. These changes were reflected in different forms of organization. Hence,
while collaboration in early civilization (up to the medieval period) was based on a
conviviality function (driven by trust and solidarity) and organized around (social)
networks, increasing population density saw a transition into an economic function
and market organization. Further increases in population brought about a political
function to cities, which in turn resulted in a transition to a hierarchical organiza-
tion—including the emergence of new cities. These organizational modes served as
a means of coordination. Legal rules were implemented to ensure the effectiveness
of this coordination (governance). Hence, while networks put emphasis on rules that
facilitated collaboration (and enhancing trust and hypotaxis), markets required ex
post assignment of negative externalities. Hierarchies, on the other hands, required
ex ante regulation of behavior. The arrival of new waves of technological innova-
tion allowed for higher density (or lower effective density) and a transition back to
networks, or at least for a transition away from hierarchy. The 21st century wave
(of technological innovation) is driven by the Internet of Things and automation
processes such as blockchains. This innovation is allowing for a return to a network
organization at a scale (population-density) never seen before. Cities will now be
able to optimize their densities (while still increasing their population), while more
cities will emerge across the globe. Legal rules need to remove barriers to collabora-
tion, rather than the current (market-based) emphasis on internalizing externalities.
The outcomes from the Sharing City Seoul Project, were analyzed to illustrate this
transition.
Hypotaxis is imperative to helping cities transition into the sharing economy
(and beyond), which itself is imperative for the sharing economy to keep growing. In
turn, this necessitates re-imagining global governance beyond the nation-state model.
Sharing cities will have to be afforded a wide margin of autonomy, including a legal
personality on the international stage. Nation-states provide the autonomy under
which the sharing economy can prosper within sharing cities. Within this autonomy,
the sharing economy emerges from non-state actors such as community and voluntary
associations. Best practices would then be diffused through international frameworks
for cooperation between sharing cities (sharing city networks).
In terms of the role of the Seoul Metropolitan Government (SMG) in stimulating
innovation in the sharing economy, while the sharing economy is a generalization
of social welfare through digital technology, unlike the traditional welfare model,
the sharing economy counsels higher orders of government against intervention in
the activities of the sharing economy. Instead, hypotaxis and solidarity guide the
evolution of sharing infrastructures at the city scale. The sharing economy requires
navigating a middle ground between laissez-faire individualism and bureaucratic col-
lectivism by identifying social responsibilities at the meso-scale between the individ-
ual and the state. To elucidate this point, the Sharing City Seoul Project is analyzed
through the lens of hypotaxism. The key proposition is that state regulation should
only facilitate the self-organization of this meso-scale (sharing) economy. Local gov-
ernments are best suited to facilitate this self-organization. Seoul illustrates the new
paradigm beyond externalizing diseconomies through regulatory intervention that
focuses on facilitating innovation in the sharing economy.
The above analysis needs to be qualified for specific legal subjects and the theories
underlying them. To understand how the sharing economy is going to influence
jurisprudence, the analysis should look at the effect of the sharing economy on
private and public law. A detailed analysis is beyond the scope of this chapter. The
chapter, therefore, provides only a sketch of possible developments.
References
Agyeman J, McLaren D (2017) Sharing cities. Environ Sci Policy Sustain Dev 59(3):22–27
Bae Y (2009) Decentralization and Transformation of Central-State in South Korea: The Role of
Local and Civil Society. Paper presented at the Global Korea: Old and New 6th Korean Studies
Association of Australasia (KKSA), Biennial Conference Sydney, 9–10 July 2009
Bernardi M (2016) Sharing city, Seoul 2016: state of the art and news. Available at LabGov (Labo-
ratory for Governance of the City as a Commons). http://www.labgov.it/2016/12/08/sharing-city-
seoul-2016-state-of-the-art-and-news/. Accessed 2 Mar 2019
Blank Y (2005–2006) The city and world. Columbia J Trans Law 44(3):875–899
Booth P (ed) (2014) Catholic social teaching and the market economy. Institute of Economic Affairs
Monographs, London
Buchanan J, Stubblebine W (1962) Externality. Economica 29(116):371–384
Caulfield J, Larsen H (eds) (2002) Local government at the millenium. Springer, Wiesbadem
Choi J, Choe C, Kim J (2013) Local government and public administration in Korea. Local
Government Officials Development Institute, Seoul
Coase R (1960) The problem of social cost. J Law Econ 3(1):1–44
Cooter R (1982) The cost of coase. Legal Stud 11(1):1–33
Cooter R, Ulen T (2014) Law and economics, 6th edn. Pearson Education Limited, Harlow, Essex
Dahlman C (1979) The problem of externality. J Law Econ 22(1):141–162
Davis K (1955) The origin and growth of urbanization in the world. Am J Sociol 60(5):429–437
Demographia (2018) World urban areas 14th annual edition. Available via Demographia. http://
www.demographia.com/db-worldua.pdf. Accessed 4 May 2019
Demsetz H (1967) Toward a theory of property rights. Am Econ Rev 57(2):347–359
Ebdon C, Franklin AL (2006) Citizen participation in budgeting theory. Public Adm Rev 66(3):437–
447
Friedmann J (1986) The world city Hypothesis. Dev Change 17(1):69–83
Frug GE (1980) The city as a legal concept. Harvard Law Rev 93(6):1057–1154
44 B. F. Gussen
Frug GE (1999) City making: building communities without building walls. Princeton University
Press, Princeton
Frug GE, Barron DJ (2006) International local government law. Urban Lawyer 38(1):1–62
Galka M (2016) A visual history of urbanization, from the world’s first city in 3700 BC to the present.
Available at Huffpost https://www.huffingtonpost.com/max-galka/a-visual-history-of-urban_b_
10547270.html. Accessed 15 Apr 2019
Greene S, McGinty J (2016) What if cities could create a truely inclusive local sharing
economy. Available at Urban Institute. https://www.urban.org/sites/default/files/publication/
81546/2000816-What-if-Cities-Could-Create-a-Truly-Inclusive-Local-Sharing-Economy.pdf.
Accessed 3 May 2019
Guerini F (2014) How Seoul became one of the world’s sharing capitals. Forbes
Gussen BF (2013) On the problem of scale: Hayek, Kohr, Jacobs and the reinvention of the political
state. Const Polit Econ 24(1):19–42
Gussen BF (2014) Subsidiarity as a constitutional principle in New Zealand. New Zealand J Public
Int Law 12(1):123–144
Gussen BF (2016a) Australian constitutionalism between subsidiarity and federalism. Monash
University Law Rev 42(2):383–418
Gussen BF (2016b) An economic model for constitutional designs: from Nation-States to an
‘Olympic World System’. Coase-Sandor Summer Institute in Law and Economics Colloquium
Series. Chicago Law School, Chicago
Gussen BF (2018) Ranking economic performance and efficiency in the global market. IGI Global,
Hershey
Gussen BF (2019) Axial shift: city subsidiarity and the world system in the 21st century. Palgrave
Macmillan, Singapore
Hayek F (1976) Law, legislation and liberty, volume 2: the mirage of social justice. University of
Chicago Press, Chicago
Heinrichs H (2013) Sharing economy: a potential new pathway to sustainability. GAIA-Ecol
Perspect Sci Soc 22(4):228–231
Humby T (2012) Maccsand: intergovernmental relations and the doctrine of usurpation. South
African Public Law J 27(2):628–638
Iaione C (2106) The CO-city: sharing, collaborating, cooperating, and communing in the city.
American J. Econ. Sociol 75(2):415–455
Johnson C (2013) Is Seoul the next great sharing city? Available at United Nations University.
https://ourworld.unu.edu/en/is-seoul-the-next-great-sharing-city. Accessed 4 May 2019
Kardashev N (1964) Transmission of information by extraterrestrial civilizations. Soviet Astron
8(2):217–221
Kim WB (2003) The evolution of regional economic disparities in Korea. Korea J 43(2):55–80
Kohr L (1977) The overdeveloped nations. Schoken Books, New York
Lee J (1996) The politics of decentralization in Korea. Local Government Studies 2(3):60–71
Lee J, Arrington C (2008) The politics of NGOs and democratic governance in South Korea and
Japan pacific. Focus 23(1):75–96
McLaren D, Agyeman J (2015) Sharing cities: a case for truly smart and sustainable cities. MIT
Press, Cambridge, MA
Oh JK-C (1999) Korean politics: the quest for democratization and economic development. Cornell
University, Ithaca, New York
Paton GW (1973) A textbook of jurisprudence. Oxford University Press, Oxford
Poteete J, Ostrom E (2010) Working together: collective action, the commons, and multiple methods
in practice. Princeton University Press, Princeton
Putnam RD, Leonardi R, Nanetti RY (1994) Making democracy work: civic traditions in modern
Italy. Princeton University Press, Princeton
Rifkin J (2011) The third industrial revolution: how lateral power is transforming energy. Palgrave
Macmillan, New York
Rifkin J (2014) The zero marginal cost socitey. Palgrave Macmillan, New York
Rose CM (1986) The comedy of the commons: commerce, custom, and inherently public property.
Available at the Yale Law School Legal Scholarship Repository. https://digitalcommons.law.yale.
edu/cgi/viewcontent.cgi?article=2827&context=fss_papers. Accessed 4 May 2019
Schor J (2016) Debating the sharing economy. J Self-Governance and Manage Econ 4(3):7–22
Schumpeter JA (1934) The Theory of Economic Development. Oxford University Press, Oxford
Schumpeter JA (1939) Business cycles: a theoretical, Historical and statistical analysis of the
capitalist process. Porcupine Press, Philadelphia
Schwab K (2016) The fourth industrial revolution. Crown Publishing Group, New York
Seong KR (2000) Delayed centralization and incomplete democaratic consolidation. In: Diamond
LJ, Shin DC (eds) Institutional reform and democratic consolidation in Korea. Hoover Institution
Press, Stanford University, Standford, pp 127–148
Seoul Metropolitan Government (2012) The sharing city seoul project. Available at
Seoul Metropolitan Government. http://english.seoul.go.kr/policy-information/key-policies/city-
initiatives/1-sharing-city/. Accessed 4 May 2019
The Five Capitals (2019) Available at forum for the future. https://www.forumforthefuture.org/
project/five-capitals/overview. Accessed 1 May 2019
The Largest Cities (2007) The largest cities in the world by land area, population and density.
Available at City Mayors Statistics. http://www.citymayors.com/statistics/largest-cities-density-
125.html. Accessed 1 May 2019
Thompson GF (2003) Hierarchies, markets, and networks: a preliminary comparison. In: Thompson
GF (ed) Between hierarchies and markets: the logic and limits of network froms of organization.
Oxford University Press, Oxford, pp 21–52
The 4037 Cities (2015) The 4037 cities in the world with over 100,000 People. Available via Brillant
Maps. https://brilliantmaps.com/4037-100000-person-cities/. Accessed 2 Mar 2019
Vermesan O, Friess P (2013) (eds) Internet of things: converging technologies for smart
environments and integrated ecosystems. River Publishers, Delft
Williamson OE (1981) The economics of organization: the transaction cost approach. Am J Sociol
87(3):548–577
Digital Age and Personal Data
The Digital Person—A New Legal
Entity? On the Role of Law
in an AI-Based Society
Cecilia Magnusson Sjöberg
Abstract The starting point is that there is a need to discuss the concept of “digital
person” as a new legal entity in light of the development and application of AI
technology. The purpose is not just to introduce yet another term to add to those
of the natural person and the legal person. The intention is rather to initiate the
notion of a new legal figure, which under certain circumstances tentatively could be
granted a legal capacity, with rights and responsibilities. The risk is otherwise an
emerging dysfunctional legal society where there is no legal entity, which can take
the role of the subject. Issues that arise concern e.g., self-driving car liability, pricing
algorithms on the competitive market and data protection when profiling consumers.
Furthermore, well-established principles on openness and transparency are also in
jeopardy. This is especially apparent in the context of machine learning and dynamic
algorithms. More specifically, there is a risk that a passive community can lead to
legal losses on behalf of both individuals and organizations. A risk scenario would be
when various transactions, which occur in digital environments, are not recognized
as legally valid because the intelligent agent applied is neither recognized as a natural
nor a legal person.
Keywords Algorithms · Artificial intelligence (AI) · Automation · Digital law ·

Digital person · Legal entity · Machine learning · Transparency
C. Magnusson Sjöberg (B)

Faculty of Law, Stockholm University, Stockholm, Sweden
e-mail: cecilia.magnussonsjoberg@juridicum.su.se

https://doi.org/10.1007/978-981-15-1350-3_4
50 C. Magnusson Sjöberg
1 Introduction
Could the emerging AI-based society entail a need for a new legal entity?1 A legal
entity here refers to a general concept that, within a given legal order, acts as a carrier
of various legal functions, which may include both duties and benefits. To provide a
few examples: A so-called third party is a legal entity, which is or can be relevant in a
legal relationship, without formally being a party to it. The same can be said to apply
for a creditor and a debtor, respectively. The first refers to an entity that is owed a debt
and the second to an entity that owes a debt to another party. Of particular interest
in this context is whether or not the long-established system with “natural persons”
and “legal persons” is dynamic enough to face the challenges that the development
and use of Artificial Intelligence (AI) entail. In short: Is there a need to discuss a new
legal entity in light of the developments in AI technology?
The historic perspective can broadly be summarized as follows. There has, of
course, been a period in the past when no distinction was made between natural and
legal persons. However, legal relationships nowadays are characterized by legal rules
and regulations aimed at individuals and/or organizations in various constellations.
Naturally, it is hard to predict the future. Even so, speculating that digitalization
will continue beyond where it currently stands does not seem overly daring. Robotic
embodiment of advanced technology is just one example of an area requiring atten-
tion. In parallel with our past transformation from an agrarian society to an industrial
one and then, in large parts of the world, on to an information society, an AI-based
society is now appearing. Obviously, this is not a matter of a swift societal change,
but something that happens gradually, albeit remarkably quickly. Of particular note
in this development is the emergence of advanced and even self-learning algorithms,
which can be programmed and used in digital environments encompassing large
amounts of data.
Against this background, a number of key questions crystallize and are described
in more detail below. The first question is if we already have an encoded law, governed
by algorithms. The second is if administration of justice can be achieved with the
help of robots. The third is if we need to introduce an analytical model in the form
of a legal entity called a “digital person.”
The format of the text is relatively free as regards both depth and breadth. There-
fore, the reader should not expect more extensive analyses, but can hopefully still be
inspired to consider these questions further. Section 1, this introduction to the topic,
is followed by Sect. 2, which describes the aforementioned key questions in detail.
Section 3 discusses the potential effects of AI on the role of law in society, as well
as if and how the creation of the concept of the digital person can alter these effects.
Section 3 also contains some suggested further reading. Section 4 concludes with
general remarks on the future contents of the legal profession and the potential that
the concept of the digital person entails.
1 This text is a reworked contribution to the essay collection “Människor och AI” (“Humans and
AI”), edited by Daniel Akenine and Jonas Stier, published by ADDAI.ORG, 2019. Linnéa Holmén
has carried out the translation from Swedish into English.
The Digital Person—A New Legal Entity? On the Role of Law … 51
2 Three Key Questions
2.1 Do We Already Have Encoded Law Governed

by Algorithms?
The hypothesis is that yes, this is the case, with reference to the fact that the devel-
opment has already gone so far that it can justifiably be said that much law, both
nationally and—to a significant extent—internationally, is dependent on algorithms
that have been encoded to be executed by computers. Of course, this is a very simplis-
tic line of reasoning, as the reality is much more complex and cannot be encompassed
in a simple yes or no. However, the fact remains that our society is dependent on legal
automation in the sense of fully or partially automated legal decision-making. This
is apparent mainly in the public sector, where administrative decisions are generated
automatically as a rule, rather than as an exception. The digital is, in other words,
the norm.
Legal automation is particularly common in areas like taxation, social security,
and student aid. Of special interest from an AI perspective is the development that
not only simpler types of mass administration are subject to automated decisions.
To an increasing extent, personalized assessments of individual characteristics etc.,
are being shifted to machine processing. Naturally, automated procedures with more
subjective implications also occur in the private sector, for instance in assessing the
creditworthiness of customers.
2.2 Can Administration of Justice Be Achieved with the Help

of Robots?
The hypothesis under this heading is also that legal dispute resolutions do not always
require a biological being, but could tentatively be performed with the help of AI
embodied in a robot. Of course, it would be naïve to suggest that this is currently
more than a vision. One explanation lies in the fundamental human functions that
we usually associated with a court’s processing of both (civil law) disputes and
(criminal law) cases. While the need varies for human interaction between involved
parties, prosecutors, judges, lay judges, arbitrators, etc., it is not called into question.
Still, administration of justice is already often performed through a written proce-
dure in itself based on digital facts and analyses created through fully or partially
automated functions. Whether the result of increasingly AI-based legal automation
will be generally positive or negative for the individual remains to be seen. In terms
of risk factors, the existence of “biased robots” challenging fundamental principles
on objectivity and equal treatment is one example of a topic already under debate.
At the same time, robots have the potential to strengthen the rule of law through
predictable justice and legality, in some sense. The latter refers, in brief, to deci-
sions being made and executed only when they are consistent with the regulations in
force. However, it will likely become increasingly difficult to ensure transparency in
the decision-making process as the complexity of the algorithms increases through
ancillary machine learning.
2.3 Is a New Legal Entity in the Form of a Digital Person

Needed?
History moves on, which is not always apparent to people living in the midst of
societal development. Still, in our time, it is noticeable that digitalization is achiev-
ing new heights, both quantitatively and qualitatively. To monitor and promote a
humanistic perspective when machine-based solutions are increasingly coming to
shape both private life, business pursuits, and the undertakings of public authorities,
new elements must be included in the ongoing dialogue. The hypothesis here, as has
already been suggested, is that there is a need for some form of analytical model—a
new legal entity—to which the reasoning on AI and law can be connected.
This approach is not based on the assumption that a new legal entity could solve all
the legally oriented questions on optimal AI applications with connected distribution
of liability and management of exclusive rights, such as copyright, patents, etc. Still,
a well-balanced juridification—or legalization, if you prefer—of AI, in order to
make use of the potential of the technology, appears interesting as a topic of further
research. Here are some starting points:
i. A digital person is obviously not the same as a biological being, though it can
be noted that traditional delimitations are being broken up through the use of
implants, etc.;
ii. A digital person is also not a person/entity that can be seen as a type of legal
person, e.g., a limited liability company or a sole proprietorship.
This background, among other things, reveals the need for an analysis model that
in a more tangible way than today can move forward the legally oriented reasoning
on AI. This approach, in turn, is based on the insight into—or rather: the acceptance
of—AI as an at least partially new phenomenon that fundamentally changes the
legal infrastructures of society. Thus, it entails a new dimension with implications
regarding intelligent agents and much more. On the Wikipedia site, an intelligent
agent has been described as “an autonomous entity which observes through sensors
and acts upon an environment using actuators (i.e., it is an agent) and directs its
activity towards achieving goals (i.e., it is ‘rational,’ as defined in economics).”
In a discussion like this, there is reason to consider what might be called the “null
alternative.” Here, this would relate to what might happen if the development of AI
were allowed to run its course without the legal creativity associated with the creation
of a new legal entity—the digital person.
Seen from a legal perspective, it is possible to roughly distinguish some negative

consequences at both the macro and the micro level. Generally speaking, a future
dysfunctional legal society appears, in which rule of law guarantees are not only
eroded, but risk disappearing entirely, as there is no functional legal entity which can
take the role of the subject. Well-established principles on openness and transparency
are also in jeopardy. This is especially apparent when dynamic algorithms provide
the foundation of machine learning. More specifically, the “null alternative” can lead
to legal losses on the part of individuals and organizations when various transactions,
which occur in digital environments, are not recognized as legally valid because the
intelligent agent is neither a natural nor a legal person. The questions are many and
reasoning regarding the digital person is resumed in closing.
3 General Reflections with Relevant References
3.1 The Role of Law in an AI-Based Society
The aforementioned hypotheses can preliminarily be considered as verified. Of

course, a more detailed legal scientific approach requires both broader and more
in-depth studies. However, it can summarily be concluded that the existence of algo-
rithms that encode and govern legal decision-making is now a reality. This fact does
not, however, equate to the encoded legal order working smoothly. On the contrary,
transformation of legal information in the form of legislation or court cases into code
brings to the fore a wide variety of different questions related to the rule of law. An
example is who—a social secretary, judge, or systems developer—has the authority
to decide how to program the meaning of a “reasonable standard of living” within the
framework of financial aid in the form of social assistance. The discussion in Sweden
during the spring of 2017 on the use of a robot (“Ernst”) in the social administration
of Trelleborg and Kungsbacka is an adjacent example of how fundamental questions
regarding job elimination are also cause for concern, as well as if human interaction
is at risk when daily AI use becomes more powerful and readily available.2
However, a topic of particular concern is (simply put) the transition from tra-
ditionally deterministic algorithms, i.e., algorithms which given input data behave
in the same way over and over again, to dynamic algorithms, which depending on
different input data change over time. The latter form of machine learning mainly
challenges fundamental legal principles on openness and transparency. It is one thing
that a certain AI application occurs within the framework of a “black box,” the con-
tents of which a rights-holder can have access to, and another that means no human
individual can fully understand various automated courses of events.
Such lines of reasoning often include the expression “code is law” as a paraphrase
of Lessig’s work in the field (see below). However, this point is—not least on the
2 Voister (2017).
back of the aforementioned—slightly misleading. While it is important to recog-

nize computer programs as norm-setting, it is more complex to make a comparison
between code as definitively governing in the same way as “law.” What is typical
of legislation is the changeability of regulations. Naturally, a constitution like the
Swedish Freedom of the Press Act is changed very seldom, unlike the legislation
on secrecy, which is more or less continuously subject to changing provisions. The
point here is that the term “law” is at least as complex and dynamic as any computer
program. There is thus every reason to avoid comparing apples and oranges when
the technological and legal semantics have this nature.
However, this does not mean that those with knowledge of IT should ignore
matters related to the rule of law, in the form of predictability, openness, equal
treatment, and legality, but rather that the discussion might need to be more nuanced.
The importance of the rule of law in the courts’ administration of justice should for
instance not be confused with the citizens’ need of legal safety. The risk of being
beaten by wandering robots on city streets must, in other words, be managed. It can
also be concluded, slightly at odds with this, that robots or AI-based surveillance with
advanced facial recognition in public places can make society safer for a person who
feels vulnerable and for that reason is willing to surrender some privacy. It should
be kept in mind that the technology itself is neither “good” nor “evil.”
Thoughts and discussions on the use of robots are nothing new per se. However,
this does not detract from the importance of, in tandem with societal development,
keeping robotification as an extension of AI high up on the agenda. One challenge
is not to get caught up in the human fascination for mechanical peers, but rather
to highlight the functionality that robots provide in a wider context. An interesting
initiative in this area is the drafted EU report, which though it has as its starting
point the “personhood status” of robots, with a recommendation to “the Commission
on Civil Law Rules on Robotics” (2015/2103(INL)) from the “Committee on Legal
Affairs,” also contains broader discussions and suggestions.
The line of reasoning presented in this text results in the need for cohesive legal
reasoning which—using among other things a new legal entity: the digital person—
can unify the law in the AI-based society. This is, as already mentioned, an analysis
model that yields a wide range of questions to which there are naturally no precise
and concise answers. The following is a small selection. The time spectrum stretches
from the current to the future and the questions are more or less closely related to
today’s lifestyle, which to an increasing extent has a digital nature.
i. Is a digital person the same thing as an algorithm or a set of algorithms which
combined give the digital person a basic algorithmic identity? This implies
that a digital person is more than a digital identity in the form of an online
actor.
ii. Can a basic algorithmic identity be the foundation for materialization of dig-
ital persons, i.e., robotization, for instance through judges with an underlying
AI-based functionality, but designed as humans? If the answer is yes, can we
then get human behaviors programmed into them, in the form of empathy,
arrogance, kindness, sleepiness, etc.?
iii. Is there a reason to, in certain contexts, for instance during a trial, retain a cog-
nitive distinction between biological (natural person) and machine entities?
Do we need to extend our catalogs of rights with a right to, in certain contexts,
be informed if communication is taking place with a natural person, a legal
person, or a digital person?
iv. Should it be possible to, with legislative support, appeal for instance an author-
ity’s use of algorithms: In other words, a right to system redress? Will this yield
a new profession based on digital powers of attorney and digital representatives
to match the digital counterparts?
v. Do legal systematics need to be modified so that we, in addition to what is
referred to as traditional “hard law” and “soft law,” which may include cus-
toms, standards, etc., need “digital law” (“agent law”)? Is traditional law not
intelligent enough when it comes to digital information resources and has it
therefore played out its role as a conflict resolution mechanism and guarantor
of the freedoms and rights of natural persons?
vi. Is the rule of law outdated as a value and replaced by ethics, which will be man-
aged not by courts but by the (partially virtual) Ethical and Legal Advisory
Boards (ELABs)?
vii. Is AI in the legal domain revolutionizing information management through
smart solutions for contract management with ancillary due diligence—a
kind of company review performed ahead of mergers and acquisitions—in
the private sector? What level of semantics will in the future be needed for
information search in the legal field? Is it a matter of course nowadays that
the search algorithms used in our legal databases will not burden the user
with search hits where no distinction is made between documents containing
the words “legislature,” “legislation,” or simply “leg” (in the context of for
instance a “leg iron”)?
viii. Is it realistic to imagine comprehensible AI for everyone, or would it be more
reasonable to strive for algorithmic representativeness with the help of digital
persons that can match the counterparts’ corresponding AI level? Is the EU
General Data Protection Regulation, (EU) 2016/679, provision on automated
individual decision-making, including profiling, on the right track, at least
during a transitional period (see Article 22)?
ix. Could fair trade algorithms in connection with assessment of creditworthi-
ness, taxation, social insurance, etc., supplement the protection of individu-
als’ freedoms and rights which is currently made manifest through traditional
legislation, administration of justice, and contract-based solutions between
parties?
x. How can education at our institutes of higher education be modernized and
modified beyond the current, often outdated, view of society as shaped by
physical infrastructures rather than digital ones?
3.2 Suggested Further Reading
The suggestions below are in no way a complete reading list, but rather a selection of
literature etc., on AI and law, seen over time. Themes common to the selected refer-
ence literature are that the legal order is a special domain for the application of AI and
the legal consequences in the form of accountability, etc. In the thesis Rättsautoma-
tion: Särskilt om statsförvaltningens datorisering, the author of this text presents
this problem area in Swedish legal science.3 Already at that early date, there was a
growing interest in AI-connected expert systems and so-called knowledge systems
within law. The prevailing conditions for succeeding with various applications were,
naturally, limited. However, this did not prevent approaches that took questions of
principle, relating to transformation of legal information into computer programs,
to their extremes. An early example of this comes from Richard Susskind with the
book Expert Systems in Law: A Jurisprudential Inquiry.4 A more contemporary
contribution is the thesis of Stanley Greenstein: Our Humanity Exposed: Predictive
Modelling in a Legal Context.5
A book that has come to be seen almost as a classic is Code and Other Laws of
Cyberspace by Lawrence Lessig.6 What is particularly interesting with this publica-
tion is how Lessig at an early stage, albeit with Anglo-Saxon starting points relatively
far from our continental European legal order, raised the question of the computer
program as a legal steering mechanism. Beyond traditional analysis of legal cases and
normative reasoning on legislation, he found structural changes regarding regulation
on the Internet. However, in practical application of law, it is common to mainly
associate Lessig’s work with the so-called Creative Commons copyright licenses.7
Societal development also means that the legislator is now approaching AI-related
matters, usually in terms of distribution of responsibility and accountability. One
example of this is Utredningen om självkörande fordon på väg and its final report.8
It is often said, usually with a negative tone, that the legislator cannot keep up and
the legislation is lagging behind. The question is if these statements are correct and
also if it not might be beneficial with a certain lag, to avoid having regulation that
grows obsolete quickly. However, this is not the same as promoting passivity or a
fear of anything related to modern IT, especially AI. An attempt to avoid ducking
too much for the view that legislation in every way possible should be technology-
neutral—and therefore at risk of being watered-down—can be found in the report of
Digitaliseringsrättsutredningen (the Inquiry on Law and Digital Transformation).9
The Inquiry, for which the author held the position as so-called Inquiry Chair, reasons
3 Magnusson Sjöberg (1992).

4 Susskind (1988).
5 Greenstein (2017).
6 Lessig (1999). A second version is available for downloading free of charge at: http://codev2.cc/
download+remix/Lessig-Codev2.pdf.
7 See https://creativecommons.org/tag/lawrence-lessig/.
8 The Swedish Ministry of Enterprise and Innovation (2018a).
9 The Swedish Ministry of Finance (2018).
about automation in public administration and suggests, among other things, that
information should be provided about how an authority in its handling of cases or
errands uses algorithms or computer programs which entirely or partially impact on
the result or standpoints of automated selections or decisions.
It is easy to get lost in the flood of literature on AI. Naturally, we as readers seek out
different kinds of references depending on our professional profiles, interests, current
needs, etc. One work that stands out thanks to its accessibility based on complexity is
Max Tegmark’s book Liv 3.0: att vara människa i den artificiella intelligensens tid.10
For a legal scholar, this read is not only educational, but also particularly interesting
with questions like “how can we update our legal systems to be more fair and efficient
and to keep pace with the rapidly changing digital landscape?”.11 In the book, other
links to the law are also presented, in particular with regard to copyright and other
intellectual property rights.
It is important to note that the law acts as both an inhibitor and a facilitator in
the development and usage of AI. In this context, it is worth mentioning digital
humanities, which is a growing research field at Stockholm University (http://dhv.
dsv.su.se). In brief, digital humanities refers to interdisciplinary studies of digital
artefacts and environments and their significance for mankind and society. This
includes, for instance (i) human science studies of parties and their roles in the
digital society; (ii) social and legal aspects, questions of liability and ethics, and;
(iii) interactions between humans in digital systems and between humans and digital
entities.
In a discussion on the role of law in the AI-based society, the focus increasingly
often falls on the legal profession. What do future lawyers and legal scholars at
our institutes of higher education, as well as law clerks at our courts, see as their
future professional roles? There is an intense news flow about jobs disappearing
and robots that can perform a due diligence both faster and better than humans,
not least in connection with international business transactions. The answer is not
given, beyond it being a particular task within our legal programs at the universities,
as well as within the training of judges, to prepare the lawyers/legal scholars for
the legal consequences of AI and the importance of having law play a proactive
role in these contexts, rather than being reactive, as it has traditionally been, when
problems have already arisen. This approach characterizes the education in Law and
Informatics at Stockholm University, where the subject is a mandatory course at the
advanced level (second cycle), with the option of taking in-depth specialized courses
and writing a thesis project. The course textbook is Rättsinformatik – Juridiken
i det digitala informationssamhället.12 See also Regeringens nationella inriktning
för artificiell intelligens,13 in which the importance of a strong connection between
research, education, and innovation in AI is emphasized, as well as the fact that
“standards in AI have the possibility of promoting technical, semantic, legal, and
10 Tegmark(2017).
11 Tegmark,pp. 123, 139 ff.
12 Magnusson Sjöberg (ed) (2018).
13 The Swedish Ministry of Enterprise and Innovation (2018b).
other interoperability both within and between companies and public institutions, as
well as contributing to increased clarity in relation to users and consumers.”14
In terms of horizon scanning, a few other references deserve mention. For instance,
the Swedish Foundation for Strategic Research has, since the spring of 2018, worked
with roughly ten authors to create the report Livet med AI.15 Among the contributions,
there is one with a specific focus on law. This is by Peter Wahlgren, who addresses
the matter of regulation of artificial intelligence.16 Vinnova’s report called Artificiell
intelligens i svenskt näringsliv och samhälle: Analys av utveckling och potential 17
also appears relevant in the context, even if it touches only briefly on the regulatory
development connected to AI.18 See further the work ongoing within the OECD
(http://www.oecd.org/going-digital/ai/oecd-initiatives).
4 Conclusion
Lastly, I personally hold high hopes for a continued dialogue, both within science
and in the everyday, not least in fora created by the initiative ADDAI.ORG. And
remember: While a legal scholar may seem dull, the law can be one of the ways to
create AI with mankind at its core. It is against this background that the discussion
on implementation of a digital person in the legal order should be held.
The concept as such is not just a term to add to those of the natural person and
the legal person. Rather, it is a legal figure which, under certain circumstances, could
be granted a legal capacity, with rights and responsibilities. Of particular interest is
the creation of space for discretion in the form of enforceability with an aim to gain
effectiveness of legal assessments in various digital environments. The legal entity,
i.e., the digital person, would as such be possible to describe as a constellation of
algorithms consisting of a basic algorithmic identity, which could be profiled and
specified with reference to various purposes. In order to facilitate the evaluation of
if a digital person with a certain algorithmic identity is working well, given tradi-
tional values like ethics, legal certainty, and trust, an assessment can be made in
terms of functional equivalency. (This has been and remains a well-known method
when traditional contracts, legislation, etc., are to be adapted for digital conditions.)
Eventually, the legislator can decide on regulations that give digital persons a legal
capacity, in both a limiting and a widening sense. Some forms of legislation might
also be self-generating. The role and future staffing of courts, as well as the contents
of the legal profession in general, appear increasingly uncertain, but it must be a
profession that cannot fear the analysis of algorithms.
14 The Swedish Ministry of Enterprise and Innovation (2018b), p. 10.

15 The Swedish Foundation for Strategic Research (2018).
16 The Swedish Foundation for Strategic Research (2018), pp. 52–57.
17 Vinnova (2018).
18 Vinnova (2018), pp. 79–81.
References
Greenstein S (2017) Our humanity exposed: predictive modelling in a legal context. Dissertation.
Stockholm University, Stockholm
Lessig L (1999) Code and other laws of cyberspace. Basic Books, New York
Magnusson Sjöberg C (1992) Rättsautomation: Särskilt om statsförvaltningens datorisering.
Dissertation. Norstedts Juridik, Stockholm
Magnusson Sjöberg C (ed) (2018) Rättsinformatik – Juridiken i det digitala informationssamhället.
Studentlitteratur, Lund
Susskind R (1988) Expert systems in law: a jurisprudential inquiry. Oxford University Press, Oxford
Tegmark M (2017) Liv 3.0: att vara människa i den artificiella intelligensens tid (English edition:
Life 3.0: being human in the age of artificial intelligence). Translated into Swedish: Sjöstrand
Svenn H, Svenn G. Volante, Stockholm
The Swedish Foundation for Strategic Research (2018) Livet med AI. SSF report no. 29
The Swedish Ministry of Enterprise and Innovation (2018a) Utredningen om självkörande fordon
på väg. SOU 2018:16
The Swedish Ministry of Enterprise and Innovation (2018b) Regeringens nationella inriktning för
artificiell intelligens. N2018.14
The Swedish Ministry of Finance (2018) Juridik som stöd för förvaltningens digitalisering. SOU
2018:25
Vinnova (2018) Artificiell intelligens i svenskt näringsliv och samhälle: Analys av utveckling och
potential. VR 2018:08
Voister (2017) Robot styr försörjningsstöd i Trelleborg. https://www.voister.se/artikel/2017/07/
robot-styr-forsorjningsstod-i-trelleborg/. Accessed 19 Sept 2018
Nudging Consent and the New Opt-Out
System to the Processing of Health Data
in England
Janos Meszaros, Chih-hsing Ho and Marcelo Corrales Compagnucci
Abstract This chapter examines the challenges of the revised opt-out system and
the secondary use of health data in England. The analysis of this data could be valu-
able for science and medical treatment as well as for the discovery of new drugs. For
this reason, the UK government established the “care.data program” in 2013. The
aim of the project was to build a central nationwide database for research and policy
planning. However, the processing of personal data was planned without proper pub-
lic engagement. Research has suggested that IT companies—such as in the Google
DeepMind deal case—had access to sensitive data and failed to comply with data
protection law. Since May 2018, the government has launched the “national data opt-
out” (ND opt-out) system with the hope of regaining public trust. Nevertheless, there
is no evidence of significant changes in the ND opt-out, compared to the previous
opt-out system. Neither in the use of secondary data, nor in the choices that patients
can make. The only notorious difference seems to be in the way that these options
are communicated and framed to the patients. Most importantly, according to the
new ND opt-out, the type-1 opt-out option—which is the only choice that truly stops
data from being shared outside direct care—will be removed in 2020. According to
the Behavioral Law and Economics literature (Nudge Theory), default rules—such
as the revised opt-out system in England—are very powerful, because people tend
to stick to the default choices made readily available to them. The crucial question
analyzed in this chapter is whether it is desirable for the UK government to stop
promoting the type-1 opt-outs, and whether this could be seen as a kind of “hard
paternalism.”
Keywords Nudge theory · Choice architectures · Opt-out system · Personal data ·

GDPR · ND opt-out · Hard paternalism
J. Meszaros (B) · C. Ho
Institute of European and American Studies, Academia Sinica, Taipei, Taiwan
e-mail: dr.janos.meszaros@gmail.com
M. Corrales Compagnucci
Center for Advanced Studies on Biomedical Innovation Law (CeBIL), Faculty of Law, University
of Copenhagen, Copenhagen, Denmark

https://doi.org/10.1007/978-981-15-1350-3_5
62 J. Meszaros et al.
1 Introduction
Governments are always actively seeking to enable efficient healthcare systems with
the aim of improving the quality of care while reducing public spending.1 The subject
of this chapter is the secondary use of health data in England, which is one way
of reaching these goals. The secondary use of health data refers to the processing
of data collected during direct care for new purposes, such as research and policy
planning.2 England has adopted a new opt-out system called “national data opt-out”
(ND opt-out) and it is available since May 2018.3
Default rules—such as the ND opt-out system in England—are very powerful. The
reason is that people tend to stick to the default option and choosing a different option
requires an active decision and further deliberation costs. In other words, people tend
to prefer the easiest option. In this case, the option which does not require mental
effort such as in the opt-out systems. However, the further use of health data poses
complex ethical,4 legal and technical challenges.5
Default rules can create a lot of good, but also do a lot of harm. This is one of
the key conceptual arguments of the Behavioral Law and Economics literature that
blend insights from cognitive psychology and economics.6 It takes into account the
psychological traits of human behavior and a variety of other factors such as emo-
tional, social and cognitive as the overarching framework to discuss legal issues.7
The subject of this chapter is about architectures, freedom of choice, and the legit-
imate ways of the UK government to nudge its citizens—as a new form of “hard
paternalism.”
The chapter is divided into 7 sections. After this introduction, Sect. 2 explains the
main tenets of Behavioral Law and Economics. It uses real life examples to illustrate
the pervasive nature of nudges—in particular, default rules—and choice architectures
which are everywhere, influencing inadvertently the decisions that people make.
Section 3, discusses nudging techniques in the healthcare sector, in particular opt-
out systems. Section 4, explains the National Health Service (NHS) and the opt-out
system in England. This section is divided into two main parts. The first part revisits
1 Deloitte (2016), p. 3.
2 Institute of Medicine (2013); Hanney and González-Block (2015), pp. 1–4.
3 See Digital NHS UK. Available at: https://digital.nhs.uk/national-data-opt-out. Accessed 10 June
2019.
4 Institute of Medicine (2013); Safran et al. (2007), pp. 1–9.
5 Safran et al. (2007), pp. 1–9.
6 Given the importance of this new field of law, the Government of the United Kingdom Cabinet
Office established the “Behavioral Insight Team” (BIT)—unofficially known as the “Nudge Unit.”
The BIT was originally a governmental organization set up to apply insights from behavioral
economics to improve public policy and services. Recently, the BIT became a limited company and
it is co-owned by the government. Since the BIT was spun off as a social purpose company, it has
given birth to a global movement that now spans 153 countries. See, The Behavioral Insight Team.
Available at: https://www.bi.team. Accessed 10 June 2019.
7 Angner and Loewenstein (2016) pp. 1–56; see generally, Zeiler and Teitelbaum (2015); Minton
and Kahle (2013).

Nudging Consent and the New Opt-Out System to the Processing … 63
the old opt-out system and explains the new types of opt-out rules set out in the
new ND-opt-out. The second part, focuses on the way this information is presented
and framed to the patients. Section 5, provides some statistics on opt-outs based
on recent data released by the NHS. Section 6, delves into details concerning the
theoretical discourse of the so-called “Libertarian Paternalism” in Cass Sunstein’s
narrative. According to Sunstein, there are different kinds of paternalisms. The main
distinction relevant to the discussion of this chapter is the difference between hard
versus soft paternalism. The first one coerces individual freedom. Hard paternalisms
are, therefore, not desirable. The latter provides freedom of choice. This is the kind of
paternalism advocated in the last section of this chapter (Sect. 7), which concludes
with the opinion that the type-1 opt-out should not be ruled-out from the revised
opt-out system in England.
2 Behavioral Law and Economics, Choice Architectures

and Default Rules as Prime Nudges
Behavioral Law and Economics became very popular and entered the mainstream of
modern law and economics thanks to the works of Richard Thaler and Cass Sunstein.
The first won the Nobel Prize in Economics with Nudge Theory in 2017. The main
postulate of this theory is that improved choices and information disclosure could
softly nudge (push or poke gently)8 individuals to improve decision-making and
welfare.9
Real-world illustrations of nudges can be found everywhere in our daily lives.
Urinals at Amsterdam Schiphol airport with images of a fly just above the drain are
one of such examples. According to Nudge Theory, the image of a fly would attract
the attention of men and prompt them with a target at which they would aim. This
experiment showed “spillage” on the bathroom floor was considerably reduced by
50–80%.10
Another typical example of a nudge in our daily life is a cafeteria. Think of the
manager of the cafeteria who has the freedom to arrange the food in certain places.
She could place the food in a place that is more visible to people in order to affect
their decisions. Putting the salad at the entrance and in a visible place, would increase
the likelihood that customers would choose the salad first as a healthier option.11 To
count as a nudge, “the intervention must be easy and cheap to avoid.”12 Placing the
salad at eye level is a nudge. Banning junk food however is a mandate.13
8 SeeEnglish Collins Dictionary (Nudge).

9 Corrales and Kousiouris (2017), p. 161.
10 Corrales and Jurčys (2016), p. 533.
12 Thaler and Sunstein (2009), p. 6.
13 Coggon et al. (2017), p. 177.
Empirical studies were carried out at two college campuses—the University of

Connecticut and Alfred University in New York—by students and managers inter-
ested in seeing how re-arranging cafeteria options would influence student behavior.
Their interest was not healthy eating as in the previous cafeteria example, but waste.
They realized that it was very easy to load trays with food that ended up as a waste.
Therefore, they ran a “trayless” experiment for a couple of days and noticed that
food and beverage waste dropped between 30 and 50%. This amounts to 2 tons of
food and about 424 L of liquid waste saved on a weekly basis.14
These examples take us directly to the definition of a “choice architect.” A choice
architect, is any person who changes “every small feature in the environment that
attracts our attention and influences the decision that we make.”15 For Thaler and
Sunstein, a nudge is “any aspect of the choice architecture that alters people’s behav-
ior in a predictable way without forbidding any options or significantly changing
their economic incentives.”16 In other words, Nudge Theory is mainly about design-
ing choices that influence and prompt individuals to improve the decisions that they
make.17
A canonical example of a nudge is the Global Positioning System (GPS) which
helps individuals find the best and shortest route. An interesting feature of the GPS
is that individuals can always take another route and the GPS can easily track the
location again and re-direct them with the route using signals from satellites. In this
respect, the GPS system does not coerce individuals to take one particular route. It
is up to the users to follow the directions of the GPS or not. This is one of the main
characteristics of a nudge. A nudge never overrides individual freedom.18
Nudging is nothing new. The private sector has been nudging consumers for
decades. Marketing agencies have always used different nudging techniques to attract
the attention of their customers and influence their behavior in order to sell their
products. Nudges can be very helpful for individuals and society. Some of these
nudges, however, may be regarded to be more controversial than others. For example,
road signs are undeniably helpful for the community. They can be hardly regarded
to be controversial. They give instructions to people and warn them to drive more
carefully. However, if road signs are put in the wrong place, they could be dangerous
and create accidents.19
Choice architects have the responsibility of organizing the context in which people
make decisions.20 Thus, choice architectures embrace the idea of nudges. The most
powerful nudges are: warning signals, information disclosures and default rules
14 See Nudging in the Cafeteria (2008). Available at: https://nudges.wordpress.com/2008/04/17/

nudging-in-the-cafeteria/. Accessed 10 June 2019.
15 Willis (2015).
16 Thaler and Sunstein (2009) p. 6.
17 Businessballs.com. Nudge theory. Available at: https://www.businessballs.com/improving-
workplace-performance/nudge-theory/. Accessed 10 June 2019.

18 Corrales and Kousiouris (2017), pp. 165–166.
19 See, e.g., generally, Jamson (2013), p. 298.
20 Thaler and Sunstein (2009), p. 3.
(emphasis added). In this chapter, we will focus on default rules as prime nudges.
According to the Behavioral Law and Economics literature, they are inevitable and
they are everywhere.21
A classic example to illustrate a nudge as a default rule is the hardware and soft-
ware of printer machines. Users can choose between single or double-sided printing.
Single-side printing would obviously require more paper and ink, while double-
sided printing would significantly reduce the costs expenditures of an organization.
Rutgers University ran an experiment in its New Brunswick campus. They simply
changed the default settings of all printers from one-sided to double-sided and saved
over eighty-nine million sheets of paper during the first years of the conservation
program. This amounted to a 44% reduction which is tantamount to 4650 trees. This
zero-cost option is a good example of how changing a small feature in the design of
the computer software and hardware architecture can make a big difference.22
3 Nudges in the Healthcare Sector & Opt-Out Systems
A cursory look at behavioral economic insights in the healthcare sector suggests that
these interventions tend to be small and often set as default rules. Subtle changes in
how choices are presented, for instance, as opt-in or opt-out can make big changes in
the behavior of participants. A very good example to illustrate this are default rules
related to the postmortem organ donation.23 Two main default system exist at the
global scale: (i) opt-in system: which requires explicit consent from the deceased,
and; (ii) opt-out system: whereby consent is automatically assumed.24 The latter
means that the deceased is a donor by default.25
The procedure for opt-in and opt-out differs greatly from country to country. In
the United States the deceased must have previously signed up in a state registry,26
whereas in other countries such as in Japan and most European Member States,
citizens have the option to check a box as an opt-in or opt-out rule when they have
to renew their driving license. Opt-out default systems make the percentage of organ
donation much higher than in opt-in systems.27 For example, countries such as Spain,
Austria, France, Hungary, Poland and Portugal, have all implemented opt-out systems
and the number of organ donation increased exponentially to 99%28 in comparison
21 Sunstein (2014), pp. 1–30, 179.

22 Stoknes (2015), p. 25; see also Sunstein (2016).
23 Ben-Porath (2010), p. 11.
24 Heshmat (2015), p. 243.
25 Corrales et al. (2019), p. 197.
26 Detels and Gulliford (2015), p. 782.
27 See John et al. (2013), p. 104; Quigley and Stokes (2015), p. 64; Thaler (2009); Hamilton and
Zufiaurre (2014), p. 18.

28 Leitzel (2015), p. 137.
to other countries such as Denmark (4.35%) and the Netherlands (27.5%)29 which
have opt-in systems.30
Decision-making using opt-in or opt-out rules could also be attributed to the
culture of a society, which may incrementally change its perception based on user
experience. Empirical studies conducted in countries such as Germany, United States
and Austria, revealed that signing up for an organ donation in an opt-in system was
generally considered to be a virtuous act of benevolence, whereas abstaining to donate
under an opt-out system was commonly viewed as egotistic and antisocial.31
In a recent blog by surgeon Ara Darzi—director of the Institute of Global Health
Innovation of the Imperial College London—he explains the potential of behavioral
economics in the field of public health and how this could be effectively applied in
cancer screening. Cancer survival is lower in the UK in comparison to other countries
and screening significantly reduces the morbidity and mortality. One approach to
raise cancer screening is to link the screening test to other tests that people are more
familiar with, such as dental check-ups or annual vehicle testing (MOT test in the
UK).32 This is just another good example of how nudging and default rules can
make a beneficial impact in the health care service. Below we explain the role of the
National Health Service and the recent developments of the opt-out system in the
England.
4 The National Health Service and Opt-Out System

in England
The National Health Service (NHS) provides universal and free public health ser-
vices in England. As a result, NHS data provide a valuable resource of routinely
collected primary (e.g., visiting general practitioner (GP) practices) and secondary
(e.g., hospital admissions, outpatient appointments, accident and emergency atten-
dances) healthcare data covering almost the whole population of England.33 Three
broad categories of data are collected from patients during direct care in England:
(i) basic personal data such as age and gender; (ii) medical information such as
diagnosis; and, (iii) administrative information (e.g., waiting times).34
This information is protected by the common law duty of confidentiality (CLDC).
Only in special cases can this information be processed for a new purpose, without
consent. These situations might be a court order or overriding public interest (e.g.,
29 Shafir (ed) (2013), p. 496.

30 Corrales et al. (2019), p. 197.
31 Zamir (2015), p. 103; see, also, generally, Davidai et al. (2012), pp. 15201–15205.
32 Darzi (2017).
33 Piel et al. (2018), pp. 594–600.
34 NHS factsheets for health and care staff, Factsheet 1B—Types of data used and legal protection
in place (2018), p. 1. Available at: https://digital.nhs.uk/services/national-data-opt-out-programme/

guidance-for-health-and-care-staff. Accessed 10 June 2019.
epidemics). Furthermore, Section 251 of the 2006 National Health Service Act allows
the Secretary of State for Health to make regulations that bypass the CLDC for defined
medical purposes. The Health Service (Control of Patient Information) Regulations
2002 play this role, as they allow the disclosure of confidential patient information
without consent.35 The main reason for these disclosures is that seeking consent
would require disproportionate effort or it would be impossible, and processing
anonymous data might be not useful for research and planning in many cases.36
Against this legal background, the care.data program was initiated in 2013, aiming
at extracting patient information from direct care providers without patient consent,
to build a central nationwide database for research and policy planning.37 However,
the processing has been planned without proper public engagement, and the central
database was aimed to be used by public and private third-party users, such as IT
companies. The care.data program has raised serious public concerns38 and it has
been paused several times.39 During this period, the National Data Guardian40 (NDG)
started to investigate the care.data program and developed models for consent and opt-
out.41 The NDG published the “Caldicott Review” in 2016, a report that highlighted
the issues about the secondary use of health data in the UK, and provided several
recommendations for processing, consent and opt-out.42 After the publication of the
report, the NHS in England canceled the care.data program.43 Despite the failure,
the government remained committed to realizing the benefits of the secondary use of
health data without consent; thus, the government implemented a new national data
opt-out (ND opt-out) system in 2018.
4.1 The Opt-Out System in England
The state has the power to bypass patient consent in some circumstances, such as
improving healthcare and medical research for the public interest. On the other hand,
the citizens’ autonomy might be maintained by giving them the choice to opt-out.
Thus, the purpose of the opt-out system is to use confidential data without the burden
35 Mészáros and Ho (2019), p. 13.

36 Lee et al. (2012), pp. 38–44.
37 For more information behind the NHS plans for the secondary use of health data, before the
creation of care.data program, see Department of Health (2006); see also Department of Health
(2015).
38 Sterckx and Cockbain (2014), pp. 227–228, Mori (2016).
39 Vezyridis and Timmons (2017), p. 2.
40 The National Data Guardian (NDG) advises and challenges the health and care system in the UK to
help ensure that citizens’ confidential information is safeguarded securely and used properly. Avail-
able at: https://www.gov.uk/government/organisations/national-data-guardian/about. Accessed 10
June 2019.
41 National Data Guardian for Health and Care (2016), pp. 1–56.
42 National Data Guardian for Health and Care (2016), pp. 6–9.
43 Department of Health and Social Care (2016).
of seeking explicit consent, while respecting peoples’ autonomy. There might be

cases, however, when opt-outs may seriously hinder data integrity, thus they cannot
be applied.44 The number of opt-outs can be decreased with nudging techniques
against this option by, for example, highlighting and framing the importance of the
secondary use of health data. This is where Nudge Theory comes to the fore. The UK
Government dedicated efforts to convince the patients about the value of research and
policy planning. Nevertheless, the public dissatisfaction with the Google DeepMind
patient data deal45 and several other studies46 indicated that the UK citizens might
have concerns about whether their sensitive data is processed for a new purpose
outside of direct care.
In the Google DeepMind patient data deal, for instance, Google’s artificial intel-
ligence firm was allowed to access health data from over 1.6 million patients to
develop an app monitoring kidney disease called “Streams.”47 Public concerns and
corroborative research studies suggested that the Google DeepMind deal had access
to other kinds of sensitive data and failed to comply with data protection law.48
The NDG report and the failure of the care.data program lead to a change in the
opt-out system in England to earn back the public trust. The “old” opt-out system
consisted of two choices for the citizens: type-1 and type-2 opt-outs. The “new”
system is the ND opt-out, and it is publicly available online since May 2018. However,
as we will point out in the conclusion of this chapter, there is neither a significant
change in the secondary use of data, nor in the choices that patients can make. The
only notorious difference is how these options are communicated to the patients.
4.1.1 The “Old” Opt-Out System
In England, there have been two options for citizens to prevent the secondary use of
their confidential information. The type 1 opt-out prevented the information being
shared outside the GPs practice for purposes other than direct care, while the type 2
opt-out prevented information being shared outside the NHS Digital49 for purposes
beyond the individual’s direct care. Since May 2018, the type 2 opt-out has been
44 Jones et al. (2017), pp. 43–50; Rothstein and Shoben (2013), p. 27.
45 The ICO has ruled the Royal Free NHS Foundation Trust failed to comply with the Data
Protection Act when it provided patient details to Google DeepMind. Available at: https://ico.
org.uk/about-the-ico/news-and-events/news-and-blogs/2017/07/royal-free-google-deepmind-trial-
failed-to-comply-with-data-protection-law/. Accessed 10 June 2019.
46 Stockdale et al. (2018), pp. 1–25; Wyatt et al. (2018), pp. 1–8; Aitken et al. (2016), p. 73.
47 See Streams. Available at: https://deepmind.com/applied/deepmind-health/working-partners/
how-were-helping-today/. Accessed 10 June 2019.

48 McGoogan (2017).
49 NHS (National Health Service) Digital is an executive non-departmental public body of the
Department of Health in the UK. The NHS Digital is the national provider of information, data
and IT systems for commissioners, analysts and clinicians in health and social care. Available at:
https://www.gov.uk/government/organisations/nhs-digital/about. Accessed 10 June 2019.
Fig. 1 The “old” opt-out System in England (NHS digital: implement type 2 patient opt-outs
(2016), p. 2)
replaced by the ND opt-out, and the type 1 is not communicated as an option for the
citizens anymore. The type 1 decisions will be removed in 2020.
As Fig. 1 shows, the type 1 opt-out is the only option, which truly stops data from
being shared outside direct care. The type 2 option had several unclear limitations.
It stopped data from being shared outside of NHS Digital for research and planning.
The citizens might have expected from the opt-out information that their confidential
information can be circulated and used only for care and planning inside the NHS.
However, pseudonymized data was an exemption, since it could be further processed
for a secondary purpose regardless of the patients’ choice. Probably the reason for
NHS to push citizens toward the type 2 opt-outs was the need for data for planning
and facilitating research. The goal was to reduce the costs of care and improve the
effectiveness of the healthcare system, thus nudging the patients for using the type
2 opt-out had a public interest.
4.1.2 The New National Data Opt-Out System
From May 2018, the type 2 opt-out has been replaced by the ND opt-out. The
previously recorded type 2 opt-outs have been automatically converted to ND opt-
outs. Existing type 1 opt-outs will be respected until 2020, when the Department of
Health will remove them. What is more important, the government stopped promoting
type-1 opt-outs for citizens. There are leaflets, posters and an information webpage
about the new ND opt-out, and neither of them conveys the fact that the patients
are still able to choose the type 1 option at the GPs, which would truly stop the
processing of their confidential information outside of direct care. The ND opt-out
is communicated for patients in several ways such as: by healthcare staff, leaflets,
posters, online, and they could also get information via telephone. The type 1 opt-out
is not publicized to citizens on these publicly available materials. Furthermore, on

the information website, which is on these materials, the type 1 opt-out is not even
mentioned. Information about the type 1 opt-out50 can be only found on the NHS
website, under the “resources for health and care staff,” by clicking on the “more
information for patients with a previous type 2 opt-out” menu.51 This communication
indicates that the government intends to slowly roll out the type 1 option, before the
final removal in 2020.
The new ND opt-out will only apply when identifiable data is shared for research
and managing the efficient and safe operation of the healthcare system,52 thus in the
case of anonymized and pseudonymized data, this choice does not apply. Another
issue is that the use of the terms “anonymized,” “pseudonymized” and “de-identified”
is not consistent in the UK legal terminology. These terms are used interchangeably
in various codes, white papers and leaflets. Therefore, the citizens might be confused
about their meaning, and they might assume that anonymization means their data
is safe. However, in the UK, the term “anonymization” has the same meaning as
“de-identification” in the General Data Protection Regulation (GDPR), thus it can
also involve “pseudonymization.” In other words, “anonymized” data might be just
“pseudonymized” data in many cases. Pseudonymization is the separation of data
from the direct identifiers (e.g., name, address, NHS number), so that re-identification
is not possible without additional information (the “key”) which is held separately.
Thus, it is still possible to re-identify the data subjects after this security measure.53
In this chapter, we use the terms “anonymization”54 and “pseudonymization”55 in
a manner consistent with the GDPR and the EU Data Protection Board. According
50 NHS Digital: Opting out of sharing your confidential patient information. Available at: https://
digital.nhs.uk/about-nhs-digital/our-work/keeping-patient-data-safe/how-we-look-after-your-
health-and-care-information/your-information-choices/opting-out-of-sharing-your-confidential-
patient-information. Accessed 15 June 2019.
“Type 1 opt-out: medical records held at your GP practice: You can also tell your GP practice
if you do not want your confidential patient information held in your GP medical record to be used
for purposes other than your individual care. This is commonly called a type 1 opt-out. This opt-out
request can only be recorded by your GP practice.”
51 NHS Digital, Opting out of sharing your confidential patient information. Available at:
https://digital.nhs.uk/about-nhs-digital/our-work/keeping-patient-data-safe/how-we-look-
after-your-health-and-care-information/your-information-choices/opting-out-of-sharing-your-
confidential-patient-information. Accessed 10 June 2019.
52 NHS Digital, About the national data opt-out. Available at: https://digital.nhs.uk/services/
national-data-opt-out-programme. Accessed 10 June 2019.

53 Article 29 Working Party, Opinion 05/2014 on Anonymisation Techniques (2014), p. 3.
54 Regulation (EU) 2016/679 of the European Parliament of 27 April 2016 on the protection of
natural persons with regard to the processing of personal data and on the free movement of such
data, and repealing Directive 95/46/EC (General Data Protection Regulation) [hereinafter “GDPR”].
Recital 26: anonymous information, namely information which does not relate to an identified or
identifiable natural person or to personal data rendered anonymous in such a manner that the data
subject is not or no longer identifiable.
55 Article 4 (5) of the GDPR: “pseudonymisation” means the processing of personal data in such
a manner that the personal data can no longer be attributed to a specific data subject without the
use of additional information, provided that such additional information is kept separately and is
to these sources, anonymized data can no longer lead to the identification of the
data subject, while pseudonymization is a useful security measure which reduces the
“linkability” of a dataset with the original data subject.56 The umbrella term for both
of these measures is “de-identification.” Another possible ground of misunderstand-
ing for the citizens is the notion of “direct care.” They might assume this activity
consists of their care by the professionals in a healthcare institution. However, direct
care is a much broader concept,57 since it may also include the assurance of safe and
high-quality care, which requires a background work that might not be apparent for
the patients.
4.2 The Presentation of the Information on the Information

Materials for Patients
4.2.1 General Information
The UK government needs the citizens’ health data for research and policy plan-
ning; thus it is crucial how the information about opt-out is presented to them. As
explained above, the ND opt-out is communicated to patients in several ways: by
healthcare staff, leaflets, posters, videos, online, and they may also obtain informa-
tion via telephone. These materials contain mostly similar information. However,
there are special versions of them for young people and minorities. In this section,
we introduce how these materials communicate the secondary use of health data for
citizens to help them to make their decision. Since the default setting is the sharing
of health data, these materials do not have to convince citizens to give their consent,
just to understand the importance of their data and accept the situation.
On all the information materials, the first thing which is presented to patients is
the value of their health information: “Information about your health and care helps
us to improve your individual care, speed up diagnosis, plan your local services and
research new treatments.”58 By starting with this information, the patients might
subject to technical and organizational measures to ensure that the personal data are not attributed
to an identified or identifiable natural person.
56 Article 29 Working Party, “Opinion 05/2014 on Anonymization Techniques” (WP216, 10 April
2014), p. 20.
57 National Data Guardian (2013), p. 128. “A clinical, social or public health activity concerned with
the prevention, investigation and treatment of illness and the alleviation of suffering of individuals.
It includes supporting individuals’ ability to function and improve their participation in life and
society. It includes the assurance of safe and high-quality care and treatment through local audit,
the management of untoward or adverse incidents, person satisfaction including measurement of
outcomes undertaken by one or more registered and regulated health or social care professionals
and their team with whom the individual has a legitimate relationship for their care.”
58 NHS (2018) Your Data Matters to the NHS, p. 1. Available at: https://digital.nhs.uk/services/
national-data-opt-out-programme/supporting-patients-information-and-resources. Accessed 25
May 2019.
realize the importance and public interest behind the secondary use of their data,
thus there might be less of a chance they choose to opt-out. There is also a possibility
that after reading this information, they might not continue to read over the whole of
the information material.
The NHS materials about the secondary use of confidential patient data continue
with the following statement: “In May 2018, the strict rules about how this data
can and cannot be used were strengthened.” This statement is debatable since the
exemptions of the opt-out system did not change: the de-identified data can be still
used for a new purpose even in the case of opt-out, and the type-1 opt-out is planned
to be canceled. The NHS also promises on the leaflet that: “The NHS is committed
to keeping patient information safe and always being clear about how it is used.”
However, the Google Deepmind case highlighted that the NHS might also share
confidential patient data with private companies for “direct care” which turned out not
to be a proper legal basis59 after the investigation by the Information Commissioner.60
The information materials state that “You can choose whether your confidential
patient information is used for research and planning.” This statement is true, how-
ever, only for the type 1 opt-out, which is about to be cancelled, and not communicated
to patients. By this presentation, the patients might be biased and assume the ND
opt-out can provide the full protection of their data. The leaflet explains the meaning
of confidential patient information as follows: “information identifies you and says
something about your health, care or treatment. Information that only identifies you,
like your name and address, is not considered confidential patient information and
may still be used: for example, to contact you if your GP practice is merging with
another.” 61
As it was highlighted in the previous section, the misunderstandings around de-
identified information may lead to misinterpretation. The leaflet clearly explains how
information can be confidential, and the processing of it is necessary for administra-
tive purposes and direct care, even in the case of opt-out. In the next part, the leaflet
explains who can use this confidential patient information: “NHS, local authorities,
universities and hospital researchers, medical colleges and pharmaceutical compa-
nies researching new treatments.”62 What this part of the information leaflet does
not specify, are other not medical-related companies, such as Google Deepmind
and other IT corporations that might also get access to health data. As healthcare is
becoming digitized, such as X-ray diagnostics using AI and machine learning, IT
59 Powles and Hodson (2017), pp. 351–367.

60 The ICO has ruled the Royal Free NHS Foundation Trust failed to comply with the Data
Protection Act when it provided patient details to Google DeepMind. Available at: https://ico.
org.uk/about-the-ico/news-and-events/news-and-blogs/2017/07/royal-free-google-deepmind-trial-
failed-to-comply-with-data-protection-law/. Accessed 10 June 2019.
May 2019.
May 2019.
companies are gaining a crucial role in providing the backbone of direct care and
medical research. For many years, medical doctors played the most important role
in improving healthcare. However, very recently programmers and analytics have
contributed significantly to improve the quality of care.63
4.2.2 Information About the Exemptions
After introducing the potential users of health data, the information leaflet turns to
situations, when the opt-outs might be ignored. It only introduces one situation with
the highest public interest, the epidemics: “You can choose to opt out of sharing your
confidential patient information for research and planning. There may still be times
when your confidential patient information is used: for example, during an epidemic
where there might be a risk to you or to other people’s health.” However, there are
many other situations when the opt-outs might be ignored, such as court orders and
with regard to the use of de-identified data. The information about other exemptions
can be only found on the NHS opt-out website.64
The information materials clarify that confidential patient information will be used
for direct care regardless of the opt-outs: “Will choosing this opt-out affect your care
and treatment? No, your confidential patient information will still be used for your
individual care. Choosing to opt-out will not affect your care and treatment. You will
still be invited for screening services, such as screenings for bowel cancer.”65 On the
one hand, this information empowers citizens to opt-out if they wish, since it does
not affect their individual care. On the other hand, this information may also inform
patients that their information will be used regardless of their decision.
The following information might be the main nudging strategy against opt-out:
“What should you do next? You do not need to do anything if you are happy about how
your confidential patient information is used. If you do not want your confidential
patient information to be used for research and planning, you can choose to opt-out
securely online or through a telephone service.”66 Most citizens do not have deep
knowledge on how their information is processed. The time and effort67 to read
through privacy policies are burdensome68 and in many cases time wasting, since
there might be no alternative choices (e.g., MS Windows, national health services).
Thus, people have their subjective perception of the quality of the health service
63 Meskó et al. (2018), pp. 1–4.

64 NHS, When your choice does not apply. Available at: https://www.nhs.uk/your-nhs-data-matters/
where-your-choice-does-not-apply/. Accessed 10 June 2019.

June 2019.
June 2019.
67 McDonald and Cranor (2009), pp. 543–568.
68 Solove (2013), pp. 1880–1903.
they directly receive, but less understanding concerning the information processing
and administrative work behind it. It is possible that the patient is satisfied with the
service, but her data is misused by private companies; and on the contrary, it is also
possible that the patient had a bad experience during treatment, and chose not to
share data, which would have been wisely used by private corporations to improve
her future care.
4.2.3 Special Opt-Out Materials
The NHS is providing special information materials for minorities and young people.
The information leaflets are slightly different from the general materials. The most
apparent difference is the design of the information documents. While the general
leaflet has a blue background with white letters, the others which target minorities
and young people look more like traditional materials, with black and blue letters
on white background. The materials do not have a significant difference in length:
they are approximately 450–500 words, stretched across two pages. The leaflets for
minorities and young people have cover pages, making the material more comfortable
to hold and read. However, the wording of the materials is slightly different, since
the leaflet for young people has the shortest general word length, which makes it
easier to understand (Table 1).
There is special information material for care givers (e.g., doctors, nurses),69
which advises them to promote the NHS website providing information for patients
on the value of health data. The NHS website addresses the concerns about privacy
and the selling of data to private companies: “While people may feel uncomfortable
with the idea of the NHS ‘selling’ data, there would also be concerns if valuable data is
given away to companies for free.”70 The other part of the leaflet with instructions to
Table 1 Comparison of NHS information materials on the ND opt-out system

General Minority information leaflet Young people
Number of pages 2 4 4
Number of words 449 504 505
Number of characters (without 2317 2548 2275
space)
Number of characters (with 2694 3033 2755
spaces)
Number of spaces 377 485 480
General word length 5.16 5.05 4.50
69 NHS (2018) A guide to the national data opt-out for carers. Available at: https://digital.nhs.
uk/services/national-data-opt-out-programme/supporting-patients-information-and-resources.
Accessed 11 May 2019.
70 Understanding Patient Data, Frequently Asked Questions. Available at: https://
understandingpatientdata.org.uk/what-you-need-know. Accessed 23 June 2019.

the healthcare staff is neutral, asking them to disclose both the risks and advantages
for patients and to support them in their choice.71 Overall, the special materials
contain similar information, but they communicate the same message in a slightly
different way.
5 Opt-Out Statistics
The NHS regularly releases statistics on opt-outs.72 From May 2018, the type 2
opt-out has been replaced by the ND opt-out, and the existing type 2 opt-outs have
been automatically converted to ND opt-outs. In March 2018, the number of type 2
opt-outs was 1.4 million representing 2.42% of the population. At the time of writing
this chapter, the latest publication of the opt-out statistics73 was in March 2019, and
the total number of ND opt-outs was 1.6 million, which represented 2.74% of the
population of England. This number is relatively low, especially if the well-advertised
ND opt-out system is compared with the previous system.
As explained before, the type-1 opt-out is no longer advertised to the public.
Furthermore, there are no statistics released on the type 1 opt-outs in 2019, and the
existing type 1 opt-outs are planned to be removed in 2020. In March 2018, the
number of type 1 opt-outs was 1.85 million, which was 3.13% of the population in
England (Fig. 2).
The type 1 and type 2 opts-outs were presented at the GP level before May
2018.74 The number of type 2 opt-outs is accurate since every opt-out choice has
been registered and reported from the GPs to NHS Digital with identifiable patient
information; thus, there are no duplications. However, the number of type 1 opt-outs
is not accurate, since one person may have been recorded several times at different
GPs, and the GPs could only report the number of opt-outs to NHS, without personal
data.75 After informing the public about the ND opt-outs in many different ways
(e.g., leaflets, posters, videos), the number of opt-outs has not raised significantly,
since only 0.3% of the population decided not to share their personal data in the
71 NHS (2018) A guide to the national data opt-out for carers, p. 1. “If you think the person you care
for is happy for their information to be shared you don’t need to do anything further.” “If you think
that the person you care for would wish to register a national data opt-out, or you are satisfied that
registering a national data opt-out is in that person’s best interest then follow step 3.”
72 NHS Digital, statistics on the volumes of national data opt-outs. Available at: https://digital.nhs.
uk/data-and-information/publications/statistical/national-data-opt-out. Accessed 12 June 2019.

73 NHS Digital, statistics on the volumes of national data opt-outs, March 2019. Available at:
https://digital.nhs.uk/data-and-information/publications/statistical/national-data-opt-out/
march-2019/ndop-mar19. Accessed 12 June 2019.
74 Type 1 opt-outs have been reported as instances (i.e., number of times the opt-out code occurs
within GP records, which may include the same patient recorded at more than one practice), therefore
the NHS Digital could not de-duplicate this information.
75 NHS Digital, statistics on the volumes of national data opt-outs, March 2018. Available at:
https://digital.nhs.uk/data-and-information/publications/statistical/care-information-choices/
mi-care-information-choices-england-march-2018. Accessed 16 May 2019.
Fig. 2 Opt-out rates in England
Table 2 The opt-out rates in England in 2018 and 2019

Number of opt-outs Rate (compared to the population of
England)
Type 1 opt-outs (March 2018) 1,846,250 3.13%
Type 2 opt-outs (March 2018) 1,422,250 2.42%
May 2018 ND opt-out is online, and the existing type-2 opt-outs are
converted to ND opt-outs
ND opt-outs (July 2018) 1,602,910 2.71%
ND opt-outs (March 2019) 1,639,012 2.74%
new system. The slight increase might indicate that the NHS’s campaign on the
importance of sharing health data was successful, and their message reached the
patients (Table 2).
6 Hard Paternalism in Healthcare?
The gist of the matter is whether the UK government may legitimately nudge its
citizens and whether this could be seen as a kind of “hard paternalism.” In Why
Nudge? The Politics of Libertarian Paternalism, Cass Sunstein focuses on finding a
justification for various nudging techniques. On the normative level, Sunstein tries
to challenge John Stuart Mill’s “harm principle,”76 which suggests that individuals
can do whatever they want as long as their actions do not harm others. If they do, the
76 Mill (1859).
government intervention can be justified to constrains the person and prevent such
harm.77
The harm principle is justified on the grounds that individuals know better what is
good for them and that governments do not have enough information and resources
to know all the needs of its citizens. Sunstein refers to this argument as the “epistemic
argument” and indicates that this way of reasoning is sometimes wrong. Sunstein goes
on and argues that in some cases, paternalistic interventions are desirable, specially
where people are likely to make a mistake and it is necessary to provide the means
for improved decision-making.78
Paternalism comes in many forms. Sunstein explains that there are mainly four
types of paternalism grouped into two main categories. One category differentiates
“hard” versus “soft” paternalism. One of the criteria for distinguishing these types of
paternalism relates to the imposition of material costs on individuals. “Hard pater-
nalism” occurs when people’s free choice is coerced by the government, whereas
“soft paternalism” is when the person is free to choose the form of action. The latter
denotes very little or no intervention from the state79 such as in the GPS example
explained above in Sect. 2.
In traditional “hard paternalism,” the so-called “nanny state” uses its coercive
power to nudge its citizens to do what is in their best interests.80 Nonetheless, “soft
paternalism” holds the view that government intervention is legitimate and justified
only when the person is consciously aware and acts voluntarily.81 Mill’s famous
example of the person who is about to cross a damaged bridge (so-called “Bridge
Exception”), illustrates this case in point clearly.82 Consider the case where the
government could not communicate the risks of a bridge that is about to collapse
because of language limitations (i.e., the intervened person does not speak the local
language and thus she can neither read the signs nor understand any warning signals
given). In this scenario, the government’s use of force to stop the person from crossing
the bridge would be justified as her liberty consists in doing what she wants, and
falling and dying is most probably not her will in this case.83
Sunstein provides some compelling arguments against those who oppose govern-
ment intervention, nudging, and autonomy. He criticizes “welfarist” objections for
failing to take into account the fact that most public policy decisions are already made
and individuals have only a limited ability to control the exercise of those underly-
ing choices. Moreover, “welfarists” fail to acknowledge the empirical findings on
behavioral economics as well as the fact that choice architectures and nudges are

80 Bishop (2009), p. 296.
81 Tanner (2007), p. 200; Hartley (2012), p. 70; Angner (2016), p. 264.
82 See, e.g., generally, Jackson (2006), pp. 68–69.
83 Sunstein (2014), pp. 63–99.
everywhere. Sunstein advocates for a choice architecture, which he labels libertar-

ian paternalism: individuals are encouraged to make active choices, which helps to
solve the shortcomings of a much criticized, one-size-fits-all approach.84
7 Conclusion
Behavioral Law and Economics has become increasingly relevant as a point of ref-
erence in policy-making and regulation over the past decade. In this regard, Nudge
Theory lends itself as powerful tool that can help us to improve the normative frame-
work. At first glance, there seems to be no major differences in the new ND-opt-out
system. However, a closer look reveals some subtle, but very important changes.
The bone of contention is that choices provided in the new ND-opt-out are going to
be more limited in 2020, as the type-1 opt-out will disappear. Moreover, the UK’s
government attempt to nudge patients by framing (or even hiding) information seems
to fall under the “hard paternalism” category in the Sunstein narrative, which is not
the most desirable one.
The linchpin and value of nudges is to bring to the fore the prospect of “choice.”
Therefore, the best way for the UK government to act paternalistic, but at the same
time respecting citizen freedom is with a “soft paternalism” approach. This approach
does not seem to run counter to the use of active choice mechanisms, nudges or
default rules. On the contrary, the paternalist actions may lead to welfare gains that
are greater than the welfare loss. Patients may feel frustrated to discover that certain
decisions have already been made for them. Therefore, it is important to leave an
option to revisit those decisions if people do not like them. This is the reason why we
advocate to keep the type-1 opt-out system as a default rule. In this way, the opt-out
choice architecture will enable patients with more options. This, in turn, will offer a
stronger data protection scheme without overriding individual freedom.85
Acknowledgements This research is supported by a Novo Nordisk Foundation grant for a sci-
entifically independent Collaborative Research Program in Biomedical Innovation Law (grant
agreement number NNF17SA0027784) and the Multidisciplinary Health Cloud Research Program:
Technology Development and the Application of Big Health Data. Academia Sinica, Taipei, Taiwan.
References
Aitken et al (2016) Public responses to the sharing and linkage of health data for research purposes:
a systematic review and thematic synthesis of qualitative studies. BMC Med Ethics 17(1):73
Angner E (2016) A course in behavioral economics, 2nd edn. MacMillan Education Palgrave,
London
84 Corrales and Jurčys (2016), pp. 534–535.

85 Corrales and Jurčys (2016), pp. 534–535.
Angner E, Loewenstein G (2016) Behavioral economics. Elsevier’s handbook of the philosophy

of science, vol 5. http://www.cmu.edu/dietrich/sds/docs/loewenstein/BehavioralEconomics.pdf.
Ben-Porath S (2010) Tough choices: structured paternalism and the landscape of choice. Princeton
University Press, Princeton
Bishop M (2009) Essential economics: an A to Z guide. The Economist, 2nd edn. Bloomberg Press,
New York
Coggon J, Syrett K, Vienns AM (2017) Public health law: ethics, governance and regulation.
Routledge, London
Corrales M, Jurčys P (2016) Sunstein, Cass, Why nudge: the politics of libertarian paternalism.
New Haven/London: Yale University Press, 2014, 208 pp, pb £10.99. (Book Review). Mod Law
Rev 79(3):533–536
Corrales M, Kousiouris G (2017) Nudging cloud providers: Improving cloud architectures through
intermediary services. In: Corrales M, Fenwick M, Forgó N (eds) New technology, big data and
the law. Springer, Singapore
Corrales M, Jurčys P, Kousiouris G (2019) Smart contracts and smart disclosure: coding a GDPR
compliance framework. In: Corrales M, Fenwick M, Haapio H (eds) Legal tech, smart contracts
and Blockchain. Springer, Singapore
Darzi A (2017) There is huge potential to apply behavioral economics in health. https://blogs.
bmj.com/bmj/2017/10/16/ara-darzi-there-is-huge-potential-to-apply-behavioural-economics-in-
health/. Accessed 10 Jun 2019
Davidai S, Gilovich T, Ross L (2012) The meaning of default options for potential organ donors.
Proc Natl Acad Sci 109(38):15201–15205
Deloitte (2016) International review, Secondary use of health and social care data and applicable leg-
islation. https://media.sitra.fi/julkaisut/Muut/International_review_secondary_use_health_data.
pdf. Accessed 10 Jun 2019
Department of Health (2006) Best research for best health: a new national health strategy, London
Department of Health and Social Care (2016) Written statement to Parliament: review of health and
care data security and consent. https://www.gov.uk/government/speeches/review-of-health-and-
care-data-security-and-consent. Accessed 10 Jun 2019
Detels R, Gulliford M (2015) Oxford textbook of global public health, vol 1, 6th edn. Oxford
University Press, Oxford
Hamilton D, Zufiaurre B (2014) Blackboards and bootstraps: revisioning education and schooling.
Sense Publishers, Rotterdam
Hanney S, González-Block M (2015) Health research improves healthcare: now we have the evi-
dence and the chance to help the WHO spread such benefits globally. Health Res Policy Syst
13(12):1–4
Hartley D (2012) Education and the culture of consumption: personalisation and the social order.
Routledge, London
Heshmat S (2015) Addiction: a behavioral economic perspective. Routledge, New York
Institute of Medicine (2013) Best care at lower cost: the path to continuously learning health care
in America. The Institute of Medicine, National Academy of Sciences, Washington DC
Jackson J (2006) Ethics in medicine: virtue, vice and medicine. Polity Press, Cambridge
Jamson S (2013) The Role of Acceptance in Behavioral Adaptation. In: Rudin-Brown C, Jamson S
(eds) Behavioral adaptation and road safety: theory, evidence and action. CRC Press, Boca Ratón
John P et al (2013) Nudge, nudge, think, think: experimenting with ways to change Civic behavior.
Bloomsbury, London
Jones K et al (2017) The other side of the coin: harm due to the non-use of health-related data. Int
J Med Inform 97(2017):43–50
Lee LM, Heilig CM, White A (2012) Ethical justification for conducting public health surveillance
without patient consent. Am J Public Health 102(1):38–44
Leitzel J (2015) Concepts in law and economics: a guide for the curious. Oxford University Press,
Oxford
McDonald A, Cranor L (2009) The cost of reading privacy policies. I/S: J Law Policy Inf Soc
4(3):543–568
McGoogan C (2017) NHS illegally handed Google firm 1.6 m patient records, UK data watch-
dog finds. https://www.telegraph.co.uk/technology/2017/07/03/googles-deepmind-nhs-misused-
patient-data-trial-watchdog-says/. Accessed 10 Jun 2019
Meskó B, Hetényi G, Győrffy Z (2018) Will artificial intelligence solve the human resource crisis
in healthcare? BMC Health Serv Res 18(545):1–4
Mészáros J, Ho C (2019) Big data and scientific research: the secondary use of personal data under
the research exemption in the GDPR. Acta Juridica Hungarica (in press)
Mill J (1859) On liberty. John W. Parker and Son, London
Minton E, Kahle L (2013) Belief systems, religion, and behavioral economics: marketing in
multicultural environments. Business Expert Press, New York
Mori I (2016) The one-way mirror: public attitudes to commercial access to health data. https://
www.ipsos.com/sites/default/files/publication/5200-03/sri-wellcome-trust-commercial-access-
to-health-data.pdf. Accessed 10 Jun 2019
National Data Guardian for Health and Care (2013) Information: to share or not to share? The
Information Governance Review
National Data Guardian for Health and Care (2016) Review of data security, consent and opt-outs
Piel F et al (2018) The challenge of opt-outs from NHS data: a small-area perspective. J Public
Health 40(4):594–600
Powles J, Hodson H (2017) Google DeepMind and healthcare in an age of algorithms. Health
Technol 7(4):351–367
Quigley M, Stokes E (2015) Nudging and Evidence-Based Policy in Europe: Problems of Normative
Legitimacy and Effectiveness. In: Alemanno A, Sibony A (eds) Nudge and the law: a European
perspective, modern studies in European Law. Hart Publishing, Oxford
Rothstein MA, Shoben AB (2013) Does consent bias research? Am J Bioeth 13(4):27
Safran C et al (2007) Toward a national framework for the secondary use of health data: an American
Medical Informatics Association white paper. J Am Med Inform Assoc 14(1):1–9
Shafir E (ed) (2013) The behavioral foundations of public policy. Princeton University Press,
Princeton
Solove D (2013) Introduction: privacy self-management and the consent dilemma. Harvard Law
Rev 126(7):1880–1903
Sterckx S, Cockbain J (2014) The UK National Health Service’s ‘Innovation Agenda’: lessons on
commercialization and trust. Med Law Rev 22(2):227–228
Stockdale J, Cassell J, Ford E (2018) Giving something back: a systematic review and ethical
enquiry of public opinions on the use of patient data for research in the United Kingdom and the
Republic of Ireland [version 1]. Wellcome Open Res 3:6
Stoknes P (2015) What we think about when we try not to think about global warming. Chelsea
Green Publishing, Vermont
Sunstein C (2014) Why nudge?. Yale University Press, New Haven, The Politics of Libertarian
Paternalism
Sunstein C (2016) Green by default: how a nudge and wink can save the planet. https://theecologist.
org/2016/sep/27/green-default-how-nudge-and-wink-can-save-planet. Accessed 10 Jun 2019
Tanner M (2007) Leviathan on the right: how big-government conservatism brought down the
Republican Revolution. Cato Institute, Washington, D.C.
Thaler R (2009) Opting in vs. opting out. The New York Times. http://www.nytimes.com/2009/09/
27/business/economy/27view.html?_r=0. Accessed 10 Jun 2019
Thaler R, Sunstein C (2009) Nudge: improving decisions about health, wealth, and happiness.
Penguin Group, New York
Vezyridis P, Timmons S (2017) Understanding the care.data conundrum: new information flows for
economic growth. Big Data Soc 4(1):2
Willis O (2015) Behavioral economics for better decisions, ABC.net. https://www.abc.net.

au/radionational/programs/allinthemind/better-life-decisions-with-behavioural-economics/
6798918. Accessed 10 Jun 2019
Wyatt D, Cook J, McKevitt C (2018) Perceptions of the uses of routine general practice data beyond
individual care in England: a qualitative study. BMJ Open 8(1):1–8
Zamir E (2015) Law, psychology, and morality: the role of loss aversion. Oxford University Press,
Oxford
Zeiler K, Teitelbaum J (eds) (2015) Research handbook on behavioral law and economics. Edward
Elgar Publishing, Northampton
National Electronic Health Record
Systems and Consent to Processing
of Health Data in the European Union
and Australia
Danuta Mendelson
Abstract This study focuses on the single most important regulatory aspect of
data processing, namely consent to data processing. It compares approaches to con-
sent under the General Data Protection Regulation (EU 2016/679) of the European
Parliament and of the Council on the protection of natural persons with regard
to the processing of personal data (and on the free movement of such) (GDPR)
in the context of European Union (EU) national electronic health record (NEHR)
schemes (also referred to as “national digital health networks”) with the approach
of the Australian national health record scheme called My Health Record (MHR).
The GDPR, subject to derogation in limited circumstances, is binding on all 27
EU member countries. Under Articles 168 (2) and (7) of the Treaty on the Func-
tioning of the European Union (2007), while the EU has a duty to “encourage
cooperation between the Member States…to improve the complementarity of their
health services in cross-border areas,” the European Union Member States retain
the power to manage their own health services. However, in doing so, subject to
narrow derogations, the management of their NEHR systems must conform to the
GDPR. The GDPR governs the processing of data in any form including data con-
tained in national electronic health systems (European Commission Recommen-
dation on a European Electronic Health Record exchange format (C(2019)800) of
6 February 2019. Available at: https://ec.europa.eu/digital-single-market/en/news/
recommendation-european-electronic-health-record-exchange-format. Accessed 13
May 2019). Given that, unlike the Australian MHR scheme, national electronic med-
ical/health records systems of EU Member States are at different stages of develop-
ment, and that derogations enable a measure of variance in compliance, individual
European systems will not be discussed. Australia is a non-EU jurisdiction, and does
not have the European Commission’s certificate of adequate level of data protec-
tion (GDPR Article 45 empowers the European Commission to determine whether
a country outside the EU offers an adequate level of data protection, whether by
its domestic legislation or of the international commitments it has entered into. For
further discussion, see below). One of the reasons for the absence of certification
D. Mendelson (B)
Deakin Law School, Deakin University, Melbourne, VIC, Australia
e-mail: danuta.mendelson@deakin.edu.au

https://doi.org/10.1007/978-981-15-1350-3_6
84 D. Mendelson
might be the effectively non-consensual nature of the My Health Record system

that administers, collects, stores, and provides access to health and clinical data of
Australians.
Keywords General data protection regulation (GDPR) · My health record ·

Consent · Data processing · Electronic health records
1 Introduction
One of the major legal attributes of personal freedom is having an enforceable right
to exercise voluntary and informed consent. A manifestation of un-freedom is legal
and/or practical absence or loss of the enforceable right to exercise one’s right of
refusal. In the world of datafication , though distinct from us as individuals in a
corporal, emotional, or spiritual sense, personal data has become an inalienable part
of our “data-subject” personhood. Modern technology has created for each of us an
electronic digital persona that does not possess clear juridical personality1 or a clear
status as res (thing) in the legal sense.2 Hence there are problems with defining legal
ownership of personal data. The legal status of personal medical data is murky, for
although information is not property, personal data about us—the data-subjects—
could be considered a “res,” that is property; but the question of whose property, is
yet to be determined. To put it another way, it is unclear what proprietary interest
we have in our personal data, particularly in our medical data, whether we, as data-
subjects, can exercise “dominium” (the legal right of ownership) over it, and if so to
what extent.
Philosophical and wider legal implications relating to juridical or practical depri-
vation of the right to refuse, and of what it means to be a free person in the datafied
world are beyond the scope of this chapter; rather, the analysis concentrates on
patients-cum-data-subjects’ consent because one of the manifestations of (the privi-
leges associated with) ownership is “an owner’s ability to make decisions about how
a thing is to be used: to specify who should use it, and when, and for what purposes.”3
Thus voluntary informed consent is the critical element in evaluating the adequacy of
protecting an individual’s rights with respect to the processing of his or her personal
health data. Whereas under the GDPR consent is pivotal, the Australian My Health
Record legislative scheme is not based on consent. All Australians have been non-
consensually “registered” (their personal data uploaded) in the national electronic
health record system; once registered, they were then, as “healthcare recipients,”
allowed to cancel that registration by actively opting out. For those who did so,
the controller—national system operator of the My Health Records scheme—was
required to remove their health information, but not all of the personal data, already
1 Roman law denied juridical personality to slaves. Helmholz (2012), p. 29.

2 Thiswould allow actions in rem concerning the title to status of the individual’s medical data as
property.
3 Hickey (2012), p. 227. See also, Honoré (2006), p. 132.
National Electronic Health Record Systems and Consent … 85
uploaded. Those who did not actively cancel their registration, have the right to
remove, or restrict access to documents uploaded on their file. However, healthcare
providers do not have any legal obligation to advise patients on when and which data
they upload to the system.
The study will briefly discuss the general structure of NEHR schemes, then the
main GDPR provisions relating to consent, which will be compared with the relevant
parts of the My Health Record scheme.
2 National Electronic Health Records Systems (NEHR)
According to the World Health Organization (WHO), a “national electronic health

records system” is:
implemented under the responsibility of the national health authority… that will typically
make a patient’s medical history available to health professionals in health care institutions
and provide linkages to related services such as pharmacies, laboratories, specialists, and
emergency and medical imaging facilities.4
Governments all over the world,5 including those of the EU Member States,6 have
developed, or are developing,7 technological infrastructure for national electronic
4 WHO, Report of the Third Global Survey on eHealth 2016, p. 94. Available at:
http://apps.who.int/iris/bitstream/10665/252529/1/9789241511780-eng.pdf#page=118.
5 As of 2 June 2017, an NEHR system, as per the definition provided in the 2015 WHO Global
eHealth Survey, existed in 29 countries of the European region: Albania, Austria, Azerbaijan,
Belgium, Bosnia and Herzegovina; Cyprus, Denmark, Estonia, Finland, Iceland, Israel, Italy, Kaza-
khstan, Kirgizstan, Lithuania, Luxemburg, Montenegro, Norway, Portugal, Montenegro, Repub-
lic of Moldova, Romania, Russian Federation, San Marino, Spain, Tajikistan, Turkey, Turk-
menistan, Uzbekistan (notably, France and Germany did not participate in the Survey). Avail-
able at: https://gateway.euro.who.int/en/indicators/ehealth_survey_84-has-a-national-ehr-system/
visualizations/#id=31759&tab=table. Accessed 13 May 2019. In the rest of the world, Australia,
Botswana, Cabo Verde, Cambodia, Canada, Chile, China, Costa Rica, El Salvador, Ethiopia, Iran,
Jamaica, Jordan, Kiribati, Lesotho, Malawi, Malaysia, Mexico, Mongolia, Oman, Pakistan, Panama,
Paraguay, Peru, Singapore, Timor-Leste, Uruguay, Zambia responded as having implemented a
NEHR system. WHO Atlas of eHealth Country Profiles. Available at: http://apps.who.int/iris/
bitstream/10665/204523/1/9789241565219_eng.pdf. Accessed 13 May 2019.
6 As of 2 June 2017, a NEHR system, as defined in the 2015 WHO Global eHealth Survey, existed in
29 countries of the European region: Albania, Austria, Azerbaijan, Belgium, Bosnia and Herzegov-
ina; Cyprus, Denmark, Estonia, Finland, Iceland, Israel, Italy, Kazakhstan, Kirgizstan, Lithuania,
Luxemburg, Montenegro, Norway, Portugal, Montenegro, Republic of Moldova, Romania, Russian
Federation, San Marino, Spain, Tajikistan, Turkey, Turkmenistan, Uzbekistan (France and Ger-
many did not participate in the Survey). Available at: https://gateway.euro.who.int/en/indicators/
ehealth_survey_84-has-a-national-ehr-system/visualizations/#id=31759&tab=table. Accessed 13
May 2019.
7 See, e.g., the German Appointment Service and Supply Act (TSVG), adopted on 14th March 2019,
which requires the German statutory health insurance funds to provide policyholders from 1 January
2021 with electronic health records.
86 D. Mendelson
health records systems.8 These systems tend to be distributive in nature, drawing

identifiable clinical information from participating repositories to form a critical set
of patient records. For example, Member States of the EU are required to include as
baseline, patient summary; ePrescription/eDispensation; laboratory results; medical
imaging and reports; as well as hospital discharge reports.9 The Australian sys-
tem collects a much wider range of documents, including letters from specialists.
Although the particulars of architecture and nomenclature regarding the nature of
electronic health records systems can differ depending on jurisdiction, the actual
structure of these schemes is not dissimilar.
Typically, the national health authority responsible for the NEHR scheme is the
controller in GDPR terminology, or the System Operator in Australia. They tend to
outsource the operation of the scheme to private platform providers to run either a sin-
gle (lake) repository, or as a system comprising of several federated electronic record
repositories.10 Patients’ clinical records, prescriptions, hospital discharge summaries,
pathology, radiology, and other reports and documents uploaded to the system are
sorted out and directed by an algorithm, called document registry, to the individ-
ual patient’s file. In the case of Australia, these patient files are called My Health
Record.11 To reach the appropriate destination, that is the patient’s file, both the indi-
vidual patient’s file and the relevant records must have the same unique identifier
(social security number; passport number, identity card number, Medicare number,
health insurance number, an especially created numerical identifier, etc.).
In the context of the GDPR, data contained in any electronic health record (EHR)
is, by definition, “data concerning health,” namely, “personal data related to the
physical or mental health of a natural person, including the provision of health care
services, which reveal information about his or her health status.”12 Doctors and other
healthcare professionals, who upload data provided by patients-cum-data subjects,
are referred to as “processors” in GDPR terminology.
3 The Requirement of Consent Under General Data

Protection Regulation (GDPR)
GDPR is primarily concerned with delineating requirements for lawful processing

of personal data. The term “processing” is defined in the GDPR Article 4 (2) as:
8 See, e.g., Fragidisn and Chatzoglou (2017), pp. 125–126; De Pietro and Francetic (2018), p. 70;
Hodge and Callahan (2017); eHealth Taskforce Report (2012) Redesigning health in Europe
for 2020. Available at: https://ec.europa.eu/information_society/newsroom/cf/dae/document.cfm?
doc_id=2650. Accessed 13 May 2019.
9 European Commission Recommendation on a European Electronic Health Record exchange format
(EU) 2019/243 of 6 February 2019 para 11.

10 See, e.g., Kierkegaard (2015), p. 151.
11 Garrety et al. (2016).
12 GDPR Article 4 (15). Data concerning health also includes genetic data, biometric data, and
information regarding the provision of health care services.

any operation or set of operations which is performed on personal data or on sets of per-
sonal data, whether or not by automated means, such as collection, recording, organization,
structuring, storage, adaptation or alteration, retrieval, consultation, use, disclosure by trans-
mission, dissemination or otherwise making available, alignment or combination, restriction,
erasure or destruction.
All functions of the authority-cum-controller operating a NEHR system come

within the scope of “processing” as defined above. This means that when appro-
priate,13 each of the listed operations has to comply with the GDPR consent
requirements.
Consent of data subjects has a “crucial role”14 in determining the lawfulness of
personal data processing by a controller or a processer (doctors, other healthcare
professionals) under the GDPR. It is defined in Article 4 (11) of the GDPR as:
any freely given, specific, informed and unambiguous indication of the data subject’s wishes
by which he or she, by a statement or by a clear affirmative action, signifies agreement to
the processing of personal data relating to him or her.
In other words, the GDPR rejects the notion of the so called “passive,” “meta-
consent” “ongoing,” “dynamic,” “passive,” “standing” or any other non-specific con-
sent15 whereby the lack of the data-subject’s express refusal can be considered a valid
permission. Indeed, Recital 32 requires that consent:
should be given by a clear affirmative act establishing a freely given, specific, informed and
unambiguous indication of the data subject’s agreement to the processing of personal data
relating to him or her, such as by a written statement, including by electronic means, or
an oral statement… Silence, pre-ticked boxes or inactivity should not therefore constitute
consent. Consent should cover all processing activities carried out for the same purpose or
purposes.
At the very minimum, national electronic health records schemes, which process
special categories of personal data, namely “data concerning health”16 must ensure
that the processing of that data:
i. Is proportionate to the aim pursued, which is to provide the best health outcomes
to the particular individual patient;
ii. Is respectful of the “essence of the right to data protection”: this presumably
includes the control by patients over “their own personal data,”17 and;
iii. Is based on the data subject’s explicit consent.
13 It would be clearly inapposite for the controller to seek a data subject’s consent to some of the
technical operations (structuring in the sense of formatting files, alignment or combination of files)
listed in Article 4 (2); however, other operations will require valid consent.
14 Guidelines on Consent under Regulation 2016/679 Revised and Adopted on 10 April 2018
(WP259), p. 3.
15 See, e.g., Wee et al. (2013), p. 344; Budin-Ljøsne et al. (2017), pp. 2–3; Johnsson and Eriksson
(2016), p. 472; Steinsbekk, Myskja and Solberg (2013), pp. 897–898; Kaye et al. (2015), pp. 142–
143.
16 GDPR Article 9 (1), “Processing of special categories of personal data.”
17 GDPR Recital 7.
88 D. Mendelson
The data-subject’s consent must be free, that is, there must be absence of
compulsion.18
Article 6 is the fulcrum of the GDPR because it sets up the lawful grounds for
processing. Thus, GDPR Article 6 (1) provides that:
processing shall be lawful only if and to the extent that at least one of the following applies:
the data subject has given consent to the processing of his or her personal data for one or
more specific purposes.19
The controller has the burden of demonstrating valid consent. It is one of the
conditions for valid consent under the GDPR Article 7 that in cases of processing
based on consent, “the controller shall be able to demonstrate that the data subject
has consented to the processing of his or her personal data.” It is a truism to note the
existence of power imbalance between patients and staff in public hospitals, with the
former anxious lest any reluctance on their part to agree with the attending doctor or
nurse lead to diminution of care. Indeed, in relation to public authorities, the GDPR
Recital 43 is based on a presumption that the imbalances of power between public
authorities as controllers and individual data subjects make it unlikely that the latter’s
“consent was freely given in all the circumstances of that specific situation.”
NEHR systems involve multiple processing operations that are outsourced to
several private subcontractors; it is still to be determined whether those subcontrac-
tors whose operations fall within definition of the GDPR processing would need to
demonstrate a specific consent, or whether the liability for compliance should rest
with the controller. However, it is clear that general consent will not provide “a valid
legal ground for the processing of personal data” unless the public authority can
demonstrate that a separate consent was given “to different personal data processing
operations”20 (if it is “appropriate in the individual case”).
The GDPR has developed the concept of “granular” consent that is particularly
relevant to the multifaceted structure of the NEHR systems. For instance, if the par-
ticular NEHR system uses uploaded personal health data for purposes that include the
sharing of health information with third parties; or for purposes that are not of immedi-
ate therapeutic benefit to the patient (for example, research; tracing drug/medication
(WP259), p. 5.
19 Five other grounds for lawful processing listed in GDPR Article 6 (1) include legal obligations
under contracts, and “the performance of a task carried out in the public interest or in the exercise
of official authority vested in the controller”; the necessity to protect “the vital interests of the
data subject or of another natural person”; and necessity based on “the legitimate interests pursued
by the controller or by a third party, except where such interests are overridden by the interests or
fundamental rights and freedoms of the data subject which require protection of personal data…Point
(f) of the first subparagraph shall not apply to processing carried out by public authorities in the
performance of their tasks.”
20 For example, collection, storage, adaptation or alteration, retrieval, disclosure by transmission,
dissemination or otherwise making available, erasure or destruction of the data subject’s personal
information.
use, health insurance claims, etc.),21 the patient must be informed of each specific
purpose, and the controller must obtain a separate additional consent for each of these
other purposes.22 Thus, under GDPR Recital 43 “consent is presumed not to be freely
given if the process/procedure for obtaining consent does not allow data subjects to
give separate consent for personal data processing operations respectively…despite
it being appropriate in the individual case.”23
Equally necessary for the validity of consent under GDPR, is the requirement that
it must be informed.24 Either the relevant NEHR controller or doctors as processors
of patients’ data must provide the patient with “minimum information.”25 Apart from
the duty to disclose the controller’s identity,26 patient must also be informed about
the purpose of each of the processing operations, for example, profiling.27 Profiling
involves automated data analysis that recognizes and categorizes correlations, for
example, identifying actual and potential drug abusers by collecting data on opioid
use as recorded in hospital, clinical and pharmaceutical documents; an algorithm is
then used to find and match patterns and correlations with the individual patient’s
characteristics.28 In relation to automated processing, Article 22 (1) provides that:
21 If “the controller (NEHR authority/agency/operator) has conflated several purposes for processing
and has not attempted to seek separate consent for each purpose, there is a lack of freedom.”
Guidelines on Consent under Regulation 2016/679, p. 10.
22 Unless there is another lawful basis under GDPR Article 6 (1) that is more appropriate in the
situation. See fn 48); though this would be very rare in cases where controller is a public authority.
(WP259), p. 10. For example, it would be inappropriate to seek consent from a person suffer-
ing from advanced Dementia. See also, Recital 32, “Consent should cover all processing activities
carried out for the same purpose or purposes. When the processing has multiple purposes, consent
should be given for all of them.”
(WP259), p. 13.
25 Guidelines on Consent under Regulation 2016/679, pp. 13–14.
26 See also, GDPR Recital 42: “For consent to be informed, the data subject should be aware at least
of the identity of the controller and the purposes of the processing for which the personal data are
intended.”
27 GDPR, Article 4 (4) defines “profiling” as: “any form of automated processing of personal data
consisting of the use of personal data to evaluate certain personal aspects relating to a natural
person, in particular to analyze or predict aspects concerning that natural person’s performance at
work, economic situation, health, personal preferences, interests, reliability, behavior, location or
movements.”
28 Controllers using automated and semi-automated processes must demonstrate that: (a) they are
necessary under certain contractual circumstances; or (b) authorization “by Union or Member State
law to which the controller is subject and which also lays down suitable measures to safeguard the
data subject’s rights and freedoms and legitimate interests”, or (c) “data subject’s explicit consent.”
90 D. Mendelson
The data subject shall have the right not to be subject to a decision based solely on automated
processing,29 including profiling, which produces legal effects30 concerning him or her or
similarly significantly affects him or her.31
Hence, the profiling that results provide data for a fully automated decision-
making process,32 which might impact on the data-subject’s access to health care,
health insurance, credit, and alike, must be disclosed to the person for the purpose
of “granular” consent,33 or refusal thereof.
In situations where algorithms making automated decisions as described in Article
22 (1) are used, the controller or the processor must tell the data subject that they
are engaging in this type of activity; they must provide meaningful information in an
accessible form (using common terminology, not samples of algorithms) about the
nature of the particular data analytics techniques; and explain the significance and
the envisaged consequences of such processing.34
Irrespective of the nature of processing (semi-automated or fully automated), the
controller or the processor must also explain to the patient what type of his or her data
will be collected and used: clinical records; letters to and from specialists; secondary
data copied from information contained in uploaded records of other patients (third
parties) and incorporated into the patient’s record.35 Where relevant, doctors, nurses
and other healthcare processors need to discuss with patients potential transfers of
29 “Solely automated decision-making is the ability to make decisions by technological means
without human involvement.” Guidelines on Automated Individual Decision-making and Profiling

for the Purposes of Regulation 2016/679 Adopted on 3 October 2017, p. 7.
30 Apart from affecting legal rights and legal status, the term “legal effects” encompasses “auto-
mated decisions that mean someone is: entitled to or denied a particular social benefit granted by
law, such as child or housing benefit; refused entry at the border; subjected to increased security
measures or surveillance by the competent authorities; or automatically disconnected from their
mobile phone service for breach of contract because they forgot to pay their bill before going on
holiday.” Guidelines on Automated Individual Decision-making and Profiling for the Purposes of
Regulation 2016/679 Adopted on 3 October 2017, p. 10.
31 The effect on data subjects can be significant (as against trivial) even where no legal (statu-
tory or contractual) rights or obligations are specifically affected, for example, denial of a bank
loan. Article 22 (4): “Decisions referred to in paragraph 2 shall not be based on special cate-
gories of personal data referred to in Article 9 (1), unless point (a) or (g) of Article 9 (2) applies
and suitable measures to safeguard the data subject’s rights and freedoms and legitimate interests
are in place.” Available at: http://www.privacy-regulation.eu/en/article-22-automated-individual-
decision-making-including-profiling-GDPR.htm. Accessed 12 May 2019.
32 Controllers using these processes must demonstrate also that: (a) they are necessary under certain
contractual circumstances; or (b) authorization “by Union or Member State law to which the con-
troller is subject and which also lays down suitable measures to safeguard the data subject’s rights
and freedoms and legitimate interests,” or (c) “data subject’s explicit consent.”
33 GDPR Recital 42.
34 Guidelines on Automated Individual Decision-Making and Profiling for the Purposes of
Regulation 2016/679 (WP 251), p. 25.

35 See also, Opinion 15/2011 of the Article 29 Data Protection Working Party on the definition of
consent, (WP187), pp. 19–20.

the latter’s health data, including international data transfers and cloud storage. The
patient must also be informed about his or her right to withdraw consent.36
Failure by the controller to comply with each element of the informed consent
requirement will result in making the principle of user control “illusory and consent
will be an invalid basis for processing” under the GDPR Article 6.37 By structuring
its protections around explicit and informed consent, the GDPR recognizes that
data-subjects have a measure of dominium over their personal data.
4 Australian Legislative Framework for My Health Record

System and Patients’ Consent to Processing of Their
Personal Data
The My Health Record system38 is directly governed by three statutes, My Health

Records Act 2012 (Cth)39 ; Healthcare Identifiers Act 2010 (Cth); the relevant pro-
visions of the Privacy Act 1988 (Cth) and other statutes40 ; in addition, there are six
separate binding legislative instruments known as Rules and Regulations.41
36 See Article 7 (3) GDPR.

37 “If the controller does not provide accessible information, user control
becomes illusory and consent will be an invalid basis for processing.” Guide-
lines on Consent under Regulation 2016/679 (WP259), p. 13. Available at:
file:///C:/Users/Home/Documents/000%20My%20Word%20Documents/ELECTRONIC
%20HEALTH%20RECORDS/2018%20GDPR%20Opinion%20on%20Consent.pdf. Accessed
12 May 2019.
38 Mendelson and Wolf (2016), pp. 288–290.
39 As amended by My Health Records Amendment (Strengthening Privacy) Act 2018 (Cth).
40 Including National Health Reform Act 2011 (Cth); Private Health Insurance Act 2007 (Cth);
National Health Security Act 2007 (Cth); Health Insurance Act 1973 (Cth); Census and Statistics
Act 1905 (Cth); National Health Act 1953 (Cth); Australian Bureau of Statistics Act 1975 (Cth);
Freedom of Information Act 1982 (Cth); Privacy Amendment (Private Sector) Act 2000 (Cth);
Human Services Legislation Amendment Act 2011(Cth); Australian Institute of Health and Welfare
Act 1987 (Cth); Australian Information Commissioner Act 2010 (Cth).
41 My Health Records Rule 2016; My Health Records (Assisted Registration) Rule 2015; and My
Health Records (National Application) Rules 2017; My Health Records Regulation 2012; Healthcare
Identifiers Regulations 2010; My Health Records (Information Commissioner Enforcement Powers)
Guidelines 2016.
92 D. Mendelson
The MHR scheme authorizes the Australian Digital Health Agency (ADHA)42 as
its System Operator,43 and controller, to collect, use and disclose44 health informa-
tion contained in any individual’s MHR.45 A private company, Accenture Australia
Holdings Pty Ltd,46 is at present contracted by the System Operator to act as the
MHR system’s National Infrastructure Operator. In particular, Accenture provides
and manages the National Repositories Service database system,47 which “holds the
key data sets” for each individual’s My Health Record,48 including “shared health
summaries, event summaries, discharge summaries, specialist letters, prescribed and
dispensed medications and healthcare recipient-only notes.”49 These key data sets
and other personal health data can be either uploaded into the National Repositories
Service or obtained from participating repositories.50 Each action involves processing
as defined in the GDPR Article 4 (2) above.
Originally, the system was an opt-in model, that is, patients had to actively reg-
ister to have My Health Record created for them by the System Operator. However,
42 ADHA was previously known as the National E-Health Transition Authority (NeHTA).
43 Under contract with the System Operator, a private company, Accenture Australia Holdings Pty
Ltd (a subsidiary of Accenture Holdings plc), acts as the National Infrastructure Operator (NIO)
of the system. Accenture provides and manages the National Repositories Service database sys-
tem, “which holds the key data sets which make up a My Health Record, including shared health
summaries, event summaries, discharge summaries, specialist letters, consumer entered health
summaries and consumer notes.”
44 Neither “collection” nor “disclosure” is defined in the legislation; however, according to My
Health Record Act 2012 (Cth) s 5, the verb “use” includes “accessing the information; viewing the
information; modifying the information and deleting the information.”
45 My Health Records Act 2012 (Cth) s 58. Other registered repository operators (entities that hold,
“or can hold, records of information included in My Health Records for the purposes of the My
Health Record system”) and portal operators (operators of an electronic interface that facilitates
access to the My Health Record system); as well as the Chief Executive of Medicare, the Department
of Veterans’ Affairs, the Department of Defence and the department for responsible for aged care
can collect use and disclose identifying information. My Health Records Act 2015 (Cth) s 5, s 49,
50D, and s 58A.
46 A subsidiary of Accenture Holdings plc.
47 In addition, the MHR system draws upon information held in other repositories around Australia,
operated by a mix of private and public sector organizations. The Department of Health, Personally
Controlled Electronic Health Record System Operator: Annual Report 2012–2013, “Registration of
repository operators.” Available at: http://www.health.gov.au/internet/publications/publishing.nsf/
Content/pcehr-system-operator-annual-report-2012-2013-toc~3-operation-of-the-eHealth-record-
system~3.2-registration~3.2.3-registration-of-repository-operators. Accessed 12 May 2019.
48 My Health Records Act 2012 (Cth) s 15(i).
49 Australian Digital Health Agency, “My Health Record”, “Glossary.” Available at:
https://www.myhealthrecord.gov.au/glossary. Accessed 12 May 2019.

50 Office of the Australian Information Commissioner, National Repositories Service: Implemen-
tation of recommendations—My Health Record System Operator Final Report (September 2016),
2.5. Available at: https://www.oaic.gov.au/privacy-law/assessments/national-repositories-service-
implementation-of-recommendations-my-health-record-system-operator. Accessed 12 May 2019.
since very few patients were willing to register for what then was called a Person-
ally Controlled Electronic Health Record (PCEHR),51 this model was changed in
December 2017, into an opt-out operation, which enabled the System Operator to
non-consensually create a My Health Record for each Australian.52
For several months in 2017–2018, the System Operator collected “preparatory”
identifying information about people who were not registered in the My Health
Record system when it was an opt-in model before allowing Australians to exercise
their option to opt-out.53 This preparatory identifying information includes, inter
alia, every Australian’s name, address, date of birth and gender, personal telephone
number; personal electronic address; and whether each person’s identity has been
verified. In relation to identity verification, the System Operator collected details of
the individual’s driver’s license or passport.54
People were given a window of a six-month period between the 16th July 2018 and
the 31st of January 2019 to opt out.55 Criticisms by the public and the Australian Sen-
ate Standing Committees on Community Affairs56 led to legislative amendments
giving patients the right to request at any time that the System Operator cancels
(permanently destroys) records containing health information in their individual My
Health Record.57 However, the actual My Health Record with its metadata (the name
and healthcare identifier of person who requested the cancellation and the date of
51 In 2015, only approximately 2.1 million individuals, about one per cent of Australian population,
registered for a PCEHR. The government (and the opposition) were concerned that not only patients,
but also healthcare providers lacked “any incentive to adopt and contribute to the system.” See,
the Parliament of the Commonwealth of Australia House of Representatives, Health Legislation
Amendment (eHealth) Bill 2015, Explanatory Memorandum, p. 6. Available at:
https://parlinfo.aph.gov.au/parlInfo/download/legislation/ems/r5534_ems_211631f6-fc59-
4890-8ab0-a99237f40152/upload_pdf/503821.pdf;fileType=application%2Fpdf. Accessed 12
May 2019.
52 Wolf and Mendelson (2019), p. 3.
53 My Health Records (National Application) Rules 2017 sch 1, r 5 and r 8(1).
54 My Health Records (National Application) Rules 2017, r 7; My Health Records Regulation 2012,
reg 1.1.7.
55 My Health Records (National Application) Rules 2017, Department of Health, My Health
Record: National Opt-out (15 November 2018). Available at: http://www.health.gov.au/internet/

main/publishing.nsf/Content/my-health-record-national-opt-out. Accessed 10 May 2019.
56 Senate Standing Committees on Community Affairs, Report on “My Health Records Amendment
(Strengthening Privacy) Bill 2018” 12 October 2018. Available at:

https://www.aph.gov.au/Parliamentary_Business/Committees/Senate/Community_Affairs/
MyHealthRecords/Report/c01 and https://www.aph.gov.au/Parliamentary_Business/Committees/
Senate/Community_Affairs/MyHealthRecords/Report/c02. Accessed 10 May 2019.
57 My Health Records Amendment (Strengthening Privacy) Act 2018 (Cth) sch 1, cls 5 and 6 inserting
ss 17 (2) (b) and 17 (3) of the My Health Records Act 2012 (Cth).
94 D. Mendelson
cancellation),58 and possibly the “preparatory” non-health related information,59 is

to be retained by the System Operator.
By the 31st of January 2019, over 2.5 million Australians had opted out of the
system60 and from that date the System Operator commenced populating individual
MHRs for the remaining 17 million patients with health information about them.61
Presumably, the majority of patients who did not inform the System Operator about
their intention to opt-out made a conscious decision to remain in the system. For at
least some of them, this decision would have been underpinned by a mistaken belief,
fostered by successive governments,62 that people need a My Health Record to save
their lives in the event of an emergency.63 The usual example for the life-saving role
of My Health Records is the scenario of an unconscious patient being brought into
an emergency department. However, given that “unconsciousness is a time-sensitive
medical emergency where early physiological stability and diagnosis are vital in
optimizing patient outcomes,”64 no reasonable emergency physician would wade
through an unconscious patient’s electronic health records before attending to him
or her.65 According to the Australian Medical Association’s Guide to Medical Prac-
titioners on the use of the Personally Controlled Electronic Health Record System
(Guide):
There is no obligation on any medical practitioner to access a patient’s PCEHR (now My
Health Record) in an emergency situation. Any access attempt should be determined by an
assessment of the clinical scenario.66
58 My Health Record Act 2012 (Cth) ss 17 (3), (4).

59 Although My Health Record 2012 (Cth) s 17 (3) provides that upon cancellation, the System
Operator “must destroy any record that includes health information that is included in the My
Health Record of the healthcare recipient” other than the person’s name and date of cancellation, it
is silent in relation to the “preparatory information.”
60 It has been argued that the number of persons who have opted out is much higher, see Australian
Privacy Foundation, “My Health Record.” Available at: https://privacy.org.au/campaigns/myhr/.

61 Gothe-Snape (2019) “My Health Record opt-outs top 2.5 m as service moves to ‘evolving’ choice”
ABC News 20 Feb 2019, 8:05 p.m. Available at: https://www.abc.net.au/news/2019-02-20/my-

health-record-opt-outs-top-2.5-million/10830220. Accessed 12 May 2019.
62 Mendelson (2010), pp. 662–663.
63 For example, the Australian Digital Health Agency on its site states “My Health Record lets you
control your health information securely, in one place. This means your important health information
is available when and where it’s needed, including in an emergency.” Available at: https://www.
myhealthrecord.gov.au/for-you-your-family. Accessed 14 May 2019.
64 Cooksley et al. (2018), p. 88.
65 However, while patient is being attended, a request for previous hospital records should be urgently
made, the next of kin or bystanders able to describe what has happened should also be contacted.
Cooksley et al. (2018), p. 89; see also, Braun et al. (2016).
66 AMA Guide to Medical Practitioners on the use of the Personally Controlled Electronic Health
Record System 2.8.2 (30 August 2012). Available at:

https://ama.com.au/system/tdf/documents/AMA%20Guide%20to%20using%20the%
20PCEHR%20Final%20June%202012%20Formatted%20300812.pdf?file=1&type=node&id=
36028. Accessed 12 May 2019.
Since there was virtually no public campaign aimed at informing the population
about their right to request cancellation, a high percentage of patients would have
been unaware of this option. Furthermore, many hospital patients, people in their
80s and 90s living alone or in special accommodation, persons in mental or other
healthcare facilities, the homeless, itinerant workers, the very poor, etc., might not
have access to a computer or sufficient digital literacy to opt-out. Pertinently, since
2012, all newborn children in Australia have been, and continue to be, automatically
registered in the MHR system, unless their parents request otherwise.67 These kinds
of silent or inactive “consent” to automatic registration in the system that is designed
to process their personal health data would be considered a flagrant breach of Article
6 (1) of the GDPR on the grounds that free, specific, and informed consent of data
subjects has not been obtained by the controller in accordance with Article 4 (11) of
the GDPR.
4.1 The “Standing” or “Ongoing” Consent
Indeed, consent as defined in the GDPR Article 4 (11) is absent from the MHR
regulatory framework.68 The very objective of the legislation is to ensure that once
“an individual is participating in the My Health Record system they would not need to
consent for their health information to be provided to the My Health Record system
by a registered repository operator,”69 so that “information from a wide range of
sources” can be included in the system.70 Thus the common interpretation of the
less than clearly drafted section 41 of the My Health Record Act 2012 (Cth), which
relates to the registration of patients on the MHR system, is that “a healthcare recipient
must give standing consent for registered healthcare provider organizations to upload
health information to the healthcare recipient’s My Health Record.”71
The Australian Digital Health Agency on the website titled “Digital health and
patient consent” states that:
Under the My Health Records Act 2012, healthcare provider organizations are authorized to
upload information to the My Health Record System. This means that, subject to the situations
described below, there is no requirement for a healthcare provider to obtain consent on each
67 The My Health Record registration form for the child is given to mothers in the first days
after childbirth. Available at: https://www.myhealthrecord.gov.au/sites/default/files/hd106_mhr_

newborn_factsheet_a4.pdf?v=1524052601. Accessed 13 May 2019.
68 My Health Record Act 2012 (Cth) refers to the Privacy Act 1988 (Cth) s 6, which states that
“consent means express consent or implied consent.”

69 The Parliament of the Commonwealth of Australia House of Representatives Health Legislation
Amendment (Ehealth) Bill 2015, Explanatory Memorandum 32. Available at:

http://www.austlii.edu.au/au/legis/cth/bill_em/hlab2015323/. Accessed 14 May 2019.
Amendment (Ehealth) Bill 2015, Explanatory Memorandum 32.

Amendment (Ehealth) Bill 2015, Explanatory Memorandum 72 (emphasis by the author).

96 D. Mendelson
occasion prior to uploading clinical information. There is also no requirement for a healthcare
consumer [the patient] to review clinical information prior to it being uploaded.72
Likewise, the Australian Medical Association’s Guide advises that once registered
in the system, “patients provide all medical practitioners’ ‘standing consent’ to upload
clinical information.”73 According to the Guide, even though it is good medical
practice to tell patients (and if so, to document this advice), that the information
will be uploaded to their MHRs, physicians are not obligated to provide such advice
either under the code of medical ethics or under the law.74 Nevertheless, a discussion
regarding the processing of data should be contemplated in cases where the physician
considers that the information is “potentially sensitive” and the patient may have
reservations about uploading it on the MHR system.75
In addition, the legislation authorizes registered healthcare providers to upload to
the patient’s MHR health information about a third party (another patient), “if the
health information about the third party is directly relevant to the healthcare of the
patient.”76 There is no requirement to notify, let alone seek consent from the relevant
third party about the uploading of their personal health data into another person’s
record.
That said, physicians and other healthcare providers are not authorized to upload
information where the patient has given an “express advice…that a particular record,
all records or a specified class of records must not be uploaded.”77 In other words, the
system’s legislative default is an unarticulated notion of “standing” or tacit consent:
collect the information as if it were consented to unless expressly advised other-
wise. For example, unless patients expressly advise the relevant “registered health-
care provider organization” that they refuse to have information about conditions
listed below uploaded into MHR system, Sections 41(3) and 41(4) of the My Health
Records Act 2012 (Cth) appear to override health privacy protections extended to
them in New South Wales, Queensland, and the Australian Capital Territory.78 In
these jurisdictions, healthcare providers must have the express or written consent of
the patient to upload and disclose identifying information relating to positive test
72 Australian Digital Health Agency. Available at: https://www.digitalhealth.gov.au/using-the-my-

health-record-system/maintaining-digital-health-in-your-practice/patient-consent. Accessed 26
March 2019.
73 AMA Guide to Medical Practitioners on the Use of the Personally Controlled Electronic Health
Record System 4.5.3.1 (30 August 2012). Available at:

https://ama.com.au/system/tdf/documents/AMA%20Guide%20to%20using%20the%
20PCEHR%20Final%20June%202012%20Formatted%20300812.pdf?file=1&type=node&id=
36028. Accessed 14 May 2019.
Record System 4.5.3.2 (30 August 2012).

Record System 4.5.3.4 (30 August 2012).

76 My Health Record Act 2012 (Cth) s 41(3A).
77 My Health Record Act 2012 (Cth) s 41(3).
78 Personally Controlled Electronic Health Records Regulation 2012 reg 3.1.1.
results for Acquired Immune Deficiency Syndrome, HIV or cervical cancer; or con-
fidential information associated with notifiable conditions, contagious conditions,
environmental health events, perinatal history, cancer history or pap smear (Cervical
Screening Test) history.79 There is clearly a tension between patients’ right to deter-
mine digital visibility of their personal medical data and the government’s desire to
collect complete health information about them.
The legislation provides registered healthcare recipients in all jurisdictions with
some measure of control by allowing them to “effectively remove” (make inaccessi-
ble) information from their My Health Record at any time. Patients can restrict access
to their entire record or to certain information within the record, including certain
documents; they can also exclude certain healthcare providers and organizations
from access.80 However, these controls can be utilized only after the often sensitive
health information has been non-consensually processed, and possibly viewed by
any of the thousands of registered healthcare providers81 and authorized employees
of the 15,600 registered healthcare organizations.82 Incidentally, the ex post facto
access controls by patients make all MHRs inherently unreliable.
The term “standing consent” is not used in the MHR legislation and subordinate
legislation, nor does it appear in the Office of the Australian Information Commis-
sioner’s list of “Key Concepts” for the Australian Privacy Principles Guidelines
(April 2015). The notion of “standing consent” is contrary to all GDPR principles
that protect data subjects’ fundamental rights to control their data through the exer-
cise of their right to consent or to refuse having all or particular aspects of their data
processed. The Australian system infringes each GDPR requirement for legitimate
personal data processing under Article 6.
The MHR System Operator is also authorized to arrange for use of computer
programs to make decisions83 ; presumably, these would include decisions that are
based solely on automated processing. Yet the legislation does not provide safeguards
that are analogous to Article 22 (1) of the GDPR. Australians have no right to know
whether and what decisions on the MHR system are made by algorithms without
human involvement; have no right to either consent to or refuse being subject to
79 Public Health Act 2010 (NSW) ss 56, 92; Public Health Act 2005 (Qld) ss 55, 77 to 79, 105 to
107, 175 to 177, 220 to 222, 238 to 240 and 266 to 268; Public Health Act 1997 (ACT) ss 110 and
111.
80 My Health Records Act 2012 (Cth) s 15 (b) (i) and ss s15 (c) (i) and (ii); My Health Records Rule
2016 (Cth) r 5(e); rr 6(1)(c), (2)(a). However, My Health Records Rule 2016 rr 7 and 8 allow, in
certain emergency circumstances, healthcare provider organizations to use the emergency access
function that overrides patient’s imposed access controls in order to read or retrieve information in
the relevant MHR. The use of the emergency access function is monitored.
81 For a more comprehensive discussion of access to MHRs by registered healthcare providers and
patients’ control over their MHRs (or effective lack thereof) see Wolf and Mendelson (2019).
82 ADHA, My Health Record Statistics. Available at:
https://www.myhealthrecord.gov.au/sites/default/files/my_health_record_dashboard_-_7_apr_
2019_0.pdf?v=1557190079. Accessed 13 May 2019.
83 My Health Record Act 2012 (Cth) s 13A (1).
98 D. Mendelson
profiling even where legal, social, and possibly health consequences of such decisions
may adversely affect them.
5 Conclusion
The GDPR has placed the individuals’ right to consent to or refuse processing of
their personal health data not just in general, but in specific instances at the center of
its data protection framework. In contrast, Australian legislators did not predicate the
lawfulness of personal health data processing on the individual data-subject’s valid
and informed consent; hence, the creators of the MHS system did not embed in its
architecture the right of patients to give free, informed, specific and unambiguous
indication that they agree to the processing of their personal health data in every
(reasonable) instance. By not allowing registered patients to exercise control over
their personal health data before it is uploaded to the MHR, the legislation effectively
nullifies any rights to the ownership of their data. As noted above, the failure to
protect fundamental rights of data subjects under the MHR scheme could be the
reason why Australia is not seeking certification for attainment of adequate level of
data protection from the European Commission. Chances are that while many more
Australians will cancel their MHR in future, those who remain—the very young,
the elderly, and other vulnerable patients—are and will remain disempowered data
subjects of the MHR processing system, or in the words of Joshua AT Fairfield, its
“digital serfs.”84
Acknowledgements This chapter was based on Mendelson (2018) The European Union Gen-
eral Data Protection Regulation (EU 2016/679) and the Australian My Health Record scheme—a
comparative study of consent to data processing. JLM 26:23–38.
References
Ben-Assuli O (2015) Electronic health records, adoption, quality of care, legal and privacy issues
and their implementation in emergency departments. Health Policy 119:287–297
Braun M et al (2016) Coma of unknown origin in the emergency department: implementation of an
in-house management routine. Scand J Trauma Resuscitation Emerg Med 24:61. https://www.
ncbi.nlm.nih.gov/pmc/articles/PMC4848793/. Accessed 12 May 2019
Budin-Ljøsne I et al (2017) Dynamic consent: a potential solution to some of the challenges of
modern biomedical research. BMC Med Ethics 18(1):4
Cooksley T, Rose S, Holland M (2018) A systematic approach to the unconscious patient. R Coll
Phys Clin Med (Lond) 18(1):88–93
De Pietro C, Francetic I (2018) E-health in Switzerland: the laborious adoption of the Federal Law
on Electronic Health Records (EHR) and Health Information Exchange (HIE) networks. Health
Policy 122:69–74
84 Fairfield (2017), p. 1.
Fairfield JAT (2017) Owned. Property, privacy, and the new digital Serfdom. cambridge University
Press, Cambridge
Fragidisn LL, Chatzoglou PD (2017) Development of Nationwide Electronic Health Record
(NEHR): an international survey health. Policy Technol 6:124–133
Helmholz RH (2012) The law of slavery and the European Ius Commune. In: Allain J (ed) The
legal understanding of slavery: from the historical to the contemporary. Oxford University Press,
Oxford, pp 17–39
Hickey R (2012) Seeking to understand the definition of slavery. In: Allain J (ed) The legal
understanding of slavery: from the historical to the contemporary. Oxford University Press,
Oxford
Garrety K et al (2016) National electronic health record systems as ‘wicked projects’: the Australian
experience. Inf Polity 21:367–381
Hodge SD Jr, Callahan J (2017) Understanding medical records in the twenty-first century. Barry
Law Rev 22:273–294
Honoré AM (2006) Property and ownership: marginal comments. In: Endicott T et al (eds) Properties
of law: essays in honour of Jim Harris. Oxford University Press, Oxford
Johnsson L, Eriksson S (2016) Autonomy is a right, not a feat: how theoretical misconceptions have
muddled the debate on dynamic consent to biobank research. Bioethics 30(7):471–478
Kaye J et al (2015) Dynamic consent: a patient interface for twenty-first century research networks.
Eur J Hum Genet 23(2):141–146
Kierkegaard P (2015) Interoperability after deployment: persistent challenges and regional strategies
in Denmark. Int J Qual Health Care 27(2):147–153
Mendelson D (2010) Healthcare identifiers legislation: a whiff of fourberie. J Law Med 17(5):660–
676
Mendelson D (2018) The European Union General Data Protection Regulation (EU 2016/679) and
the Australian My Health Record Scheme—a comparative study of consent to data processing.
JLM 26:23–38
Mendelson D, Wolf G (2016) My [electronic] health record—Cui Bono (for whose benefit)? JLM
24:283–296
Mendelson D, Wolf G (2017) Privacy and confidentiality. In: Freckelton I, Petersen I (eds) Tensions
and traumas in health law. Federation Press, Sydney, pp 266–282
Steinsbekk KS, Myskja BK, Solberg B (2013) Broad consent versus dynamic consent in biobank
research: is passive participation an ethical problem? Eur J Hum Genet 21(9):897–902
Wee R, Henaghan M, Henaghan M (2013) Dynamic consent in the digital age of biology: online
initiatives and regulatory considerations. J Primary Health Care 5(4):341–347
Wolf G, Mendelson D (2019) The my health record system: potential to undermine the paradigm
of patient confidentiality? UNSWLJ 42(2):619–651
Blockchain and Code
Legal Education in a Digital Age
Why Coding Matters for the Lawyer of the Future
Mark Fenwick, Wulf A. Kaal and Erik P. M. Vermeulen
In short developers will be at the center of solving the world’s

most pressing challenges. However the real power comes when
every developer can create together collaborate share code and
build on each other’s work. In all walks of life we see the power
of communities and this is true for software development and
developers. Satya Nadella (CEO Microsoft), Nadella (2018)
Abstract In this chapter, we explain the benefits of introducing a Coding for

Lawyers course in the legal curriculum and present our initial experiences with the
course. The chapter outlines the broader context of the transformation of education
in a digital age; describes the importance of computer code in a legal context, partic-
ularly in terms of the on-going disruption of the legal profession; and, introduces the
main features of the course and its initial reception. The main argument is to suggest
that the lawyers of the future will be transaction engineers managing the responsible
deployment of new technologies and the design of a new global architecture, and that
to perform this function effectively, legal professionals need to develop a number
of new skills and capacities, including an understanding of the basic concepts and
power of coding.
Keywords Blockchain · Coding · Computer code · Digital transformation ·

Ethereum · Lawyers · Legal education · Legal tech · Smart contracts · Software ·
Transaction engineers · Trust
M. Fenwick (B)
e-mail: mark@law.kyushu-u.ac.jp
W. A. Kaal
University of St. Thomas School of Law, Minneapolis, USA
E. P. M. Vermeulen
Department of Business Law, Tilburg University, Tilburg, The Netherlands
Legal Department, Signify (formerly Philips Lighting), Amsterdam, The Netherlands

https://doi.org/10.1007/978-981-15-1350-3_7
104 M. Fenwick et al.
1 Introduction
We all now live in a world of “ubiquitous computing.”1 Computers are embedded in

all aspects of our everyday lives. As a result, computer code now provides the unseen
and unnoticed architecture structuring our whole existence: work, recreation, com-
munication, consumption, travel, and education/research.2 Think about how much
of our time is spent interacting with devices that are, at some level, structured by
computer code. Such experiences can be direct and proximate—a Google search
on a smartphone or computer, for instance—or more “distant”—driving to work in
a car that relies on code-based systems in various aspects of its performance (e.g.,
cruise control, automated braking, GPS). In all cases, it is code that, in large part,
makes the experience possible and code that, ultimately, provides the structure for
that experience.
The world today is a world structured by computer code or—as the venture capital-
ist and serial entrepreneur, Marc Andreessen, puts it—“software is eating reality.”3
Code provides the fundamental architecture that sets the terms on which life in a
digital age is lived. Code determines how easy it is to protect personal information
or express ourselves. It determines whether access to information is open or whether
specific information or space is zoned, and access limited. Code affects who sees
what, or who or what is monitored. It determines how machines communicate in the
Internet of Things. The deep architecture of a digital world can regulate in a host
of ways, ways that one cannot begin to see (or understand) unless you consider the
nature of such code and how it operates.
Code raises all manner of substantive legal issues. But computer code and digital
technologies are also transforming what it means to be a lawyer. In that respect, all
lawyers will be affected by the on-going digital transformation. If Satya Nadella is
right that developers will be at the “center” of “solving the problems of a today’s
world,” then lawyers must be able to speak the language of code to participate in the
crucial task of engaging with and solving these problems.
In this chapter, we pursue the thought that lawyers have, historically speaking,
been at their most effective—and socially useful—when they operate as transaction
engineers, i.e., facilitating new forms of business and other social relationships. It
seems obvious that the engineering of the near-future will, to a large extent, be code-
based. This does not mean, as some have suggested, that lawyers are irrelevant or
that they will disappear.4 Instead, the digital transformation will disrupt the legal
profession and demand a different set of skills and capacities than have traditionally
been taught in law school or used in legal practice.
It is for this reason that we decided to introduce a Coding for Lawyers course in our
legal education programs. We are convinced that coding can help us in solving many
1 The concept of “ubiquitous computing” was first coined by Mark Weiser, chief scientist at XEROX,
in the late 1980s.

2 See Lessig (2006), Mitchell (1996).
3 Andreessen (2011).
4 See, e.g., Mangan (2017).
Legal Education in a Digital Age 105
contemporary economic, environmental and social issues and that to participate in

the multi-disciplinary teams of the future, lawyers will need to develop some coding
skills.
It is perhaps not entirely surprising to discover that this initiative has been met
with a certain amount of skepticism. “Is it necessary for law students to learn how to
code?”; “What is the value-added for us (non-technologists) of understanding code?”;
or, “Isn’t this all just a waste of time?” These are some of the typical responses of
colleagues.
No doubt, there is something to these concerns. For a start, we don’t need to have
any understanding of code to successfully navigate the digital world. Users of digital
devices don’t need to be able to code for themselves. After all, we use technology all
the time, without ever really understanding it. The overwhelming majority of users
of the Internet don’t necessarily understand Transmission Control Protocol/Internet
Protocols (TCP/IP protocols). Moreover, coding isn’t easy. Acquiring competency
takes a significant investment of time and even if law students do feel the need to
learn how to code, there are thousands of available resources (on- and off-line) out
there that can teach them. It isn’t effective or necessary to add “another course” to
the curriculum.
To be clear from the start, however, we are not suggesting that law students
should become professional coders. After all, coding is complicated and to become
a serious coder does take time. However, we do believe that the ability to understand
and communicate with coders is a necessary skill for the lawyer of the future. As
such, we think that law students will benefit from understanding the basic concepts
and power of coding. Not only by reading or hearing about it but by participating in
and experimenting with coding projects in a classroom environment.
The Coding for Lawyers course has generated a lot of interest, so this chapter
aims to briefly introduce the background, content and initial experience of the course.
Section 2 outlines the broader context of the transformation of education in a digital
age; Sect. 3 suggests that lawyers have been at their most effective when operating as
“transaction engineers” that create opportunities for new forms of business and other
social relationships; Sect. 4 describes the importance of computer code in a legal
context particularly in terms of technology-driven changes to the legal profession
(i.e., Legal Tech) and introduces our main argument, namely that to perform the
transaction engineer function effectively lawyers require some basic grasp of the
concepts and capacities of coding; Sect. 5 briefly introduces the main features of the
new course; and Sect. 6 concludes.
2 Education in a Digital Age
Over the last half-century, digital technologies have transformed the world. The emer-
gence of computer-based devices and communication infrastructures have triggered
multiple social, economic and cultural effects. Here, we use the term digital trans-
formation to refer to this shift from analog, electronic and mechanical machines to
networked, digital devices and the on-going social effects associated with the prolif-
eration of these new technologies. The process of digitization began in earnest with
the launch of the Intel microprocessor in California in the early 1970s and is driven
by an ongoing series of technological innovations. Most significantly cheaper and
smaller digital hardware (first PCs and, more recently, smartphones); global com-
munication networks and mass connectivity (i.e., the Internet); and cloud-based data
storage & automated algorithms.
With ever-shorter innovation cycles, it seems obvious that new technologies are
going to continue to transform every aspect of how we live and work. Constant
technological disruption is now the “new normal” and, as a consequence, “old world”
concepts, paradigms and ideas are becoming less relevant. At least, they are being
challenged and disrupted by this new networked world of interconnected digital
technologies.
The resulting uncertainties create a massive challenge for all educators and not
just lawyers. “What should we be teaching our students today? How can we prepare
students for the complex and uncertain world of tomorrow?” Teaching has always
tended to be “backward-looking” and knowledge-based. Transmitting the settled and
authoritative knowledge of the past has been the starting point for our whole approach
to education. For instance, in a legal context, students have traditionally analyzed
existing laws, regulations, and cases. The idea has been that if you understand and
examine historical developments, you would be able to solve future problems by
applying old doctrines and precedents to the new situation. Similar logic can be
seen in other fields. MBA programs, for example, employed the same “case study”
approach in a business context. “Old” solutions are studied, analyzed and then applied
to new situations.
The responsibility of the educator in this approach to education was to acquire
and then transmit this settled information/knowledge. In a world of information
asymmetries, the educator-student relationship was, by necessity, a hierarchical one.
After all, the teacher had all the knowledge. This was the source of their authority
and credibility as educators. But this model seems less well suited to a world of fast-
paced change and easy access to information. Prior experience may not be relevant
to a fast-changing reality, and all the information is only ever one Google search
away.
If the future is radically different from the present, it doesn’t make sense to focus
too much attention on transmitting information that is already readily available and
(anyway) seems likely to be less and less relevant.
Instead, education needs to become much more forward-looking and skills—
rather than content—based. How then do we prepare the next generation for dealing
with unknown future problems? What skills are going to be most useful in this
new world? And what is the role and the responsibility of educators in this new
model? These are the questions that everyone involved with education needs to be
asking. Denying this change is only doing a disservice to our students and those
programs/faculties/universities that are quickest to adapt stand to benefit. In the
same way, business has been transformed over the last decade by the arrival of
digital technologies, and it is not unreasonable to believe that education may face
similar levels of disruption over the next one or two decades. For a start, everyone
is going to need a much better technical grasp of the core technologies surrounding
computers, communication networks, and artificial intelligence (AI). For many of
us, the underlying technologies that are driving the digital transformation remain a
mystery, and that (surely) is a problem. Practical technical knowledge needs to be
integrated into many fields of education and from a young age. Coding and data
analysis seem a good starting point. But we also need to identify other subject-
specific skills and capacities that are important in a world of constant change. The
focus should be on building skills that will assist the next generation in making better
decisions under conditions of cognitive and normative uncertainty.
3 Lawyers as Transaction Engineers
When thinking about the skills that legal education needs to be transmitting to stu-
dents, it is important to consider the function and needs of the legal profession. In
this regard, it might be helpful to examine the role of lawyers in earlier periods of
technological change. Crucially, lawyers can play a vital role in co-creating solutions
around the deployment of new technologies. In this context, the history of the legal
profession can provide some guidance as to the type of role that lawyers can usefully
play.
From a historical perspective, lawyers have been most important when they have
operated as transaction engineers that create opportunities for new forms of busi-
ness and other social relationships. Consider the development and growth of Silicon
Valley as a center for digital technologies in the early 1970s. While the idea of the
clustering of similar businesses was a significant source of innovation, there is a broad
consensus that the legal industry was also important in the development of technol-
ogy firms and facilitating innovation.5 For example, lawyers were responsible for
drafting the innovative contractual provisions that protected high-risk investors—
for instance, angel investors and venture capitalists—from the relational and per-
formance risks associated with investing in young companies and inexperienced
founder-entrepreneurs. 6 Moreover, the involvement of lawyers in both non-legal
and legal activities, such as deal-making, matchmaking, gatekeeping, and conciliat-
ing, also served as an important sorting device for entrepreneurs that needed more
than just investors to start and scale their young businesses. Finally, the contractual
mechanisms and the lawyer-dominated market for reputation reduced information
5 Chander (2014), p. 639.

6 Bernstein (1995), pp. 245–251.
asymmetries between the entrepreneurs and investors and, as such, were necessary
to bring the demand- and supply-side of venture capital together in a way that was
effective and mutually advantageous.7
We can see from the Silicon Valley example how lawyers functioned as socially-
useful transaction engineers. Law firms operated as crucial intermediaries that
brought together, in a safe space, various parties with different but mutually com-
patible interests and novel forms of expertise. Lawyers were not only acting in the
interests of their clients, but a secondary effect of their work was to open a new space
for innovative young start-ups to flourish. On this type of account, the often-neglected
contribution of local law firms to the institutionalization of venture capital and ven-
ture capital contracting goes some way in explaining the success of Silicon Valley.8
The legal profession was a crucial factor in creating an exciting and economically
important new world.
The problem, however, is that lawyers have often failed to perform this function
of being proactive transaction engineers that add value. Instead, they often become
a hindrance or obstacle to any form of transacting, let alone creative or innovative
transacting.9 This can happen for multiple different reasons, but the tendency to
proceduralize solutions and to employ standard form templates is one major factor.
Fixed and standardized solutions are often imposed on complex, dynamic transac-
tions resulting in frustration and difficulties, especially for clients.10 The list of com-
plaints is familiar: lawyers are verbose, they do not listen, they are unresponsive, they
are constantly saying “no,” they charge too much, they are not commercially minded,
they spend too much time on trivial issues, they do not keep clients informed, they
constantly “over-lawyer,” and they do not communicate clearly and concisely. As a
result, lawyers have developed a reputation as one of the least trusted professions.11
In a highly competitive and fast-changing environment, law firms need to focus
on re-discovering their function as effective transaction engineers that can help par-
ties to facilitate interactions and reduce costs. After all, the Fin Tech revolution is
premised on the multiple costs that could and should be cut: agency costs, transaction
costs, monitoring costs, regulatory costs and (the increasingly important category of)
compliance costs. In that respect, the potential for lawyers to add value is enormous.
As such, lawyers are well-positioned to play a crucial role in the co-creation of
the infrastructure for the deployment of new technologies. Such co-creation involv-
ing partnerships between multiple actors can be crucial to building a better digital
future.12
7 See Mehta (2004), Schmidt (2013).

8 McCahery (2012), p. 211.
9 Rubinstein and Roth (2007), p. 3.
10 Open Law (2017), Parsonanikas (1980), p. 605.
11 See, e.g., Nelson (2015), Thompson (2012).
12 Callison et al. (2018), p. 766, Fenwick and Vermeulen (2015), p. 611.
4 The “Lawyer of the Future”
Many commentators—both from practice and academia—can recently be found sug-

gesting that the digital transformation may mark the beginning of the “end” for
lawyers and other legal advisors.13 We certainly do not agree with this argument.
Nevertheless, we do think that if lawyers of the future are to function as effective
transaction engineers playing a socially productive role in the deployment of digital
technologies, it is clear that the legal profession is going to need to adapt in various
ways. Here we identify a number of features of the “lawyer of the future” and outline
how code will be intimately connected with the evolving profession.
4.1 The Evolving Character of Legal Work
Before we turn to a more code-specific discussion, here a couple of general obser-

vations about the evolving character of legal work and how a different skill-set is
required in the context of the digital transformation and the performance of this
transaction engineer function.
Firstly, lawyers of the future will need to be able to assume the role of project
managers or, at least, active participants in multi-disciplinary teams that will design
solutions for the problems of the future. As such, a capacity to operate effectively
in diverse teams will take on a much greater significance than has previously been
the case for lawyers. The ability to work with and communicate with a diverse
range of partners will be increasingly required. In the digital world, this means
that lawyers will not only have to work closely with their traditional professional
partners such as accountants or financial advisors but also with engineers, designers,
architects and other technical experts and specialists (depending on the particular
project/transaction). In this new model of legal services, law firms will become more
like platforms with an emphasis on connecting legal and other experts and managing
the resulting collaboration and transactions.14 In a new world of platforms, this type of
matchmaking and project-based partnership, will mean that lawyers and other legal
advisors need to be aware of the way network technology, and other code-based
technologies operate.
Secondly, in pursuing these new solutions, lawyers will be confronted with a
very different type of client with a different set of needs. Fast-growth technology
companies with few assets and fewer employees are increasingly central in a digital
environment. The most successful companies today have leveraged the opportunities
of networked digital technologies to develop new business models. Trust, value and
wealth are created through such platforms, connections, and networks, instead of
the (more traditional) management of workers or physical assets.15 Crucially, many
13 See,e.g., Koebler (2017), Susskind (2010).

14 For more on the emergence of platforms, see Fenwick et al. (2018).
15 See Vermeulen (2018a, b).
firms in this new innovation-driven economy adopt new organizational forms and
governance structures to deliver their latest products and services.16
What then are the main features of these new organizations? To appeal to Millen-
nial talent and consumers, younger firms have often embraced more mission-driven
and inclusive organizational cultures and practices in which a “best-idea-wins” cul-
ture replaces formal hierarchies. Significantly, however, many such new firms have
struggled to maintain this new governance model and fulfill their initial promise. As
firms scale, they often find themselves transforming into the very thing that they were
initially designed to avoid, namely a corporate dinosaur.17 Lawyers of the future will
need to understand the opportunities and challenges of the digital world and help
firms to maintain their more open and inclusive governance structures even after
they have become more prominent and successful.
Moreover, as traditional concepts of a “career” become much less relevant in
the so-called “Gig Economy,” it will become increasingly important to build and
communicate a personal “brand” by telling the right kind of story.18 More open
organizations will mean having to work in teams of strangers, often from diverse
national or disciplinary backgrounds. The ability to work in such a group, continually
adapting to new situations and working patterns, becomes crucial.
4.2 Legal Tech
Turning now to the importance of code, the performance of this transaction engineer
function will require a knowledge of coding. A first reason for this claim is the
increased importance of information technologies in legal practice. Legal technology,
or Legal Tech, is changing the way that lawyers practice law. In this context, Legal
Tech refers to platforms, IT services, and software that make law firms and lawyers
more efficient in performing their activities.19 Legal Tech has evolved from support
systems to fully integrated and automated services for lawyers that increasingly
disrupt the practice of law. As such, Legal Tech can be defined as the integration
of information technology services and software in a legal context, as well as the
development of legal platforms and their applications.20 Since the 1970s, with the
invention of the first legal databases, Legal Tech has supported lawyers in their work,
reducing costs and improving performance. The success of these earlier efforts and
developments in information technology have attracted further investment in this
field, further boosting the potential efficiency gains.21
16 See,more generally, Callison et al. (2018).

17 See Fenwick et al. (2018).
18 The “Gig Economy” is a labor market characterized by the prevalence of short-term contracts or
freelance work, as opposed to permanent jobs.

19 See Vermeulen (2017a).
20 Lenon and Tarling (2017).
21 Catalyst Investors (2018), Rayo (2017).
As such, Legal Tech has made law firms and lawyers more efficient in performing
their activities. Early examples of such technologies include automated billing, doc-
ument storage, practice management, and accounting software.22 But, from the early
2010s, Legal Tech became more advanced and started to incorporate technology
that assisted legal professionals in due diligence and e-discovery processes.23 Since
around 2015, Legal Tech has continued to evolve in unprecedented ways. Multiple
start-up companies and their investors have started to capitalize on technologies, and
their applications are already replacing some junior lawyers and disrupting the exist-
ing parameters for the practice of law.24 We would suggest that to play a meaningful
role in designing and implementing these solutions. It becomes essential for prac-
ticing lawyers to have, at least, some understanding of the underlying technologies,
specifically code. Not every lawyer will be involved in the design of such technolo-
gies, but almost all lawyers will be users. Moreover, more feedback that can be
provided the more such technologies will be able to develop and improve with each
iteration.
Four types of Legal Tech start-up can be distinguished. The first category includes
start-up companies that offer a range of online legal services, removing the “in-
person” legal consultation process and guidance for clients.25 The second legal start-
up category involves online matching platforms that connect lawyers with clients.26
Such platform start-ups help consumers find an appropriate lawyer without the
involvement of a law firm. The third category entails start-ups that use AI tools to
take over their lawyer time-consuming and expensive legal research activities such as
reviewing, understanding, evaluating, and reapplying contracts.27 Finally, start-ups
with expertise in blockchain technology attempt to replace lawyers as intermediaries
in certain types of transactions.28
A central purpose of these start-up companies seems to be the disruption of exist-
ing legal practices, and this has broad repercussions for the legal profession. Extant
legal services are rendered increasingly irrelevant or seem likely to be replaced by
Legal Tech.29 In particular, junior legal professionals and legal support staff are
likely the first victims of the development of Legal Tech.30 Legal Tech applications
will soon be able to perform much of the work of junior lawyers—such as document
review—without the human elements that introduce imprecision, flaws, inaccuracies,
possible lawsuits, and delay.31 Second, and more importantly, the legal profession
22 Anderson and Hathaway (2014), O’Loughlin (2018).

23 Ambrogi (2015).
24 Cornell Tech Blog (2016).
25 Andreae (2017).
26 Andreae (2017).
27 Andreae (2017).
28 Vermeulen (2017b).
29 Lohr (2017).
30 Vermeulen (2017a).
31 Vermeulen (2017a).
will be forced by such start-up companies to innovate, a task that is not easily accom-
plished by overextended and often cumbersome legal organizations that have deeply
entrenched operating procedures and have long since lost the capacity for rapid re-
invention.32 Nevertheless, the potential benefits seem to make such a transformation
inevitable for most law firms.
More speculatively, Legal Tech has the potential to rapidly transform law firms
and legal departments into virtual law firms. Virtual law firms may dominate in the
future. A virtual law firm is a platform with an emphasis on connecting legal and other
professionals in a collaborative online environment involving human and machine
actors. When implemented successfully, the effect of the platform model will be the
creation of a flexible and accessible community of professionals with different skills
and experience. The bigger the community, the easier it is to offer solutions tailored
to the needs of the clients. The virtual law firm model attracts a broad spectrum of
law firms. One extreme is represented by the traditional law firm characterized by a
hierarchy with partners at the top and varying levels of associates, paralegals, and non-
lawyers below them. On the other end of the spectrum are those firms that adopt an
Airbnb-type platform organization, mainly providing a matchmaking/coordination
service. Enormous variations exist between the two extremes, depending on the level
of implementation of Legal Tech.
Some start-ups are already looking to develop as legal platforms that add value
by connecting lawyers and clients. At the moment, the focus is primarily on match-
making but integrating AI or machine learning are explicitly mentioned as a key
near-future goal. For instance, UpCounsel offers entrepreneurs on-demand access to
experienced lawyers.33 LawyerlinQ34 in the Netherlands and Digitorney35 in Ger-
many offer law firms the possibility to “insource”36 special knowledge and skills
for more complex projects. Digitorney aims to disrupt the traditional relationship
between companies and corporate law firms by providing a platform or matchmak-
ing function that connects companies with the law firm that is best placed to meet a
company’s specific needs. As such it helps companies overcome the lack of trans-
parency or comparability in the legal market services, as well as the greater degree
of specialization in a business law context.
In various ways, Legal Tech is replacing—or at least, supplementing—the tradi-
tional role of law professionals. Law professionals play a crucial role in establishing
trust and truth in legal transactions. They negotiate, draft, and interpret contracts and
help enforce them; they create laws and regulations that protect the weaker parties,
and they design structures that enable the registration and transfer of tangible prop-
erty and intellectual property. Well-drafted legal contracts help establish confidence
32 Winick (2017).
33 Upcounsel https://www.upcounsel.com/. Accessed 21 January 2019.
34 LawyerlinQ https://about.lawyerlinq.com/. Accessed 21 January 2019.
35 Digitorney https://www.digitorney.com/. Accessed 21 January 2019.
36 “Insourcing,” in this context, refers to the practice of using an organization’s own staff or other
resources to accomplish a task that was previously outsourced.

in the validity of the transaction and the economic benefits of the deal for the con-
tracting parties. Important matters, such as the truth about ownership and control, the
transfer of property, and the allocation of risk and control, are typically covered in
a contract. However, such deal-making, matchmaking, gatekeeping, and enforcing
roles are increasingly performed by—or, at least, with the assistance of—technol-
ogy. This trend is likely to accelerate soon, enabled by new technologies (such as
blockchain technology) and smart contracts.
But it will not stop here. Near future technological advances—most obviously,
machine learning and deep learning—have already started to replace lawyers and
other legal professionals. Artificial intelligence tools help clients to review, under-
stand and even draft legal documents.37 Data analytics, machine learning, and deep
learning are not only used to do legal research but also assist in legal decision-making
and the prediction of legal cases.38 As such, there is no doubt that Legal Tech will
automate legal work, such as contract drafting, legal risk management, and dispute
resolution. If legal work is dependent on and performed by algorithms in the future,
it will soon be crucial for all lawyers to have a better understanding of data analytics
and AI.
4.3 Designing a New Global Architecture
Many of the solutions that the lawyer of the future will be expected to help design
will be technology-based, i.e., the transactions that lawyers will be facilitating will be
dependent on computer code. 39 It is clear that the global architecture of the future will
revolve around digital technologies and digitization. Emerging technologies, such as
artificial intelligence, the Internet of Things, and big data analytics, are enabling the
emergence of economies with new types of organizations and innovative business
models. Digital technologies have changed and will continue to change consumer
behavior. Employer-employee relationships are changing. We think differently about
status and ownership. Automation of manual and knowledge work is happening and
cannot be stopped. Algorithms increasingly determine our choices. Sensors and bio-
metrics will enable us to do business in new ways. Companies are witnessing the
transformation from products to services. We will increasingly assume digital iden-
tities. Assets, such as real estate, cars, etc., will also be digitized in the future. We
are creating a digital world that mimics the “real world.” Technology leads to flatter
and more decentralized organizations and business models. Platform companies are
transforming industries. Nanotechnology will have a significant impact on our life
37 For example, Beagle (2019) https://www.beagle.ai/, Legalzoom (2019) https://www.legalzoom.

com/, and Legal Robot (2019) https://www.legalrobot.com/. Accessed 21 January 2019.
38 Rich (2018).
39 For more, see Fenwick et al. (2018a).
expectancy. The list goes on, and it is clear that in our digital society, the coordi-
nation and incentive structures will be significantly different from the hierarchical,
proceduralized and regulated structures in our current more centralized societies.
So, why are lawyers so important in the design and architecture of this flatter
world? Steve Jobs alluded to an answer in his “lost interview” in 1995:
Everybody in this country should learn how to program a computer, should learn a computer
language because it teaches you how to think. It’s like going to law school. I don’t think
anybody should be a lawyer but going to law school can actually be useful because it teaches
you how to think in a certain way… I view computer science as a liberal art.40
Jobs didn’t elaborate on what he meant by this “certain way” of thinking or why
that might be a powerful quality/skill in a digital world. But our experience makes
us think that Steve Jobs was correct, and that legal reasoning can play a crucial role
in designing the new infrastructure for a digital age. Here are three skills that—
that taken together—distinguish legal thinking and which are relevant for the digital
transformation.
i. Complex Problem-Solving: Legal thinking is all about creative problem-solving
involving highly complex fact patterns. This means critical thinking, analyzing,
and applying. Lawyers are trained to think in and with sophisticated concepts,
and this is a helpful skill when solving the known and unknown challenges of
the future;
ii. Incentive-Based Thinking: Legal thinking traditionally involves the analysis and
study of the impact of rules and regulation on human behavior. This incentive-
based thinking can also be powerful in the digital age in which traditional
systems, models and assumptions will be overhauled and thinking about the
effects of technological architecture on human behavior will be crucial in making
choices about the best architecture;
iii. Storytelling and Persuasion: There are often more answers to legal problems,
and a premium is on constructing a convincing story and persuading others as
to your point of view. These values seem consistent with the needs of a digital
world in which the best solution is by no means visible and storytelling comes
to the fore.
Of course, legal reasoning traditionally involves looking backward and building on
precedents and analogies with an emphasis on previous experience and rules. Legal
reasoning in a digital age, therefore, needs to be adjusted and creativity needs to be
incorporated.
So, what can be done to facilitate law firms in meeting these new challenges and
re-focusing on being active and effective transaction engineers? What seems clear is
that, in the context of the digital transformation, an understanding of code is going
to be crucial for the lawyer of the future to perform this function effectively. The
development of blockchain technologies and smart contracts are particularly relevant
in this regard.
40 Jobs (1995).
4.4 The Building Blocks of a Digital World? Blockchain &

Smart Contracts
To understand blockchain technology, it makes sense first to consider the Internet.

The Internet enabled a free, fast and global exchange of information and ideas. The
blockchain adds another dimension by making it possible to transfer and exchange
value (and assets) without the involvement of intermediaries.
So, what is a blockchain? A blockchain is a shared and distributed digital ledger
or database that maintains a continuously growing list of blocks. A block could
contain records of transactions regarding digital assets but could also include facts
and information. Once the record is verified and validated, a block is added to the
chain with previous records in linear and chronological order.
What makes the blockchain such a revolutionary technology is that the ledger or
database is distributed to a countless number of participants (nodes) around the world
in public peer-to-peer networks (similar to the Internet) or private (or permissioned)
peer-to-peer networks (similar to an intranet).41 These participants can be individuals
or organizations (and even things). The only condition is that they have a mobile
phone and Internet connection. Everyone with a mobile phone can create a real
digital ID and interact with other people in the blockchain network. Blockchain
technology thus enables and facilitates access to finance, insurance services, stock
markets, etc. Such peer-to-peer transactions are possible because the technology uses
a distributed consensus model where the network nodes verify, validate and audit
transactions before and after they are executed. This is safer than a traditional model
in which transactions can only be accomplished through third-party intermediaries,
such as a bank, judiciary or notary.
Network connectivity is also crucial because it allows for multiple copies of the
blockchain to be available across a distributed network. This makes it practically
impossible to alter or erase information in the blockchain. The use of cryptographic
hashes makes tampering with blockchain records even more difficult, if not impos-
sible. Cryptographic hashes comprise complex algorithms and even a minuscule
change to the blockchain will result in a different hash value, making manipulation
instantly and readily detectable. As such, digital signatures help establish the identity
and authenticity of the parties involved in the transaction. These security measures
make blockchain validation technologies more transparent and less prone to error
and corruption than existing methods of verifying and validating transactions via
third-party intermediaries.42
In short, blockchain technology creates an independent and transparent plat-
form for establishing truth and building trust.43 Intermediaries, bureaucracy and old-
fashioned procedures are replaced by code, connectivity, crowd, and collaboration.44
41 Gupta (2017).
42 White (2017).
43 White (2017).
44 Vermeulen (2018a, b).
The technology increases openness and speed, while at the same time significantly
reducing costs.
But perhaps the most significant feature of blockchain is that it is so adaptable.
There are multiple possible applications relevant in a business context. Most obvi-
ously, blockchain can be used to provide new methods of processing digital transac-
tions.45 But blockchain can also be used for cryptocurrencies, records management
(for example, real estate, corporate or medical records), e-voting and identity man-
agement. It is for this reason that blockchain technology has been mentioned as one
of the most significant disruptive technological innovations since the emergence of
the Internet.46
Blockchain becomes particularly attractive in the legal context when combined
with smart contracts. In this context, a smart contract refers to a computer program
code or protocol that automates the verification, execution and enforcement of certain
terms and conditions of a “contractual” arrangement. Nick Szabo, the computer
scientist and lawyer, first introduced the term in 1994.47 In a business context, a smart
contract could be an essential part of, for instance, a car loan. For example, if the
borrower misses a payment (tracked via a blockchain-like technology), the contract
would not allow the use and operation of the car (enforced via networked technologies
that disable the car automatically, rather than a repo man physically depriving a
driver of access to their car).48 Such smart contracts will become more prevalent in
the growing world of the Internet of Things. The more devices are connected, smart
contracts will be more frequently used in the execution and enforcement of legal
transactions. There is no doubt that smart contracts are already disrupting traditional
legal assumptions, doctrines, and concepts. For instance, it will give a boost to the
sharing economy (with its implications for property law).
A more complex example of the use of blockchain technologies is the set-up of
so-called decentralized autonomous organizations (DAOs) built on software, code,
and smart contracts, challenging traditional corporation laws.49 DAOs are merely
computer code. They do not have any directors, managers or employees. The gov-
ernance structure is built with and on software, code and smart contracts that run
on a public decentralized blockchain platform (in most cases Ethereum). This auto-
mated structure is intended to give participants/investors in the DAO direct real-time
control over contributed funds and where such funds would be distributed. DAOs
fit in the flatter, decentralized and automated world. With software code automating
procedures and tasks, the focus of knowledge work is shifting from the routine appli-
cation of procedures to designing the systems and standardized functions that are
then performed by machines. Unfortunately, law school programs have been slow
to adapt to these technological developments. Most students are still being prepared
45 Gupta (2017).
46 Hiesboeck (2016).
47 Szabo (1994).
48 Küster (2017).
49 For more on DAOs and the broader technology-driven transformation of corporate governance,
see Fenwick and Vermeulen (2018).

for a hierarchical, centralized and “proceduralized” world. In the labour market of

the future, however, a premium will be placed on a person’s capacity to design and
communicate innovative solutions, rather than comply with pre-established proce-
dures. And since these new solutions will be code-based, an understanding of code
and coding will be essential to participate effectively in our digital world.
Teaching students and experienced lawyers the basics of how to code and inspiring
them to get out of their comfort zone, will be a necessary first step to help them
embrace the many future opportunities of a software-based environment. Blockchain
and smart contracts can solve multiple societal challenges and—in doing so—to
facilitate new opportunities for disruptive business models. Consider the following
real-world implementations of these technologies that will impact on legal practice:
i. Health and wellbeing: Blockchain technology has the potential to transform

healthcare, giving the patient more control in the healthcare ecosystem by
increasing the security, privacy, and interoperability of their health data;
ii. Agriculture and food security: Consumers increasingly favor clean food, but it
can be difficult to verify the integrity of products. A distributed ledger replacing
the current supply chain would provide greater transparency, disclosure and
trust. Fair price-setting and fast payment systems would also be facilitated;
iii. Safe, clean and efficient energy supplies: We are facing a rapid growth in dis-
tributed energy resources. Think rooftop solar and electric vehicles, for example.
Governments, utilities, and other stakeholders need to find new ways to regulate
better and manage the electricity grid. Blockchain has the potential to offer a
reliable, low-cost solution for financial or operational transactions to be recorded
and validated across a distributed network.
In each of the above examples, a major social problem could potentially be addressed
more effectively by the use of blockchain technology and smart contracts.
From a legal perspective, an interesting feature of such technologies is how they
deliver a new mechanism for trusting. In a 20th-century business context, trust was
typically created and maintained by rules, regulations, or contracts.50 One way of
thinking about the law is as a mechanism for stabilizing expectations and building
trust and reputation when interacting with strangers.51 “I may not know you or even
like you, but the fact that we have a contract means that I can (to a certain extent)
trust you.” In a digital environment, trust can be achieved through software code
agreed upon between the parties that reduce the need for (or at least, the scope of)
traditional contracts. Recent interest in smart contracts suggests that this is going to
be a significant growth area in the near future. Moreover, building trust via code is
also crucial in machine-to-machine (M2M) based transactions. As M2M interaction
becomes normalized in an Internet of Things environment, the issue of trust is re-
imagined as a technical and design problem.
Such code-based solutions are increasingly delivered through online/cloud-based

services. Again, this trend will only increase with the proliferation of M2M interac-
tions. But instead of combatting cybersecurity with the introduction of more law in
the books, the lawyer of the future will need to engage with more technology-based
solutions. At the very least, lawyers will need to acquire the necessary knowledge
to evaluate technology-based solutions and compare them with the more traditional
paper-based alternatives.
Finally, there are the myriad ethical issues that are created in a code-based world
and being aware of these issues is again essential for the lawyer of the future. Consider
the example of the driverless car. How do we want our driverless vehicle to react when
confronted with an unavoidable accident? Should it minimize the loss of life, even if
that means sacrificing the occupants of the car or should it prioritize the lives of the
occupants at any cost? Alternatively, should the choice be a random one? This is just
one (well-known) example of an ethical challenge associated with new technologies.
There are already multiple ethical questions involved with the dominant position of
software code in our society.52 Building the capacity of the lawyers of the future to
think about the social and ethical implications of code is both essential and inevitable.
But, to say something sensible about the ethical aspects of technology, it is necessary
to understand more about the capacities and limits of coding.
5 Coding for Lawyers
The above discussion provides the background for why coding will become increas-
ingly important for lawyers. So, what does the Coding for Lawyers course entail
and what lessons has the experience of offering such a course taught us, the course
organizers?
Two defining features of this course are its interdisciplinary and dynamic character.
Most obviously, the course is taught by a team of lawyers, coders, and mathemati-
cians, and the curriculum has been constantly adapted in pursuit of an appropriate
balance between the various disciplinary elements. The need for constant experi-
mentation by multidisciplinary teams—both regarding course content and teaching
method—is one of the most important takeaways we have gained from the experience
of offering this course. And, this is something all educators (as well as law firms and
policymakers) should consider integrating into their practice.53
In the course, we first introduce the main features and advantages of blockchain
and smart contract-based applications. We explain how these applications inhibit
rent-seeking and offer greater transparency and security. In particular, the organiza-
tion of DAOs is introduced in more detail to show how traditional forms of business
organization are being disrupted by blockchain based technology. The open-source
52 See, e.g., Bossmann (2016).

53 Fenwick et al. (2018b).
governance protocols used in DAOs are outlined to highlight how an openly read-
able ledger means anyone can monitor the integrity of transactions. The distributed
cooperation component highlights a key advantage of such technologies, namely that
hackers must be able to out-compute the entire network (which is difficult. Moreover,
since DAOs are cheap and straightforward to clone, this will potentially lead to more
competition and improvement in the model. The distributed and anonymous nature
of the organizations prevents natural and political monopolies.
Of course, DAOs and other blockchain-based technologies still have significant
technical and operational shortcomings, and, in our discussion, the following weak-
nesses are highlighted. There is still a lack of decentralization (there are currently
no true DAOs). For instance, Bitcoin‘s proof of work protocol has led to mining
pools because of economies of scale and unbalanced reward structures. Also, the
anonymity in blockchain organizations means that they are prone to 51% attacks.54
We also discussed the many examples in which the anonymity (and autonomy) have
led to hacks.
Next, we introduce the mathematics behind cryptography, namely the role of
hash functions, and Merkle Trees. Several recent real-world blockchain initiatives
are introduced. For instance, we present a blockchain-based “reputation verification
platform” (Semada.io) that rewards trustful parties and punishes bad actors.55 The
Semada Proof of Stake protocol (SPoS) is presented as a recent example that uses its
reputation-verification platform to solve the centralization, efficiency and security
problems that afflict existing blockchain consensus protocols.
Finally, students are introduced to the Ethereum platform and given a basic intro-
duction to coding on that platform. Assessment for the course involves students
coding for themselves. Here are some examples of the kind of project that students
have submitted:
• Blockchain solutions for the purchase of used cars.
• A blockchain-based dispute resolution procedure.
• A blockchain-based land and real estate registration system.
• Blockchain-based digital identities for refugees or other stateless persons.
• Blockchain-based ride-sharing.
• Energy trading schemes.
• Smart contracts for political campaign funding.
• Blockchain and smart contracts in compliance.
• A decentralized social media platform.
• Blockchain solutions for art traders.
In an increasingly software-driven world, we need to remain smart about tech-
nology. And this is what the Coding for Lawyers course is aiming to achieve. It is
not about teaching students how to become coders, but about making them realize
54 A 51% attack refers to an attack on a blockchain in which hackers gain control of more than 51%
of the networks mining rate allowing them to prevent new transactions from receiving confirmation
or reverse transactions that had already been completed.
55 For more, see Calcaterra and Kaal (2018).
how important it is to think about our relationship with new technology and tech-
nology experts. It is about encouraging a new level of technological literacy. In this
way, students can see the new opportunities that technology creates, but also think
about the new issues (practical and ethical) that such technologies create. Such an
informed, but critical approach toward technology seems important for all lawyers. In
this respect, law firms also have some obligation to institutionalize life-long learning
mechanisms that ensure all of their employees are given the necessary core knowl-
edge to understand, deploy and provide meaningful feedback on technology-based
solutions.
6 Conclusion
Lawyers should not feel threatened by the exponential growth of new technology and
the subsequent social and economic change that it brings. But nor should they deny
such change and cling to traditional ways of operating. Instead, lawyers should view
emerging technologies as a source of tremendous opportunity and growth. If, as seems
likely, machines can reduce standardized legal work, there will be more time for
assisting the client with the new and specific challenges of navigating the complexities
of a digital environment. However, to enjoy the benefits of such opportunities, it will
be necessary to possess a new level of literacy in the various basic building blocks
of this new world. In an age of ubiquitous computing, a crucially important element
is code and coding. In this chapter, we explored the benefits of introducing a Coding
for Lawyers course in the legal curriculum and its potential benefits for the legal
profession. The main argument was to suggest that the lawyers of the future will
operate as transaction engineers and that to perform this function effectively, legal
professionals will all need to be able to understand—at some level—the concepts
and power of coding.
References
Ambrogi R (2015) The 10 most important legal technology developments of 2015. Law Sites.
Available at: https://www.lawsitesblog.com/2015/12/the-10-most-important-legal-technology-
developments-of-2015.html. Accessed 21 Jan 2019
Anderson WH, Hathaway JA (2014) All-in-one practice management applications. GP
Solo. Available at: https://www.americanbar.org/publications/gp_solo/2014/July-August-2014/
allinone_practice_management_applications.html. Accessed 21 Jan 2019
Andreae Z (2017) Legal tech start-ups. Medium. Available at: https://medium.com/legal-tech/legal-
tech-startups-9755b18f93ac. Accessed 21 Jan 2019
Andreessen M (2011) Why software is eating the world. Wall Street J. Available at: https://www.
wsj.com/articles/SB10001424053111903480904576512250915629460. Accessed 21 Jan 2019
Bernstein L (1995) The Silicon Valley lawyer as transaction cost engineer? Oregon Law Rev 74:239–
255
Bossmann J (2016) Top 9 ethical issues in artificial intelligence. World Economic

Forum: agenda. Available at: https://www.weforum.org/agenda/2016/10/top-10-ethical-issues-
in-artificial-intelligence. Accessed 21 Jan 2019
Callison W, Fenwick M, McCahery JA, Vermeulen EPM (2018) Corporate disruption: the law and
design of organizations in the twenty-first century. Eur Bus Organ Law Rev 19:737–769
Calcaterra C, Kaal W (2018) Semada technical whitepaper: blockchain infrastructure for measuring
domain specific reputation in autonomous decentralized and anonymous systems. Available at:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3125822. Accessed 21 Jan 2019
Chander A (2014) How law made Silicon Valley. Emory Law Rev 63:239–255
Catalyst Investors (2018) Legal tech is primed for growth investments. Available at: https://
rossintelligence.com/legaltech-growth-investments/. Accessed 21 Jan 2019
Cornell Tech Blog (2016) 3 Reasons why tech companies need a new kind of lawyer. Cornell Tech:
Law Tech Blog. Available at: https://tech.cornell.edu/news/3-reasons-why-tech-companies-need-
a-new-kind-of-lawyer/. Accessed 21 Jan 2019
Digitorney (2018) White paper. Available at: https://group.digitorney.com/wp-content/uploads/
2018/12/Digitorney-STO-Whitepaper_20181214.pdf. Accessed 21 Jan 2019
Fenwick M, Kaal WA, Vermeulen EPM (2018a) Regulation tomorrow: what happens when
technology is faster than the law. Am Univ Bus Law Rev 6:561–584
Fenwick M, Kaal WA, Vermeulen, EPM (2018b) Legal education in a digital age: why coding for
lawyers matters. U of St. Thomas (Minnesota) Legal Studies Research Paper No. 18–21. Available
at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3227967. Accessed 21 Jan 2019
Fenwick M, McCahery JA, Vermeulen EPM (2018) The end of “corporate” governance: Hello new
world of platform governance. European Corporate Governance Institute Working Paper No.
430/2018. Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3232663. Accessed
21 Jan 2019
Fenwick M, Vermeulen EPM (2015) The new firm. Eur Bus Organization Law Rev 17:595–623
Fenwick M, Vermeulen EPM (2018) Technology & corporate governance: Blockchain, crypto, and
artificial intelligence. European Corporate Governance Institute—Working Paper No. 424/2018.
Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3263222. Accessed 21 Jan
2019
Gupta S (2017) Blockchain: The next big revolution. Medium. Available at: https://medium.com/
blockchain-musings/blockchain-the-next-big-revolution-dabb748d33fa. Accessed 21 Jan 2019
Hiesboeck M (2016) Blockchain is the most disruptive invention since the internet itself: not just in
finance. Digital Doughnut. Available at: https://www.digitaldoughnut.com/articles/2016/April/
blockchain-is-the-most-disruptive-invention-since. Accessed 21 Jan 2019
Jobs S (1995) The lost interview. Available at: https://www.youtube.com/watch?v=TRZAJY23xio.
Accessed 21 Jan 2019
Koebler J (2017) Is AI making lawyers a disappearing profession. Financial Review. Available
at: https://www.afr.com/business/legal/is-artificial-intelligence-making-lawyers-a-disappearing-
profession-20170418-gvmzbs. Accessed 21 Jan 2019
Küster F (2017) What are smart contracts in blockchain technology. Captain Altcoin. Available at:
https://captainaltcoin.com/blockchain-smart-contracts. Accessed 21 Jan 2019
Lenon J, Tarling B (2017) The next phase of legal technology has already been built. Above the
Law: Evolve the Law. Available at: https://abovethelaw.com/legal-innovation-center/2017/12/07/
the-next-phase-of-legal-technology-has-already-been-built/. Accessed 21 Jan 2019
Lessig L (2006) Code, and other laws of cyberspace, Version 2.0. Basic Books, New York
Lohr S (2017) I, robot, Esq.? Not just yet. New York Times. 19 March 2017, B7
Mangan D (2017) Lawyers could be the next profession to be replaced by computers. CNBC, Avail-
able at: https://www.cnbc.com/2017/02/17/lawyers-could-be-replaced-by-artificial-intelligence.
html. Accessed 21 Jan 2019
McCahery JA (2012) Corporate venture capital: from venturing to partnering. Oxford Handbook
of Venture Capital. Oxford University Press, Oxford
Mehta V (2004) Principle-agent issues in private equity and venture capital. Available at:
https://repository.upenn.edu/cgi/viewcontent.cgi?referer=https://www.google.com/&httpsredir
=1&article=1013&context=wharton_research_scholars. Accessed 21 Jan 2019
Mitchell W (1996) City of bits: space, Place and the Infobahn. MIT Press, Boston
Nadella S (2018) Microsoft + Github = Empowering developers. Microsoft Official Blog. Avail-
able at:https://blogs.microsoft.com/blog/2018/06/04/microsoft-github-empowering-developers/.
Accessed 21 Jan 2019
Nelson F (2015) Why do people hate lawyers so much? Lawyers Weekly. Available at: https://www.
lawyersweekly.com.au/folklaw/16179-why-do-people-hate-lawyers. Accessed 21 Jan 2019
O’Loughlin E (2018) Legal document management software buyer’s guide: Software advice. Avail-
able at: https://www.softwareadvice.com/legal/document-management-comparison/. Accessed
21 Jan 2019
Open Law (2017) Decentralizing the deal. Medium. Available at: https://media.consensys.net/
decentralizing-the-deal-e6af1c0cfdab. Accessed 21 Jan 2019
Parsomanikas MG (1980) Frustration of contract in international trade law and comparative law.
Duquesne Law Rev 18:551–599
Rayo EA (2017) AI in law and legal practice: a comprehensive view of 35 current applications.
Tech Emergence. Available at: https://www.techemergence.com/ai-in-law-legal-practice-current-
applications/. Accessed 21 Jan 2019
Rich B (2018) How AI is changing contracts. Harvard Business Rev. Available at: https://hbr.org/
2018/02/how-ai-is-changing-contracts. Accessed 21 Jan 2019
Rubinstein F, Roth AM (2007) The life cycle of a venture-backed company, Deal strategies for
venture capital and private equity lawyers. Aspatore Books, New York
Schmidt D (2013) Entrepreneur’s choice between venture capitalist and business angel for start-up
financing. GRIN. Available at: https://www.grin.com/document/230839. Accessed 21 Jan 2019
Susskind R (2010) The end of lawyers: rethinking the nature of legal services. Oxford University
Press, Oxford
Szabo N (1994) Smart contracts. Available at: http://www.fon.hum.uva.nl/rob/Courses/
InformationInSpeech/CDROM/Literature/LOTwinterschool2006/szabo.best.vwh.net/smart.
contracts.html. Accessed 21 Jan 2019
Thompson D (2012) The least-trusted jobs in America: congress members and car salespeople, The
Atlantic. Available at: https://www.theatlantic.com/business/archive/2012/12/the-least-trusted-
jobs-in-america-congress-members-and-car-salespeople/265843/. Accessed 21 Jan 2019
Vermeulen EPM (2017a) The future of law, lawyers and law professors … and the exponential
growth of disruptive technology. Medium. Available at: https://chatbotslife.com/the-future-
of-law-lawyers-and-law-professors-and-the-exponential-growth-of-disruptive-technology-
b5c979608c9c. Accessed 21 Jan 2019
Vermeulen (2017b) There is no escape from blockchain and AI … Lawyers better be pre-
pared! Medium. Available at: https://medium.com/@erikpmvermeulen/there-is-no-escape-from-
blockchains-and-artificial-intelligence-lawyers-better-be-prepared-2d7a8221c627. Accessed 21
Jan 2019
Vermeulen EPM (2018) We all struggle! Medium: hackernoon. Available at: https://hackernoon.
com/we-all-struggle-adb0cfc65773. Accessed 21 Jan 2019
White M (2017) Will blockchain transform the public sector? Deloitte Insights. Avail-
able at: https://www2.deloitte.com/insights/us/en/industry/public-sector/understanding-basics-
of-blockchain-in-government.html. Accessed 21 Jan 2019
Winick E (2017). Lawyer-bots are shaking up jobs. MIT Technology Review. Available at: https://
www.technologyreview.com/s/609556/lawyer-bots-are-shaking-up-jobs/. Accessed 21 Jan 2019
Reputation Protocol for the Internet
of Trust
Craig Calcaterra and Wulf A. Kaal
a complete absence of trust would prevent us from even getting

up in the morning. (Niklas Luhmann). (Luhmann (1979).
A very large portion of ‘gig economy’ startups are at the core
basically just a dispute resolution system a reputation system
and a search engine.
If you’re looking to build a decentralized version of one maybe
consider focusing on one component. (Vitalik Buterin). (Buterin
(2018).
Abstract Internet-based platform businesses outcompete traditional brick and mor-

tar business on nearly all measures of comparison. Similarly, distributed ledger tech-
nology (DLT) businesses depend on the Internet and show a lot of promise in multiple
business verticals. Yet, Internet-based platform businesses and DLT businesses have
not reached their full potential. Multiple studies, data, and anecdotal evidence sug-
gest that one core factor that undermines their evolution is the worldwide decreasing
trust in the Internet and under-developed trust in decentralized technology solutions.
Semada and the Semada Research Institute (SRI) believe that a decentralized reputa-
tion network can reverse that trend to increase trust in the Internet and increase decen-
tralized technology adoption. We call this solution the Semada Internet of Trust—a
network that uniquely captures real world information, context, and value in crypto-
graphic transactions generating transparently validated consensus on truth. Through
the creation of unconscious and conscious trust in decentralized networks adoption
becomes a desirable outcome and increases. Semada is a unique blockchain platform
well suited for the decentralized web. Semada’s core architecture continually resolves
the blockchain trilemma by combining: (i) decentralization; (ii) scale, and; (iii) secu-
rity. Semada’s version of proof-of-stake—a.k.a. the Anchor Protocol—is a unique
consensus algorithm that enables block propagation through reputation staking and
C. Calcaterra (B)
Department of Mathematics, Metropolitan State University, Minneapolis, USA
e-mail: craig.calcaterra@metrostate.edu
W. A. Kaal
University of St. Thomas School of Law, Minneapolis, USA

https://doi.org/10.1007/978-981-15-1350-3_8
124 C. Calcaterra and W. A. Kaal
verification. Through its Sockpuppet Protocol, Semada is resilient to deceptive fake

Internet identities, so-called sockpuppet accounts, and 51% attacks. Through its Per-
sona Protocol, Semada facilitates digital sovereignty, self-sovereign identity, privacy,
security and decentralized reputation. Trust enabling reputation can only be earned
over time on the Semada network. The Semada Internet of Trust creates a framework
of trust through reputation and incentive optimization that enables market condi-
tions for unprecedented business models and market capitalization while reducing
adverse selection and moral hazards for transacting parties. The network will pro-
vide information symmetry while reducing transaction costs for market participants.
Businesses will leverage the platform to achieve efficiencies across verticals.
Keywords Trust · Internet · Blockchain · Reputation · Protocol
1 Introduction
Internet-based platform businesses have been disrupting centralized brick and mortar
businesses since the inception of the Internet. Yet, in recent years, increasing evidence
suggests that the Internet is experiencing a trust crisis, e.g., Internet users no longer
trust one another as they have in the past. To restore confidence in the Internet, efforts
have tended to focus on introducing and implementing more regulations, processes,
and procedures. Yet, the traditional trust and cohesion mechanisms have reached
their limits.
The purpose of encryption technology on the blockchain is to safeguard individu-
als’ digital sovereignty and power over their personal privacy. Superficial identifiers
such as physicality, sex, gender, race, culture, language, location, or group iden-
tification, which have traditionally disenfranchised minorities, are less relevant in
Decentralized Autonomous Organizations (DAOs) with pseudonymous geographi-
cally distributed members. This encourages more opportunities in business and serves
the goal of just distribution of power, creating an efficient and responsive meritocracy.
Distributed reputation networks can help make a difference for trust on the Inter-
net. The enduring nature of reputation linked to blockchain promotes transparent
and fair power distribution to those who contribute to the common good of a DAO
and away from those who detract from this common good or cause harm to the least
powerful. Creating and maintaining a secure and meaningful reputation motivates
productive collaboration in a decentralized economy by focusing members’ actions
toward improving their DAOs for the promise of future currency rewards.
Reputation Protocol for the Internet of Trust 125
2 Problem
2.1 Internet-Based Platforms Proliferate
Internet-based platform businesses have been proliferating since the dot.com boom
in the 2000s. With the increasing availability of big data, deep learning, machine
learning, sensors and sensor data, as well as DLT solutions, such technology-
driven evolution of Internet-based platform businesses can be expected to expand
exponentially.
Internet-based platform services are so successful because their peer-to-peer capa-
bilities make them compelling, efficient, and attractive in a fast-changing world. That
is why people are so willing to place their trust in new businesses and brands.
But before they were widely adopted, trust issues that obviously exist in a peer-
to-peer environment had to be solved. Why would consumers trust somebody they
do not know or someone who is anonymous? Search algorithms, data analytics,
and consensus protocols have helped digitize trust by minimizing bad experiences,
maximizing great experiences, and avoiding unwanted issues (such as the double
spending of cryptocurrencies). Yet, consumers continue to struggle with trust issues
with these platforms. Digital trust may sound more attractive than institutionalized
trust, but it still has its shortcomings. The most serious challenge for them are the
trust crises on the Internet.
2.2 Trust Crises on the Internet
Human lives are defined by who, what, when and how humans trust. While we tend
to think of “trust” and “trusting” as a feeling or emotion, a better way to think of trust
may be the particular types of decisions humans make. In essence, humans “trust”
when we decide to take some kind of risk. This trust can involve banal decisions
such as what local supermarket to choose or which physician to select for a particular
treatment. Trusting also involves a decision to expose ourselves to a danger of some
kind. We decide to place our fate in another person or organization. As such, trust
can be seen as the triumph of hope over fear. We overcome our natural aversion to
risk in anticipation of a better future (a tasty meal, good health etc.). Every day, all
of us make many different decisions to trust. Together, these choices define who we
are and what matters to us.1 But who, what, when and how we trust is changing.
The public trust in institutions and institutional governance, e.g., institutionalized
trust, has been eroding for years. Scandals, corruption, slow procedures, human error,
1 Throughout human history, personal trust is first established within closed families and com-
munities. Yet, in an industrialized and globalized world, trust is established by organizations
(governments, businesses, schools, and other intermediaries) and procedures (rules, contracts, and
hierarchies). Modern economy and society are built on the success of these more impersonal or
institutionalized forms of trust.
inefficient legacy systems, etc., increasingly destroy trust in the organizations and
procedures of the modern world. As a result, we live in an age of diminishing trust
in organizations and procedures. We are less and less willing to place our fate in the
hands of incumbents and traditional ways of doing things.
To restore confidence in the “old world,” efforts have tended to focus on introduc-
ing and implementing more regulations, processes, and procedures. Yet, the tradi-
tional trust and cohesion mechanisms have reached their limits. Who and how we trust
has already changed. The world of organizations (corporations, government, etc.),
and procedures is being replaced by new forms of trust. Slow and expensive experi-
ences have resulted in greater distrust in established institutions and their products
and services.
This has driven the emergence of new and disruptive Internet-based platform busi-
ness models. Trust has become digitized and automated. We are much less trusting
of organizations or procedures, but instead trust machines and algorithms. Yet, this
digitized trust is experiencing its own crisis which undermines the proliferation of
value-enhancing Internet-based platform businesses and DLT businesses.
The trust crises on the Internet cannot be directly traced and have several origins
and may in fact have a generational dimension.2 Because of the dominance of a
few select broadband Internet service providers, changes in the net neutrality rule,
and concerted efforts at misinformation of the public via Internet campaigns, the
Internet has become a network of vertically and horizontally integrated monopolies
that create core deficiencies including: “(i) constrained knowledge exchange because
of information silos; (ii) lack of competition impedes innovation (including at the
protocol level); (iii) increased exposure to cyber and socially engineered attacks,
and; (iv) diminished consumer protection and rights. These deficiencies in today’s
Internet age consistently and progressively undermine trust on the web.
Human interaction with machines on the Internet exacerbates the Internet trust
crises. Humans trust machines on the Internet to store and process information and
to transact with each other and with machines. Yet no centralized or decentralized
authentication engine has typically verified the trust humans place in machines or
otherwise enabled a trusting environment for Internet-based transactions. Worse yet,
today’s Internet is designed for hierarchical societal structures and with an underlying
authoritative trust model. Many inefficiencies are associated with the traditional
hierarchical trust model including serious cyber security vulnerabilities.
Several studies have provided evidence on the acceleration of trust issues on
the Internet around the globe.3 Trust issues in the United States can clearly be
distinguished as exacerbated in international comparison.4
2 PEW Research (2014). (“standing social science survey question, “Generally speaking, would you
say that most people can be trusted or that you can’t be too careful in dealing with people,” just 19%
of Millennials say most people can be trusted, compared with 31% of Gen Xers, 37% of Silents
and 40% of Boomers.”)
3 Edelman Report (2018).
4 Ibid.
Cyber security attacks may be among the factors that help explain the decline in
trust on the Internet. Anecdotal evidence suggests that the cost of instigating cyber-
attacks has declined significantly. Cyber security attacks can now be launched by
unsophisticated parties at minimal cost with maximal potential for damage.5
By the year 2020 the majority of experts estimate that a total of 30 to 50 billion
devices will connect to the Internet.6 The prevailing opinion of these researchers
suggests that the centralized version of the Internet cannot scale to that level. This
creates unprecedented cyber security issues.7 Decentralized authentication solutions
for IoT devices are needed.
The Edelman Report also suggests that trust is on the decline worldwide among
multiple metrics. Fake news is contributing to this world-wide trend.
3 Internet-Based Platforms Use of Reputation Metrics
Internet-based platforms use reputation metrics in an effort to improve their cus-

tomer experience and their own business models. Those include Facebook’s “like”
mechanism as well as reputation rankings on Amazon, Uber, and Airbnb. Several
studies have demonstrated that existing reputation solutions for Internet-based plat-
form businesses are subjected to significant shortcomings.8 Resnick and Zeckhauser
find that: “One of the earliest and best-known Internet reputation systems is run by
eBay, which gathers comments from buyers and sellers about each other after each
transaction. Examination of a large data set from 1999 reveals several interesting
features of this system, which facilitates many millions of sales each month. First,
despite incentives to free ride, feedback was provided more than half of the time.
Second, well beyond reasonable expectation, it was almost always positive. Third,
reputation profiles were predictive of future performance. However, the net feedback
scores that eBay displays encourages Pollyanna assessments of reputations, and is
far from the best predictor available. Fourth, although sellers with better reputations
were more likely to sell their items, they enjoyed no boost in price, at least for the
two sets of items that we examined. Fifth, there was a high correlation.”9 (Fig. 1).
5 Fortune (2016).
6 Nordrum (2016).
7 McKinsey & Company (2015).
8 Resnick and Zeckhauser (2001) “The presumptive challenge to Internet-based feedback systems
is to get buyers to provide feedback with reasonably high frequency, and to provide it honestly.
Frequency is not a problem, presuming the feedback is unbiased. More than half of transactions
receive feedback. However, the 0.3% negative feedback rate on transactions (0.6% of those that
provided feedback) and 0.3% neutral feedback numbers from eBay, our principal data source, are
highly suspicious.”
9 Resnick and Zeckhauser (2001).
Fig. 1 Internet-based platform’s use of reputation solutions

3.1 The Shortcomings of This Use of Reputation

in Internet-Based Platform Businesses Include
the Following
i. All centralized reputation scores can be sybil attacked, e.g., fake Internet
accounts (sockpuppets) or fictitious ratings disrupt true reputation scoring;
ii. Users earned their reputation but do NOT own reputation—e.g., if Face-
book and other Internet-based platform businesses accidentally deleted user
accounts, years of data would be removed and users have near zero ability to
reclaim the data;
iii. All financial and reputation scores are trapped in Silos. Such reputation data
incompatibility undermines interoperability between platforms. There is also
no way to aggregate user reputation in disparate platforms;
iv. Users cannot build reputation-based social capital—the platforms control the
reputation and users cannot otherwise engage with the reputation built on such
platforms;
v. No Privacy: Employers have unfettered access and are checking people on
social media and credit checks. No way to curtail for users;
vi. No Reputation Data Autonomy for users;
vii. No Reputation-Based Network Effect—lacking decentralized aggregation of
reputation. Reputation currency created on the Platforms only benefits the
respective platforms;
viii. No Self-Sovereign Identity.
3.2 Internet of Trust
While trust in old world organizations and procedures has been eroding for years,
society is not convinced that distributed or decentralized peer-to-peer networks can
deliver enhanced digitized trust. To truly enable a more distributed society, it is
necessary to personalize and humanize digital trust.
The authors believe that a decentralized reputation infrastructure can increase
and humanize trust on the Internet and in decentralized technology solutions. We
call this solution the Internet of Trust. Core building blocks of the Internet of Trust
are protocols that enable a trust verification network.
4 Semada Research Institute
The Semada Research Institute (SRI) is a blockchain research laboratory that devel-
ops and promotes research built on the core reputation verification algorithm of the
Semada Platform. Combining existing Semada protocols, such as Persona and Sock-
puppet Protocols, among others, SRI‘s initiatives and research agenda is directed by
the resolutions passed by the Semada Research Network (SRN). Together, SRI and
SRN develop a research agenda to implement the Semada core as a truly decentralized
reputation-based trust engine for the Internet of Trust in the distributed economy.
4.1 Founding Principle
A founding principle of the SRI is its recognition and affirmance of reputation-based

networks as the inevitable and central methodology for any evolution of decentralized
and centralized digital networks.
4.2 Reputation as True Decentralization
The transferability of stores of value entailed in cryptocurrencies creates core points

of attack and undermines the very nature of decentralization. Because of the store
of value entailed in cryptocurrencies (and really any currencies or store of value),
exercise of power over such value inevitably leads to centralization due to economies
of scale.
Reputation as a metric and store of value does not suffer the same consequences
described above and does not lead to centralization because it must be earned, it
can be lost, yet it cannot be turned into a fungible store of value that allows the
exchange of goods etc. Hence, reputation does not naturally lead to centralization
like currencies do.
4.3 Motivation
Does a need for a decentralized reputation-based research platform exist?

Conglomerates like Facebook, LinkedIn, Twitter, Uber, Airbnb, Yelp and other
platforms have built platforms that capture reputation. Yet, nobody has created a
truly Sybil-attack resistant reputation engine that cannot be corrupted, regardless of
the economics at stake.
4.3.1 Smart Contracts Create New Solutions and Have Multiple Core
Benefits
The merciless mathematical logic of computerized code that makes up a smart

contract is extremely valuable for clarifying intent, as well as making business
transactions more efficient and certain.
Yet, smart contracting has several significant downsides:
• Bugs: Newly-coded programs, alas, always have the potential for bugs;
• Human business interactions require a flexibility that is at odds with the merciless
mathematical logic of smart contracts. Business interactions require:
– Flexibility in interpreting intent;

– Ability to continue collaboration when unanticipated eventualities arise;
– Possibility that each party may fulfill only a portion of any intended collabo-
ration.
4.3.2 Business Degenerates with Smart Contracting if Left to Its Own

Devices
The original vision for smart contracts is between businesses and clients following
a “code is law” concept that self-executes and self-regulates, with often anonymous
parties. This creates a zero-sum scenario where both parties are entirely incentivized
to provide the minimum possible to fulfill the perfectly rigid self-executing contract.
This type of business ultimately degenerates as people will spend extra effort to
deliver the minimum acceptable products. In other words, both parties are bound by
common sense to push the contract as far as possible in their own favor. The parties
no longer create a harmonious long-term business relationship.
This effect changes if reputation of the counterparties is at stake. With the oppor-
tunity to create new valuable reputation tokens, members strive to act in ways which
improve the platform for the long term instead of exploiting short-term arbitrage
opportunities.
5 Semada Research Network
The development and evolution of Semada’s Internet of Trust requires a comprehen-

sive research agenda. The Semada Research Institute (SRI) is a blockchain research
laboratory that develops and promotes research, architecture, and design built on
the core reputation verification algorithm of the Semada Platform. Combining exist-
ing Semada protocols, such as Persona and Sockpuppet Protocols, among others,
SRI‘s initiatives and research agendas are directed by the Semada Research Network
(SRN), a decentralized autonomous organization (DAO) on the Semada Platform that

originates and implements research proposals for the Semada Platform.
SRN is the research engine within the Semada platform. Its purpose is the iden-
tification of valuable new research ideas and protocol upgrades. SRN is a DAO on
the Semada Platform and an expertise tag within the Semada core.
Members of the SRN DAO develop and implement the research agenda for SRI
by:
(i) Posting architecture expansion agendas or individual proposals, including via
new use cases of Semada Core, and providing evidence of their research for
protocol expansion, and;
(ii) Writing and posting code on Github or Bitbucket that directly enhances or
further expands the Semada Core architecture and/or its associated protocols,
including its use cases and agenda.
5.1 SRN DAO Candidates
Membership of the SRN DAO is contingent on being selected by the existing SRN
DAO members. Candidates for SRN DAO membership only qualify if they have
posted valuable protocol upgrades that were upvoted by the existing members.
SRN DAO members get paid for their upvoted protocol contributions in both SEM
Tokens and SRN Tokens. SEM tokens are fungible and traded on cryptocurrency
exchanges. SRN Tokens are not fungible. SRN Tokens are the membership tokens
that grant access to the SRN DAO. Only the SRN Tokens are used for voting on
incoming protocol improvement proposals.
5.2 Semada Platform
Semada is a 4th generation blockchain platform for the decentralized web. The core
components of the platform are: (i) core protocol stack to achieve consensus, repel
network attacks and propagate transaction blocks; (ii) immutable decentralized mem-
ory ledger to store transaction blocks; (iii) network resource market/exchange to stake
CPU, memory, bandwidth, etc. for rent seeking; (iv) programmable smart contract
interface for transaction and resource manipulation.
The architecture is intended to solve the blockchain trilemma, combining scala-
bility, security, and decentralization. The underlying protocols create inherent trust
on the network based on a decentralized reputation system of trust scores. Semada
is modular, has an advanced smart contracting platform and on-chain governance
for swift protocol upgrades. Semada is resilient to Sybil attacks, tyranny of majority
and 51% attacks. Through its universal trust protocol, Semada removes corruptible
intermediaries and creates trusted paths to form an unprecedented trusted mesh as

the foundation for the web 4.0.
Through cryptoeconomic incentives, Semada adds network resources to process
and propagate. Semada is a decentralized platform that methodically builds trust
for smart contracting parties and the smart contracting code itself. Semada properly
rewards good-faith business behavior and smart contract development, providing an
ideal evolutionary environment where business can continually improve and grow.
This evolutionary business environment is achieved by employing the basic
strengths of blockchain technology: decentralization, an eternal record, and trans-
parency. This allows Semada to create meaningful and secure SEM tokens which
represent the reputation of members, with different types of tokens for every different
type of expertise.
These foundational objects of secure and meaningful SEM tokens solve many
basic problems in blockchain technology, such as giving the proper incentives
for guaranteeing secure proof-of-stake consensus protocols for block production.
This eliminates the unsustainable inefficiencies of proof-of-work based blockchains.
Meaningful reputation tokens also naturally incentivize stable and productive on-
chain governance for all protocol upgrades. With on-chain governance true DAOs
are achieved for any type of business.
This general environment for creating DAOs provides the level playing field many
independent companies require to negotiate and collaborate in developing the IoT.
SEM tokens provide the membership identification layer for the IoT, where any
device with the proper reputation can be trusted to add or download information, and
use smart contracts to automatically settle remuneration debts. The proper reputation
is provided to a device if it is verified as being installed following the protocols
specified by the particular DAO for the specific device.
5.3 Protocols
5.3.1 Anchor Protocol
Block propagation in the Semada platform is facilitated by staking reputation, not a

fungible currency as in Proof of Stake (PoS). Accordingly, the consensus algorithm
is called Semada Proof of Reputation (PoR). In the Semada infrastructure, this PoR
protocol is called the Anchor Protocol because it is the first use of the underlying
voting algorithm. Moreover, staking in the Anchor Protocol means anchoring your
reputation to a block. In other words, Semada block producers anchor their reputation
to a block, and if the block is invalid or cancelled out, their reputation depreciates.
How Semada uses reputation for block production and propagation.
Core comparison:
Figure 2 shows the core comparison of Proof of Work, Proof of Stake and Semada
Proof of Reputation (otherwise known within the Semada infrastructure as the Anchor
Protocol). While PoW necessitates slow and energy intensive hash mining, PoS
Fig. 2 Core comparison of PoW, PoS and PoR
enables some level of enhanced throughput through staking fungible currency that is
used to qualify for block propagation. The fungibility of currency is the main point
of attack on PoS and leads inevitably to centralization. Unlike traditional PoS, PoR,
e.g., Semada’s Anchor Protocol, uses reputation scores as a non-fungible currency to
qualify for block propagation. As such, the Anchor Protocol (PoR) is attack resistant,
fully decentralized, scalable and allows evolutionary protocol upgrades.
Figure 3 illustrates the staking mechanism in the Anchor Protocol and all Semada
DAOs. In T1, a DAO member has 100% of its DAO Reputation score at T1. When
staking at T2, The DAO member takes part of her Persona, e.g., reputation score in
that DAO, and stakes it either to qualify as a block producer or to participate in a
validation pool. After the validation pool was decided in T3, if the DAO member
sided with the winners of the validation pool, her reputation was enhanced from
100% in T1 to 120% in T2. This process repeats itself as long as the DAO member
actively participates on the Semada Platform by staking in her respective DAOs.
Anyone with any Anchor tokens has the potential to be a block producer. Semada
Core (pseudo) randomly selects the block producers weighted by their holdings,
meaning if you have more reputation, as evidenced by the Anchor holdings, you
are more likely to be selected. This gives anyone on the planet, at any time, the
opportunity to be a block producer.
Block production is scheduled ahead of time where a list of randomly selected
block producers is selected and weighted by their Anchor token holdings in the
specific DAO related to block production. When a block producer’s time comes, they
produce and publish the block, sending its reference and all fees included to Semada
Core to open a validation pool (betting pool) where the other SEM Token holders can
check that the block is created according to protocol and vote accordingly. Consensus
follows a slight variation on the GHOST protocol (greedy heaviest observed subtree)
that bitcoin and Ethereum use for proof of work, meaning the subtree with the greatest
weight of SEM Token-backed votes is canonical. Then all fees collected from the
block of transactions that are distributed in the reputation-weighted salary.
Fig. 3 Staking mechanism in the anchor protocol and all Semada DAOs
So a successful block producer doesn’t win the lottery with a great deal of fees if
they are chosen, they win the lottery of half of a great deal of new reputation tokens
if they are chosen while the rest of the members share the other half of newly minted
reputation tokens for policing the block in the validation pool. This makes the whole
system more stable (salary of fungible fees is regular and predictable as opposed to
winning the lottery) and gives better incentives (e.g., there is less reason to join a
mining pool, since all fungible fees are already shared in proper proportion, so it is
more decentralized).
Active participation is encouraged because new reputation tokens are minted in
every validation pool for every block. So block production is strongly encouraged
(because a greater percentage of the new tokens are given to a successful producer)
and policing is gently encouraged (you share in part of the new tokens with everyone
who is active, which encourages activity but doesn’t unstably slash non-participation
(which can be innocent if the network is down, etc.).
Producing bad blocks is slashed because the bad block producer will lose their
availability stakes (the tokens the producer staked to be considered for the random
selection of block producers) in the validation pool.
The computing resources are whatever it takes to make a block (no resources
needed to hash mine). Currently, making a block is very easy for a laptop or even a
good cellphone to do.
Fig. 4 Selection procedure for block producer in Semada proof of stake
Figure 4 in the Semada Protocol, random selection of experts is decided by relative

weight of reputation. Before a smart contract is engaged, experts have the opportunity
to stake SEM tokens to signal their availability for work. These availability stakes
will be added as the chosen experts’ upvote bet on their evidence-of-work post. In
this example, the yellow disk stops randomly along the bar, but is most likely to stop
on the 2nd expert.
Figure 5: After the random selection of the Anchor token staked block producer
as illustrated in Fig. 4, the selected block producer collects all valid SEM-token gas-
paying transactions in a block and publishes it. This proposed block is then sent to
the Anchor token-holder DAO validation pool. If the pool upvotes the block as valid,
it is added to the Semada blockchain. The collected SEM token fees are distributed
to all members of the DAO, as described below.
Figure 6 illustrates that after the block producer was awarded her SEM tokens for
successful and validated block production, as illustrated in Fig. 5, all Anchor token
holders in the Anchor Protocol participate in the SEM token distribution from the
block in proportion to their respective SEM token holdings (Fig. 7).
Fig. 5 Validation pool block verification
Fig. 6 Block production salaries

Anch Proportional SEM

or Anchor tokens Payout
Alice 404 1020 Total / 2.524752475 20 Total SEM / 7.92156863

404 Anchor = Proportional Anchor
=
Bob 111 1020 Total / 9.189189189 20 Total SEM / 2.17647059

=
Carlo 303 1020 Total / 3.366336634 20 Total SEM / 5.94117647

=
Donna 202 1020 Total / 5.04950495 20 Total SEM / 3.96078431

=
Total 1020 20
Fig. 7 Semada reputation salary calculation example
How does Joe earn tradeable currency on the Semada Platform? SEM tokens
(traded on binance) are the tradable currency used for and within transactions (includ-
ing as gas). Anchor tokens are the Semada block propagation reputation tokens
available to the Semada block propagation DAO members.
The Anchor tokens determine who has power in the system for deciding:
(i) who gets to make the blocks;
(ii) whether the blocks are valid in the validation pool, and;
(iii) what percentage of the SEM token denominated reputation salary the Anchor
token holders get.
The SEM tokens are the fungible (perfectly transferrable) currency of the system that
are used in transactions contained in blocks, like bitcoin or ether. SEM is added to
fees for each transaction by public users so block producers will add their transactions
to the next block. These fees determine how many new Anchor tokens are minted
for a block, then the fees are distributed to all Anchor holders in SEM denominated
reputational salary.
New Anchor tokens are created in proportion to the number of SEM denominated
transaction fees collected by the system in each block. Therefore, Anchor tokens are
highly inflationary. To maintain his relative power, Bob must continue to participate.
Each time a block is produced, the SEM Token denominated fees from the trans-
actions in the block are shared with the entire Anchor protocol DAO members as
SEM Token denominated reputation salary (SEM Token is earned for each block in
proportion to DAO members’ Anchor token holdings). If Bob owns 1% of Anchor
tokens, he earns 1% of all SEM token fees generated by transactions in the block
(if block 1 has 50 SEM tokens and block 2 has 100 SEM, Bob always makes 1% of
the SEM token fees in proportion to his 1% of total available Anchor tokens). Block
producers do not gain immediate SEM token fees, only more Anchor tokens. If Bob
doesn’t participate in block production, e.g. Bob does not participate in the valida-
tion pool or does not stake an availability stake to be chosen for block production,
he still shares the SEM token denominated reputation salary (for holding proportion
of Anchor tokens as an Anchor DAO member) from the DAO member validation
pool, but he does not share in the additionally minted Anchor tokens. Thus, Bob’s
relative power (as measured by his Anchor token holdings out of total Anchor tokens
in circulation) and SEM token salary decreases.
5.3.2 Sockpuppet Protocol
The basic Web of Trust (WoT) works as follows: Keep track of your network transac-
tions, and ask how well the people who are part of your transactions are satisfied with
the interaction. If they also have a big reputation and are satisfied with the transaction
then your reputation goes up. If you get a lot of these good transactions then you
have a good reputation. Early adopters who are all trustworthy. If someone behaves
badly their reputation will drop.
Here’s the problem. If I use a lot of sockpuppet accounts, I can raise my reputation
arbitrarily high, by behaving well for a while, then making a lot of transactions with
myself and rating myself high.
So the solution that is always offered is to control entry by identity verification. The
problem with this is, if the reputation is genuinely valuable, a sockpuppet account
can afford to go through whatever hoops you have in place to create false identi-
ties (including stealing biometric data if necessary), then increase their reputation
arbitrarily, as described above.
If the reputation is not valuable, then you won’t be able to get honest users to go
through the hoops required to identify themselves securely, since it’s not worth it.
So the only time WoT works is when the service is not valuable, such as PGP
(email which is essentially free). Then it’s not worth it to create sock puppet accounts,
so in that case you can trust the WoT network.
If you are trying to create an economic solution that is worth real money, where
you need to be able to trust the other people in the network will behave well and
follow protocol, you can’t just assume their historical good behavior will prove their
future behavior will also be good. Sockpuppet accounts allow us to game the system
in an automated manner and falsely create a valuable reputation, so we can leach
whatever value there is out of the system.
The voting algorithm of Semada core is designed to combat that very system, so
that it is not economically feasible to game the system without adding genuinely
valuable improvements, as proven with the fees that are added to the system and the
fair validation pool that every fee is subject to.
Fig. 8 Multiple sockpuppet accounts with 1 token each
In summary, the Web of Trust is a traditional attempt to create decentralized

reputation which is critically flawed and should not be used when fungible currency
is at stake:
(i) It counts the number of transactions that are positive/honest and how much each
member supports the other members (web of trust)—many DLT startups use
this approach;
(ii) However, sockpuppet accounts can grow their value much quicker in the web
of trust by validating each other. Honest users are much slower than the
sockpuppets validating each other. Hence, the system is flawed.
Here is how Semada fixes this web of trust sockpuppet flaw:

Figure 8 demonstrates that multiple sockpuppet accounts with 1 token each are
still equal to a higher token amount of the same DAO member with only one account.
In the Semada Persona Protocol, people, e.g., DAO members, can use sockpuppet
accounts but they are wasting their efforts. Semada uses validation pools to make any
change in rewards. All power comes from validation pools in relation to what DAO
members stake, and all fungible currency rewards are shared fairly with the group
in proportion to their individual reputation. The sockpuppet Protocol thus breaks the
incentives for Sybil attacks.
5.3.3 Persona Protocol
In the Semada Platform, anonymous DAO memberships define the reputation of their
members by the amount of tokens the members hold in the respective DAO. The more
DAO tokens members own the more the DAO, the system, and the platform respect
the DAO member. The token is based exclusively on meritocracy. Physicality, group
Fig. 9 Respectively aggregated DAO memberships individuals on the Semada platform
identification, race, culture language etc., are relatively superficial identifiers and do
not play the same role in Semada as they have society historically. Semada may be
seen as a forum for meritocracy, providing the most honest valuation of individual
actions and contributions to a DAO.
Figure 9 illustrates the respectively aggregated DAO memberships individuals on
the Semada Platform may have that define their Self Sovereign Identity on Semada
and ultimately the Internet. In other words, Semada users are identified by their
reputation scores in the respective DAOs they choose to become members of and
participate in by staking their respective DAO reputations. Other commonly used
identifiers such as social media provides, credit scores in centralized systems, etc.,
do not matter. The core identifiers are DAO token scores.
Different users may utilize such Persona Protocol Scores in different settings.
The use cases for persona protocol scores are very wide ranging. Semada DAO
scores are a conversation starter for people to engage with the platform and the
individuals on the platform. For example, if person A has a very high reputation
score in Semada Solidity Programming template Expertise tag/DAO, certain people
will want to engage with person A just because of A’s Semada Score in the respective
DAO. Over time, the Semada team will use the Semada reputation scoring to engage
with and emulate existing social identity networks.
From a social justice perspective, the Semada Persona Protocol takes power and
spreads it out much further to people who have no agency in the centralized systems.
In the existing systems very few audits exist that derive from the community itself.
In any Semada DAO, community audits are at the core.
To facilitate DAO member onboarding and enable fully verified external wallets,
SRI is developing proprietary mobile biometrics technology that allows users to
sign in with their biometrics into a particular wallet if they so choose. Semada uses
zero knowledge proofs for its biometrics onboarding technology that guarantees
continuing anonymity for users, once onboarded. Users who wish to forego biometric
identifiers via zero knowledge proofs on their mobile etc., devices can opt to become
Semada DAO members in a completely anonymized way.
5.3.4 Evolutionary Blockchain Governance Protocol
The Semada protocol uses a weighted directed acyclic graph for a precedent system
and DAO governance.
A graph is a collection of points, called vertices, with optional connections
between vertices, called edges. The vertices of our graph are the posts to the forum.
The edges are citations directed from one post (the referrer) to another (the refer-
ence). The edges give the graph direction, since references always point backward
in time. Since there can be no time loops of references, the graph has no cycles, so
it is acyclic. All in all, the forum is a directed acyclic graph (DAG) (Fig. 10).
In the Semada Protocol, evolutionary DAO governance is enabled by the Weighted
DAG precedent system. As Fig. 11 above demonstrates, more reputation weight and
salaries are allocated to posts on the Semada Forum that get continuously referenced
by other posts. As a precedent dissipates over time, new precedence in the Semada
Forum emerges naturally to replace older precedent. This is the essence of the evolu-
tionary nature of the platform. As a matter of fact, entire expertise tags and DAOs on
the Semada platforms can emerge and be replaced over time with new and improved
systems naturally within the existing architecture of the platform.
Fig. 10 Graph properties

Fig. 11 Weighted DAG Precedent system
5.3.5 Core Semada DAOs
Anchor DAO
The Anchor Protocol DAO is the Semada DAO responsible for block production.
Each time a block is produced, the SEM Token denominated fees from the transac-
tions in the block are shared with the entire Anchor DAO members as SEM Token
denominated reputation salary (SEM Token is earned for each block in propor-
tion to DAO members’ Anchor token holdings). As Anchor DAO members par-
ticipate in the block validation pools and join winning pools their Anchor token
increases. Conversely, if they more often lose in validation pools, their Anchor
holdings decrease.
SRN DAO
The Semada Research Network (SRN) DAO is the ideation and implementation
engine for SRI.Membership of the SRN DAO is contingent on being selected by the
existing SRN DAO members. Candidates for SRN DAO membership only qualify if
they posted valuable protocol upgrades that were upvoted by the existing members.
SRN DAO members get paid for their upvoted protocol contributions in both SEM
Tokens and SRN Tokens. SEM tokens are fungible and traded on cryptocurrency
exchanges. SRN Tokens are not fungible. SRN Tokens are the membership tokens
that grant access to the SRN DAO. Only the SRN Tokens are used for voting on
incoming protocol improvement proposals.
Arbitration DAO
Disputes that may arise pertaining to Semada products, investments, and services
will be governed by the Semada Arbitration DAO. The Arbitration DAO is decen-
tralized and organized using the core principles outlined above using the Sockpuppet
and Persona Protocols. Its evolutionary governance structure will change in time to
optimize its efficiency and effectiveness, but it will begin according to the structure
outlined in the Semada Technical Whitepaper.
Network Infrastructure for Decentralized Economies
Through its Internet of Trust protocols, SRI creates a network infrastructure for
decentralized economies. The four core infrastructure solutions SRI created to facil-
itate decentralized economies include the DAO infrastructure for a network that con-
tinuously creates consensus for propagating transactions, the associated verification
platform, the decentralized reputation economy, and the tokenization infrastructure.
DAO Infrastructure
Core to decentralized economies is a network that continuously creates consensus

for propagating transactions irrespective of the device or its hardware capabilities.
Over time, the Semada DAO Infrastructure will become the environment for hosting
decentralized companies.
SRI‘s protocols facilitate autonomous self-governing networks of DAOs.
A good example that helps illustrate the DAO improvements over existing Internet-
based platform companies is Uber, the ride-sharing platform company. In essence,
the DAO is a collective of people who decide to follow a certain protocol. Semada
is the platform that helps the members of DAOs adhere to their own protocols. The
Uber DAO on the Semada Platform can be seen as Uber the company with all its
constituents except without the company, e.g., the entity, itself and its hierarchical
governance structures. If Uber were a Semada DAO, the DAO collective of Uber
drivers would become Uber, e.g., a fully decentralized company without hierarchies.
The control and power over the Uber DAO would be completely in the hands of the
DAO Uber collective. Yet, the staking mechanisms in the Semada protocols make the
voting structure different than any previous attempts at creating liquid democracies.
Several core benefits are associated with the DAO infrastructure for decentralized
platform businesses. First and foremost, the pricing should not be determined by the
company Uber. The drivers and their passengers know best what pricing can work
among them. The pricing for rides would be determined by the DAO members and
their passengers. Hence, the marketplace is in the DAO itself. Second, the decentral-
ized network of DAOs is significantly more efficient because the centralized fees,
e.g., in Uber’s case 25%, can be shared between the drivers/DAO members and the
passengers. Third, the decentralized DAO business infrastructure is more efficient
because the drivers themselves know best how to assess other drivers as members of
the DAO. Fourth, the DAO collective enables unprecedented information symmetry
among all constituents because unlike centralized Internet-based platform businesses,

DAO members continuously, iteratively, and incrementally evaluate all information
pertaining to decision-making metrics and protocols that govern the DAO. More
complex DAO arrangements can be further governed by smart contracts. Because of
the information history of the smart contracts, users can further get paid for improving
smart contracts.
The combined efficiencies of DAOs over existing businesses suggest that DAO
businesses could effectively overcome collective action problems and outcompete
Internet-based businesses such as Uber.
Verification Platform
Through its DAO infrastructure, the Semada platform enables expertise tags, e.g.,
individual DAOs on the Semada platform. Such DAOs can be used to verify oth-
erwise uncertain or questionable information in both centralized and decentralized
businesses. Verification DAOs help create direct and conscious as well as indirect
and unconscious trust in the Internet.
Examples of the DAOs that could benefit from the unified frontend UI that
functions as Semada’s verification engine include but are not limited to the following:
• Oracle DAO for decentralized business and blockchain connectivity and interop-
erability;
• Wikipedia DAO. For example, compare Wikipedia today with Wikipedia on
Semada. The Semada Wikipedia DAO would allow all DAO members to vote on
each person’s contributions. This would create a self-policed ranking of valuable
contributions;
• Fake News DAO;
• Decentralized credit score DAO;
• Underwriting DAO;
• Rideshare DAO;
• Meetup DAO;
• Publication Board DAO.
Reputation Economy
SRI recognizes the shortcomings in Internet-based platforms’ attempts to use reputa-

tion. Based on that recognition, SRI has developed the core Semada Architecture as a
new platform that enables a decentralized reputation economy.Current Internet-based
platforms use reputation to improve their centralized control structures.10 By con-
trast, SRI‘s core protocols facilitate the infrastructure for a decentralized reputation
economy.
SRI‘s decentralized reputation solutions have several advantages over existing
Internet-based platforms’ attempts at utilizing reputation to improve their business
models:
10 See critique supra at Fig. 1 and accompanying text.

• Interoperable contextual reputation scores remove reputation data silos;

• Reputation scores and reputation personas, e.g., profiles, cannot be created by fake
Internet accounts. The Semada Sockpuppet Protocol enables that through its sybil
attack resistance
• Users build and actually own and control their respective reputations through the
respective Semada DAOs in a fully decentralized way. The system is designed to
avoid and disable any form of centralized corporate control over reputation;
• Decentralized reputation scores allow users to build reputation-based social capi-
tal. Since the users own their own scores, they can now use their reputation scores
to gain traction in various business contexts;
• Complete privacy through anonymity of scores;
• Users gain full reputation data autonomy.
These core benefits of Semada’s decentralized reputation solutions can create

unprecedented business models and new forms of economic output that facilitate
a decentralized reputation economy:
• Network Effects: are created by removing data silos via interoperable contextual
reputation scores, combined with users’ social capital, new network data, and
users’ self-sovereign identity;
• New Network Data: The reputation solutions create an unprecedented kind of
data SRI called Rich Context Reputation GRAPHS which create new networks
based on different kinds of reputation data;
• Social Capital: Users’ control over their reputation creates social capital that can
be applied and utilized in unprecedented settings, facilitating new business lines
and options for capital formation;
• Self-Sovereign Identity: Semada Persona Protocol Human Uniqueness Identifiers
enable users to control their own digital identity to identify or de-identify from the
network.
Tokenization Infrastructure
Asset tokenization is a necessary and inevitable part of the crypto evolution. The
existing crypto economy has produced several notable tokenization platforms. Yet,
an enterprise level user interface for asset tokenization is missing.
The Semada protocols provide an ideal infrastructure and technology solution
for tokenization which allow Semada to create an easy to bootstrap standard for
tokenization.
On the Semada platform each asset or service has its own tokens in a separate
DAO/expertise tag. Each asset DAO will have its own standards for monitoring and
policing the valuation of their asset, in line with the specific requirements of their
field. The Semada protocols safeguard these DAOs from sockpuppet attacks and 51%
attacks, ensuring productive collaboration on the level playing field of a decentralized
platform.
6 Conclusion
Lacking decentralized technology solutions are a core factor that helps explain the
decreasing trust in the Internet. A decentralized reputation network can help reverse
that trend to increase trust in the Internet and increase decentralized technology
adoption. As a foundational technology, decentralized reputation networks have the
potential to create lasting change in the Internet technology infrastructure. Many
infrastructure solutions will be required until such potential can materialize.
References
Buterin V (2018) Twit about Gig Economy https://twitter.com/VitalikButerin/status/

1023923633175945222. Accessed 27 May 2019
Fortune (2016) How hackers make money from DDoS attacks http://fortune.com/2016/10/22/ddos-
attack-hacker-profit/. Accessed 27 May 2019
McKinsey & Company (2015) The internet of things: mapping the value beyond the hype (2015)
https://www.mckinsey.com/~/media/McKinsey/Business%20Functions/McKinsey%20Digital/
Our%20Insights/The%20Internet%20of%20Things%20The%20value%20of%20digitizing%
20the%20physical%20world/The-Internet-of-things-Mapping-the-value-beyond-the-hype.ashx.
Accessed 27 May 2019
Luhmann N (1979) Trust and power. Wiley, Chichester
Nordrum A (2016) Popular internet of things forecast of 50 billion devices by
2020 is outdated https://spectrum.ieee.org/tech-talk/telecom/internet/popular-internet-of-things-
forecast-of-50-billion-devices-by-2020-is-outdated. Accessed 27 May 2019
PEW Research (2014) http://www.pewresearch.org/fact-tank/2014/03/07/6-new-findings-about-
millennials/. Accessed 27 May 2019
Resnick P, Zeckhauser R (2001) Trust among strangers in internet transactions: empirical anal-
ysis of eBay’s reputation system https://cseweb.ucsd.edu/groups/csag/html/teaching/cse225s04/
Reading%20List/E-bay-Empirical-BodegaBay.pdf. Accessed 27 May 2019
Legal Issues Related to Blockchain
Technology—Examples from Korea
Gyooho Lee
Abstract Blockchains are used to record all transactions into online ledgers, which
are subsequently saved onto many blocks across the Internet. The popularity of
cryptocurrencies, which make use of blockchain technology, has made the public
interested in blockchain technology. Blockchain technology, however, is used in
many fields, including cryptocurrencies. This chapter deals with two issues raised by
the Korean legal community in terms of blockchain technology. One is intellectual
property-related issues of open source software on which blockchain technology is
based. The other issue concerns how to guarantee the authenticity of e-Appostilles
by using blockchain technology. The lack of authenticity of e-Appostilles is a big
legal hurdle which prevents e-Appostilles from gaining popularity in many countries.
Blockchain computing is a good solution for guaranteeing the authenticity of e-
Appostilles. In this regard, the chapter illustrates an inherent legal issue of blockchain
technology and blockchain technology as a method to solve a current legal problem.
At the outset, this chapter indicates that blockchain technology based on open source
software can be subject to intellectual property-related issues. Also, this chapter
ensures that blockchain technology can enhance the world-wide application of e-
Appostilles. However, it is necessary for each country to employ the technology
which complies with its domestic law on digital signatures, on electronic documents,
and on public notaries prior to application of the new technology. In cases where
the underlying public document is paper, compared to a digital file and overseas
Koreans request issuance of e-Apostilles, the issuing authority’s legal system needs
to be adopted to issue English electronic public documents as its underlying public
document. Otherwise, overseas Koreans should have to notarize the translated version
of paper-typed public documents. It is an inefficient and cumbersome process. Also,
this paper illustrates that blockchain technology can establish a decentralized and
secure system of e-Register for e-Apostilles by preventing each Contracting Party
to the Apostille Convention from joining unified e-Register system advised by an
international organization.
G. Lee (B)
Institute for Culture, Media, and Entertainment Laws, School of Law, Chung-Ang University,
Seoul, Republic of Korea
e-mail: cion2004@hanmail.net

https://doi.org/10.1007/978-981-15-1350-3_9
150 G. Lee
Keywords Blockchain technology · E-Apostilles · Open source software · Trade

secret · Copyright
1 Introduction
Blockchains are used to record all transactions into online ledgers, which are saved
onto many blocks across the Internet. Hence, blockchains are inherently hard to mod-
ify, and recorded transactions are permanent due to the redundancy and accountabil-
ity of their recording systems.1 Blockchains have decentralized,2 consensus-based,3
secure,4 and trustless features.5
The popularity of cryptocurrencies, which make use of blockchain technology,
has made the public interested in blockchain technology. Blockchain technology,
however, is used in many fields, including cryptocurrencies. For example, some
United States taxes might become payable by blockchain-based cryptocurrencies.6
A New Hampshire bill required acceptance of Bitcoin, inclusive of payment of state
taxes.
Also, blockchain technology can create a decentralized, reliable, secure decision-
making process for an organization, which is encoded directly into source code.7 In
addition, blockchain technology can be used in voting.8
This chapter deals with two issues raised by the Korean legal community in terms
of blockchain technology.
One is intellectual property-related issues of open source software on which
blockchain technology is based. When it comes to open source software, one piv-
otal court case needs to be explored. It is related to conflicts between patentee and
licensee of open source software and between the possessor of a trade secret and the
copyright owner of open source software.
The other issue concerns how to guarantee the authenticity of e-Appostilles by
using blockchain technology. The lack of authenticity of e-Appostilles is a big legal
hurdle which prevents e-Appostilles from gaining popularity in many countries.
Blockchain computing is a good solution to guarantee authenticity of e-Appostilles.
The chapter explores an IP-related issue which inherently concerns blockchain
technology and proposes how to ensure authenticity of e-Appostilles by using
blockchain technology.
In sum, the chapter illustrates an inherent legal issue of blockchain technology
and blockchain technology as a method to solve a current legal problem.
1 Galavis (2019), p. 563.

2 See, e.g., Jeffery (2017).
3 See, e.g., Hammerschmidt (2017).
4 See Miles (2017).
5 Kasireddy (2018).
6 2015 Bill Text NH H.B. 552.
7 Ehrsam (2017).
8 Wright and De Filippi (2015), pp. 27–28.
Legal Issues Related to Blockchain Technology—Examples from Korea 151
2 IP Law Issues Concerning Free and Open Source

Software
2.1 Blockchain Technology and Open Source Software
Most of the original blockchain developers rely on free and open source software.9
That is, a legal issue related to open source software is directly or indirectly connected
with blockchain technology. In this context, the following case needs to be examined.
2.2 Patented Blockchain Technology and Free and Open

Source Software (FOSS)
If a blockchain technology is protected by patent, the licensee will need to get per-
mission from the patentee through a term in the license before implementing the
technology, irrespective of whether it is open source software or not. However, the
issue of patents can be more complicated because of FOSS licenses. Furthermore,
since the rapid increase in software patents recently has made these complications
more common, patent-related provisions are increasingly found in the newer versions
of open source software licenses.10
Blockchain developers are likely to be exposed to liability by patent grants on
FOSS. 11 Many Blockchain developers who do not have full legal knowledge are
likely to include third party software into FOSS and subsequently disclose the source
code, making them vulnerable to patent infringement actions brought by the third
party patentees.12
As for case law concerning the relationship between patents and FOSS, a good
illustration is Daum Communication, Corp. v. DidioNet, Corp.13 This case is directly
applicable to blockchain technology .
In Daum Communication, Corp. v. DidioNet, Corp.,14 the plaintiff, Daum, devel-
oped a program called PotEncoder using open source software called the X.264
Codec. DidioNet, the defendant, had a patent similar to the open source software.
The plaintiff brought an invalidation action against the defendant before the Korea
Intellectual Property Tribunal. The case was decided in favor of the defendant. On
9 Sung (2018), p. 72.

10 Korea Copyright Commission, Open Source SW License Information Systems. Available at:
https://www.olis.or.kr/en/LicenseIntroduction.do. Accessed 30 June 2019.
11 Lee (2007), p. 408; Lee et al. (2012), p. 15.
12 Lee (2006), pp. 953–956.
13 Korean Patent Court Decision on August 29, 2013, pp. 1–42 (Case No. 2013 Heo 1023).
14 Korean Patent Court Decision on August 29, 2013, pp. 1–42 (Case No. 2013 Heo 1023).
152 G. Lee
appeal, the plaintiff submitted new evidence showing that the patent was invalid and
the Korean Patent Court invalidated the defendant’s patent.15
2.3 A Clash Between Trade Secret and Copyright
Blockchain developers based on FOSS may end up with copyright-related disputes

even though they hold trade secrets on their derivative works based on FOSS. The
following Korean case shows a copyright issue with which a blockchain developer,
a possessor of a trade secret, can face in the near future.
In ElimNet, Inc. v. HaionNet, Inc.,16 the plaintiff had developed a derivative work
based on GPL-licensed software and subsequently held this as a trade secret in
violation of GPL’s terms. The defendant, HaionNet, created another derivative work
on basis of the plaintiff’s derivative work. In ElimNet, the issues regard whether the
defendant violated the plaintiff’s trade secret and whether the plaintiff violated the
GPL and the underlying copyright in the GPL-based software. The first issue was
settled by the court’s judgment mentioned below and the latter was settled by parties’
agreement.17
In what was the first civil case involving GPL license, the Korean Supreme Court
held that “the source code of a modified computer program can be a trade secret
in the case where the copyright owner of the derivative computer program, which
amended the virtual private network OSP to which the GPL condition was linked,
refused to disclose the source code of the modified computer program in violation of
the General Public License.”18 ETUND 1.00 was the revised version of VTUND19
which was initially created by Maxim Kransnyansky and disclosed under the GPL.
As long as ETUND 1.00 and ETUND 1.0420 demonstrated originality as the deriva-
tive works of VTUND, the source code of ETUND 1.04 could be considered a trade
15 Korean Patent Court Decision on August 29, 2013, pp. 3, 13 (Case No. 2013 Heo 1023).
16 Korean Supreme Court Decision on February 12, 2009, pp. 1–4 (Case No. 2006 Do 8369).
17 Dong and Kim (2010), p. 15.
18 Dong and Kim (2010), p. 15.
19 VTUN is the open source software which improves network transfer speed by implementing
channel bonding.
20 ETUN is a VPN software mainly developed by defendant A together with B, C, and D under the
employment of ElimNet from September 2002 to March 2004 based on the VTUN. A accepted a
suggestion by HaionNet to join its VPN business on October 2004 and turned VTUN. HL which
adds some improvement of ETUN over to an employee of HaionNet on November 2004. Afterward,
A resigned from ElimNet without transferring the source code of ETUN to ElimNet while keeping
the copy personally and by the request of the employee of HaionNet fixed some bug of ETUN and
named it HAI, and the employee of HaionNet commenced the commercial service using HAI since
January 2005. While ElimNet and HaionNet are the only companies providing VPN service based
on VTUN code in Korea, and HaionNet’s HAI and ElimNet’s ETUN share much of their source
codes, HaionNet dishonestly advertised on its homepage and in business promotion that HAI to
was an innovative technology developed on June 2004 on its own.
secret of its creator because the source code was not known to the public, had inde-
pendent economic value, and was preserved as a secret by the substantial efforts of
its creator.21 In ElimNet, the plaintiff won the case on the basis of the defendant’s
violation of the trade secret. However, in accordance with the obiter dictum in the
ElimNet case,22 the copyright owner of ETUND 1.04 could still be held liable for
copyright infringement against the copyright owner of VTUND due to the fact that
refusal to disclose the source code of ETUND 1.04 (i.e., the trade secret) would
constitute a violation of the obligations in the GPL.
2.4 Copyleft Provisions and Competition Law Issues of FOSS

Licenses
When it comes to blockchain technology based on FOSS, copyleft provisions and

competition law issues of FOSS licenses that need to be taken into account.
According to copyleft provisions, licensees may modify the covered program
or any part thereof, thus making a work based on the program, provided that any
work that they distribute or publish, which in whole or in part contains or is derived
from the covered work or any part thereof, is licensed as a whole at no charge to
all third parties under the terms of the same license. Furthermore, the licensee may
copy and distribute the FOSS in object code provided the license accompanies it
with the complete corresponding machine-readable source code. Hence, Samsung
Electronics Co., Ltd.23 and LG Electronics Co., Ltd.24 provide the source codes of
their goods based on open source software, such as cell phones and TVs. 25 These
copyleft provisions may be considered anti-competitive according to the Monopoly
Regulation and Fair Trade Act as they could potentially result in cartel effect due to the
strong ‘viral’ nature of their provisions.26 Open source companies can produce better
software than closed source firms due to cost sharing. However, in reality, the cost
sharing can create a de facto quality cartel which encourages open source companies
not to invest their resources at the early stage of their business.27 If completely
differentiated, all open source companies can monopolize their submarkets through
their cost sharing. On the other hand, if they can substitute each other perfectly, the
market for open source software becomes highly competitive, so that this situation
21 Korean Supreme Court Decision on February 12, 2009, p. 4 (Case No. 2006 Do 8369).
22 Korean Supreme Court Decision on February 12, 2009, p. 4 (Case No. 2006 Do 8369).
23 Samsung Open Source Release Center, available at http://opensource.samsung.com/ (accessed
on June 22, 2019).

24 LG OpenSource CodeDistribution. Available at: http://www.lg.com/global/support/opensource/
opensource.jsp. Accessed June 22, 2019.

25 Lee et al. (2012), p. 23.
26 Sohn (2012), pp. 87 and 114.
27 Offerman, Open versus Closed Source: A Delicate Balance. Available at: https://joinup.ec.europa.
eu/elibrary/case/open-versus-closed-source-delicate-balance.
154 G. Lee
can decrease benefits and incentives in quality improvements, resulting in a quality

cartel.28
3 e-Apostilles Project Based on Blockchain Technology
3.1 Apostilles Convention
The HCCH Convention of 5 October 1961, on Abolishing the Requirement of Legal-

ization for Foreign Public Documents (hereinafter “Apostille Convention”), provides
the circulation of public documents executed in one Contracting Party to the Con-
vention and to be submitted in another.29 It has been effective since January 24,
1965. As of September 12, 2018, the number of Contracting Parties to the Apos-
tille Convention is 117. The Republic of Korea (hereinafter “Korea”) acceded to
the Apostille Convention on October 25, 2006 and it has been effective in Korea
since July 14, 2007.30 Chain certification of public documents, which require the
costly and time-consuming formalities of a full legalization process of public docu-
ments, is substituted with the mere issuance of an apostille among the Contracting
Party to Apostille Convention.31 Apostille Convention is applicable only to public
documents.32
3.2 Apostilles from the Republic of Korea
The issuance of apostilles in Korea has increased 241% in 2017 as compared to

2009.33 Sixty-four percent of apostilles have been issued for notarized documents.34
Almost all public documents are written in Korean. Generally, foreign government
officers request a translated version of a public document (Fig. 1).35
28 Offerman, Open versus Closed Source: a Delicate Balance. Available at: https://joinup.ec.europa.
eu/elibrary/case/open-versus-closed-source-delicate-balance. Accessed 24 June 2019.

29 HCCH, HCCH Convention of 5 October 1961 Abolishing the Requirement of Legalisation for
Foreign Public Documents, p. 1. Available at: https://assets.hcch.net/docs/80d0e86f-7da8-46f8-
8164-df046285bcdd.pdf. Accessed 20 July 2019.
30 HCCH, Status Table: Convention of 5 October 1961 Abolishing the Requirement of Legalisa-
tion for Foreign Public Documents. Available at: https://www.hcch.net/en/instruments/conventions/

status-table/?cid=41. Accessed 20 July 2019.
31 HCCH, Status Table: Convention of 5 October 1961 Abolishing the Requirement of Legalisa-
tion for Foreign Public Documents. Available at: https://www.hcch.net/en/instruments/conventions/

status-table/?cid=41. Accessed 20 July 2019.
32 Article 1 of Apostille Convention.
33 Lee et al. (2017), p. 39.
34 Lee et al. (2017), p. 39.
35 Article 1 of Apostille Convention.
100,000
80,000
60,000
40,000
20,000
2009 2010 2011 2012 2013 2014 2015 2016 2017
Apostilles for Public Documents

Apostilles for Notarized Documents (Notarial for Translation)
Fig. 1 Apostilles for public documents in Korea (source Lee 2017)
On November 30, 2016, the Korean government opened an online apostille Service
to promote the viability of apostille services.36 Most of public documents can be
issued online. In 2016, more than 65,000,000 public documents had been issued and
printed out at home (Fig. 2).
The apostille system is able to check the validation of an underlying public
document in real time while it is connected with other organization’s system (Fig. 3).
Over 4000 apostilles for public documents are issued every month.37 Eighteen
percent of apostilles are issued within a fully automated process via the Internet.
Issuing the document at home Automated public document issuance machine
Fig. 2 Issuance of public document at home and an automated machine for issuance of public
documents (source Lee 2017)
36 Ministry of Foreign Affairs, Republic of Korea e-Apostille Service. Available at: http://www.
apostille.go.kr. Accessed 20 July 2019.
37 Lee et al. (2017), p. 39.
156 G. Lee
STEP. 1 STEP. 2 STEP. 3
Applicant Applicant Applicant

1. Select an Organization 1. Input Information 1. Print out apostille
2. Input the Issued (e-mail, Phone no)
Number of underlying 2. Select an Destination
document. Country
Fig. 3 Online apostilles (source Lee 2017)
6,000
5,000
Online Apostilles
4,000
Offline Apostilles
3,000
2,000
1,000
Dec. 2016 Jan. 2017 Feb. 2017 Mar. 2017 Apr. 2017 May. 2017 Jun. 2017
Fig. 4 Number of issuance of apostilles (source Lee 2017)
Koreans can get an Apostille and its underlying public document abroad. Govern-
mental officers in foreign country can verify the authenticity of the apostille on the
Internet (Fig. 4).
The issuance of apostilles via online abroad are also reliable due to the fact
that an e-Register service is available. Overseas Koreans can save time and money
in obtaining an apostille for public documents (e.g., family certificate). It will take
30 min to get an Apostille for family certificate, marriage certificate, criminal records
and other public documents, including the sign up process. Foreign government
officers can easily figure out how to verify apostilles (Figs. 5 and 6).
Verification of over 1000 Apostilles is requested by foreign countries or parties
concerned every month as shown in Table 1.
Top 10 countries Top 20 countries

offline apostille online apostille
1 France 4,550 1Top 20 countries
France 1,047 11 U.K. 91
online apostille
2 Germany 3,178 2 U.S.A 975 12 Japan 82
3 Spain 2,528 3 Germany 846 13 Italy 71
4 Russia 1,437 4 Mexico 250 14 Swiss 70
5 U.S.A 1,312 5 Austria 238 15 New Zealand 70
6 Czech 1,242 6 Netherlands 161 16 Kazakhstan 69
7 India 897 7 Czech 159 17 Belgium 67
8 Mexico 826 8 Russia 128 18 India 49
9 Italy 793 9 Spain 110 19 Dominica 44
10 Kazakhstan 715 10 Australia 104 20 Hong Kong 39
Fig. 5 Online apostilles of top 20 countries (source Lee 2017)
1 Advanced Lv. e-Register 2 QR-Code 3 2D Barcode
Fig. 6 Verification process of apostilles (source Lee 2017)
Table 1 Number of success

Date Number of success of Number of failure of
and failure of verification of
verification verification
apostilles (source Lee et al.
2017, p. 42) 2017-11 1140 241
2017-10 1065 176
2017-09 1085 198
2017-08 1309 204
2017-07 1278 145
2017-06 1819 129
2017-05 1319 106
2017-04 1308 103
2017-03 1594 144
2017-02 1295 116
2017-01 1108 109
2016-12 1449 183
158 G. Lee
3.3 Legal Issues Related to e-Apostilles
3.3.1 Failure to Offer e-Apostilles
The Korean government has failed to offer e-Apostille (Digital File Type) service due
to one legal obstacle under the Digital Signature Act and the other under the Frame-
work Act on Electronic Documents and Transactions even though the e-Apostille sys-
tem assists foreign governments and overseas Koreans in easily verifying e-Apostilles
as compared to online apostilles, which still mandate a separate verification process.
Under the Digital Signature Act, a re-signing problem is raised by the public sec-
tor in Korea. When the Ministry of Foreign Affairs issues e-Apostilles with digital
signatures, the digital signature of its underlying public documents such as criminal
records will disappear. This issue is called “re-signing problem.” Also, a scanned
underlying public document is not regarded to be an original public document under
the Framework Act on Electronic Documents and Transactions.
In addition to the two legal problems, several technical issues arise in terms of
e-Apostilles:
i. Necessity for a standard specification for supporting paper and digital types of
apostilles at the same time;
ii. How to implement an advanced level of e-Register for paper apostilles;
iii. How to deliver information contained in paper Apostilles to other countries;
iv. How to secure its system from cracking (security); and,
v. How to comply with domestic law on personal information protection.
In particular, this chapter focuses on legal issues, proposing that blockchain
technology can solve the legal problems.
3.3.2 Re-signing Issue Under Digital Signature Act
The term “digital signature” means a piece of information in digital form affixed on,
or logically combined to, an electronic message in order to identify the signer and
verify that the electronic message has been signed by that signer.38 The term “certified
digital signature” means a digital signature that satisfies the following requirements
and is grounded upon an authorized certificate:
(a) That the digital signature creating key shall be only held by and known only to
the subscriber;
(b) That the subscriber shall be controlling and managing the digital signature
creating key at the time of signing;
(c) That it shall be ascertained whether there has been any alteration in the digital
signature concerned since it was affixed; and,
38 Article 2, subparagraph 2 of Digital Signature Act (Act No. 14839, July 26, 2017).
(d) That it shall be ascertained whether there has been any alteration in the electronic
message concerned since digital signature was affixed.39
The certified digital signature must meet the foregoing requirements.
An e-Apostille needs a digital signature of its competent authority. However, e-
Apostilles break the digital signature of the issuing authority as shown in Figs. 7 and
8.
Apostille along with valid digital signature of its competent authority in a Con-
tracting State to Apostille Convention is valid but the underlying public document
does not have any valid digital signature of its issuing authority. The underlying
public document is not a lawful public document in Korea. Is there any way how a
combination of Apostille allonge and its underlying public document is lawful? A
Contracting State to the Apostille Convention must have its domestic law deeming
the combination of Apostille allonge and its underlying public document as lawful.
Apostille Allonge
Underlying Doc.
Sign A
Sign A Sign A
Natural born electronic or scanned E-Apostille needs an e-signature

underlying document with e-signature of its competent authority, but it
of issuing authority of a State. breaks the e-signature of the issuing
authority.
Fig. 7 Problem of re-signing for e-Apostilles (1) (source Lee 2017)
An e-Apostille file
Apostille allonge with a valid e -signature of its
competent authority in contracting state is valid.
But its underlying document does not have a

valid e-signature of its issuing authority.
It is not a lawful document under Korean law.
Sign A
Sign A Is a combination of an apostille allonge and its

underlying document lawful?
Fig. 8 Problem of re-signing for e-Apostille (2) (source Lee 2017)
39 Article 2, subparagraph 3 (a) to (d) of Digital Signature Act.

160 G. Lee
Ministry of Justice provides an electronic notarial document

Notarial Doc. with a notary’s e-signature guaranteed by government.
And its records are fully managed by online service of the

Ministry of Justice.
Notary’s e-sign Sign A
Ministry of Foreign Affairs can not attach an electronic

Apostille Allonge
apostille allonge to its e-notarial document.
Because it will invalidate the e-signature of the notary.
And the Ministry of Foreign Affairs is not permitted to

MOFA’s e-sign Sign A make a legal document through a scanning process.
Fig. 9 Possible scenario of the re-signing problem (source Lee 2017)
In this regard, under the Korean Digital Signature Act, the combination of Apostille
allonge and its underlying public document can not be lawful.
For example, the Ministry of Justice provides an electronic notarial document with
notary’s e-signature guaranteed by government and its public records are fully man-
aged by online service of the Ministry of Justice. The Ministry of Foreign Affairs can
not attach electronic Apostille allonge to e-notarial public documents. It is because it
will invalidate the e-signature of the notary. Also, the Ministry of Foreign Affairs is
not permitted to make a legal public document through a scanning process (Fig. 9).
3.3.3 The Authenticity of Scanned Public Documents Under

the Framework Act on Electronic Documents and Transactions
Under the Framework Act on Electronic Documents and Transactions, The term
“electronic document” means information, prepared, transmitted, received, or stored
in an electronic form by an information processing system.40 No electronic document
shall be denied legal effect as a document solely because it is in an electronic form,
except as otherwise expressly provided for in other Acts.41 Matters concerning dig-
ital signatures in electronic transactions shall be governed by the Digital Signature
Act.42 The content of electronic documents stored in a certified electronic document
center shall be presumed unmodified during the period of storage.43 Where a certified
electronic document center issues a certificate of matters on the storage, an origina-
tor, an addressee and the date and time of transmission and receipt of an electronic
document stored in the relevant certified electronic document center according to
methods and procedures prescribed by Presidential Decree, matters stated on the
40 Article 2, subparagraph 1 of the Framework Act on Electronic Documents and Transactions.

41 Article 4, paragraph 1 of the Framework Act on Electronic Documents and Transactions.
42 Article 11 of the Framework Act on Electronic Documents and Transactions.
43 Article 31-7, paragraph 1 of the Framework Act on Electronic Documents and Transactions.
certificate shall be presumed true and correct.44 This Act shall also apply to foreign-
ers and foreign corporations: Provided, That with respect to foreigners or foreign
corporations of a State which does not provide protection corresponding to this Act
to citizens or corporations of the Republic of Korea, protection under this Act or
treaties to which the Republic of Korea acceded or concluded by the Republic of
Korea may be restricted commensurately therewith.45
3.3.4 Main Concepts of Another Way for e-Apostille by Using

Blockchain Technology
According to a Handbook on the Practical Operation of the Apostille Convention

published by the Hague Conference on Private International Law Permanent Bureau
in 2013, “Competent Authorities may employ a variety of methods to “attach” an
e-Apostille by logically associating it with the underlying public document. In the
case of e-apostilles issued using PDF technology, the e-Apostille may be attached
by incorporating the e-Apostille and electronic public document into a single PDF
document. Alternatively, the e-Apostille may be attached to the electronic public doc-
ument file as a separate file (although in practice it is the electronic public document
that is attached to the e-Apostille).”46 In particular, this opens a door for a Contracting
Party to the Apostille Convention to legalize a combination of e-Apostille allonge
and its underlying public document by stating that “Alternatively, the e-Apostille
may be attached to the electronic public document file as a separate file (although in
practice it is the electronic public document that is attached to the e-Apostille).”47
Apostilles should not be detached from the underlying public document (Fig. 10).48
Even though an e-Apostille package is composed of separate files, it is permitted
to be legalized when a Contracting Party to the Apostille Convention can legalize the
44 Article 31-7, paragraph 2 of the Framework Act on Electronic Documents and Transactions.
45 Article 40 of the Framework Act on Electronic Documents and Transactions.
46 The Hague Conference on Private International Law Permanent Bureau (2013), p. 63, paragraph
270.
271 (“For a multi-page document, the Apostille should be placed on the signature page of the
document. If an allonge is used, this should be affixed to the front or the back of the document (see
C&R No 17 of the 2003 SC). For practical reasons, an Apostille should be placed on the underlying
public document in a way that does not conceal the matters being certified (e.g., the signature), or
any content of the document.”); The Hague Conference on Private International Law Permanent
Bureau (2013), p. 64, paragraph 272 (“If attaching the Apostille to a particular document is not
practical (or indeed not permitted by the law of its State), the Competent Authority may wish to
instruct the applicant to obtain a certified copy of the document to be apostillized instead.”).
273 (“Competent Authorities should inform applicants that the Apostille must remain attached
to the underlying public document. In particular, they should advise applicants wishing to make
photocopies of apostillized documents that detaching the Apostille from the underlying public
document invalidates the Apostille.”).
162 G. Lee
Package typed e-Apostille e-Register supported by blockchains
Proposal at 10th e-APP Forum
Fig. 10 Package of e-Apostille (source Lee 2017)
e-Apostille Package A PDF file

with e-signature
Underlying
Document
Sign A
E-Sign A
allonge
A PDF file
with e-signature
Sign A
metadata
Sign A
E-Sign A
Fig. 11 Basic structure of e-Apostille package (1) (source Lee 2017)
e-Apostille package. An e-Apostille package is composed of an Apostille allonge

and its underlying document file (Figs. 11 and 12).
E-Apostille package’s metadata has information on underlying document and
apostille allonge. Metadata (machine readable data) can be delivered on the Internet
(Fig. 13).
InterPARES Trust suggested to apply blockchain technology to e-Apostille on
November 1, 2016 at 10th e-APP Forum. Blockchain technology can solve the
following research questions:
XML type metadata for e-Apostille

E-Apostille package
<Apostille>
<title>
Underlying APOSTILLE (Convention de La Haye du t octobre 1961)
Document </title>
<underlying document>
<country> Republic of KOREA </country>
Sign A <signed by>System Operation Officer< </signed by>
<capacity of>Supreme Court of KOREA< </capacity of>
<stamp of>Supreme Court of KOREA< <stamp of>
</underlying document>
allonge <competent authority>
<at>Seoul< </at>
<the>2017.Sept.03< </the>
<by>Ministry of Foreign Affairs< </by>
Sign A <no>XXA2017G2VKOR< </no>
<stamp>Ministry of Foreign Affairs< </stamp>
<sign>Kim Byung Ho< </sign>
</competent authority>
<additional information>
metadata <apostille type> digital file </apostille type>
<e-register> advanced </e-register>
</additional information>
</Apostille>
Sign A
Fig. 12 Basic structure of e-Apostille package (2) (source Lee 2017)
E-Apostille
package
Paper-typed
Paper
Underlying
documents
Document
Paper
Allonge
XML type metadata for a paper-typed apostille
<Apostille>
text information of underlying document and apostille allonge
metadata <additional information>

<apostille type> Paper </apostille type>
<e-register> basic level </e-register>
</additional information>
Sign A </Apostille>
Fig. 13 How to support for paper-typed apostille (source Lee 2017)
i. Can the data be trusted?;

ii.Can the records from which the data are derived be trusted or even traceable?;
iii.Are digital records complete? Are they authentic?;
iv. How were they generated and by whom (human, computer, program, or
protocol)?;
v. How are digital records stored and under what jurisdiction?; and
vi. Who has access to digital records? How secure are they?
164 G. Lee
In this context, blockchain technology is applicable to e-Apostille package tech-

nology. Also, blockchain technology is applicable to paper-typed public documents.
In addition, blockchain technology can guarantee security of the e-Apostille system
of each Contracting State to Apostille Convention and be applied for non-repudiation
policy on issuance of e-Apostille. Foreign government officers can verify an e-
Apostille with a blockchain hash code. Under the Korean Act on Personal Infor-
mation Protection, to transfer personal information of an individual to a foreign
country without his/her permission is strictly prohibited, for example, unless there
is international agreement between Korea and the foreign country. Hash code does
not contain personal information. In this regard, blockchain technology can protect
personal information and solve re-signing problem (Fig. 14 and Table 2).
State 1
State 2
State 3
hash code of issued number

of normal apostille
hash code of metadata of e-Apostille package

for paper-typed documents
hash code of e-Apostille file

hash code of e-Apostille package file
Fig. 14 Blockchain technology with e-Apostille package (source Lee 2017)
Table 2 Hash codes of e-apostille package (source Lee 2017)

Hash code of e-register category Information displayed
Issued number Basic “Yes”/“No”
Metadata Additional “Yes”/“No” + information on Apostille and/or
underlying document (possibly visual check)
e-apostille package Advanced “Yes”/“No” + information on Apostille and/or
e-apostille file underlying document (possibly visual check) +
digital verification of Apostille and/or underlying
document
4 Conclusion
Blockchain technology can enhance the world-wide application of e-Appostilles.

However, it is necessary for each country to employ the technology which complies
with its domestic law on digital signatures, on electronic documents, and on public
notaries prior to application of the new technology. In cases where the underlying
public document is paper, compared to a digital file and overseas Koreans request
issuance of e-Apostilles, the issuing authority’s legal system needs to be adopted to
issue English electronic public documents as its underlying public document.
Otherwise, overseas Koreans should have to notarize the translated version of
paper-typed public documents. It is an inefficient and cumbersome process.49
Also, blockchain technology can establish a decentralized and secure system
of e-Register for e-Apostilles by preventing each Contracting Party to the Apos-
tille Convention from joining unified e-Register system advised by an international
organization.50
In addition, it should be noted that blockchain technology based on open source
software can be subject to intellectual property-related issues.
References
Dong J-S, Kim S-H (2010) Understanding of open source software license. Copyright Culture
194:11
Ehrsam F (2017) Blockchain governance: programming Our future, MEDIUM. https://medium.
com/@FEhrsam/blockchain-governance-programming-our-future-c3bfe30f2d74
Galavis J (2019) Blame it on the blockchain: cryptocurrencies boom amidst global regulations. U
Miami Int Comp L Rev 26:561
Hammerschmidt C (2017) Consensus in Blockchains. In Short, MEDIUM. https://medium.com/
@chrshmmmr/consensus-in-blockchain-systems-in-short-691fc7d1fefe. Accessed 9 Aug 2019
HCCH, HCCH convention of 5 October 1961 abolishing the requirement of legalisation for
foreign public documents. Available at https://assets.hcch.net/docs/80d0e86f-7da8-46f8-8164-
df046285bcdd.pdf. Accessed 10 Aug 2019
HCCH, Status table: Convention of 5 October 1961 abolishing the requirement of legalisation for
foreign public documents. Available at https://www.hcch.net/en/instruments/conventions/status-
table/?cid=41. Accessed 10 Aug 2019
LG OpenSource CodeDistribution, http://www.lg.com/global/support/opensource/opensource.jsp.
Jeffery A (2017) US SEC denies a second application to list Bitcoin product, CNBC. https://
www.cnbc.com/2017/03/29/us-sec-denies-a-second-application-to-list-bitcoin-product.html.
Accessed 8 Aug 2019
Kasireddy P (2018) ELI5: what do we mean by “blockchains are trustless”? MEDIUM.
https://medium.com/@preethikasireddy/eli5-what-do-we-mean-by-blockchains-are-trustless-
aa420635d5f6. Accessed 10 Jun 2019
Korea Copyright Commission, Open source SW license information systems. https://www.olis.or.
kr/en/LicenseIntroduction.do. Accessed 10 Jun 2019
49 Lee et al. (2017), at 44.

50 Id.
166 G. Lee
Lee C-N (2006) Legal implication on software patent on open source software. Admin Law 17(1–
2):953–956
Lee C-N (2007) Necessity for patent protection of software. Ind Property Right 23:395
Lee C-N et al (2012) Open source software license guide 2.0 (Intermediate Level) (Ministry of
Culture, Sports and Tourism and Korea Copyright Commission)
Lee G (2017) e-Apostille for a better international cooperation, a presentation delivered on July 4,
2017 at International Conference to Take Place to Celebrate 20th Anniversary of ROK’s Accession
to Hague Conference on Private International Law (HCCH), HCCH Asia Pacific Week 2017
Lee G et al (2017) A study on standardization of e-Apostilles, Final report submitted by Korea
Association of Private International Law to Ministry of Foreign Affairs
Miles C (2017) Blockchain security: what keeps your transaction data safe? IBM. Avail-
able at https://www.ibm.com/blogs/blockchain/2017/12/blockchain-security-what-keeps-your-
transaction-data-safe/. Accessed on 9 Aug 2019
Ministry of Foreign Affairs, Republic of Korea e-Apostille Service. Available at http://www.
apostille.go.kr. Accessed 10 Jun 2019
Offerman A, Open versus closed source: a delicate balance. https://joinup.ec.europa.eu/elibrary/
case/open-versus-closed-source-delicate-balance. Accessed 8 Aug 2019
Samsung Open Source Release Center, http://opensource.samsung.com/. Accessed 10 Jun 2019
Sohn H-S (2012) Antitrust aspects of open source commercialization. Informedia Law Rev
15(3):87–114
Sung H-C (2018) When open source software encounters patents: blockchain as an example to
explore the dilemma and solutions. J Marshall Rev Intell Prop L 18:55
The Hague Conference on Private International Law Permanent Bureau (2013) A handbook on the
practical operation of the Apostille convention (hereinafter “Apostille Handbook”) https://assets.
hcch.net/docs/ff5ad106-3573-495b-be94-7d66b7da7721.pdf. Accessed 10 Jun 2019
Wright A, De Filippi A (2015) Decentralized blockchain technology and the rise of lex
cryptographia. SSRN 1:27–28
Autonomous Systems
and Future Challenges
Lawyers’ Views on Autonomous Driving
Shinto Teramoto
Abstract The sharing of vehicles is likely to lead to less human drivers. When multi-
ple passengers synchronously or asynchronously share one vehicle, what they want to
use is the vehicle. The sharing of vehicular resources inevitably involves autonomous
driving technology as it is an important element. The development of full autonomous
driving on public roads is still on its way. At this stage, we should consider how we can
achieve safe and user friendly autonomous driving on public roads. So, the current
issue must be who can effectively prevent traffic incidents and accidents involving
autonomous driving by investing financial, human, and technological resources. Safe
autonomous driving is a product of a well-organized network system. The impor-
tance of telecommunication predicts that telecom and network service industries
will play major roles in realizing safe road traffic involving autonomous driving. If
lawyers discuss the allocation of cost caused by traffic incidents or accidents involv-
ing autonomous driving without considering the involvement of telecom and network
service industries, it simply shows the carelessness of lawyers. Autonomous driv-
ing is the way to realize the sharing of surplus resources, which have been unused
or wasted, by means of aggressive involvement of information and communication
technology (ICT).
Keywords Autonomous driving · Liability · Road traffic as a network
1 Introduction
Unused or wasted resources are omnipresent in vehicular road use.1 Most vehicles
on the road have unused space inside. Roads and parking lots are often occupied
by vehicles that are not being used to transport passengers or cargo. Considerable
1 See, e.g., MLIT (2016a), p. 2; MLIT (2016b), p. 55; European Environment Agency (2010); and
Federal Highway Administration (2018).
S. Teramoto (B)
e-mail: teramoto.shinto.717@m.kyushu-u.ac.jp

https://doi.org/10.1007/978-981-15-1350-3_10
170 S. Teramoto
energy resources are consumed inefficiently by vehicles. How can we reduce the
amount of unused or wasted resources in vehicular road use?
There would be two possible strategies to reduce unused or wasted space in
vehicles:
i. One strategy is to have one vehicle transport simultaneously as many passengers
as possible. Typically, a mass transit system such as a bus or a tram employs this
strategy. This can be paraphrased as the synchronous sharing of one vehicle and
one driver by multiple passengers;
ii. Another strategy is to reduce the number of vehicles that are just parking without
transporting passengers. Taxi and limousine companies employ this strategy. This
can be paraphrased as the asynchronous sharing of one vehicle and one driver
by multiple passengers. However, in the case of car rentals or car sharing, only
the vehicles are shared.
Also we can consider the strategies to reduce the unused or wasted space of roads
and parking lots. One strategy is to reduce the unused or wasted space in vehicles,
as discussed above. Another strategy is to reduce the number of vehicles that are
parked without transporting passengers. In sum, we can reduce the unused or wasted
space of roads and parking lots by means of synchronous or asynchronous sharing
of vehicles and drivers by passengers.
Moreover, we can consider the strategies to save energy resources consumed
inefficiently by vehicles. One strategy is to reduce the number of power sources (i.e.,
engines and electric motors, etc.) per passenger. This means synchronous sharing
of one vehicle by multiple passengers. Another strategy is to reduce the idle weight
transported by a vehicle. For this purpose, we can reduce deadhead vehicles by means
of asynchronous sharing of vehicles by passengers. A driver who is not a passenger is
also deemed as idle weight. Getting rid of such a driver also reduces the transportation
of idle weight.
In sum, the strategies to reduce the quantity of unused or wasted resources in
vehicular road use are either the following or a combination thereof:
i. Synchronous sharing of one vehicle and/or one driver by multiple passengers;
ii. Asynchronous sharing of one vehicle and/or one driver by multiple passengers;
and,
iii. Elimination of human drivers.
Ideal sharing of vehicles is likely to lead to less human drivers. When multiple
passengers synchronously or asynchronously share one vehicle, what they want to
use is the vehicle. Although passengers also share a human driver, it is only because
the vehicle cannot be driven without a human driver. The driver is not an essential
resource if the vehicle can be driven without a human driver. In sum, sharing of
vehicular resources inevitably involves autonomous driving technology as it is an
important element.
Lawyers’ Views on Autonomous Driving 171
2 The Favorite Question Asked by Lawyers
Lawyers tend to emphasize the debate on who should be liable for accidents. Actually,
the favorite question asked by lawyers is who should be liable to compensate injured
pedestrians, bikers or drivers in the case of a traffic accident involving an autonomous
driven car.
According to the typical arguments of the lawyers, those that should be held liable
often include one or more of the following2 :
(i) the passenger of the autonomous driven car, who should have overridden the
autonomous driving system in an emergency. Of course, the author does not
think such an override is practicable;
(ii) those who own or operate the autonomous driving car involved in an accident;
and/or,
(iii) the manufacturer and/or the distributor of the autonomous driving car or
autonomous driving system.3
However, presumably, we, lawyers, are not well prepared to talk about liability.
Making somebody liable must be beneficial to society. However, to the best knowl-
edge of the author, it is rarely discussed how and why making any of the said persons
liable will make roads safe. The human skill necessary to drive a car is likely to be
quickly lost by depending on autonomous driving. It is fanciful to hope a human
driver can override an autonomous driving system in an emergency. Table 1 suggests
that not a few accidents involving bodily injuries are caused by drivers’ operating
error. Making “should have been” drivers liable is not likely to make them more
cautious in preventing traffic accidents.
Numerous elements such as signals, signs, behaviors of pedestrians, bikes
and other vehicles, and communication, miscommunication or dis-communication
among them constitute essential parts of road use and its safety. A traffic incident or
accident involving an autonomous driving car is likely to be caused by a very com-
plex combination of these elements. Presumably, any of these elements can increase
or decrease the probability of such an incident or accident happening.
Moreover, our consideration of who should be liable is greatly affected by the
degree of consumers’ acceptance of new technology. However, such degree of accep-
tance of autonomous driving technology is quickly changing every day or even every
time according to the progress of the technology and citizens’ own experience. If
we talk about liability based on the current limited degree of acceptance of our-
selves affected by current imperfect autonomous driving technology, we are likely
to mislead ourselves.
We should reconsider the role of the legal concept of liability. Of course, the
concept of liability may, to a certain degree, be related to the concept of moral
2 See, e.g., Marchant and Lindor (2012); Harris (2015).

3 See, e.g., Faure et al. (2016), which makes detailed discussion concerning related issues.
172
Table 1 Motor vehicle accidents by type of driver operating errors (source adapted from ITARDA 2014)
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
All accidents causing bodily injuries 839,343 824,002 784,086 734,394 676,138 654,207 647,510 618,693 597,045 566,357
Accidents caused by driver’s operating error 6.9 7.3 7.2 6.7 7.0 6.7 6.8 7.2 7.3 7.4
(%)
Type of driver operating error
Steering control error (%) 21.2 19.2 18.8 18.6 17.4 17.5 16.8 15.2 14.9 14.2
Pedal misapplication (%) 13.2 12.3 13.0 14.4 13.9 15.0 14.3 14.5 14.3 15.3
Brake control error (%) 39.6 41.3 42.2 40.8 41.8 41.4 42.8 44.8 46.9 48.1
Other driver errors (%) 26.0 27.1 26.0 26.3 27.0 26.1 26.2 25.5 24.0 22.4
S. Teramoto
responsibility. However, liability can also be deemed as a tool to share social costs
among those that have suffered damage and other participants of the society.4
Suppose that a traffic accident occurs and a pedestrian is hit by a car and injured.
By making the driver of the car liable to compensate the pedestrian for his/her injuries,
the pedestrian can partly shift the cost necessary to recover from his/her injuries to
the driver. Moreover, such cost will be shared by a very large number of drivers
through the mechanism of insurance. In other words, each of a very large number of
drivers contributes a small surplus of money to share the cost that would have been
solely borne by the injured person.5
Suppose also that the probability of the occurrence of a traffic incident and accident
can be reduced by preventive means. Of course, the development, installment or
preparation of such preventive means requires the investment of resources including
time, money and labor. Also, certain parties are in a good position to prepare such
preventive means, while others are not. For example, automobile makers can possibly
prevent traffic incidents and accidents by developing an automatic braking system or
other safety devices and installing them in their automobile products.6 Human drivers
can possibly prevent traffic incidents and accidents by going through safety driving
training and information provision.7 They can compare the cost to be incurred by
themselves when they become liable to compensate for the damage caused by a traffic
incident or an accident, with the cost of reducing the probability that such an incident
or an accident will occur. If the latter is substantially lower, or the predictability of
the latter is much higher than that of the former, they would be more likely to choose
assuming the cost of preventing incidents or accidents in advance.
The development of full autonomous driving on public roads is still on its way. It
is obvious that, at this stage, we should consider how we can achieve safe and user
friendly autonomous driving on public roads. So, the current issue must be who can
effectively prevent traffic incidents and accidents involving autonomous driving by
investing their own financial, human, and technological resources.
A passenger “who should have overridden the autonomous driving controls” is not
in a good position to prevent traffic incidents and accidents involving autonomous
driving. As Table 1 suggests, the skill of non-professional drivers could be quite
poor. Also, they are not well trained to deal with emergencies. Even if we, ordinary
drivers, could override autonomous driving controls, the situation would not likely
be better. Moreover, obviously, it is fanciful to expect a “passenger” depending on
autonomous driving every day to be able to override the autonomous driving controls
in an emergency.
4 See, e.g., Cooter (1991).

5 See, e.g., Trowbridge (1975).
6 See, e.g., MLIT (2017), outlining the safety devices equipped in currently sold automobiles in
Japan.
7 For example, several automobile manufacturers are providing safe driving training includ-
ing TOYOTA (https://www.toyota.co.jp/mobilitas/), Honda (https://www.honda.co.jp/safetyinfo/),

and BMW (https://www.bmw-drivingexperience.com/en/trainings/bmw-driving-experience/bmw-
academy-experience/safety-training.html).
174 S. Teramoto
Many safety devices are employed both in human driven cars to assist human
drivers and also in autonomous driven cars.8 So, it makes sense to nudge human
drivers to share the cost of developing and procuring such safety devices, by making
them liable in traffic accidents. Also, the insurance premium payable by the owner,
operator or the driver of a vehicle equipped with such safety devices will be dis-
counted.9 Such favorable insurance premium is likely to nudge citizens and business
companies to procure and operate or drive such safety cars.
However, when fully-autonomous driven cars become a majority of the vehicles
on the roads, such concept of liability is likely to become obsolete. First of all, no
more drivers exist in the vehicles. Moreover, those who own or operate autonomous
driving cars also are not in a good position to prevent traffic incidents and accidents
involving autonomous driving. Modern automobiles are increasingly computerized
and becoming black boxes. What the owners and operators of automobiles can do
is almost limited to having a contract with the maintenance service provider and
requesting them to replace old or failed assemblies with new assemblies provided by
the suppliers that are authorized by the automobile production makers. Therefore, we
cannot expect that they can contribute much to prevent traffic incidents and accidents
involving autonomous driving vehicles.
In sum, the current issue must be who can effectively prevent traffic incidents and
accidents involving autonomous driving by investing their own financial, human, and
technological resources. Making those who own or operate autonomous vehicles
liable in traffic accidents is likely to nudge them to be very careful in choosing
and purchasing autonomous driving cars while the degree of “user acceptance” of
autonomous driving is very low. However, when autonomous driving cars become
the majority of the vehicles on the road, such impact of liability would be lost because
owners and operators will no longer be held liable because of choosing autonomous
driving cars.
Obviously, they can contribute to preventing traffic incidents and accidents involv-
ing autonomous driving, by investing funds, labor, and technology to improve
autonomous driving systems and affiliated safety devices. However, greater road
safety cannot be achieved only by a single automobile. It is achieved and maintained
only by the combination of the building and good maintenance of safe roads, sig-
nals, signs, the behavior of pedestrians and bikers, etc. The manufacturers and/or the
distributors of autonomous driving cars or autonomous driving systems can prevent
traffic incidents and accidents only by means of cooperating with other road traffic
participants, as human drivers have to learn such cooperation to get driver’s license.
We, lawyers, should not forget the essential contribution to safety of all road users.
Road traffic is a networked system.
8 See,e.g., Trowbridge (1975).

9 Forexample, GIROJ (2018) amended insurance premium and made it favorable for the operators
and drivers of automobiles equipped with advanced safety devices, effective on and from January
1, 2020.
In addition, we have to consider the relationship between the degree of “user

acceptance” of autonomous driving. If the degree of user acceptance of new technol-
ogy is very low, citizens and industries are likely to refuse to pay the costs incurred
by the employment of such technology, and consider that only those who employ
such technology should cover the cost. In such case, it would become impractical
to allocate the cost among a very large number of citizens and industries through
the combination of liability and voluntary and/or obligatory insurance. However, it
is often impossible for such early adopters to solely cover every such cost. So, an
intervention by the government, by means of the introduction of a law, becomes
necessary in order to make every citizen and industry share such cost through tax
and governmental expenditure. In contrast, if the degree of user acceptance is very
high, citizens and industries are often willing to share such cost through the com-
bination of liability and insurance, as we experience by contracting and benefitting
from voluntary and/or obligatory auto insurance.
Needless to say, at the moment, the degree of user acceptance of autonomous
driving is not high. However, such degree of acceptance is likely to become higher
incrementally through the development of autonomous driving technology and dis-
semination of knowledge and experience among citizens and industries. It would not
be fair if lawyers talk as if they (or, we) have sufficient knowledge about the degree of
user acceptance so that they (or, we) can design an appropriate and practical combi-
nation of liability and insurance to cover and share the cost incurred by autonomous
driving. We, lawyers, have no crystal ball to predict the degree of user acceptance of
autonomous driving, even if it is concerned with the very near future.
3 Safe Autonomous Driving Is a Product

of a Well-Organized Network System
Irrespective of whether or not it involves autonomous driving, the safety of road

traffic is being achieved by the relationships between all of its participants. When
we deem road traffic as a network, each of such participants is a node, and the rela-
tionships between each pair of nodes is an edge. For example, a human driver is
continuously receiving information from signals, signs, road surfaces, pedestrians,
bikers, other vehicles and their drivers, police, and others, and responding (or, some-
times, failing to respond) to them, who in turn, are receiving (or, sometimes, failing
to receive) such responses. Each of the vehicles and its driver, human or not, is a node
participating in a very large networked system called road traffic. The nodes that can
contribute to preventing incidents and accidents involving autonomous driving are
not limited to autonomous driving vehicles or systems and passengers who should
have overridden the autonomous driving system. Also, they cannot prevent incidents
and accidents without mutual communication with other nodes participating in road
traffic. Contending hastily that only specific classes of nodes should be liable does
not make sense.
176 S. Teramoto
Fig. 1 Peer to peer (left) and hub and spokes (right)
Assuming that mutual communication among nodes contributes to road traffic

safety, it is very natural to expect that telecommunication constitutes an essential
piece of safe autonomous driving. In road traffic, mutual communication among
nodes is essential in order to prevent them from conflicting or colliding with one
another.
The very classic type of peer-to-peer communication employed by vessels and
aircrafts is a pair of navigation lights (Fig. 1). The green light is starboard, and the
red light is portside. In contrast, air traffic controls and Centralized Train Controls
(CTC) are typical examples of “hub and spokes” communication networks designed
to ensure the safety of traffic. Maritime traffic controls and road signal systems also
employ “hub and spokes” networks in highly crowded areas. This situation is not
likely to be changed by introducing autonomous driving instead of human drivers.
The safety of road traffic is heavily dependent on peer-to-peer communications,
such as turn signals, eye contact, hand gestures, and so on. It is likely that autonomous
driving vehicles have to transmit signals that can be easily received and identified
by every other node of road traffic. Also, every other node of road traffic has to
transmit signals that can be easily received and identified by every autonomous
driving vehicle. Such generation, transmission, reception, and processing of signals
can only be realized through telecommunication between nodes, because passengers
of autonomously driven vehicles cannot commit themselves in such communication
as transmitters or receivers of signs and signals. Not only autonomous driving cars, but
also traffic signals, signs, perpetual or temporary obstacles on the roads, pedestrians,
bikes and bikers, police, and any other nodes should have devices that enable such
mutual telecommunication to realize safe autonomous driving.
The importance of telecommunication predicts that telecom and network service
industries will play major roles in realizing safe road traffic involving autonomous
driving. If lawyers discuss the allocation of cost caused by traffic incidents or acci-
dents involving autonomous driving without considering the involvement of tele-
com and network service industries, it simply shows the carelessness of lawyers.
Autonomous driving is the way to realize the sharing of surplus resources, which
have been unused or wasted, by means of aggressive involvement of information and
communication technology (ICT). If positive resources are shared by more and more
road traffic participants with the help of telecom and network service industries, the
corresponding cost of autonomous driving should also be shared by a wider scope

of participants.
4 How Can We, Lawyers, Contribute to the Safety of Road

Traffic Involving Autonomous Driving?
We, lawyers, tend to overly devote ourselves to the issue of who should be liable
to compensate for the damage caused by a traffic accidents involving autonomous
driving. Such questions emphasize the conflicts of interest among citizens, industries
and governments and retards and suffocates their collaboration. However, they (or,
we) should collaborate with one another to develop an interface and standards for
every road traffic participant to effectively and efficiently communicate with each
other.
There are many duties to be borne by lawyers before hastily debating who should
be liable. The standardization of telecommunication among road traffic participants,
promoting industries to develop signals and signs that can communicate with both
human drivers and autonomous driving system, to develop smart devices affixed to
pedestrians and bikers that can communicate with autonomous driving systems, and
to develop industry standards for smart roads.
References
Cooter RD (1991) Economic theories of legal liability. J Econ Perspect 5(3):11–30

European Environment Agency (2010) Occupancy rates of passenger vehicles, EUR. https://www.
eea.europa.eu/data-and-maps/indicators/occupancy-rates-of-passenger-vehicles/occupancy-
rates-of-passenger-vehicles. Accessed 31 May 2019
Faure M et al (2016) Liability for unknown risks: a law and economics perspective. J Eur Tort Law
7(2):198–228
Federal Highway Administration (2018) Average vehicle occupancy factors for computing travel
time reliability measures and total peak hour excessive delay metrics. https://www.fhwa.dot.gov/
tpm/guidance/avo_factors.pdf. Accessed 31 May 2019
GIROJ (General Insurance Rating Organization of Japan) (2018) Jidōsha Hoken Sankō Junritsu
Kaitei no Goannai (Announcement on revision of reference loss cost rates for voluntary automo-
bile insurance). https://www.giroj.or.jp/ratemaking/automobile/pdf/201809_announcement.pdf#
view=fitV. Accessed 31 May 2019
Harris M (2015) Why shouldn’t worry about liability for self-driving car accidents. Avail-
able via IEEE SPECTRUM. https://spectrum.ieee.org/cars-that-think/transportation/self-driving/
why-you-shouldnt-worry-about-liability-for-selfdriving-car-accidents. Accessed 31 May 2019
ITARDA (Institute for Traffic Accident Research & Data Analysis) (2014) Untenzouza no Yamari o
Fuzegu (Preventing drivers conducting errors) https://www.itarda.or.jp/itardainfomation/info107.
pdf. Accessed 19 July 2019
Marchant GE, Lindor RA (2012) The coming collision between autonomous vehicles and The
liability system. Santa Clara Law Rev 52:1321–1340. https://spectrum.ieee.org/cars-that-
178 S. Teramoto
think/transportation/self-driving/why-you-shouldnt-worry-about-liability-for-selfdriving-car-
accidents. Accessed 31 May 2019
MLIT (Ministry of Land, Infrastructure, Transport and Tourism of Japan) (2016a) Suuji-de-miru
Jidousha (Automobiles in figures). Automobile Business Association of Japan
MLIT (Ministry of Land, Infrastructure, Transport and Tourism of Japan) (2016b) Safe measures
for road traffic (Dourokotsu no Anzenshisaku), Reviews of the Policies (Assessment) in FY2015
http://www.mlit.go.jp/common/001125128.pdf and http://www.mlit.go.jp/common/001125130.
pdf. Accessed 31 May 2019
MLIT (Ministry of Land, Infrastructure, Transport and Tourism of Japan) (2017) Jōyōsha Me-Ka-
Niyoru Jitsuyōka ASV Gijutsu no Ichiran (List of advanced safety vehicle features by pas-
senger car manufacturers). http://www.mlit.go.jp/jidosha/anzen/01asv/resourse/data/2017_03_
ASVgijutsu.pdf. Accessed 31 May 2019
Trowbridge CL (1975) Insurance as a transfer mechanism. J Risk Insur 42(1):1–15
Adversarial Machine Learning: A Blow
to the Transportation Sharing Economy
Steven Van Uytsel and Danilo Vasconcellos Vargas
Abstract Adversarial machine learning has indicated that perturbations to a picture

may disable a deep neural network from correctly qualifying the content of a picture.
The progressing research has even revealed that the perturbations do not necessarily
have to be large in size. This research has been transplanted to traffic signs. The
test results were disastrous. For example, a perturbated stop sign was recognized
as a speeding sign. Because visualization technology is not able to overcome this
problem yet, the question arises who should be liable for accidents caused by this
technology. Manufacturers are being pointed at and for that reason it has been claimed
that the commercialization of autonomous vehicles may stall. Without autonomous
vehicles, the sharing economy may not fully develop either. This chapter shows that
there are alternatives for the unpredictable financial burden on the car manufacturers
for accidents with autonomous cars. This chapter refers to operator liability, but
argues that for reasons of fairness, this is not a viable choice. A more viable choice
is a no-fault liability on the manufacturer, as this kind of scheme forces the car
manufacturer to be careful but keeps the financial risk predicable. Another option is
to be found outside law. Engineers could build infrastructure enabling automation.
Such infrastructure may overcome the problems of the visualization technology, but
could potentially create a complex web of product and service providers. Legislators
should prevent that the victims of an accident, if it were still to occur, would face years
in court with the various actors of this complex web in order to receive compensation.
Keywords Adversarial machine learning · Deep neural network · Product liability

law · Operator liability · No-fault liability · Infrastructure enabled automation
S. Van Uytsel (B)

e-mail: uytsel@law.kyushu-u.ac.jp
D. V. Vargas
Faculty of Information Science and Electrical Engineering, Kyushu University, Fukuoka, Japan

https://doi.org/10.1007/978-981-15-1350-3_11
180 S. Van Uytsel and D. V. Vargas
1 Introduction
Uber and Lyft are among the best known examples of the sharing economy. A com-
bination of mobile communication devices connected to the Internet and excess
capacity in cars allowed these firms to develop an application to link potential pas-
sengers to drivers. Whereas the car-sharing apps were initially looked at with some
attempt by the traditional car manufacturers, the manufacturers are now “falling over
themselves to get involved in the sharing economy.”1 The reason for the shift towards
the sharing economy is obvious. The business model of the traditional car manufac-
turers is based on the understanding that customers attribute value towards owning
a specific product from a specific brand. Car manufacturers therefore design vehi-
cles to build loyalty among its customers. Once customers value mobility solutions
more, i.e., the ability to get from one point to another seamlessly, the manufacturers
need “changes, rebrands and new businesses”2 in order to “generate another kind of
brand loyalty.”3 Participation in the sharing economy may offer this new business
opportunity.
It is further predicted that the share-centric mobility landscape will get a new
impetus with autonomous vehicles. Combining autonomous vehicles with the prin-
ciples of the sharing economy will allow the establishment of “shared autonomous
vehicle fleets.”4 In this new business model “fleet members would not own a car
but instead will have a subscription to transportation services. If they need transport,
they would use a location-aware app to arrange for it.”5 The advantages of this new
business model would be vast. The Boston Consulting Group refers to “not having
to find a parking place…, being able to multitask or be more productive while trav-
eling.”6 Not only travel will become more convenient, the consumer will not have
to care about all aspects related to vehicle maintenance. Of course, this requires that
autonomous vehicles hit the road.
Recent studies on adversarial machine learning7 made Michael Grossman, a
Texas-based injury lawyer, skeptical of the viability of autonomous vehicles.8 These
studies had pointed out that adversarial attacks or perturbations on pictures makes
it difficult for the algorithm to correctly classify the content of that picture. If this
is applied to traffic sign recognition, simple graffiti on the sign could mislead the
1 Gagan (2018).
2 Gagan (2018).
3 Gagan (2018).
4 Deloitte (2019).
5 Deloitte (2019).
6 Boston Consulting Group (2019).
7 Adversarial machine learning is a discipline in machine learning focusing on misleading the
machine learning’s algorithms by providing manipulated inputs. See Kurakin, Goodfellow and
Bengio (2017), p. 1.
8 Grossman (2017).
Adversarial Machine Learning: A Blow to the Transportation … 181
algorithm that is analyzing the picture of the traffic sign captured by the camera.9
Rather than recognizing the traffic sign for what it is, the algorithm could attribute
a different meaning to the traffic sign. The consequences could be disastrous, espe-
cially if, for example, a stop sign would be read as a speeding sign.10 When rational
car manufacturers know this defect, they will not proceed with the marketing of
autonomous vehicles.
This chapter will argue that the claim by Grossman may be true from a tra-
ditional product liability law point of view.11 However, such a claim neglects three
elements. First, autonomous vehicle technology is still in full progress. Second, fully
autonomous vehicles will not necessarily drive on roads as we currently know them.
Third, legislation can be adapted to support technological innovation. It is this last
point that will be highlighted in this study. This chapter will review three different
liability regimes and investigate whether the issue posed by adversarial perturbations
on algorithms, which in our example are deep neural networks,12 would mean the
end of the development of autonomous vehicles.
The chapter is structured as follows. In Sect. 2, the chapter highlights the current
state of technology of autonomous vehicles. The research on adversarial attacks
is detailed in Sect. 3. Section 4 builds on this issue and explains the reason why
algorithms can be fooled. Based on the perception that technology is unreliable,
Sect. 5 argues that control through technology is something we should not yet rely
on. In order not to trump technological innovation in this situation, Sect. 6 will
explore the evolution that law should make. Section 7 will conclude.
Two limitations should be mentioned. The analyses presented in this chapter
are applicable to Level 3 autonomous vehicles driving in autonomous mode and
above. Further, this chapter does not look at whether the legislator needs to formulate
standards for autonomous vehicles and, if they do, what it could mean for liability.13
2 The Current State of Autonomous Driving
The debate on autonomous vehicles has been greatly influenced by the Society of
Automotive Engineers that classified autonomous vehicles in six different levels of
autonomy.14 The lowest level, Level 0 (no automation), demands that the driver per-
forms all driving tasks. The driver is in sole control of the vehicle. If technology is
9 See, e.g., Eykholt et al. (2018); Carlini and Wagner (2017); Evtimov et al. (2017); Houben et al.
(2013); Karungaru et al. (2013); Stallkampa et al. (2012).
10 Eykholt et al. (2018); Sitawarin (2018), p. 2.
11 Others support this view, see, e.g., Noussia (2019), p. 82.
12 The text uses algorithms and deep neural networks (DNN) interchangeably. Deep neural networks
are a class of machine learning algorithms.

13 See on this aspect, Adkisson (2018); London (2019); London and Danks (2018).
14 Society of Automative Engineers (2018); See also Channon et al. (2019), pp. 4–5.
assisting the driver at this level, it is limited to providing a warning.15 Level 1 vehicles
offer one type of driving assistance, such as adaptive cruise control or lane center-
ing. Control by the driver of these driving-assistance systems is necessary.16 Partial
automation is offered at Level 2 and means that the driver can rely on a combination
of two types of driving-assistance systems. This could, for example, be a combi-
nation of adaptive cruise control, lane-keeping systems, or an emergency breaking
system.17 Automated driving starts at Level 3. Level 3, which is characterized as con-
ditional automation, enables the vehicle to monitor the driving environment under
normal conditions. The driver is required, however, to control the system and take
over whenever the system requests to do so.18 The control by the driver is not required
anymore for Level 4 vehicles. Under this high automation mode, the vehicle is able to
interpret the driving environment and take the decisions necessary for driving under
certain conditions. The driver can still take control when he desires or when the
system fails due to, for example road works, road diversions, or off-road driving.19
Level 5 will offer full automation.20 This means that the system can drive by itself
under all circumstances. The system will even be able to drive in “unpredictable or
changing physical environments.”21
It is not enough to only understand the different levels of automation. An under-
standing of the technology is also required to discuss autonomous vehicles. It is
technology that will allow the system to locate the vehicle, plan what the vehicle
should do taking into consideration what is around the vehicle, and then execute the
plan. This sensing, planning and acting stage is complex, since a vehicle is moving
in an environment with “other moving vehicles on the road, pedestrians, cyclists and
motorcyclists, all of which move at different speeds, as well as animals and even
debris, such as that from traffic accidents.”22 Hannah YeeFen Lim continues that
“weather conditions can also hamper an autonomous vehicle’s perception of its sur-
15 Lim (2018), p. 4.
16 Fuchs (2019).
17 Pearl (2017b), p. 27.
18 Lim (2018), pp. 4–5.
19 Pearl (2017b), pp. 28–29.
20 Fuchs (2019).
21 Lim (2018), p. 5. Many predictions have been made on when autonomous vehicles will hit the
road (for a summary of these predictions, see Gurney (2013), pp. 248–251 (especially footnotes 8
to 15); Geistfeld (2017), p. 1615. However, among the vehicles currently marketed to the public,
Tesla is offering the most advanced autonomous vehicles, see Lim (2018). For an overview of the
historical development, see Kellerman (2018), pp. 106–109. Audi is said to start commercialize a
Level 3 vehicle in 2019, see Audi MediaCenter (2017); See also Abe (2018), p. 3. Hannah YeeFen
Lim purports that in early 2018 the race for a Level 5 vehicle is being dominated by Google’s Waymo
subsidiary and General Motors’ Cruise Automation subsidiary, see Lim (2018), p. 5. Ryosuke Abe
reports that Nissan will test autonomously driving taxis in the high-density city of Yokohama and
that Toyota would like to provide autonomous vehicles at the sites built for the Tokyo Olympics in
2020. No statement is made on the level of these vehicles, see Abe (2018), p. 3.
22 Lim (2018), p. 5.
roundings, for example, rain and snow will affect its ability to distinguish objects.
If there are roadworks, or slight detours resulting from construction or an accident,
or poorly marked roads, all of these will negatively affect the autonomous vehicle’s
ability to function.”23
For surveying the complex surroundings, Harry Surden and Mary-Anne Williams
describe in their article, How Self-Driving Cars Work,24 that vehicles are using both
hardware and annotated digital maps.25 The hardware, a combination of sensors and
cameras, provides real time data to the vehicle. The annotated digital maps, road-
level images annotated with information on the road surroundings, allows the vehicle
to relate back to pre-recorded information. The annotated digital maps and different
sensors will now be considered.
The annotated digital maps, by others also termed high-definition digital maps,
provide the systems of autonomous vehicles with “detailed, road-level images of most
street locations, such as overhead and street level layouts of roads, including traffic
signs and signals, footpaths and of course the associated GPS [Global Positional
System] coordinates.”26 The annotated maps are constructed by 360 degree laser
scans of roads and must constantly be updated and maintained. The function of
the annotated digital map could be threefold. First, the map allows the vehicle to
anticipate what is ahead, such as a traffic light. Second, the annotations could steer
the system to take a certain action, like stopping at a stop sign. Third, the map could
be a reference point for the system to check whether the real time data corresponds
with what was previously recorded. Annotated digital maps are lauded for their high
precision. The flipside of these maps is that they need to constantly be updated and
maintained.
Annotated digital maps are assisted by a GPS. A GPS functions through receiving
signals broadcasted by satellites in space. The fact that the position of the satellites is
known allows for the calculation of the latitude and longitude based upon the timing
of the signals. Such a calculation will provide a vehicle with a rough approximation
of where it is located. The problem however with GPS is its inaccuracy. Surden
and Williams indicate that a GPS can be inaccurate “by as much as five meters.”27
The consequences of such inaccuracy could be disastrous. Therefore, a more precise
means of positioning a vehicle needs to supplement a GPS system.
Accuracy could be given by radar systems. Currently, there are two radar systems
used for making vehicles more autonomous. On the one hand, there are the classical
radar sensors that emit radio waves. On the other hand, there is the LIDAR, short
for light detection and ranging, operating by reflecting laser beams off of nearby
objects. The classical radar sensors are being used “to detect the position and the
speed of surrounding objects.”28 Classical radars are being used for their range. They
23 Lim (2018), p. 6.
24 Surden and Williams (2017).
25 Surden and Williams (2017), pp. 11–12.
26 Lim (2018), p. 13.
27 Surden and Williams (2017), p. 17.
28 Surden and Williams (2017), p 19.
can reach up to several hundred meters, can assess the speed of multiple moving
objects—even nearby—and are relatively cheap compared to other radars. Though
the classical radar is more accurate than the GPS, it is less accurate than the LIDAR
type of radar.29 The precision is achieved by the fact that laser beams are smaller than
radio waves. The LIDAR has the capacity to calculate speed, position and distance to
nearby objects. The nearby object can even be the “white lane boundary lines painted
on road surfaces.”30 LIDAR, however, faces difficulties during heavy rain and low
hanging clouds.31
Besides the radar systems, there is also the possibility to install infrared systems
and ultrasonic sensors. Infrared systems are useful to detect heat differences, to
operate at night and to visualize in abrupt light changes or direct light. Ultrasonic
sensors work at short-range and are useful in a warning system or to help in parking
assistance.32
None of the hardware discussed up until now allows for the recognition of pictures,
colors, and letters. It is for that reason that cameras may be installed in autonomous
vehicles.33 Cameras in combination with the technique of machine vision, “the field
that studies algorithmic approaches to making sense of visuals,”34 allows autonomous
vehicles to interpret traffic signs. This technology can distinguish different traffic
signs, “determine whether a traffic signal is green or red,”35 and also recognize
words on traffic signs.36
The information gathered by all devices described above need to be processed in
order to understand “what has been physically sensed.”37 This process is often done
by algorithms. For example, the interpretation of a picture captured by a camera is
performed by an algorithm called a deep neural network (DNN). A DNN can be
weak in the recognition of pictures when there are alterations, termed noise. The
extent to which this forms a problem is explained in the next paragraphs.
3 Machines Are Easily Fooled
In 2013, researchers discovered that convolutional neural networks, the most widely
used type of DNN, could be easily fooled by changing a few pixels in the image
29 Surden and Williams (2017), p 19. The mistake can be “by several inches to feet.” See Lim (2018),
p. 11.
30 Lim (2018), p. 10.
31 Fuchs (2019); Lim (2018), p. 10.
32 Lim (2018), p. 13.
33 Cameras can also be used for distance calculation. Pictures of the same object from different
angles allows a computer to estimate the distance. This is called stereopsis.

36 Surden and Williams (2017), p. 20; see also Lim (2018), p. 8.
37 Lim (2018), p. 15.
Image 1 Original sample (left row), added mask (middle row) and resulting image (right row).
The resulting image is misclassified by the algorithm (Szegedy et al. 2013)
(see Image 1).38 This finding was followed by a series of papers showing that cur-
rent machine learning can misclassify under various scenarios. A brief overview of
adversarial machine learning will be presented, followed by some recent findings.
3.1 Adversarial Machine Learning
Regarding attacks against machine learning algorithms, there are basically two types
of attack scenarios:
38 Szegedy et al. (2013).

i. Evasion Attacks: Evasion attacks are the most common type of attack scenario. In
this scenario the attacker aims to evade the correct classification of the algorithm
by some technique. For example, some letters or words can be included in an
email to confuse and evade anti-spam filters, allowing a spam to be classified as
non-spam. This scenario is not dependent on the training type used and therefore
is more general than poison attacks.39
ii. Poison Attacks: Most machine learning algorithms need to be trained using some
dataset. Usually these datasets are large and the information contained may not
be completely secure. Moreover, in some cases the algorithms use their input
throughout the application to retrain themselves. In both cases, an insertion of
misleading samples in the dataset may lead to a poorly trained algorithm that
will fail to recognize or correctly classify some portion of the input.40
In this chapter, we will focus on the most prevalent scenario type, evasion
attacks, especially attacks that fool the currently most used type of machine learning,
convolutional neural networks.
3.2 Recent Findings
In Image 1, it was shown that by adding small nearly imperceptible noise in images
can make convolutional neural networks misclassify.41 Therefore, even if the change
in the image is imperceptible for human beings that slight change can fool neural
networks. In Image 2, for example, Nguyen et al. have shown that random noise could
be recognized as almost anything with high confidence.42 This shows the possibility
of using random noise which has no meaning for human beings, as an attack to make
neural networks misclassify.
Lately, it was shown that a stamp of random noise can be introduced in images
to make them misclassify (Image 3).43 This is in part a follow up to the finding
of Image 2 in which random noise is already recognized as a horse, car, airplane,
etc. However, the image modification this time is limited to a small circle which
the authors refer as a patch. This small but significant change in approach shows an
effective way for attackers to fool neural networks. They could for example perform
an attack against the machine learning in autonomous cars by adding a patch over a
traffic light which would make it be recognized as a human being. This would stop
the car causing a long traffic jam. This type of attack received the name of adversarial
patch.
In 2017, beyond the adversarial patch there were two important findings that
deserve mention. The first finding was an abrupt change in the concept of how

42 Nguyen et al. (2015).
43 Brown et al. (2017).
Image 2 Random noise which are unrecognizable for human beings fool deep neural networks.
Although being purely noise, they are recognized as many types of objects/animals among other
classes (Nguyen et al. 2015)
evasion attacks are executed. Usually, an attacker needs to automatically search for
an image that can fool the algorithm. However, in (Image 4) scholars found the
possibility of creating one noise mask that when added on top of any image would
make a given classifier misclassify it. In this manner, there is no need to search for
each attack, only the mask needs to be identified.
Regarding the second finding, until recently it was thought that the presence of
imperceptible noise all over the image would be necessary to fool a machine learning
Image 3 A real world attack using a patch designed to make the classifier change the classification
from banana to toaster (Brown et al. 2017)
algorithm. However, in Image 5 it was shown that actually by changing only one pixel
it is possible to fool most convolutional neural networks. This is called one-pixel
attack.
All these recent findings reveal several vulnerabilities of current machine learning.
In fact, they also show that neural networks behave somewhat chaotically in which a
small perturbation can cause great changes in the output (one-pixel attack). They also
show that these neural networks can achieve high accuracy on many difficult datasets,
the images are still not correctly interpreted by convolutional neural networks, rather
the images are processed without taking into account structure, shape, among other
features that human beings as well as other animals perceive without much effort.
Therefore, there are two main problems which derive as a consequence of current
findings in adversarial machine learning:
i. Unsafe Applications: Since all machine learning algorithms in general are vul-
nerable to many of the discovered attacks, it becomes easy for attackers to fool
the algorithm. In fact, it is possible for attackers to change the classification of
algorithms to a given target classification which is called target attack. Lastly, as
seen in some real world applications, algorithms may just happen to fail under a
certain input without any external intended interference.
ii. Crisis in the Foundations of Machine Learning: It is true that there is a high
variety of machine learning algorithms and that there are many new algorithms
available every day. However, the foundation of machine learning is usually
shared among the mostly used algorithms. And this type of foundation poses
Image 4 A single mask when added to an image produces a respective image which fools the
classifier with high probability (Moosavi-Dezfooli et al. 2017). This is called universal perturbation
some limitations on their models and learning algorithms. There is a risk that
the mostly used current paradigm will not be able to solve these issues. Having
said that, there are a few new types of algorithms appearing which do not share
the same paradigm. They are, for example, neuroevolution algorithms which
can learn the model and parameters and are not limited in the complexity of
the models used.44 Another type of paradigm are the self-organized classifiers45
44 Vargas and Murata (2017).

45 Vargas et al. (2013).
Image 5 Extreme type of attack where only one pixel is changed to make the classifier misclassify.
This is an illustration of the One-Pixel Attack (Su et al. 2017)
which can adapt and learn under noisy and dynamic scenarios. For example,
it can learn to navigate in a maze even when the maze changes its shape from
time to time.46 There are many other types of algorithms that go beyond current
paradigm limitations but their discussion here goes beyond the scope of this
chapter.
Thus, there are many problems to tackle before machine learning can be part of
our daily lives. Current machine learning systems are not safe and might be either
exploited or just malfunction on their own. It is impossible to say now how deep lies
the problem but it is possible that the problem may derive from the very foundations
of machine learning. Nonetheless, machine learning can be very helpful in various
areas and it is expected to change how we interact with the world in the years to come.
Current challenges may be frightening at first but there are many areas of research
showing paths that could overcome it. It is just a question of time.
3.3 Why Are Machines Fooled?
It may be intriguing that something that was designed to be precise can fail in simple
attacks. More intriguing yet is the knowledge that something that can achieve 98%
accuracy on recognition tasks may still fail in simple attacks such as the one-pixel
attack. The natural question is why?
The reason why this happens lies in how machine learning works. Most systems
have some parameters which must be learned and to learn these parameters a learning
algorithm (optimization algorithm) is needed. The optimization algorithm will search
for parameters that decreases the error (called loss function) and thus improves the
classification accuracy. The problem arrives from the fact that we are searching for
billions of parameters in a very high dimensional model. Therefore, there will always
46 Vargas et al. (2015).

be a given input which may make it misbehave. Even if we test it over billions of
images, there are many more possibilities of images and many more points that it
will never see in the training stage and may misbehave.
A tricky part of machine learning is that the algorithm must learn by itself. Pro-
gramming how an algorithm must behave is not machine learning because the algo-
rithm is not learning anything. Therefore, in machine learning, there is no direct
explicit control over what is going to be learned. There is, naturally, an indirect con-
trol over what it wants to maximize or minimize. Most of the algorithms want to
minimize error in the classification and by doing so it will do its best to make no
mistakes in the training set. This is exactly why a bigger dataset leads to less errors
and a more robust algorithm because the training dataset will resemble the real world
scenario.
4 Adversarial Machine Learning and Autonomous Vehicles
In the research described above, the noise in the picture to be recognized resulted
from changing one or a few pixels. When adversarial machine learning focuses on
traffic sign detection and recognition, the noise on the traffic sign is graffiti, a sticker
or a complete cover of the traffic sign. Various experiments have been set up in which
traffic signs were presented to a camera, after which a classification module had to
determine the category of the traffic sign. This module does so by breaking down the
picture into color, shape and pictogram. The results of the various experiments were
remarkable.47
A study of scholars from the University of Michigan, Washington and California
focused on two types of alterations.48 The first type of alteration was to cover the
traffic sign with a printed poster of an identical traffic sign. The visual change was
only noticeable from very close. However, the machine classified in all cases the stop
sign as a 45-mile-per hour speed limit sign and a right-turn sign as a stop sign. The
second type of alteration was to attach stickers to traffic signs. Some of the stickers
were just black and white stickers, other were black and white but in the form of
the words “love” and “hate.” Whereas the former were misqualified all the time, the
latter had a nearly 70% qualification of a stop sign as a speed limitation sign.
Concurrent work by Princeton and Purdue University scholars has shown that a
speed limitation sign with a slight alteration in the format of some paint spots made
the DNN qualify the content of traffic signs differently than their real content.49
Another example in this study is the interpretation of a speed limitation sign of
120 km per hour as a speed limitation sign of 30 km per hour.50
47 See, e.g., Eykholt et al. (2018); Carlini and Wagner (2017); Evtimov et al. (2017); Houben et al.
(2013); Karungaru et al. (2013); Stallkampa et al. (2012).
48 Eykholt et al. (2018).
49 Sitawarin (2018), p. 2.
50 Sitawarin (2018), p. 11.
A post by Berkley Artificial Intelligence Research, a group of Ph.D. scholars at the

University of Berkley, summarizes lots of the research done in this regard.51 At the
end, these young scholars make a reference to a study of Nicolas Carlini and David
Wagner. That study concludes that “adversarial examples are much more difficult to
detect than previously recognized.”52 In other words, the misinterpretation is more
frequent than expected and there is currently no effective method to overcome that
problem. The current state-of-the-art computer visualization techniques are thus not
reliable enough to provide the necessary trust in autonomous vehicles. The ques-
tion is, however, whether the potential to misinterpret traffic signs will affect the
commercialization of autonomous vehicles.
5 Adversarial Machine Learning and the Law: The End

of Autonomous Vehicles?
Vehicles of which the technology fails may put a burden on the car manufacturers
who market these vehicles. It is most likely that these manufacturers will be held
liable for the damage that is caused by their vehicles. If the chance for accidents is
low, the manufacturer may take a calculated risk. There is data on the accuracy of
machines recognizing traffic signs correctly.
A study by researchers from Germany and Denmark have compared the accu-
racy of a human being to correctly recognize traffic signs with that of a machine.
Their study concludes that “the best individual in the human performance experi-
ment achieved a close-to-perfect accuracy of 99.22%, it was outperformed in this
challenging task by the best-performing machine learning approach, a committee
of convolutional neural networks, with 99.46% correct classification rate.”53 Even
with a “computationally cheap classifier,”54 one that is often used in driving assis-
tance systems, the correct recognition rate was 95.68%.55 These results seems to
be promising. Even if cheap technology is able to get such a high percentage of
correct recognition, marketing autonomous vehicles may be feasible. The study has
one caution, however. These results can only be obtained if the traffic signs are of
the same size and not perturbated. Hence, manufacturers still have no exact data on
the risk for accidents.
Being in the dark on the risk they take, the conceptualization of the liability regime
may determine on whether car manufacturers are willing to commercialize higher
level autonomous vehicles, i.e., Level 3 or above.
51 Evtimov et al. (2017).

52 Carliniand Wagner (2017), p. 11.
53 Stallkampa et al. (2012), p. 10.
5.1 Product Liability Law as the First Obvious Choice
When machine learning technique is driving a vehicle instead of a human, product

liability law is often regarded as the field of law that should be applicable in case the
technique fails. In nearly any publication that has been published about autonomous
vehicles that contains chapters on liability, product liability takes a central role.56
Lim for example argues that “strict liability is the only way to guarantee public
safety. It will curb any misleading or deceptive conduct by autonomous vehicle
manufacturers and will also engender a responsible culture in the development of
autonomous vehicles, and not one driven by profits and kudos.”57 At the same time,
she admits that strict liability may slow down innovation. The manufacturers are
the ones who best know the capabilities of the techniques available for autonomous
vehicles so they should bear the financial and other risks.58 Another reason she
puts forward is that the new and novel products “have not been tested openly and
sufficiently for the general public to trust them and for the general public to bear the
risk of their presence and use.”59
Product liability law has several difficulties in its application. In order to turn to
this field of law, there needs to be a clear understanding that the algorithm can be
regarded as a product.60 This question goes back to the long standing discussion
in product liability law on whether software is a product or not.61 Traditionally,
the theoretical discussion on this issue, as there has not been any court decision on
this issue, looked at whether the software was delivered on a tangible medium or
presented as a downloadable from the Internet. Only the former would be considered
as a product.62 Some point out that the court need to deliver an interpretation in the
future, others say it would be better that the legislator creates clarity.63
Under the presumption that AI can be a product, the other issue that the legal
discourse has to deal with is determining what a defect is. The determination of defect
is a complex issue. Not only does this issue slightly differ along jurisdictions, even
within some jurisdictions there exist more than one test to determine a defect. In the
United States, for example, jurisprudence has developed both a consumer expectation
test and a risk-utility test. The former looks at the expectations of a reasonable
consumer, while the latter requires to see whether there is “reasonable” alternative
56 See, e.g., Channon et al. (2019), pp. 34–46; Evas (2018); Lim (2018), pp. 106–109; Crane et al.
(2017); Geistfeld (2017), pp. 1619 and 1634–1647; Maurer et al. (2015), pp. 553–570.
57 Lim (2018), p. 105.
58 Lim (2018), p. 105.
59 Lim (2018), p. 105.
60 McCormick (2019), pp. 36–37.
61 See, e.g., Rohr et al. (2018), p. 150; Rowland and Macdonald (2005), pp. 213–219.
62 McCormick (2019), p. 37.
63 McCormick (2019), p. 37. Note that Karni Chagal-Feferkorn develops criteria to exclude algo-
rithms autonomously making decisions from product liability law. She does so to make sure that
this field of law should not be applicable to advanced algorithms. See Chagal-Feferkorn (2018a).
Limiting the role of product liability law is also propondered by Ryan Abbott. See Abbott (2018).
for the product that caused the accident. For complex design issues, jurisprudence
has a tendency to employ the risk-utility test to define a defect. The downside of this
test is the need to have an insight in the technology used for, in this case, autonomous
vehicles. It goes without saying that this is beyond the ability of an average person
and therefore experts need to intervene in any procedure leading to the determination
of a defect.64
In relation to the failure of the DNN to correctly recognize traffic signs, Lim
purports that it is extremely difficult to define, what she calls, the “standard of care”65
for developing reliable algorithms. This implies that she is looking for a solution
within the field of artificial intelligence for the problems adversarial machine learning
is dealing with. Her analysis of the training process of algorithms confirms this.
Poorly functioning algorithms could be related to the size or the quality of the dataset.
In other words, experts will have to check whether the dataset of a manufacturer
was sufficient in size and quality to teach the algorithm to correctly interpret its
environment.66 Surden has developed another perspective and advocated for a fault
tolerance standard.67 Fault tolerance means that the autonomous vehicle should be
designed to enable the system to detect an error and to respond to it. Unlike Lim’s
approach, this does not necessarily require to situate the solution to the error of the
algorithms. It is perfectly possible that an algorithm misinterprets a traffic sign. If
the processing computer can than rely on an alternative system, for example the
annotated digital map, it is possible for the autonomous vehicle to detect the error. In
such a case, the system needs to decide which of the two interpretations of the reality
is the most reasonable one and proceed on that interpretation. If the system is not
able to make that decision, the fault tolerance standard could require the system to
enter in a graceful degradation. In other words, the vehicle should be able to switch
to a safe mode either by relying on the driver or by stopping.
European legislation has a more detailed provision on what a defect is. Rather than
being two separate tests, Article 6 of the EU Directive 85/374/EEC emphasizes the
safety a person is entitled to expect. This expectation will be influenced by, among
other, the presentation of the product, its reasonable expected use and the timing
of putting it into circulation. The Directive further stipulates in Article 7 (3) that
a manufacturer can only escape liability if he has used the state of scientific and
technical knowledge at the time the product was put into circulation. Combining
these two Articles with the research on interpreting traffic signs by algorithms, there
would only be one option for manufacturers. To avoid liability, manufacturers will
need to address the possible problem of the system misqualifying traffic signs.68 Just
like in Surden’s example, this does not necessarily mean that the solution should be
64 Lengthy court proceedings may be the consequence of using product liability law for the traffic
accidents. See, e.g., Chagal-Feferkorn (2018b), pp. 17–22; Pearl (2018), p. 19; Ravid (2014), p. 200.
65 Lim (2018), pp. 82–98.
66 Lim (2018), pp. 82–98.
67 Surden (2019).
68 Lim (2018), pp. 107–108.
related to the algorithms. Installing a state-of-the-art backup system could suffice. If

required, the software or the back-up systems need to be updated.69
In any of the legislative interpretations above, whether it is the one described by
Lim, the one elaborated by Surden, or the one applying the European product liability
directive, manufacturers of autonomous vehicles could have a defense against their
liability. Lim puts forward that the manufacturer could assert that the current state
of technology does not allow them to detect the failure of the algorithms.70 If the
manufacturer has installed a backup system as suggested by Surden, the question of
liability grows more complex. The appreciation that needs to be made is whether the
combination of an algorithm and a backup system is sufficient. Hence, no matter what
approach is taken, a simple accident may end up in a lengthy trial before court.71
If the victim of the accident is not able to prove, most likely with the assistance
of expensive experts, that an alternative safer technique existed or that the backup
system was not state-of-the-art, she will not be compensated for her damage.
Even if the victim of the accident is able to show that the autonomous vehicle is
defective, and the manufacturer is not able to rely on any of the defenses, the manu-
facturer may still call upon contributory negligence. Three scenarios are thinkable.
The manufacturer could point to the contributory negligence of the driver, the road
infrastructure management office, or the vandal. It could be argued that the driver
should be careful when he sees vandalized traffic signs, that the road infrastructure
management office will have the duty to keep the traffic signs clean, and that the van-
dal has committed a crime by vandalizing the public domain. Of all three, the road
infrastructure management office is the most likely to have to (partly) compensate
the manufacturer.
From the above explanation, we could deduce that a car manufacturer who neglects
to adapt the visualization technology would face liable claims under product liability
law.72 In line with the predictions of Grossman, this may prevent car manufactur-
ers from proceeding to implement this innovative technology. On a more general
level, a similar argument is put forward by scholars discussing product liability and
autonomous vehicles. Innovation will be stifled or the introduction of new products
will be delayed if looming liability claims surround the car manufacturers.73 Whether
all these claims will be successful is another question. Realizing the complex burden
of proof for the victim and the defenses available to the producer, one can ques-
tion the suitability of product liability law for traffic accidents. Other systems for
compensating victims of autonomous vehicle crashes need to be explored.
69 Lohmann (2016), pp. 337–338.

70 Lim (2018), p. 100.
71 See, e.g., Chagal-Feferkorn (2018b), pp. 17–22; Pearl (2018), p. 19; Ravid (2014), p. 200.
72 For a general discussion, see Wu (2015), p. 565.
73 See, e.g., Pearl (2018), p. 20; Smith (2017), p. 6; Lohmann (2016), p. 338; Colonna (2012),
p. 114. This is an often-heard critique. However, there is no clear evidence that product liability law
in general generates more safety. See Smith (2017), pp. 4–5.
5.2 The Operator or Overriding Provisions, Shifting Liability

to a Human Driver
Autonomous vehicles, as we have described above, do not necessarily exclude a

human driver. Up until Level 4, a human driver has even a duty to control or intervene.
In early 2019, there was no autonomous vehicle commercialized that has a higher
level of automation than Level 3. Even though there have been predictions that
high level autonomous vehicles will hit the road soon, a shift to fully autonomous
vehicles is not going to be for the near future. The presence of the human has inspired
legislators to keep the focus of liability on the human. Very often, the legislation
focusing on the human does not make distinction between the different levels of
automation. The human is brought into the ambit by stipulating that the operator
of the vehicle will be liable or the one that should be able to retake control of the
autonomous vehicle.
States like Florida, California, Oregon, Nevada, Texas and New York have legisla-
tion for autonomous vehicles in which the operator has a crucial role in the liability.74
In most of these states, the liability extends both to civil and criminal liability. Thus,
liability could extend to a case in which the autonomous vehicle hits a pedestrian or
another vehicle. In terms of defining the operator of the autonomous vehicle, the law
in Florida stipulates for example that it will be the “person [that] causes the vehicle’s
autonomous technology to engage.”75 California formulates it as the “person who is
seated in the driver’s seat, or if no person in the driver’s seat, causes the autonomous
technology to engage.”76 In other words, the person pressing the start button will be
the person legally responsible for the actions of the autonomous vehicle. In some
states, like Nevada, Florida and California, the legislation does even not require the
operator to be in the vehicle to be responsible.77
The operator provision is being criticized for its lack of precision.78 Due to the fact
that there is no distinction made between the human driven mode and the autonomous
mode, the scope of liability is regarded as too broad.79 Why should an operator
be liable in case the cruise control of a Level 2 vehicle malfunctions?80 Is there
a reason to point at the operator in case the alarm system in a Level 3 vehicle
fails to send a warning?81 A Level 4 vehicle does not even require a human to do
anything in the car, besides setting the destination. Should the operator therefore
be liable?82 To what extent does it matter for society to blame the operator for
74 Pearl (2017b), p. 49.

75 Pearl (2017b), p. 49.
76 Pearl (2017b), p. 49.
77 Pearl (2017b), p. 49.
78 Pearl (2017b), pp. 50–57.
79 Pearl (2017b), p. 51.
80 Pearl (2017b), p. 51.
81 Pearl (2017b), p. 53.
82 Pearl (2017b), p. 53.
something he cannot control?83 Moreover, people may be deterred from owning

an autonomous vehicle. Arguments that indemnification by the manufacturer may
alleviate the burden imposed on the human operator have been refuted by scholars.
They argue that the operator has seldomly been successful in proving a defect in the
product.
A variation of the operator provision is the override provision. States like Nevada,
California, Florida and the District of Columbia have this kind of provision.84 It
basically means that the vehicle needs to have a driver who can retake control when-
ever necessary. The District of Columbia requires that an autonomous vehicle has
a driver “seated in the control seat of the vehicle while in operation who is pre-
pared to take control of the autonomous vehicle at any moment.”85 Oregon is more
concrete and stipulates that an autonomous vehicle should give a driver the ability
to override the car using the brake, accelerator or the steering wheel. Colorado’s
law has a similar provision but adds to the list of equipment to control the car an
override switch. “Easily accessible means of engaging or disengaging the vehicles’
autonomous technologies”86 is the requirement put forward in the legislation of New
York.
The criticism towards the override provision stems from the assumptions underly-
ing such a provision.87 When the law prescribes that a human operator should retake
control of the vehicle, the law presumes that the operator is adequately supervising
the autonomous vehicle and that the operator has the capacity to retake control when
necessary.88 Various studies have shown that these presumptions are utopian.89 The
operators of Level 3 vehicles got easily distracted and started looking around.90 Some
of the operators engaged in secondary tasks like emailing or reading.91 If the operator
was paying attention, it still took some time for him to understand the warning and
properly react.92 There are also studies indicating that the less an operator is engaging
with the vehicle, the less aware he will become of what skills are required to drive
a vehicle.93 Making an abstraction of the ability to correctly intervene, the override
provision forces the operator to be all time vigilant. Consequently, this provision will
diminish the benefits that autonomous vehicles will bring to society. This provision
may also hamper full innovative deployment of Level 4 vehicles, because, indirectly,
the override provision imposes standards on the autonomous vehicles.
83 Pearl (2017b), p. 52.

84 Pearl (2017b), p. 57.
85 Pearl (2017b), p. 57.
86 Pearl (2017b), p. 57.
87 Pearl (2017b), pp. 58-62.
88 Pearl (2017b), p. 58.
89 See, e.g., Llaneras et al. (2013); Pearl (2017a, b).
90 Llaneras et al. (2013), p. 94.
91 Goodall (2014).
92 Goodall (2014).
93 Jamson et al. (2013), pp. 116–125.
Imposing liability on the operator or on the one who should be able to take control
would be taking the focus away from the manufacturer. The manufacturer would be
able to innovate and, to the extent possible, rely on the vigilance of the operator for its
imperfect technology. Manufacturers could even give an extra warning that operators
should be extra careful in some circumstances. In our example of the visualization
technology, the manufacturer could argue that the technology is near to perfect as
long as the traffic signs to be recognized are in perfect state. It is when the traffic
sign is altered that the technology potentially faces problems. Manufacturers could
warn vehicle operators about this problem and require to take control of the vehicle
if alterations to a traffic sign, whether it is due to graffiti or stickers, is detected. This
opens the question of course whether operators will be constantly paying attention to
traffic signs for all types of autonomous vehicles and that they are aware to respond
correctly.
Besides the critique of whether this is an appropriate way of addressing liability
for autonomous vehicles, it should also be mentioned that innovation is not nec-
essarily stimulated by this liability regime. At first sight, the manufacturer may be
able to innovate. The manufacturer will not directly bear the consequences if his
innovation goes wrong. However, the space for innovation is limited. When Google
had put forward a model of its Waymo, no steering wheel was provided in the vehi-
cle. Though this kind of innovation would be technologically possible, legislation
imposing liability on the operator or on the one who has a duty to retake control
will inhibit this innovation. Indirectly, this kind of legislation puts forward standards
for the autonomous vehicles. Standard setting is something against which several
scholars have argued.94 The technology for autonomous vehicles is still in full devel-
opment. Without having a full understanding of the available options for developing
autonomous vehicles, it may be better to judge the trustworthiness of the vehicle at
the moment it is developed, rather than a priori determine what kind of technology
should be incorporated in the vehicle.
5.3 Beyond Liability Regimes: No-Fault Compensation

Schemes and Victim Compensation Funds
Product liability law has the risk of transforming a simple traffic accident into a
complex law suit. If no defect could be defined, third party victims would even face
no compensation for their damage. Legislation imposing liability on the operator
or the one in control is directed to someone who is not always to blame for the
accident. Further, both liability regimes may inhibit innovation. An alternative for
94 See on this aspect, Adkisson (2018); London (2019); London and Danks (2018).
both liability regimes may be the no-fault compensation scheme.95 Similar to this
no-fault compensation scheme is the victim compensation fund.96
The no-fault compensation scheme has already been tested in jurisdictions like
Israel, New Zealand, Quebec and Sweden, be it in the framework of human-driven
vehicles.97 The idea behind a no-fault compensation scheme is that an insurer is
responsible for providing compensation in case a “defined, uncertain factual event
occurs.”98 Typical for this kind of compensation will be provided irrespective of a
defect in the product, a fault of the manufacturer, or negligence of the operator of the
vehicle. Maurice Schellekens also notes that the amount of compensation is often
determined in advance in the law.99 Victim compensation funds have also been tested
in the United States, mainly in the aftermath of big disasters such as the September
11 attacks and the Deep Water Horizon oil spill.100 This kind of fund also does not
decide on a fault. However, unlike the non-fault compensation scheme, the existence
of a fund would not exclude the application of normal tort law.101 If the victim does
not opt for the application of the fund, tort law will be by default the basis to judge
the liability.
No-fault compensation schemes can take different forms.102 The insurance taker
could be either the possessor of the vehicle or its manufacturer. The rationale to
choose the former is that he benefits from the use of the vehicle. The rational for
the latter is that the accident is most likely caused by non-adequately functioning
technology. It needs to be further elaborated how the manufacturer contributes for
the insurance. A lump sum will prevent the difficulty of knowing for which vehicles a
contribution needs to be paid. However, with increasing connectivity, a manufacturer
may be able to control for which vehicles contributions need to be paid.103 It is
furthermore imaginable that private ownership of vehicles will cease to exist and
that the manufacturers pool their vehicles to be dispatched when someone applies
for it through an online application. In such a case, the insurance may not be such a
big issue anymore. The contributions to the victim compensation fund could come
from more sources than the just described ones. Victim compensation funds have
also been funded by tax revenue or charitable contributions.104
95 See, e.g., Engelhard and de Bruin (2017), pp. 111–115; Schellekens (2018). We read that this
can be realized by giving the self-learning robot an electronic legal entity, see, e.g., Herrmann
et al. (2018), p. 238. In the United Kingdom, an insurance for the vehicle driving autonomously is
elaborated. See Channon (2019), pp. 22–33.
96 See, e.g., Pearl (2018); Schroll (2015).
97 Schellekens (2018), p. 320.
100 Pearl (2018), p. 4.
101 Pearl (2018), p. 22.
102 Schellekens (2018), pp. 324–327. Compare Pearl (2018), pp. 29–36 for a discussion on the
different forms of contribution.

103 Schellekens (2018), pp. 324–325.
104 Pearl (2018), pp. 31–32 and 35.
There is also no fixed format for the insurer in a no-fault compensation scheme
or for the administrator of the victim compensation fund.105 The insurer106 or the
administrator could be the state,107 third-party private insurance companies or fund
administrators,108 or the car manufacturers themselves.109 A charity is still another
option for a victim compensation fund.110
A choice between the different no-fault compensation schemes or victim com-
pensation funds could be approached on the basis of the purpose to be achieved.111
Two purposes are often advanced: compensation of the victim and prevention of
accidents. Both purposes are tampered by the criteria that innovation should still be
possible. It is accepted that any kind of no-fault compensation scheme or victim com-
pensation fund facilitates compensation more than any of the other liability based
regimes.112 If the no-fault compensation scheme or the victim compensation fund
also needs to contribute to prevention, in other words stimulate the manufacturer not
to market unsafe products, it is probably better to shift the cost of the scheme or the
fund to the manufacturer.113 This could be both realized if the manufacturer engages
in self-insurance, pays to private insurance companies or contributes to a fund.114
The question still not answered is whether the burden on the manufacturer will have a
chilling effect on innovation. It is true that the financial burden may have to be borne
by the manufacturer, but unlike with other liability schemes, a no-fault compensation
scheme or a victim compensation fund operates with more transparency: first, the
scope of the duty to compensate is a priori defined; second, the amount to be paid
is also stipulated. Therefore, manufacturers face a more manageable compensation
scheme.115
Without having expressed a preference for the no-fault compensation scheme or a
victim compensation fund, an increasing number of manufacturers are moving in this
105 Schellekens (2018), pp. 325. According to Pearl, this would also apply to the victim compensation
fund. See Pearl (2018), pp. 28–37.

106 On the insurers, see Schellekens (2018), pp. 325–326.
107 The victim compensation fund can even be administered by the court or by an administrative
government agency. See Pearl (2018), pp. 29–30.

108 Pearl (2018), p. 33.
109 Pearl (2018), p. 33; Schroll suggests one run by the Federal Government. State-run and private
insurance are alternatives. See Schroll (2015), pp. 822–827.

110 Pearl (2018), p. 35.
111 Schellekens (2018), pp. 325–326.
112 Pearl (2018), p. 23; Schellekens (2018), pp. 325.
113 Schellekens (2018), p. 328. The idea of a fund is supported by Davola (2018), and Abraham and
Rabin (2017). The latter two disagree on the funding. The former contemplates that the government
should intervene with funding, while the latter impose manufacturer enterprise responsibility and
thus require funding from the manufacturers.
114 Pearl states that a contribution to the fund by all manufacturers in an industry is unprecedent in
the United States. Therefore, it may not be a viable solution. Pearl (2018), pp. 36–37.
115 The need to make it predictable has been stressed by several scholars. See, e.g., Pearl (2018),
pp. 20–21; Schellekens (2018), p. 329; Smith (2017), p. 6.

direction. The Volvo CEO has “publicly indicated that they want to take responsibil-
ity for accidents with their self-driving vehicles.”116 CEOs of other car manufacturers
have made similar statements.117 Not a single accident with a vehicle with advanced
automation technology has reached the court. Instead, the manufacturers have inter-
nalized the costs of the accident by settling the case. In doing so, the manufacturers
prevent the case from attracting negative publicity. However, these settlements have
also positive effects. Imagine what would happen if the accidents would have gone
to court and the victims were not able to show a defect to the advanced technology
of the automation. Anyone interested in a vehicle with advanced automation tech-
nology would have lost trust. Yet, the current situation does not allow to say that the
current stance of the manufacturers would lead to a general acceptance of a no-fault
compensation scheme within the industry and among legislators.
6 Discussing Liability in an Area in Evolution
The legislator can steer the safety of products with the choice for a specific liability
regime. The discussion above is based on a status quo in relation to car ownership
and to road infrastructure. Ride sharing apps are showing that the future of mobility
is not necessarily in the direction of private ownership of cars. If a shift towards full
autonomous vehicles is possible, the need for private car ownership may even further
decrease. This is not to say that a different car ownership will solve the problems
we have described in relation to visualization technology. However, a proper attri-
bution of liability could force the manufacturers to be careful in marketing defective
products.
Compared to the future of ownership, road infrastructure has not yet seen a big
change. This may be explained by the fact that car manufacturers and other tech-
nology firms were emphasizing the development of autonomous vehicles by them-
selves. However, the burden of moving towards electrified vehicles in combination
with automation has created an increased interest in assisted automation. If assisted
automation rolls out, technology may address the problems with visualization tech-
nology. Other instruments could reduce the need for a vehicle to correctly recognize
traditional traffic signs.
6.1 Changed Car Ownership Structures
One of the papers addressing the potential for ownership change is the article by
Carrie Schroll, Splitting the Bill: Creating a National Car Insurance Fund to Pay
116 Schellekens (2018), p. 326.

117 Geistfeld (2017), p. 1629.
for Accidents in Autonomous Vehicles.118 Schroll develops the idea that ownership
of cars may shift to car-sharing companies.119 The vehicles of these car sharing
companies would be dispatched to the people ordering one through a ride sharing
app. All the necessary information for the ride could be transferred through this app:
the number of passengers, the destination, and eventually also the preferred route.
The vehicle dispatched would therefore have all information prior to the entering of
the person ordering the vehicle. It would thus be hard to argue that this person would
be the operator of the vehicle, if there is a shift towards fully autonomous vehicles. It
could be argued that the car sharing companies are the one engaging the vehicles and
should therefore be seen as the operator. The operator liability scheme could apply
to these car-sharing companies, which could take a first-party insurance for all its
vehicles.
Schroll sees two reasons for agreeing with liability on the car-sharing companies:
“First, they are in a somewhat better position than drivers to improve the safety of
(Autonomous Vehicles) because they can put pressure on manufacturers by their car-
buying decisions. If they need to buy the safest cars because they fear liability, they
will encourage competition among manufacturers to improve the safety of the cars
they make of companies. Second, the companies will reap a substantial benefit from
putting AVs on the road, but in doing so the companies create the risk that the cars
may malfunction and cause an accident. Because these companies are willing to take
this risk in the name of profit making, they should also pay the costs when the harm
from this risk occurs.”120 The upside of imposing the liability on the car-sharing
industry may be that it will make the cost for entering the business prohibitive. To
overcome the latter, Schroll advocates for a no-fault compensation fund to which the
car-sharing companies would contribute.121
6.2 Infrastructure Enabled Autonomy
Another option for the future is being explained by Swaminathan Gopalswamy and
Sivakumar Rathinam.122 Both authors indicate that “while the physical components
of the modern automobile have become quite safe and reliable (failures of engines,
transmissions or other such systems are quite rare), the safety-focus has been chal-
lenged by the rapid growth in both scope and complexity of embedded software
118 Schroll (2015). A schematic difference between human driven cars and autonomous vehicles
provided by Kellerman (2018), p. 137. The fact that drivers become passengers, there may be a
different ownership structure.
119 Schroll (2015), pp. 818–819.
120 Schroll (2015), pp. 818–819.
121 Schroll (2015), pp. 819–820.
122 Gopalswamy and Rathinam (2018). For an alternative explanation, see Herrmann et al. (2018),
pp. 129–139.
functionality in cars. The number of software related recalls are growing exponen-
tially. Such safety concerns are exacerbated for autonomous vehicles, where we are
trying to replace human decision making with algorithms. The use of machine learn-
ing for both perception and decision making brings in an inherent non-determinism
to the system performance making it nearly extremely difficult, if not impossible, to
assert performance safety of the autonomous vehicles.”123
To avoid that such a situation will lead to the end of autonomous vehicles, both
authors suggest to establish an infrastructure enabled autonomy. Rather than mak-
ing the vehicle responsible for generating the situational awareness and the decision
making based upon that awareness, these tasks should be transferred to infrastruc-
ture providers. More concrete, “the situational awareness will be generated through
sensors that are embedded in the infrastructure. …Decision making is provided
by yet another third party that takes the situational awareness information com-
ing from the infrastructure operators…, and uses standardized Application Program-
ming Interfaces…to interface with the [drive-by-wire] capabilities…to drive the cars
autonomously.”124 The driving process of the autonomous vehicles would so be split
between three different actors: the car manufacturer, the infrastructure operator and
the decision making provider. From an engineering point of view, this may facilitate
the technological development towards fully autonomous vehicles. However, from a
legal point, the situation will only become more complex.
The complexity of an infrastructure enabled autonomy is created by an increased
number of actors, many of them relying on extra technology or products. Besides,
not all of these actors are delivering a product. Some of them will engage in provid-
ing a service. The car manufacturer provides the vehicle, which is a product. The
infrastructure operator overviews a special road infrastructure that can communicate
with participants in the traffic. The operators will provide a service of connectivity,
but underlying this service is a product made up of sensors and receivers. The latter
will be most likely provided by another manufacturer. The information gathered by
the system will be analyzed to render a decision to the vehicles on the road. The
decision making provider is delivering a service. To render that service, this provider
also has to rely on products purchased elsewhere. This means that what is now a
simple accident between two vehicles for which the behavior of the drivers will be
analyzed, will become something in which it has to be assessed whether the vehicle
showed a defect, the road/decision making infrastructure was defective, the operator
was negligent, or the provider was negligent. On top of that, the connectivity, so the
authors point out, creates the risk of hacking.
Due to the separation of products and services, the Gopalswamy and Rathinam
model of infrastructure enabled automation does not allow to make an argument that
‘being connected’ is part of the package offered by the car manufacturer.125 Hence, if
product liability law were to apply to infrastructure enabled automation, each product
123 Gopalswamy and Rathinam (2018), p. 1.

124 Gopalswamy and Rathinam (2018), p. 3.
125 Gopalswamy and Rathinam (2018). Compare Geistfeld (2017), pp. 1660–1674 (discussing the
understanding that the technology for connected vehicles is offered by the car manufacturer).
within the infrastructure could lead to liability for its manufacturer on the condition
that a defect could be found. Besides, the products need to be maintained, updated
and provided. If, due to negligence, something goes wrong within this process, a
tort case could be started against the infrastructure operator or the decision making
provider. It could be questioned whether a traffic accident should lead to a complex
investigation on who is actually responsible in order to than compensate the victim.
Operator liability would make an abstraction of the technology behind infrastructure
enabled automation. When the infrastructure fails, the idea would be that the operator
would retake control of the vehicle. But why should an operator carry the burden of
liability? He or she has totally no control over the products and services provided.
The burden of product liability law or the unfairness of operator liability could be
taken away by a no-fault compensation scheme or victim compensation fund. Most
likely, this is also underlying the Gopalswamy and Rathinam model.126 Their idea is
to distribute the risks of accidents from autonomous driving among several actors.
Each of the different actors in the realization of autonomous driving would contribute
to the scheme or the fund. Without doubt these costs will finally be spread over the
customers of the mobility service. A part of the fee that the customers have to pay
to be transported from one location to another will entail the insurance costs.127
7 Conclusion
The feasibility of autonomous vehicles hitting the road has been questioned. The
reason for this observation is that car manufacturers are facing a problem with visu-
alization technology necessary to interpret, for example, traffic signs. Research has
shown that even small alterations to a traffic sign may prevent this technology from
correctly recognizing the content of that traffic sign. If car manufacturers do not
address this problem, liability may be imposed on them based upon product liability
law. This liability may be too much of a burden to proceed with the commercial-
ization of autonomous vehicles. Whether this should be the end of the development
of autonomous vehicles is dubious. Liability could either be taken away from the
car manufacturers and shifted to the operator or the person in control of the vehi-
cle. Another option would be to create a no-fault liability regime. The advantage
of this regime is that the liability is more transparent. None of the liability regimes
address the underlying technological problem. It only addresses the incentive of the
car manufacturers to take innovative risks. If car manufacturers do not succeed in
overcoming the visualization technology problem, the introduction of other infras-
tructure enabling automation through connection to the environment may be the
126 Gopalswamy and Rathinam (2018).

127 An estimation would be that insurance per vehicle would be 675 dollars for 12 years. Therefore,
a shift towards the consumers would be possible. See Herrmann et al. (2018).
solution. However, this option will require a regime to tackle the liability for acci-
dents. At first sight, a no-fault regime to which all actors contributing to the risk will
contribute.
Acknowledgements This research has been presented at Congress in Congress II during the
20th Annual Congress of the International Academy of Comparative Law (Fukuoka, July 22–
28, 2019) and at the workshop The Future of Autonomous Vehicles (Fukuoka, February 1, 2019).
The research further received financial support from Kyushu University’s QR Program: Tsubasa
Project [Interdisciplinary Research by Young Researchers] 2018–2020.
References
Abbott R (2018) The reasonable computer: disrupting the paradigm of tort liability. George
Washington Law Rev 81(1):2–45
Abe R (2018) Transition to autonomous public transportation: impacts on trip service levels
and potential benefits for transportation systems (Working Paper). https://papers.ssrn.com/sol3/
papers.cfm?abstract_id=3270747. Accessed 10 Apr 2019
Abraham KS, Rabin R (2017) Automated vehicles and manufacturer responsibility for accidents:
a new legal regime for a new era. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3151133.
Accessed 10 Apr 2019
Adkisson SD (2018) System-level standards: driverless cars and the future of regulatory design.
Univ Hawaii Law Rev 40(1):1–52
Audi MediaCenter (2017) The new Audi A8—conditional automated at Level 3. https://www.audi-
mediacenter.com/en/on-autopilot-into-the-future-the-audi-vision-of-autonomous-driving-9305/
the-new-audi-a8-conditional-automated-at-level-3-9307. Accessed 10 Apr 2019
Boston Consulting Group (2019) Self-driving vehicles, car sharing, and the urban mobility revo-
lution. https://www.bcg.com/ja-jp/industries/automotive/self-driving-vehicles-car-sharing.aspx.
Brown TB et al (2017) Adversarial patch. https://arxiv.org/pdf/1712.09665.pdf. Accessed 10 Apr
2019
Carlini N, Wagner D (2017) Adversarial examples are not easily detected: bypassing ten detection
methods. https://arxiv.org/pdf/1705.07263.pdf. Accessed 10 Apr 2019
Chagal-Feferkorn KA (2018a) Am I an algorithm or a product? When products liability should apply
to algorithmic decision-makers. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3241200.
Chagal-Feferkorn KA (2018b) The reasonable algorithm. J Law Technol Policy 1:111–147
Channon M (2019) Insurance. In: Channon M, McCormick L, Noussia K (eds) The law and
autonomous vehicles. Informa Law from Routledge, Abingdon, pp 14–33
Channon M, McCormick L, Noussia K (eds) (2019) The law and autonomous vehicles. Informa
Law from Routledge, Abingdon
Colonna K (2012) Autonomous cars and tort liability: why the market will ‘drive’ autonomous cars
out of the marketplace, 4 Case W. Res. J. L. Tech. & Internet 81
Crane DA, Logue KD, Pilz BC (2017) A survey of legal issues arising from the deployment of
autonomous and connected vehicles. Mich Telecommun Technol Law Rev 23:91–319
Davola A (2018) A model for tort liability in a world of driverless cars: establishing a framework
for the upcoming technology. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3120679.
Deloitte (2019) On demand car services and carpooling: providing more options for non-drivers.
https://www2.deloitte.com/us/en/pages/consulting/solutions/smart-mobility.html. Accessed 10
July 2019
Engelhard EFD, de Bruin RW (2017) EU common approach on the liability rules and insurance
related to connected and autonomous vehicles. In: Evans T (ed) European added value assessment:
EU common approach on the liability rules and insurance related to connected and autonomous
vehicles: accompanying the European Parliament’s legislative own-initiative report. European
Added Value Unit, Brussels, pp 38–131
Evas T (ed) (2018) European added value assessment: EU common approach on the liability
rules and insurance related to connected and autonomous vehicles: accompanying the European
Parliament’s legislative own-initiative report. European Added Value Unit, Brussels
Evtimov et al (2017) Physical adversarial examples against deep neural networks. https://bair.
berkeley.edu/blog/2017/12/30/yolo-attack/. Accessed 10 Apr 2019
Eykholt K et al (2018) Robust physical-world attacks on deep learning visual classification. https://
arxiv.org/pdf/1707.08945.pdf. Accessed 10 Apr 2019
Fuchs A (2019) Sensors for autonomous vehicles—how upcoming vehicles perceive their environ-
ment. In: Presentation at the future of autonomous vehicles conference, Kyushu University, 1 Feb
2019
Gagan O (2018) How car-sharing services transformed manufacturing: servitisation and the shar-
ing economy are shaping the future of the car industry as easy access looks more attractive
than ownership. https://www.raconteur.net/manufacturing/car-sharing-manufacturing. Accessed
10 Apr 2019
Geistfeld MA (2017) A roadmap for autonomous vehicles: state tort liability, automobile insurance,
and federal safety regulation, 105 California Law Review 1611
Goodall NJ (2014) Machine ethics & automated vehicles. In: Meyer G, Beiker S (eds) Road vehicle
automation. Springer, Heidelberg, pp 93–102
Gopalswamy S, Rathinam S (2018) Infrastructured enabled autonomy: a distributed intelligence
architecture for autonomous vehicles. https://arxiv.org/pdf/1802.04112.pdf. Accessed 10 Apr
2018
Grossman M (2017) Another reason self-driving cars aren’t happening. https://www.injuryrelief.
com/blog/another-reason-self-driving-cars-arent-happening/. Accessed 5 Mar 2018 (link is
disconnected at 10 Apr 2019)
Gurney JK (2013) Sue my car not me; Product liability and accidents involving autonomous vehicles.
J Law Technol Policy 2:247–277
Herrmann A, Brenner W, Stadler R (2018) Autonomous driving: how the driverless revolution will
change the world. Emerald Publishing, Bingley
Houben S et al (2013) Detection of traffic signs in real-world images: the German
traffic sign detection benchmark. https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=
6706807. Accessed 10 Apr 2019
Jamson AH et al (2013) Behavioral changes in drivers experiencing high-automated vehicle control
in varying traffic conditions. Trans Res Part C Emerg Tech 30:116–125
Karungaru S, Nakano H, Fukumi M (2013) Road traffic signs recognition using genetic algorithms
and neural networks. Int J Mach Learn Comput 3(3):313–317
Kellerman A (2018) Automated and autonomous spatial mobilities. Edward Elgar Publishing,
Cheltenham
Kurakin A, Goodfellow I, Bengio S (2017) Adversarial machine learning at scale. https://arxiv.org/
pdf/1611.01236.pdf. Accessed 10 Apr 2019
Lim YFH (2018) Autonomous vehicles and the law: technology, algorithms and ethics. Edward
Elgar Publishing, Cheltenham
Llaneras RE, Salinger J, Green CA (2013) Human factors issues associated with limited ability
autonomous driving systems: drivers’ allocation of visual attention to the forward roadway, pp
92–98. In: Proceedings of the seventh international driving symposium on human factors in
driver assessment, training and vehicle design. https://doi.org/10.17077/drivingassessment.1472.

Lohmann MF (2016) Liability issues concerning self-driving vehicles. Eur J Risk Regul 2:335–340
London AJ (2019) Regulating autonomous vehicles: from fixed standards to dynamic oversight. In:
Presentation at the future of autonomous vehicles conference, Kyushu University, 1 Feb 2019
London AJ, Danks D (2018) Regulating autonomous vehicles: a policy proposal. In: Pro-
ceedings of the 2018 AAAI/ACM conference on artificial intelligence, ethics, and soci-
ety. https://www.cmu.edu/dietrich/philosophy/docs/london/London%20Danks%20Regulating%
20Autonomous%20Systems%20-%20flattened.pdf. Accessed 10 Apr 2019
Maurer M et al (eds) (2015) Autonomous driving: technical, legal and social aspects. SpringerOpen,
Heidelberg
McCormick L (2019) Product liability. In: Channon M, McCormick L, Noussia K (eds) The law
and autonomous vehicles. Informa Law from Routledge, Abingdon
Moosavi-Dezfooli et al (2017) Analysis of Universal Adversarial Perturbations, Available at: https://
arxiv.org/pdf/1705.09554.pdf
Nguyen A, Yosinski J, Clune J (2015) Deep neural networks are easily fooled: high confidence
predictions for unrecognizable images. In: Proceedings of the IEEE conference on computer
vision and pattern recognition, pp 427–436. https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&
arnumber=7298640. Accessed 10 Apr 2019
Noussia K (2019) International comparisons. In: Channon M, McCormick L, Noussia K (eds) The
law and autonomous vehicles. Informa Law from Routledge, Abingdon
Pearl TH (2017a) Hands on the wheel: a call for greater Regulation of semi-autonomous cars.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2930125. Accessed 10 Apr 2019
Pearl TH (2017b) Fast & furious: the misregulation of driverless cars. NYU Ann Survey Am Law
73(24):19–72
Pearl TH (2018) Compensation at the crossroads: autonomous vehicles & alternative voctim com-
pensation schemes. In: 29th European regional conference of the international telecommunica-
tions society—towards a digital future: turning technology into markets?, pp 1–48
Ravid O (2014) Don’t sue me, I was just lawfully texting & drunk when my autonomous car crashing
into you, 44 Sw. L. Rev. 175 (2014–2015)
Rohr C, Dunkerley F, Howarth D (2018) Socio-economic analysis of the EU common approach
on the liability rules and insurance related to connected and autonomous vehicles. In: Evans T
(ed) European added value assessment: EU common approach on the liability rules and insur-
ance related to connected and autonomous vehicles: accompanying the European Parliament’s
legislative own-initiative report. European Added Value Unit, Brussels, pp 134–194
Rowland D, Macdonald E (2005) Information technology law, 3rd edn. Cavendish Publishing,
London
Schellekens M (2018) No-fault compensation schemes for self-driving vehicles. Law Innov Technol
10(2):314–333
Schroll C (2015) Splitting the bill: Creating a national insurance fund to pay for accidents in
autonomous vehicles. Northwest Univ Law Rev 109:803–833
Sitawarin C et al (2018) DARTS: deceiving autonomous cars with toxic signs. https://arxiv.org/pdf/
1802.06430.pdf. Accessed 10 Apr 2019
Smith BW (2017) Automated driving and product liability, 1 Michigan State Law Review 1
Society of Automative Engineers (2018) SAE international releases Updated visual chart for
its “levels of driving automation” standard for self-Driving vehicles. https://www.sae.org/
news/press-room/2018/12/sae-international-releases-updated-visual-chart-for-its-%E2%80%
9Clevels-of-driving-automation%E2%80%9D-standard-for-self-driving-vehicles. Accessed 10
Apr 2019
Stallkampa J et al (2012) Man vs. computer: benchmarking machine learning algorithms for traffic
sign recognition. https://christian-igel.github.io/paper/MvCBMLAfTSR.pdf. Accessed 10 Apr
2019
Su J, Vargas DV, Kouichi (2017) One pixel attack for fooling Deep neural networks. https://arxiv.
org/pdf/1710.08864.pdf. Accessed 10 Apr 2019
Surden H (2019) Autonomous vehicles, predictability, and liability. In: Presentation at the future of
autonomous vehicles conference, Kyushu University, 1 Feb 2019
Surden H, Williams MA (2017) How self driving cars work. https://papers.ssrn.com/sol3/papers.
cfm?abstract_id=2784465. Accessed 10 Apr 2019
Szegedy C et al (2013) Intriguing properties of neural networks. https://arxiv.org/pdf/1312.6199.
pdf. Accessed 10 Apr 10 2019
Vargas DV, Murata J (2017) Spectrum-diverse neuroevolution with unified neural models. IEEE
Trans Neural Netw Learn Syst 28(8):1759–1773
Vargas DV, Takano H, Murata J (2013) Self-organizing Classifiers and niched fitness. In: Proceedings
of the 15th annual conference on genetic and evolutionary computation, pp 1109–1116. https://
arxiv.org/pdf/1811.08226.pdf. Accessed 10 Apr 2019
Vargas DV, Takano H, Murata J (2015) Novelty-organizing team of classifiers in noisy and dynamic
environments. In: IEEE congress on evolutionary computation, CEC 2015—Proceedings, pp
2937–2944. https://arxiv.org/pdf/1809.07098.pdf. Accessed 10 Apr 2019
Wu SS (2015) Product liability issues in the U.S. and associated risk management. In: Maurer M
et al (eds) Autonomous driving: technical, legal and social aspects. SpringerOpen, Heidelberg
The New EU Directive on the Protection
of Trade Secrets and Its Implementation
Thomas Hoeren
Abstract In June 2018, the deadline for the implementation of a new act reforming
the trade secret law expired. The author describes the underlying EU Directive on the
protection of trade secrets (Directive (EU) 2016/943, hereinafter “Directive”) and
outlines the options of implementation. Special attention is paid to the producer’s
liability in accordance with Article 4 (5) of the Directive.
Keywords Know-how protection · EU directive · Trade secrets · Implementation
1 Introduction
Valuable knowledge can leak from a company in several ways: Employees with
special knowledge leave the company or disclose trade secrets by photographing
and storing documents. Furthermore, companies can be victims of cyber-attacks and
e-mails can be intercepted and read. The estimated damage caused by industrial
espionage amounts to around 50 billion euros per year in Germany. Internationally,
the industrialized countries have taken precautions in consideration of Article 39 (2)
of the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS
Agreement). As a minimum standard, the Member States of TRIPS have commit-
ted themselves to protect information against unauthorized outflow as long as the
information is confidential. Information is assumed to be confidential, if it:
i. is not generally known or readily accessible to persons within the usual business
circles;
ii. has commercial value because it is secret, and;
iii. has been subject to reasonable steps under the circumstances to keep it secret.
In Germany, a special protection for trade secrets existed in the traditional crimi-
nal law provision of Section 17 to 19 of the Unfair Competition Act (UWG). These
T. Hoeren (B)
Institute for Information, Telecommunication and Media Law, Faculty of Law, University of
Münster, Münster, Germany
e-mail: hoeren@uni-muenster.de

https://doi.org/10.1007/978-981-15-1350-3_12
210 T. Hoeren
offences are as well sanctioned under civil law [Section 3 UWG, Section 823 (2)
German Civil Code (BGB)]. The standards of the Unfair Competition Act (UWG)
already existed when the TRIPS Agreement became effective. Accordingly, the reg-
ulatory objectives of the TRIPS Agreement and Sections 17 to 19 UWG are not
identical. New challenges occur concerning the proof of the confidential character of
information and concerning obligations to provide evidence of appropriate confiden-
tiality agreements. Compared to other European countries, Germany has an average
level of trade secret protection. While the United Kingdom has a sophisticated system
of confidentiality obligations, in countries such as France or Malta, confidentiality
agreements are virtually unknown. This persuaded the Commission to order two
studies on the protection of trade secret in a European comparison. The studies came
to the conclusion that the protection of secrecy in Europe resembles a patchwork car-
pet.1 Thereupon, the Commission adopted a first draft of a harmonization Directive
in November 2013.2 A common text followed in May 2013, which was adopted by
the Council of Ministers. One year later—in June 2015—the Parliament presented
its report on the draft Directive.3 Based on the report, the European Commission,
the Council and Parliament drafted a compromise proposal through (unofficial) tri-
logue negotiations, which was published in December 2015.4 Finally, the Directive
was adopted in June 20165 and is now to be transposed into national law by June
2018.6 This led to discussion at national levels whether the protection of trade secrets
should be included in a comprehensive set of rules of intellectual property rights,7
or at least regulated by a special law.8 In Germany, the election for the Bundestag
caused a discontinuity of the preliminary drafts of the former coalition. Thus, the
Federal Ministry of Justice (BMJV) had to reintroduce a draft bill into the parlia-
mentary debate. It is already certain that the deadline of implementation, which is
June 2018, cannot be met. Concerning the current level of information, the ministry
plans an implementation by creating a special law, but without touching Section 17
of the Unfair Competition Act (UWG).9 An internal draft, based on the participation
1 Study on Trade Secrets and Confidential Business Information in the Internal Market (2013),
pp. 12–23.
2 Proposal for a Directive of the European Parliament and of the Council on the protection of
undisclosed know-how and business information (trade secrets) against their unlawful acquisition,
use and disclosure.
3 Report on the proposal for a Directive of the European Parliament and of the Council on the
protection of disclosed know-how and business information (trade secrets) against their unlawful
acquisition, use and disclosure.
4 Proposal for a Directive of the European Parliament and of the Council on the protection of
undisclosed know-how and business information (trade secrets) against their unlawful acquisition,
use and disclosure—Analysis of the final compromise text with a view to agreement.
5 Directive 2016/943/EU from June 6, 2016.
6 Kalbfus (2016), p. 1009; Christoph (2016), p. 465.
7 With reference to Italy and Portugal, see McGuire (2016), pp. 1000–1007.
8 Like in Sweden for instance: Act on the Protection of Trade Secrets (1990).
9 Different McGuire (2016), pp. 1000–1008, who naturally acts on the assumption that Section 17 of
the Unfair Competition Act (UWG) must be overruled or changed; Kalbfus (2016), pp. 1009–1016,
The New EU Directive on the Protection … 211
of the Federal Ministry of Justice (BMJV) and the Ministry of Labor and the Min-
istry of Economics, exists since 2017. The draft provides a direct implementation of
the Directive into national law, but can also be supplemented by other elements, if
necessary.
2 Regulatory Objectives
The Directive aims at harmonizing the various existing national rules.10 It is supposed
to serve as a complementary or alternative measure to intellectual property rights
(Recital 2). The Directive keeps the relation to the Unfair Competition Act (UWG)
open so that a future review by the Commission is possible (Recital 10). Besides,
the Directive should only contain a minimum standard of harmonization for now11
[Article 1 (1)]. It should also be considered that the EU has no competence for
criminal law and does not intend to regulate work council constitution law or press
law [Article 1 (2)]. It is astounding that the Directive does not contain any regulation
on private international law (PIL). It could be linked to the criminal law, intellectual
property rights or the Unfair Competition Act (UWG). This problem of classification
underlines that the legal nature of the newly created system is unclear. If the system
is classified in terms of intellectual property rights, the Directive applies to all acts
of use with reference to the EU. Instead, the Unfair Competition Act (UWG) asks
for the final market intervention or the intended use of the product.
3 The Concept of Trade Secrets
The cornerstone of the Directive is the concept of trade secrets.12 It is broadly defined,
wider than with accordance to Section 17 of the Unfair Competition Act (UWG).13 A
secret is already supposed to be existent if the information is generally unknown or not
known by the relevant circle of experts.14 Even the existence of an outsider or a hacker
makes the information factually accessible. However, this issue was already known
pleads for a general law with reserved civil provisions and for the incorporation of the criminal law
especially in the event of blatant intentional form of breaches in the general law.
10 Christoph (2016), p. 465; Koós (2016), p. 224.
11 Different the Commission draft (supra note 2); see also Kalbfus and Harte-Bavendamm (2014),
p. 453.
12 Klein and Wegener (2017), p. 394.
13 Kalbfus (2016), pp. 1009–1011, presents a slightly different thesis, according to which the German
jurisdiction of Section 17 of the Unfair Competition Act (UWG) essentially complies with the
Directive. Nevertheless, he considers the transposition of the Directive into a special German law
as necessary.
14 McGuire (2016), p. 1000, with the main proposal for embedding the protection of secrecy into
the system of intellectual property.

212 T. Hoeren
concerning Section 17 of the Unfair Competition Act (UWG). The jurisprudence has
based its decision on the fact that gaining information requires a great expenditure
of time and cost and therefore, it focuses on the perspective of the circle of experts.15
In addition, the trade secret must embody a commercial value, lying precisely in its
secrecy. Therefore, the value of trade secrets must lie precisely in the secret.16 This
is completely unknown to German readers, as Section 17 of the Unfair Competition
Act (UWG) neither asks for the value of the information nor the existence of a secret.
Instead, the German law has always been based on the requirement of an objectively
existing economic interest. Thus, secrets without values are protected on condition
that the disclosure of these secrets can cause damage.17 The definition is taken from
Article 39 of the TRIPS agreement. Henceforth, it will be necessary to document the
value of an information steadily before the courts.18
On the one hand, this postulates a financial accounting of trade secrets. And on
the other hand, a proof of confidentiality must be provided as a status quo. Even
with this element, one has to prove a comprehensive secret management system that
goes all the way to encryption of e-mails and IT security.19 But that is not all: It
must be proven that the information has been subject to reasonable steps to keep it
secret [Article 2 (1) (c)]. Therefore, proactive and continuous secrecy measures are
needed in order to preserve the confidential character of the information. This requires
comprehensive and above all, long-term measures that are state-of-the-art.20 It is a
novelty in Germany that legitimate confidentiality measures are a requirement of
protection. So far, they only played a role in evidence law. The Federal Supreme Court
(BGH) had so far decided that a desire for secrecy resulting from the nature of the
matter was sufficient and even the absence of any explicit confidentiality agreement
was regarded as unproblematic.21 For this reason, many people are concerned about
the meaning of the term “reasonable steps.”22 The Directive does not clarify this
concept. It does not require the best possible or most effective protective measure,
but legitimate standard of protection.23 Thus, concrete steps have to be defined for the
intra-corporate implementation. The first step is to define clear responsibilities for
protecting trade secrets within the company. Subsequently, the potentially relevant
15 Federal Supreme Court (BGH), judgment of February 23, 2012—case no. I ZR 136/10; Federal
Supreme Court (BGH), judgment of February 12, 1980—case no. KZR 7/79.
16 Souso e Silva (2014), p. 923.
17 Kalbfus (2016), pp. 1009–1011.
18 See also Herrmann (2016), pp. 368–369.
19 Pacini et al. (2008), p. 121, with the presentation of a “trade secret compliance plan” includ-
ing different measures that can be taken as well as different variables that have to be considered
depending on the company.
20 Kalbfus (2017), p. 391; Börger and Rein (2017), p. 118; to the special feature in the automotive
sector Steinmann and Schubmehl (2017), p. 194.

21 Federal Supreme Court (BGH), judgment of April 27, 2006—case no. I ZR 126/03.
22 See Kalbfus (2011), pp. 65 et seq.; Kalbfus (2017), p. 391.
23 Kalbfus (2017), pp. 391, 392; inapplicable Steinmann and Schubmehl (2017), pp. 194–198, who
are aiming for a purely factual protection of secrecy and fail to recognize that the Directive has
quite normative ideas about the protection of secrets.
know-how must be identified, evaluated and categorized according to the company’s

size and industrial sector. Ultimately, precise protective measures must be defined,
including special contractual agreements, IT security measures and the organization
of workflows.24 In the automotive industry, for example, care is taken to ensure that
technical building services, information security and organizational measures are
carried out and audited. The certification of the IT security sector is carried out in
accordance with ISO 27001.25
Moreover, the dogmatic question remains unanswered whether or not the trade
secret is recognized as a subjective right, similar to traditional intellectual property
rights. The legal nature of trade secrets is highly disputed in Germany.26 The focus
lies on the personal attribution of the right and its exclusionary effect, which is
rejected by the majority concerning secrets because the character of secrecy can easily
disappear.27 Secrecy protection does not intend to grant a right of exploitation, but
rather a protection against unfair access. Recital 16 of the Directive states that the new
provisions of the protection of innovation and competition do not create an exclusive
right of know-how or information, which is protected as a trade secret. However, there
are several reasons to define the character of trade secrets as a minor intellectual
property right. The preamble in Recital 2 openly characterizes the protection of
secrecy as a supplement or an alternative to the traditional intellectual property rights.
Concurrently, recital 3 states that trade secret are one of the most common forms
of the protection of intellectual creations and innovative know-how. Even the scope
of optional sanctions argues in favor of a minor intellectual property right, which
is for example protected by a triple damage calculation like traditional intellectual
property rights. This secures the trade secret as a large intellectual property right.
Pursuant to Article 2 No. 2, the proprietor of a trade secret is the natural or legal
person who exercises legal control over the trade secret. Therefore, a trade secret
is protected just like an intellectual property right. Pursuant to Article 2 No. 2,
the proprietor of a trade secret is the natural or legal person lawfully controlling a
trade secret. Thus, it is not the creator who is essential, but the person exercising
the lawful control. Yet, the meaning of the term “control” remains unclear. If the
factual access possibility is taken into account, the employee would be the original
proprietor as long as the employer has not received the information. However, the
Directive focuses on the protection of companies; that is why there are no regulations
about secondary acquisitions of rights. Through the characteristic of reasonable steps
of protection, an original acquisition by the entrepreneur is established. This entails
the obligation for companies to create confidentiality structures by contractual and
operational attendance and thereby establish matters of original acquisition.28
24 Kalbfus (2017), pp. 391 et seq.

25 Steinmann and Schubmehl (2017), pp. 194–197.
26 For the classification of trade secrets as absolute rights, see Köhler (2019), §17 ref. 53; Ohly (2014),
pp. 8–9; different opinion Christoph (2016), p. 465; Hauck (2016), pp. 2218–2221; McGuire (2015),
pp. 424–426.
27 Christoph (2016), p. 465; Hauck (2016), pp. 2218–2221; McGuire (2015), pp. 424–426.
28 Klein and Wegener (2017), pp. 394–396.
214 T. Hoeren
4 The Central Prohibition
The central prohibition can be found in Article 4 (1) of the Directive. It should be
noted that Article 3 emphasizes the principle of freedom of access. Therefore, the
Directive firstly lists fair commercial practices. In any case, Article 4 includes only
exceptions for certain constellations in which the principle of freedom of access
cannot be applied. This requires the Member States to ensure that the trade secret
holders are entitled to apply for all the measures, procedures, and remedies to prevent,
or obtain redress for, the unlawful acquisition, use or disclosure of their trade secret.
These measures are specified in Article 4 (2) and (3).
First of all, the acquisition of secrets is prominent [Article 4 (2)]. The English
version of the Directive refers to the term acquisition. However, in this case the
official German translation (“Erwerb”) is not quite correct. It is not a question of
acquisition in return for payment, but rather of a factual obtaining. If certain matters
are given, the obtainment is illegal without the consent of the copyright holder.
This includes unauthorized access to, appropriation of, or copying of documents,
materials or electronic files that contain trade secret or from which the trade secret
can be deduced.
These restrictive prohibitions of appropriation are extended in Article 4 (2) (b),
where any other conduct will be sanctioned if it is considered contrary to honest com-
mercial practices. Yet, this term is very imprecise and hard to define. It is proposed
to outline it according to Article 39 (2) of the TRIPS Agreement.29
The list of prohibitions does not regulate the transfer of individual information
from an employee’s brain. If an employee working as a programmer reads company-
owned programs and saves the source code, this would not constitute a prohibited con-
duct. Therefore, it can be assumed30 that the use of information from an employee’s
brain is free. However, it may be possible to apply Article 4 (2) (b) in certain cases of
abuse. Yet, Article 1 (3) of the Directive dictates that no provision of this Directive
may be interpreted as a basis for restriction to the mobility of workers. Especially,
the Directive is not a basis for any restriction of the use of experience and skills
honestly acquired by workers through the standard procedures of their employment.
Moreover, it is not a basis for imposing additional contractual restrictions.
The use and disclosure of trade secret is regulated in Article 4 (3) of the Directive.
People who have unlawfully acquired the trade secret or have breached a confiden-
tiality agreement or other confidentiality obligations are prohibited from using the
trade secret without the consent of the rights holder. It remains unclear why the
Directive includes two different breaches of contract. This regulation results in the
enforcement of contractual obligations by means of the UWG. The validity of the
agreement is irrelevant. The term “legally binding,” which was originally included in
29 See Koós (2016), pp. 224–226.

30 Federal Supreme Court (BGH), judgment of May 3, 2001—case no. I ZR 153/99; Federal Supreme
Court (BGH), judgment of March 15, 1955—case no. I ZR 111/53; Federal Labor Court (BAG),
judgment of June 15, 1993—case no. 9 AZR 558/91; Federal Labor Court (BAG), judgment of
December 15, 1987—case no. 3 AZR 474/86.
earlier drafts of the Directive, has been deleted without substitution.31 It is also new
that an unlawful interference on Article 4 of the Directive is sufficient for sanctions.
The limitation of intent and gross negligence, which were originally intended in the
draft, are not included in the final text.32 For this reason, the question of intention
or negligence only plays a role in the compensation of damages according to Article
14 of the Directive. This also does not correlate with the applicable German law,
which always presupposes at least negligence of the defendant. In accordance with
the Directive, the secret holder can enforce claims for injunctive relief and omission
in the event of a mere breach of secrecy. This corresponds with the current legal
situation in the case of breach of intellectual property rights (cf. Sec. 14 (5) of the
trademark law (MarkenG), Sec. 139 (1) of the patent law (PatG)). Consequently,
the Directive extends the protection of trade secret and brings them into line with
intellectual property rights.
5 The Extended Liability of the Manufacturer
Further new regulations are included in Article 4 (4) and (5) of the Directive. These
extended matters of liability are generally unknown to German law and therefore
dangerous.
Accordingly, the acquisition of secret information is prohibited, if the person knew
or negligently did not know that the trade secret had been obtained directly or indi-
rectly from another person who was using or disclosing the trade secret unlawfully
(Article 4 (4) of the Directive). It is astounding that even the indirect use of the exter-
nal trade secret is sufficient. Thus, the new system is different from the old system of
the UWG. Due to contractual protection, third parties can only be held responsible
in the matters of Section 17 (1) or (2) No. 1 of the UWG. The new regulation is now
directed against any unlawful pre-acquisition of third parties. Section 17 focuses on
stricter subjective elements (intent) than the Directive (intent or negligence).33
In accordance with Article 4 (5), the production, offering or placing on the market
of infringing goods or the importation, export or storage of infringing goods for those
purposes, are considered an unlawful use of a trade secret. However, this only applies
if the person who carried out these activities had knowledge of the fact or should have
known the fact under the specific circumstances that a trade secret had been used
illegally. In some cases, such breaches of secrecy are prohibited by Section 17 (2) No.
2 of the UWG, because the production of a good is considered an exploitation along
31 Kalbfus (2016), pp. 1009–1014, sees Section 18 of the Unfair Competition Act (UWG) as a
counterpart to the new regulation and demands its abolition in favor of a general regulation of trade
secrets.
32 This modification was demanded among others by the Max Planck Institute: Knaak et al. (2014),
p. 953.
33 In addition to this: Wiese (2018), pp. 141 et seq.
216 T. Hoeren
the meaning of the provision.34 In accordance with the Directive the use of supplier-
related data, obtained illegally by third parties, is related to the production. Thus,
legal protection exists against any person who is part of the downstream distribution
chain, even if he or she is unaware of the secret and even if the secret is not embodied
in the product.35
6 Exceptions and Limitations
These are the innovations that evidently go beyond the matter of fact of Section 17
of the UWG. In compensatory terms, the matter of fact of the limitations and excep-
tions of the directive has grown exorbitantly in the course of the negotiations. Such
limitations can be found in Article 3 and 5 of the Directive. The placement of the
central prohibition in Article 4 is unfamiliar, surrounded by Article 3 and Article
5. The delimitation of the specific matter of facts is also unclear. While Article 5
states “exceptions,” Article 3 mentions “lawful acquisition, lawful use and lawful
disclosure.” It is problematic how these linguistic nuances can be reflected in the
implementation. For example, Article 3 of the Directive could be an exempting ele-
ment, while Article 5 could be seen as a justification. Therefore, Article 3 would be
understood as a limitation and Article 5 as an exception. The current ministry draft
follows this evaluation.
6.1 Limitations and Exempting Elements (Article 3)
Article 3 (1) (a) defines independent discovery or creation as legitimate. The concept
of discovery refers to patent law, while creation is a typical term in copyright law.
In fact, in the event that an identical invention has been made independently of one
another by several persons, patent law regulates that the right is vested in the person
who first applied for the invention at the Patent Office (Sec. 6 sentence 3 of the
patent law (PatG), Article 60 (2) EPC). The subsequent applicant is then protected,
if necessary, by a right of prior use (Sec. 12 (1) patent law (PatG)). The phenomenon
of double-creations is also known in copyright law. This is not an infringement of
rights, but a black mark on the copyright map and a bizarre constellation of exceptions
in which both authors can assert their rights.
Article 3 (1) (b) regulates the general freedom of testing. The regulation permits
the analysis and testing of a product which has been made public or is lawfully in the
possession of the purchaser of the information. Thus, the Directive is breaking new
34 Kalbfus (2016), pp. 1009–1014.

35 Kalbfus (2016), pp. 1009–1014.
ground. It is a European regulation on reverse engineering, which is known from US

law. In Germany, reverse engineering is traditionally considered as prohibited.36
Traces of reverse engineering can be found in copyright law [Sections 69d (3),
69e of copyright law (UrhG)] and are regulated in detail in patent law (Section 11
No. 1 and 2 PatG). The semiconductor protection law implies a detailed permis-
sion of reverse engineering (Section 6 (2) No. 1 and 2 HalblSchG). It regulates the
reproduction of topographies for the purpose of analysis, evaluation or education,
as well as the commercial exploitation of topography as a result of an analysis or
evaluation. The freedom of testing refers to a product that has been made accessible
to the public or is lawfully owned by the acquirer of the information. The principle
goes far beyond the exhaustion doctrine of the copyright and patent law and also
relates to rented products. According to an astonishing small addition, the acquirer
of the information must be exempt from legal obligations in order to use the secret.
Therefore, it is allowed to limit the freedom of testing contractually. Thus, an indica-
tion on reverse engineering should be included in supply contracts and cooperation
agreements and such conduct should be excluded.
In addition, in recital 17 of the Directive, the Commission has given some thought
to cases of parasitic copying, where a regulation in the UWG (product piracy or
slavish counterfeiting) should also be considered. However, contractual limitations
are usually pointless if there are no contracts at all. No one can prevent a producer
from purchasing a competing product on the free market and rebuilding it by means
of testing. Article 3 (1) (c) regulates the right of work council to pass on information
under the application of other European or national laws. There is a provision for this
in Germany in Section 84 of the Works Constitution Act (BetrVG), which provides
an explicit right for the employee to complain to the work council. However, the
Directive explicitly allows in recital 12 that work councils may have agreed to secrecy
with the employer. Nevertheless, Section 79 of the BetrVG binds the members of the
work council to not disclose or exploit trade secrets. There is a similar provision for
individual personal measures in Section 99 (1) (2), 102 (2) (5) of the BetrVG.
Article 3 (1) (d) permits all other forms of conduct, which are consistent with legit-
imate business practices. Therein lies an opening clause for the benefit of the judiciary
based on the fair-use limitation of US law. Recital 18 mentions the disclosure of trade
secrets in audit law as an example.
6.2 Exceptions and Justifications (Article 5)
The regulation of exceptions in Article 5 of the Directive is new. Firstly, companies

are exempted in the field of press and freedom of information. A reference is made
36 Imperial Court (RG), judgment of November 22, 1935—case no. II 128/35; also compare Higher
Regional Court Hamburg (OLG Hamburg), judgment of October 19, 2000—§U 191/98; in addition
to this: Beater (1995), p. 136; Kochmann (2009), p. 140.
218 T. Hoeren
expressly to Article 11 of the Charter of Fundamental Rights. Thereby, not only

traditional press companies are addressed but possibly bloggers as well.
Article 5 (b) contains the second institution, which is still foreign to European
law, namely whistleblowing.37 Whistleblowing is defined as the disclosure of mis-
conduct, wrongdoing or illegal activities. The distinction between these matters of
fact remains unclear. It is also unclear whether a minor infringement is sufficient
as a justification of disclosure has acted in the general interest.38 It is undetermined
whether the intended purpose should be considered as subjective or objective. Recital
20 refers to a conduct which objectively serves the general interest. In addition, the
preamble of Recital 20 excludes conduct which, according to national court judg-
ments, can be qualified as good faith. The Directive is in line with European and
national initiatives to strengthen the whistleblower protection. For a long time, the
only existing regulation in Germany was the Occupational Health and Safety Act
(Arbeitsschutzgesetz). Furthermore, in individual cases, Section 34 of the Criminal
Code (StGB) and the duty to disclose in accordance with Section 138 of the StGB
have been considered. Apart from this, the labor law jurisdiction was very restrictive
and prohibited the complaint, in particular in the public eye, before making use of
internal compliance systems. Thus, on July 2nd 2016 the Financial Services Supervi-
sion Act (Finanzdienstleistungsaufsichtsgesetz) already exempted employees of the
financial sector from prosecution who reported maladministration in the company
to the Federal Supervisory Office for Insurance. Contrary to previous and restrictive
German law, the Directive allows the disclosure of maladministration in the press
without the consideration of company-internal remedies.39
The burden of proof for the accuracy of whistleblowing is difficult to ascertain.
The exceptional nature of Article 5 of the Directive indicates that the whistleblower
must prove that the disclosure of the questionable trade secrets serves the protection
of public interest. However, it is also conceivable that the company has to provide
evidence for the accuracy or inaccuracy of the whistleblowing. This evidence is
incumbent on the company, while the whistleblower must prove that he acted in the
public interest.
Article 5 (c) mostly excludes the communication between employees and work
council from the protection of trade secrets. A limitation of complaints or the noti-
fication of maladministration is not conducted. In any event, all communication is
exempted if it is necessary for the practice of work councils. There is a general clause
in Article 5 (d) according to which any legitimate interest justifies a limitation that
is permitted by Union law or national law.
37 Eufinger (2016), p. 229; Groß and Platzer (2017), p. 1097.

38 Lapousterle et al. (2015), pp. 8 et seq.
39 Federal Labor Court (BAG), judgment of December 7, 2006—case no. 2 AZR 400/05; implemen-
tation of the case law of the Federal Constitution Court (BVerfG), judgment of July 2, 2001—case no.
I BvR 2049/00; in the decision of the Federal Labor Court (BAG), judgment of July 3, 2003—case
no. 2 AZR 235/02; Ohly (2014), pp. 1–7.
7 Sanctions
No criminal sanctions are regulated as the EU has no competence in this regard. It

is rather intended to bind the member states to introduce a civil sanction system.
This involves numerous legal consequences that have so far only been granted for
monopoly rights and an extensive protection against illegal acquisition, as well as
illegal use and disclosure of trade secrets. Included are claims for destruction or
release of documents, objects, materials, substances, or electronic files which contain
the secret. Furthermore, it also includes a recall claim for infringing products and the
removal as well as the destruction of them. Claims for damages are also regulated in
detail. However, there is no right for information, which is provided by other property
rights (see Section 140b Patent Law (PatG) and Section 242 BGB).
It is now up to the Ministry of Justice to clarify to what extent individual sanctions
already exist in German law. The Federal Supreme Court (BGH) recognized the
option of triple damage calculation40 for Section 17 of the UWG. A new system of
recall and removal from the distribution channels is likely.
Compensatory to this bouquet of claims is the new extensive regulation of abuse
control (Article 7). The Member States must establish opportunities to protect the
defendant from abusive claims. These measures go beyond Section 8 (4) of the
UWG. These sanctions of abuse include damages for the defendant, sanctions against
the plaintiff, or the dissemination of information concerning court decisions. Such
measures are currently unknown in Germany. Known is only a claim for damages
in the case of unjustified protective right warnings, i.e., in the case of warnings
from special industrial protective rights. In the UWG it is recognized that a careless
warning, as a false assertion, justifies a claim for damages as an illegal encroachment
on the established and exercised business.41
The measures of legal remedies include in particular the interim injunction (Article
12), which is focused on injunctive omission, cancellation, and prohibition of sale.
Instead of the injunctive relief a right of compensation (license analogy) can be
considered, regardless of negligence or fault [Article 13 (3)]. The compensation for
damages (Article 14) applies to the triple damage calculation, including lost profit,
infringing profit of the violator, or license analogy. The option of increase in the
event of moral risks of harm is remarkable. It was not noted in the Directive that
additional claims may occur for example in Germany from Section 812 and 687 (2)
of the BGB. Moreover, the statutory limitation shall not exceed six years.42
40 Federal Supreme Court (BGH), judgment of February 18, 1977—case no. I ZR 112/75.
41 Regional Court Hamburg (LG Hamburg), judgment of May 8, 2012—case no. 407 HKO 15/12.
42 It is not intended to deal with the procedural changes made in the Directive, such as the correction
of the Düsseldorf model.

220 T. Hoeren
8 Conclusion
The Directive will permanently change European secrecy law. Compared to current
German regulations, important differences can be observed, concerning for example
the definition of a trade secret or the legitimacy of reverse engineering. Therefore, the
final implementation of the Directive can be suspenseful. In the meantime, companies
are summoned to conduct concrete confidentiality measures and to adjust to the
changed field of secrets. In this context, extended nondisclosure agreements and a
strategy to deal with the freedom of reverse engineering are most important. A new
culture of secrecy in companies and with suppliers must be added.
References
Christoph A (2016) EU-Richtlinie zum Schutz vertraulichen Know-hows—Wann kommt das neue
deutsche Recht, wie sieht es aus, was ist noch offen? GRUR-Prax 2016(21):465–467
Beater A (1995) Nachahmen im Wettbewerb: Eine rechtsvergleichende Untersuchung zu Paragraph
1 UWG
Börger S, Rein S (2017) Step-by-step. In zehn Schritten zu wirksamem Geheimnisschutz. CB
2017(4):118–123
Eufinger A (2016) EU-Geheimnisschutzrichtlinie und Schutz von Whistleblowern. ZRP
2016(8):229–231
Groß N, Platzer M (2017) Keine Klarlheit beim Umgang mit Informationen und Daten. NZA
17:1097–1104
Hauck R (2016) Geheimnisschutz im Zivilprozess—was bringt die neue EU-Richtlinie für das
deutsche Recht? NJW 2016(31):2218–2221
Herrmann V (2016) Praktische Auswirkungen der neuen EU-Richtlinie zum Schutz von ver-
traulichem Know-how und Geschäftsgeheimnissen. CB 2016(10):368–370
Kalbfus B (2011) Know-how-Schutz in Deutschland zwischen Strafrecht und Zivilrecht - welcher
Reformbedarf besteht?. Carl Heymanns, Cologne
Kalbfu B (2017) Angemessene Geheimhaltungsmaßnahmen nach der Geschäftsgeheimnis-
Richtlinie. GRUR-Prax 2017(17):391–393
Kalbfus B (2016) Die EU-Geschäftsgeheimnis-Richtlinie. Welcher Umsetzungsbedarf besteht in
Deutschland? GRUR 2016(10):1009–1017
Kalbfus B, Harte-Bavendamm H (2014) Protokoll der Sitzung des Fachausschusses für
Wettbewerbs- und Markenrecht zum Richtlinienvorschlag über den Schutz von Geschäftsge-
heimnissen. GRUR 5:453–457
Klein F, Wegener T (2017) Wem gehören Geschäftsgeheimnisse? GRUR-Prax 17:394–396
Knaak R, Kur A, Hilty R (2014) Comments of the Max Planck Institute for Innovation and Competi-
tion of 3 June 2014 on the Proposal of the European Commission for a Directive on the Protection
of Undisclosed Know-How and Business Information (Trade Secrets) Against Their Unlawful
Acquisition, Use and Disclosure of 28 November 2013, COM(2013) 813 Final. Int Rev Intellect
Property Competition Law 45(8):953–967
Kochmann K (2009) Schutz des “Know-how” gegen ausspähende Produktanalysen (“Reverse
Engineering”). De Gruyter, Berlin
Köhler H (2019) §17 UWG. In: Köhler et al. (eds) Gesetz gegen den unlauteren Wettbewerb, 37th
edn. C. H. Beck, Munich
Koós C (2016) Die europäische Geschäftsgeheimnis-Richtlinie—ein gelungener Wurf? Schutz von

Know-how und Geschäftsinformationen—Änderungen im deutschen Wettbewerbsrecht. MMR
2016(4):224–228
Lapousterle J et al (2015) What Protection for Trade Secrets in the European Union? CEIPI´s
observations on the proposal for a directive on the protection of undisclosed know-how and
business information, in Centre for International Intellectual Property Studies Research Paper
Series. Res Pap 02:1–15
McGuire MR (2016) Der Schutz von know-how im system des Immaterialgüterrechts. GRUR
2016(10):1000–1008
McGuire MR (2015) Know-how: Stiefkind, Störenfried oder Sorgenkind? GRUR 2015(5):424–436
Ohly A (2014) Der Geheimnisschutz im deutschen Recht: heutiger Stand und Perspektiven. GRUR
2014(1):1–11
Pacini C, Placid R, Wright-Isak C (2008) Fighting economic espionage with state trade secret laws.
Int J Law Manage 50(3):121–135
Proposal for a Directive of the European Parliament and of the Council on the protection of undis-
closed know-how and business information (trade secrets) against their unlawful acquisition,
use and disclosure, COM/2013/0813 final—2013/0402 (COD). https://eur-lex.europa.eu/legal-
content/EN/TXT/PDF/?uri=CELEX:52013PC0813&from=de. Accessed 22 May 2019
Proposal for a Directive of the European Parliament and of the Council on the protection of undis-
closed know-how and business information (trade secrets) against their unlawful acquisition,
use and disclosure—analysis of the final compromise text with a view to agreement. http://data.
consilium.europa.eu/doc/document/ST-15382-2015-REV-1/en/pdf. Accessed 22 May 2019
Report on the proposal for a Directive of the European Parliament and of the Council on the
protection of disclosed know-how and business information (trade secrets) against their unlaw-
ful acquisition, use and disclosure (2015). http://www.europarl.europa.eu/doceo/document/A-8-
2015-0199_EN.html. Accessed 22 May 2019
Souso e Silva N (2014) What exactly is a trade secret under the proposed directive? J Intellect
Property Law Pract 9(11):923–932
Steinmann S, Schubmehl S (2017) Vertraglicher Geheimnisschutz im Kunden-Lieferanten-
Verhältnis – Auswirkungen der EU-Geheimnisschutz-RL am Beispiel der Automobilindustrie.
CCZ 5:194–198
Study on Trade Secrets and Confidential Business Information in the Internal Market
(2013) MARKT/2011/128/D. https://ec.europa.eu/docsroom/documents/14838/attachments/1/
translations/en/renditions/pdf. Accessed 22 May 2019
Tonell M (2017) Sweden. In: Këllezi P, Kilpatrick B, Kobel P (eds) Abuse of dominant position
and globalization & protection and disclosure of trade secrets and know-how. Springer, Berlin,
pp 541–556
Wiese E (2018) Die EU-Richtlinie über den Schutz vertraulichen know-hows und ver-
traulicher Geschäftsinformationen. Inhalt und Auswirkung auf den gesetzlichen Schutz des
Unternehmensgeheimnisses. Peter Lang, Bern
Index
A Code, 3, 4, 7, 8, 53, 54, 56, 75, 96, 103–105,

Adversarial machine learning, 9, 179, 180, 109–111, 114–118, 120, 131–133,
185, 188, 191, 194 150–153, 164, 210, 214, 218
Agreement on Trade-Related Aspects of Coding, 7, 8, 103–105, 107, 110, 117–120
Intellectual Property Rights (TRIPS Common Law Duty of Confidentiality
Agreement), 209 (CLDC), 66, 67
Airbnb, 1, 2, 17, 112, 127, 130 Computer code, 8, 103–105, 113, 116
Algorithm, 1, 3, 54, 86, 89, 124, 129, 131, Consent, 4, 7, 65–68, 71, 83–85, 87–91,
133, 139, 180, 181, 184–188, 190, 95–98, 214
191, 193–195 Cooley-Eaton-McQuillin thesis, 22, 38
Artificial Intelligence (AI), 1, 2, 4–6, 8, 25, Copyright, 8, 52, 56, 57, 150–153, 214, 216,
49–58, 72, 107, 111–113, 192, 193 217
Australia, 4, 7, 83, 85, 86, 92, 93, 95, 98 Copyright Law (Germany) (UrhG), 217
Australian Digital Health Agency (ADHA),
92, 95
Automation, 9, 23, 26, 42, 51, 57, 113, 179,
181, 182, 196, 201, 203, 204 D
Autonomous driving, 4, 9, 169–171, 173– Data, 1–4, 6–8, 16, 24, 50, 53, 61–63, 66–
177, 204 75, 83–92, 95–98, 106, 107, 113, 123,
Autonomous Vehicles (AVs), 182, 202, 205 125, 127, 129, 139, 146, 162, 163,
183, 192, 216
Data processing, 83, 87–89, 97, 98
B Data protection, 4, 6, 7, 49, 55, 61, 68, 70,
Bitcoin, 3, 119, 134, 138, 150 72, 78, 83, 87, 90, 98
Blockchain, 1–4, 8, 9, 111, 113–119, 123, Decentralized Autonomous Organizations
124, 129, 131–133, 136, 145, 149– (DAOs), 116, 118, 119, 124, 133–
154, 158, 161, 162, 164, 165 135, 141, 142, 144–146
Business-to-Peer (B2P), 24, 29 Deep Learning (DL), 2
Deep neural network, 9, 179, 184
Default rules, 7, 61–66, 78
C Digital law, 55
Centralized Train Control (CTC), 176 Digital person, 4, 6, 49, 50, 52–55, 58
Cities, 1, 4–6, 21–29, 35–43, 54, 182 Digital transformation, 56, 104, 105, 107,
Civic Movement for Decentralization 109, 114
(CMD), 41 Distributed Ledger Technology (DLT), 3, 8,
Clinical records, 86, 90 123, 125, 126, 140
https://doi.org/10.1007/978-981-15-1350-3
224 Index
E Legal education, 4, 104, 107

E-Apostilles, 8, 149, 154, 158, 159, 161–165 Legal entity, 4, 6, 49, 50, 52–54, 58
Electronic health records, 4, 85–87, 94, 96 Liability, 1, 6, 9, 10, 49, 52, 57, 88, 151, 171,
Ethereum, 3, 116, 119, 134 173–175, 181, 192–196, 198–202,
Ethical and Legal Advisory Boards 204, 205, 209, 215
(ELABs), 55 Limitations, 69, 77, 181, 189, 190, 216, 217
European Patent Convention (EPC), 216 Local Autonomy Act 1947 (Korea) (LAA),
European Union General Data Protection 40
Regulation (EU GDPR), 7, 98 Lyft, 2, 180
Exceptions, 214, 216, 217
G M
General Practitioner (GP), 66, 70, 72, 75 Machine Learning (ML), 2, 180, 188
Geographical Positioning System (GPS), 64, My Health Record (MHR), 7, 83–86, 91–98
77, 104, 183, 184
Governance, 4, 5, 21–27, 29, 32–34, 37–40,
42, 110, 116, 119, 125, 132, 133, 142, N
144 National Data Guardian (NDG), 67, 68
National Data Opt-out System (ND opt-out),
7, 61, 62, 67, 69
H National E-Health Transition Authority
Health data, 6, 7, 61, 62, 67, 68, 71–74, 76, (NeHTA), 92
78, 84, 88, 91, 92, 95, 96, 98, 117 National Electronic Record Schemes
Hubs, 37, 176 (NEHR), 7, 83, 85–89
National Health Service (NHS), 6, 62, 63,
66–70, 72–76
I
No-fault liability, 9, 179, 204
Implementation, 5, 10, 19, 35, 58, 92, 112,
Nudges, 62–65, 77, 78
143, 209–212, 216, 220
Information and Communication Technol- Nudge theory, 7, 61, 63, 64, 68, 78
ogy (ICT), 9, 169, 176
Information Technology (IT), 4, 6, 54, 56,
61, 67, 72, 110, 212, 213 O
Insurance, 5, 55, 85, 86, 89–91, 115, 173– Open source software, 8, 149–153, 165
175, 199–202, 204, 218 Operator liability, 9, 179, 202, 204
Internet, 8, 22, 24–26, 33, 34, 41, 56, 105, Opt-in system, 65, 66
106, 115, 116, 123–131, 141, 144– Opt-out system, 6, 7, 61–63, 65–69, 72, 74,
147, 149, 150, 155, 156, 162, 180, 75, 78
193
Internet of Things (IoT), 2, 22, 24, 27, 34,
35, 42, 104, 113, 116, 117, 127, 133
Internet Protocols (IP), 8, 105, 150 P
People’s Solidarity for Participatory Democ-
racy (PSPD), 41
J Privacy, 54, 73, 74, 91, 93–97, 117, 124, 129,
Japanese Housing Accommodation Busi- 146
ness Act (Act No. 65 of 2017) Product liability law, 181, 193–195, 198,
(JHABA), 6 203, 204
Proof of Reputation (PoR), 133, 134
Proof of Stake (PoS), 133, 134
L Proof of Work (PoW), 133, 134
Lawyers, 4, 7–9, 16–18, 38, 57, 103–114, Protocol, 8, 116, 119, 123, 124, 126, 132–
116–120, 169, 171, 174–177, 180 136, 138–144, 146, 163
Index 225
R Transaction engineer, 105, 109, 110

Reputation, 8, 107, 108, 117, 119, 123, 124, Transmission Control Protocol (TCP), 105
127–136, 138–147 Transparency, 4, 49, 52, 53, 112, 117, 118,
Road traffic as a network, 175 133, 200
Trust, 2–4, 7, 8, 16, 26, 29, 33–35, 37, 42, 58,
61, 68, 72, 109, 112, 115, 117, 123–
S 127, 129–133, 139, 140, 144, 145,
Sanctions, 213, 215, 219 147, 162, 192, 193, 201
Semada Proof of Stake Protocol (SPoS), 119
Semada Research Institute (SRI), 8, 123,
129–132, 142–146
Seoul, 4, 5, 21, 23, 26, 35–43 U
Seoul Metropolitan Government (SMG), 36, Uber, 1, 2, 17, 127, 130, 144, 145, 180
40, 42 Unfair Competition Act (Germany) (UWG),
Sharing economies, 1–6, 9, 21–29, 31, 32, 210–212, 214–217, 219
34–38, 40–43, 116, 179, 180 United Kingdom (UK), 4, 6, 7, 61, 62, 66–68,
Smart contracts, 3, 8, 113–117, 119, 131, 70, 71, 76, 78, 210
133, 145 United States (US), 30, 65, 66, 126, 150, 193,
Sunstein, Cass, 63, 76 199, 200, 217
T
Thaler, Richard, 63 W
Trade secret(s), 5, 8, 10, 150, 152, 153, Web of Trust (WoT), 139, 140
209–215, 217–220 World Health Organization (WHO), 38, 85

Legal Tech and The New Sharing Economy

Uploaded by

Copyright:

Available Formats

You might also like

Legal Tech and The New Sharing Economy

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Legal Tech and The New Sharing Economy

Uploaded by

Copyright:

Available Formats

Perspectives in Law, Business and Innovation

Marcelo Corrales Compagnucci

Legal Tech and

More information about this series at http://www.springer.com/series/15440

Nikolaus Forgó Toshiyuki Kono

Shinto Teramoto Erik P. M. Vermeulen

Legal Tech and the New

Toshiyuki Kono Shinto Teramoto

ISSN 2520-1875 ISSN 2520-1883 (electronic)

Copenhagen, Denmark Marcelo Corrales Compagnucci

Legal Aspects of Decentralized and Platform-Driven Economies . . . . . . 1

Sharing Economy and Platforms

Digital Age and Personal Data

Blockchain and Code

Legal Issues Related to Blockchain Technology—Examples

Autonomous Systems and Future Challenges

About the Editors

Marcelo Corrales Compagnucci is attorney-at-law specializing in intellectual

Toshiyuki Kono is a distinguished professor at Kyushu University (Fukuoka,

Erik P. M. Vermeulen is a Professor of Business and Financial Law at Tilburg

Craig Calcaterra Department of Mathematics, Metropolitan State University,

Wulf A. Kaal University of St. Thomas School of Law, Minneapolis, USA

ADHA Australian Digital Health Agency

ICCPR United Nations International Covenant on Civil and Political

TRIPS Agreement Agreement on Trade-Related Aspects of Intellectual Property

Marcelo Corrales Compagnucci, Toshiyuki Kono and Shinto Teramoto

Keywords Sharing economy · Platforms · AI · Blockchain · Data protection ·

M. Corrales Compagnucci (B)

© Springer Nature Singapore Pte Ltd. 2020 1

Part III of the book—Blockchain and Code—contains three contributions.

blockchain technology. In sum, the chapter illustrates an inherent legal issue of

Part IV of the book—Autonomous Systems and Future Challenges—comprises three

Keywords Collaboration · Platforms · Comparative legal studies · Expertise

© Springer Nature Singapore Pte Ltd. 2020 15

2 Oihab and Salvator (2011).

Collaboration is in practice painstaking, riddled with glitches, laced with potential

7 See Riles (2013a, b), Hansen (2009).

3 Experimenting with Platforms: Meridian 180

Beginning in 2011, I began experimenting with a prototype of this kind of platform as

ideas to be gained from struggling to understand points of view foreign to one’s

Benjamen Franklen Gussen

Keywords Sharing economy · Seoul · Cities · Governance ·

© Springer Nature Singapore Pte Ltd. 2020 21

1 Buchanan and Stubblebine (1962).

2 The Historical Context of Sharing Cities

of any network are interconnected.” See Thompson (2003), p. 34.

There is no one universal definition of what we refer to as the “sharing economy.”

This model is driven by the Internet of Things and presumes an access-neutral

environment of open-source access and peer-to-peer collaboration breeds prosumers

This understanding of the sharing economy identifies an adjustment of taxation

The general role of regulation in this Fourth Revolution is to facilitate technical

Hence, some define the sharing economy by anchoring it in civilization, more

25 Schwab (2016), p. 74.

30 Poteete and Ostrom (2010).

35 Kohr (1977), p. 20.

3 The Role of Legal Systems

The utility of an individual A, u A , depends on a set of activities: (X 1 , X 2 , . . . , X m ).

Fig. 1 The basic