SPECIAL ISSUE: CRITICAL REALISM IN IS RESEARCH

THE GENERATIVE MECHANISMS OF DIGITAL

INFRASTRUCTURE EVOLUTION1
Ola Henfridsson
Warwick Business School, The University of Warwick,
Coventry CV4 7AL UNITED KINGDOM {ola.henfridsson@wbs.ac.uk}

Bendik Bygstad
Norwegian School of IT, Schweigaards gt. 14, 0185 Oslo NORWAY and
Department of Informatics, University of Oslo, Oslo NORWAY {bendik.bygstad@nith.no}

The current literature on digital infrastructure offers powerful lenses for conceptualizing the increasingly
interconnected information system collectives found in contemporary organizations. However, little attention
has been paid to the generative mechanisms of digital infrastructure, that is, the causal powers that explain
how and why such infrastructure evolves over time. This is unfortunate, since more knowledge about what
drives digital infrastructures would be highly valuable for managers and IT professionals confronted by the
complexity of managing them. To this end, this paper adopts a critical realist view for developing a configu-
rational perspective of infrastructure evolution. Our theorizing draws on a multimethod research design
comprising an in-depth case study and a case survey. The in-depth case study, conducted at a Scandinavian
airline, distinguishes three key mechanisms of digital infrastructure evolution: adoption, innovation, and
scaling. The case survey research of 41 cases of digital infrastructure then identifies and analyzes causal paths
through which configurations of these mechanisms lead to successful evolution outcomes. The study reported
in this paper contributes to the infrastructure literature in two ways. First, we identify three generative mech-
anisms of digital infrastructure and how they contingently lead to evolution outcomes. Second, we use these
mechanisms as a basis for developing a configurational perspective that advances current knowledge about
why some digital infrastructures evolve successfully while others do not. In addition, the paper demonstrates
and discusses the efficacy of critical realism as a philosophical tradition for developing substantive contribu-
tions in the field of information systems.

Keywords: Digital infrastructure, case study, case survey, configuration theory, critical realism, generative
mechanism, information infrastructure, multimethod, adoption, innovation, scaling

Introduction1 the possibility of effectively managing an enterprise. Yet, it

No idea in our field is more enduring than the notion that the
introduction of a new IT-based information system improves
1
John Mingers, Alistair Mutch, and Leslie Willcocks served as the senior
editors for this special issue and were responsible for accepting this paper.
The appendix for this paper is located in the “Online Supplements” section
of the MIS Quarterly’s website (http://www.misq.org).
is becoming increasingly recognized that the pervasive adop-
tion and use of information technology in contemporary
organizations makes the relationship between information
systems and organization increasingly complex (Zammuto et
al. 2007). As information systems become interconnected,
most organizations face the challenges of controlling an entire
array of systems and technologies, typically introduced over
many years and for different purposes (Ciborra et al. 2000).
As a result, the effectiveness of the single system is largely

MIS Quarterly Vol. 37 No. 3, pp. 907-931/September 2013 907

Henfridsson & Bygstad/Digital Infrastructure Evolution
conditioned by an installed base of extant socio-technical
arrangements, making it useful to apply modes of inquiry that
offer a conceptual basis for going beyond the scope of the
single system.
To understand this phenomenon, there is an emerging litera-
ture that has adopted the notion of infrastructure as a way of
conceptualizing interconnected system collectives (rather than
stand-alone information systems). In fact, the past 15 years
or so have witnessed research on digital infrastructure2
covering different settings (e.g., health, telecom, natural
resources, government, and manufacturing), levels of analysis
(e.g., group, organization, industry, and society), and tech-
nologies (e.g., standards, platforms, and the Internet). Yet, as
highlighted in Tilson et al.’s (2010) recent research commen-
tary, there is an urgent need to theorize the evolution of digital
infrastructures as our “field’s attention moves beyond admin-
istrative systems and individual tools” (p. 748). Such a need
calls for approaches with the ambition to explicate the inner
workings of digital infrastructure. It also indicates the use-
fulness of a philosophical tradition that would build the
intellectual structure for such explanation.
Covering the two main philosophical traditions in IS research
(Mingers 2004), typically referred to as positivism and inter-
pretivism,3 extant infrastructure research displays slightly
different foci. It tends to be occupied with either situated con-
texts of practice, or directly observable managerial aspects.
Adhering to interpretivism, considerable attention has been
paid to the evolution of digital infrastructure as it plays out in
the complex interdependencies between socio-technical ele-
ments (Braa et al. 2007); networks of human and nonhuman
actors (Hanseth and Monteiro 1997); and the relationships
between organized practices (Star and Ruhleder 1996). In
studies underpinned by positivist assumptions, the research
has primarily dealt with strategic IT portfolio management
and the alignment of IT imperatives with business strategy
(Broadbent and Weill 1997).
As an alternative intellectual structure for theorizing digital
infrastructure, we propose critical realism (Archer et al. 1998;
2
The current literature uses different concepts for capturing this phenomenon
including information infrastructure, IT infrastructure, e-infrastructure, and
so on. Following Tilson et al.’s (2010) call for infrastructure research, we use
the term digital infrastructure throughout this paper.
3
We use the terms interpretivism and positivism to align our terminology with
previous writings on philosophical traditions in the information systems
literature (see Orlikowski and Baroudi 2001; Walsham 1995). However, it
should be emphasized that both labels span multiple philosophical strands.
For instance, as suggested by one of our anonymous reviewers, interpretivism
includes both idealists and realists, taking different ontological positions.
908 MIS Quarterly Vol. 37 No. 3/September 2013
Bhaskar 1997; Sayer 1992) for its emphasis on generative
mechanisms (Bhaskar 1997, 1998). While there are a few
studies identifying specific mechanisms of digital infrastruc-
ture (see Bygstad 2010), little, if any, research has been
geared toward developing a comprehensive understanding of
the range and contingencies of causal structures in its evolu-
tion. Understanding these issues is important, since more
knowledge about what drives digital infrastructures would be
highly valuable for managers and IT professionals confronted
by the complexity of managing them.
Our paper deals with the following research question: Which
mechanisms contingently cause digital infrastructure evolu-
tion? We address this research question using a multimethod
research design (Mingers 2001), recently proposed as an
important principle for conducting critical realist case study
research (Wynn and Williams 2012). We first conducted in-
depth case study research (George and Bennet 2005; Gerring
2007) at a Scandinavian airline to identify key mechanisms of
digital infrastructure. Then, we conducted case survey re-
search (Larsson 1993) based on a sample of 41 cases to ana-
lyze the causal paths through which these mechanisms are
combined to produce successful digital infrastructure
evolution.
Our research makes a number of contributions. First, we
identify three generative mechanisms of digital infrastructure
and how they contingently lead to evolution outcomes.
Second, we use these mechanisms as a basis for developing
a configurational perspective that describes infrastructure
evolution as an outcome of multiple paths of interconnected
contextual conditions and mechanisms (El Sawy et al. 2010;
Meyer et al. 1993; Pawson and Tilley 1997). This perspective
advances current knowledge about why some digital infra-
structures evolve successfully while others do not. Finally,
the paper demonstrates and discusses the efficacy of critical
realism as a philosophical tradition for making substantive
contributions in the field of information systems.
Related Research and
Conceptual Basis
Digital infrastructure evolution can be broadly referred to as
a gradual process by which a digitally enabled infrastructure
changes into a more complex form. Viewing digital infra-
structure as the collection of technological and human compo-
nents, networks, systems, and processes that contribute to the
functioning of an information system (Braa et al. 2007; Tilson
et al. 2010), this evolutionary process entails both social and
technical elements (Vaast and Walsham 2009). However, re-
viewing the literature, it is clear that definitions vary, and,

over the years, the concept itself has largely served as the
lowest common denominator for IS researchers who have
shifted attention from single organizations to organizational
networks, and from systems to infrastructures (Ciborra et al.
2000).
Views on Digital Infrastructure4
There exists a relatively large volume of research that uses the
notion of infrastructure without necessarily dealing substan-
tively with the phenomenon itself (Tilson et al. 2010). In
such cases, infrastructure is typically used as an independent
variable to explain something else (for examples, see Bharad-
waj 2000; Malhotra et al. 2005; Rai et al. 2006; Tanverdi et
al. 2007), and it often refers to a collection of information
technologies and systems that jointly produce a desired out-
come. For instance, Malhotra et al.’s (2005) study of infor-
mation sharing between actors in the supply chain under-
stands IT infrastructure as partner interface-directed informa-
tion systems that “enable an enterprise to process information
collected from its supply chain partners so as to create new
knowledge” (p. 156).
In prior literature where digital infrastructure in fact is at the
center of attention, four streams of research can be distin-
guished (see Table 1). Three of these streams manifest inter-
pretivist assumptions, while the fourth reflects positivist
assumptions.5 First, complexity models presume that there is
4
Drawing on Webster and Watson (2002), we conducted a concept-centric
literature review of research on digital infrastructure. First, our ambition was
to include most, if not all, relevant articles published in the AIS basket-of-
eight IS journals, science and technology journals, and other journals (e.g.,
Information & Organization, Information Society, Information Technology
& People, and Computer Supported Cooperative Work) that were likely to
publish infrastructure studies. In addition, we added articles from conference
proceedings and books on an ad hoc basis. We used the Citeseer ABI/Inform
research database and the selected articles had the phrases “digital infra-
structure,” “information infrastructure,” “information technology infra-
structure,” “IT infrastructure,” “information systems infrastructure,” or “IS
infrastructure” in the title, abstract, keyword, or body of the publication.
From the large number of publications that met this criterion, we briefly read
the papers to identify articles with digital infrastructure as an important object
of study. We analyzed the selected publications to identify key concepts that
were used to characterize the nature of digital infrastructure. These concepts
were then clustered with the intention of identifying dominant research
streams in the digital infrastructure literature. The identified research streams
were then labeled complexity, network, relational, and strategic asset to
reflect their main theoretical emphasis.
5
Some readers may feel that typifying infrastructure research into families of
philosophical traditions such as positivism and interpretivism faces the risk
of simplifying research in a substantive area by enforcing grand thought
structures on it and its representative scholars. Appreciating this risk, we
value the possibility to surface philosophical assumptions and their conse-
quences for theorizing. As suggested by Lee (2004), recognizing philo-
sophical underpinnings “can lead to findings that would help the information
Henfridsson & Bygstad/Digital Infrastructure Evolution
no single source of digital infrastructure evolution. Grounded
in literature on complexity (Holland 1995; Mol and Law
2002; Urry 2003), such models typically highlight the com-
plexity of digital infrastructures as a multitude of actors
simultaneously enact their own goals. In other words, infra-
structure evolution is seen as the process by which hetero-
geneous and autonomous human, or organizational, actors
seek to use information technology in their adaptation to each
other and their external environments (Braa et al. 2007;
Ciborra and Failla 2000; Hanseth et al. 2006). For instance,
Braa et al. (2007) advance the notion of flexible standardiza-
tion as a key process for addressing the complexity of accom-
modating both global needs of scalability of infrastructure
standards and local needs of sensitivity to contextual
differences.
Second, network models assume that networks of human and
technical elements drive digital infrastructure evolution. This
stream of research is typically grounded in some of the early
writings of actor–network theorists such as Callon (1986) and
Latour (1987). It views infrastructure evolution as a process
by which multiple human actors translate and inscribe their
interests into a technology, creating an evolving network of
human and nonhuman actors (Aanestad and Blegind Jensen
2011; Hanseth and Monteiro 1997; Yoo et al. 2005). For
instance, Hanseth and Monteiro (1997) note how barriers to
end-user involvement were inscribed by human actors in-
volved in the integration of the EDIFACT standard in Nor-
wegian healthcare. These barriers, the authors claim, were the
result of an actor–network beyond any single stakeholder’s
control.
Third, relational models premise that infrastructure should be
appreciated through the sensemaking of its users and stake-
holders. This stream of literature has its intellectual basis in
theories on learning and work practices (Engeström 1990;
Lave and Wenger 1992). As noted in Star and Ruhleder’s
(1996) seminal article, digital infrastructure is a relational
property that becomes meaningful as an element of organized
activity. In this regard, infrastructure evolution is seen as a
process by which socio-technical relations emerge from infor-
mation technology-mediated activities meaningful in a given
community-of-practice (Pipek and Wulf 2009; Star and Ruh-
leder 1996; Vaast and Walsham 2009). For instance, Vaast
and Walsham (2009) propose a perspective on trans-situated
learning and how such learning is “supported by the local uni-
versality of an information infrastructure whose use becomes
embedded with other infrastructures” (p. 547).
systems research community do better information systems research” (p. 18).
Also, it should be emphasized that in a number of cases the philosophical
tradition is explicitly indicated in the methods section of individual articles
and/or in their reference literature.
MIS Quarterly Vol. 37 No. 3/September 2013 909
Henfridsson & Bygstad/Digital Infrastructure Evolution

Table 1. Research Streams and Definition

Research Philosophical Foundational
Streams Tradition Literature Definition (of DI Evolution) Example References
Complexity Interpretivist Complexity theory The process by which heterogeneous • Braa et al. (2007)
• Holland (1995) and autonomous human, or • Ciborra and Failla
• Mol and Law (2002) organizational, actors seek to use (2000)
• Urry (2003) information technology in their • Hanseth et al.(2006)
adaptation to each other and their
external environments.
Network Interpretivist Actor–network theory The process by which multiple human • Aanestad and Blegind
• Callon (1986) actors translate and inscribe their Jensen (2011)
• Latour (1987) interests into a technology, creating • Hanseth and Monteiro
an evolving network of human and (1997)
nonhuman actors. • Yoo et al. (2005)
Relational Interpretivist Work practice and The process by which socio-technical • Pipek and Wulf (2009)
learning theory: relations emerge from IT-mediated • Star and Ruhleder
• Engeström (1990) activities that become meaningful in a (1996)
• Lave and Wenger given community-of-practice. • Vaast and Walsham
(1992) (2009)

Strategic Positivist Strategic choice The process by which managers • Broadbent and Weill
Asset theory initiate and implement changes in an (1997)
• Beckert (1999) organization’s portfolio of systems • Broadbent et al. (1999)
• Child (1972, 1997) and tools for increasing the alignment
between its IT resources and
strategic imperatives.

Finally, strategic asset models view digital infrastructure
evolution as the process by which managers initiate and
implement changes in an organization’s portfolio of systems
and tools for increasing the alignment between its information
technology resources and strategic imperatives. In this
regard, a strategic choice view (Beckert 1999; Child 1972,
1997) is implied, that is, political action is given primacy in
analyzing organizational responses to environmental contin-
gencies. For instance, Broadbent et al. (1999) explain how
the creation of a certain level of IT infrastructure capability is
needed to successfully implement business process redesign.
The Promise of Critical Realism
In this research, we seek to understand which mechanisms
contingently cause the evolution of digital infrastructure. We
argue that the explanatory power of such mechanisms has
been masked in prior research by the adoption of philosoph-
ical assumptions inattentive to structures operating beyond
(1) the rich texture of people’s meaning-making of the socio-
technical world (interpretivist streams), or (2) events directly
observable in the empirical domain of infrastructures (the
positivist stream). Interpretivism invites attention to actors’
sensemaking and the generation of process accounts on digital
infrastructure evolution. In the relational stream of research
(e.g., Pipek and Wulf 2009; Star and Ruhleder 1996; Vaast
and Walsham 2009), for instance, this attention to sense-
making has been materialized as a focus on the patterned
activity that results from situated actors’ interaction and
dealing with technology in their work settings. The positivist
assumptions underpinning the strategic assets stream of
research (Broadbent and Weill 1997; Broadbent et al. 1999)
imply attention to characteristics of strategy and a portfolio of
systems that can be directly observed and measured. This
may lead to overly simplistic assumptions about the rela-
tionship between digital infrastructures and business success.
Although divided by two broad, but significantly different,
sets of philosophical assumptions, we argue that the extant
literature on digital infrastructure generally tends to shy away
from causality. To pick up on Tilson et al.’s (2010) call for
infrastructure theory, we should seek explanations that take
into account both the dynamic character of digital infra-
structures and the contingent causality characterizing their
evolution.
We adopt critical realism as an intellectual structure for
reconciling existing perspectives of digital infrastructure.

910 MIS Quarterly Vol. 37 No. 3/September 2013


Critical realism has been increasingly recognized in our dis-
cipline as a promising philosophical tradition to overcome
objectivism–relativism chasms (Mingers 2004; Smith 2010;
Volkoff et al. 2007). It combines a realist ontology with an
interpretive epistemology (Archer et al. 1998); although our
knowledge of the world is socially constructed and fallible,
that world exists, often independent of human beings
(Mingers 2004). Contrary to judgmental relativism, critical
realism therefore holds that some theories approximate reality
better than others, making methodological approaches to
assess knowledge claims meaningful. In addition, research
should not only concern recording of constant conjunctions of
observable events (Bhaskar 1997; Mingers 2004), as sug-
gested in so-called covering law theories of causality
grounded in Hume’s skeptic discussion of causality (Elder-
Vass 2010, Lee 2004).
Generative Mechanisms
Following Bhaskar (1997, 1998), we define generative mech-
anisms as causal structures that generate observable events.
In contrast to Humean causality, critical realists typically
ascribe such structures causal powers (Sayer 1992). Causality
is contingent; that is, the outcome of a mechanism depends on
other mechanisms (Elder-Vass 2010; Sayer 1992). In this
regard, mechanisms “act transfactually. The event or events
that they are the powers to instantiate may never actually be
instantiated; the powers may remain unactualized, yet these
powers remain in existence” (Fleetwood 2009, p. 362-363).6
A generative mechanism is “one of the processes in a concrete
system that makes it what it is—for example, metabolism in
cells, interneuronal connections in brains, work in factories
and offices, research in laboratories, and litigation in courts of
law” (Bunge 2004, p. 182). In this vein, our research ques-
tion, about which mechanisms contingently cause the evolu-
tion of digital infrastructure, is partially geared toward
defining what constitutes a digital infrastructure. Previous
literature reviews suggest that this is a fundamental issue for
furthering research in the area (see Bygstad 2008), not least
by directing attention to the underlying mechanisms that
produce observable events.
Drawing on Hedström and Swedberg’s (1998) work on mech-
anisms, we make three assumptions about mechanisms of
6
There is an extensive literature on the nature and definition of mechanisms,
both in the philosophy of social sciences (Demetriou 2009; Glennan 2009;
Hedstrom 2008), and in the critical realist community (Fleetwood 2009,
2011; Mingers 2010). While we relate broadly to this literature, it is beyond
the scope of this article to engage fully in this discourse.
Henfridsson & Bygstad/Digital Infrastructure Evolution
digital infrastructure and their generic structure (see Figure 1).
First, digital infrastructure mechanisms are self-reinforcing
(Hanseth and Aanestad 2003). A self-reinforcing mechanism
recursively feeds on itself. Since the control of an infra-
structure typically is distributed across multiple actors, infra-
structures are—for practical and economic reasons—difficult
to govern. It is partly relying on positive, or negative, feed-
back loops beyond single stakeholders’ control (Hanseth and
Braa 2000). The phenomenon of self-reinforcement is well-
known in technology and diffusion research (Katz and
Shapiro 1985), and has been attributed a central role for
understanding organizational stability and change (Sydow et
al. 2009).
Second, digital infrastructure mechanisms are composites.
They interconnect three types of mechanisms: situational
mechanisms (macro–micro level), action-formation mecha-
nisms (socio-technical action), and transformational mech-
anisms (micro–macro level) (DeLanda 2006; Hedström and
Swedberg 1998). Macro–micro mechanisms explain how the
infrastructure as a whole enables and constrains its various
components. For instance, the Internet as an infrastructure
has proved to enable unprecedented innovation possibilities
for individual entrepreneurs, as long as they are following its
standard interfaces (Hanseth and Lyytinen 2010; Zittrain
2006). Action-formation mechanisms explain “how a specific
combination of individual desires, beliefs, and action oppor-
tunities generate a specific action” (Hedström and Swedberg
1998, p. 23). Continuing our example, Internet entrepreneurs
in Silicon Valley display new forms of learning in innovative
path creation (Hagel et al. 2010). Micro–macro mechanisms
explain emergent behavior, that is, how different components
interact in order to produce an outcome at a macro level.
Completing our example, new innovation path creation leads
to new services and products that reinforce the Internet as a
basis for innovative activity.
Third, although most established work on mechanisms only
addresses the social (Hedström and Swedberg 1998; Merton
1967), it goes without saying that a necessary element in digi-
tal infrastructures is technology. Technology plays an active
role at both the structural level and the action level (Volkoff
et al. 2007), and the interaction between social and technical
elements is the constituting process of the mechanism
A Configurational Perspective
Mechanisms are characterized by contingent causality (Elder-
Vass 2010; Sayer 1992). The actualization (or lack of actuali-
zation) of the powers of a mechanism may lead to one out-
come in a particular context, and another in a different context
MIS Quarterly Vol. 37 No. 3/September 2013 911
Henfridsson & Bygstad/Digital Infrastructure Evolution
Digital Infrastructure
Macro to
micro
Socio-technical
action
Figure 1. The Self-Reinforcing Socio-Technical Mechanism
(see Ragin 2008). This multifinality (George and Bennett
2005) largely depends on the existence of other mechanisms
in the same context. Such dependence also indicates that
there are multiple causal paths through which a particular out-
come can occur, and suggests that investigations of digital
infrastructure should seek to analyze how different mecha-
nisms are configured and triggered to produce successful
outcomes.
A configurational perspective (El Sawy et al. 2010; Fiss 2007;
Pawson and Tilley 1997) can be seen as a viable approach to
investigate the causal complexity associated with such equi-
finality7 (George and Bennett 2005). It allows analysis of
possible configurations of mechanisms and relevant context-
variation to explain a particular outcome (Pawson and Tilley
2009). However, it should not be considered as a covering
law (cf. Elder-Vass 2010), but as a conjectural explanation,
being the basis of further refinement.
Using Pawson and Tilley’s (2009) context–mechanism–
outcome (CMO) scheme as a basis, Figure 2 illustrates the
starting-point for our configurational analysis, where we
derived four possible configurations of relevant context
variation from the literature and suggested a hypothetical
space of 2n configurations of mechanisms that together lead to
the same outcome (success8). This configurational perspec-
tive provides the basis for our analysis of the causal paths that
explain how, in certain contexts, a digital infrastructure
mechanism (or a combination of mechanisms) may lead to
successful evolution. Our configurational perspective rests on
two assumptions. First, we assume that the key mechanisms
7
Equifinality refers to situations where different causal paths may lead to the
same outcome (Fiss 2009; George and Bennett 2005; Rihoux and Ragin
2009). Multiple determination is a similar term used by Bhaskar (1979) (see
also Elder-Vass 2010).
8
Of course, a similar space of configurations can be created for failures.
912 MIS Quarterly Vol. 37 No. 3/September 2013
Micro to
macro
of digital infrastructure interact, that is, a mechanism may
produce different outcomes, depending on its actualization in
combination with other mechanisms. Second, we buy into the
current wisdom that infrastructures consist of both social and
technical elements (Ciborra et al. 2000; Edwards et al. 2009;
Vaast and Walsham 2009). Our literature review of con-
textual conditions was, therefore, geared toward covering
both types of elements in our configurational perspective. On
the social side, type of control stood out as an important con-
textual condition (e.g., Ciborra 2000; Hanseth and Braa 2000;
Hanseth et al. 1996; Pipek and Wulf 2009; Rolland and
Monteiro 2002; Sahay and Walsham 2006). In particular, the
idea of decentralized control of digital infrastructures (Ciborra
et al. 2000) has been considered a strong alternative to pre-
vailing centralized approaches (see Broadbent and Weill
1997). On the technical side, architecture, in particular
loosely coupled architectures, has come to the fore as an
important condition for infrastructure evolution (Aanestad and
Blegind Jensen 2010; Fabri 2009). Based on the extant infra-
structure literature, we therefore propose that decentralized
control and loosely coupled architecture work as key con-
textual conditions of digital infrastructure evolution.
Before presenting our empirical study, we also need to clarify
the outcome dimension of our configurational perspective.
Ever since DeLone and McLean’s (1992; see also 2003)
seminal article, it has been generally agreed that IS success is
a multidimensional construct involving measures such as
system quality, information quality, service quality, organi-
zational impact, and user satisfaction. However, defining
success for digital infrastructures requires consideration of the
fundamental difference between traditional in-house informa-
tion systems and infrastructures. While the former typically
involves a relatively well-defined evaluation context in terms
of objectives and end-user group, the latter refers to open,
evolving networks of interconnected systems having many
stakeholders, for whom success may be interpreted differently
(Hanseth and Lyytinen 2010). Infrastructures often evolve

Figure 2. A Configurational Perspective on Digital Infrastructure
into larger and more complex structures (such as the Internet),
without any predefined end state as they are continuously
extended and typically operate outside the control of a single
stakeholder. It therefore makes sense to relate an infra-
structure’s success to its role in, and fit with, the environment
it inhabits. Using a biological metaphor, success can then be
seen as a question of survival in a volatile business eco-
system. In this vein, we view infrastructure evolution success
as an outcome realized when (1) the infrastructure survives in
a business ecosystem by filling a relevant role over time, and
(2) the infrastructure’s affordances cannot be escaped
endogenously but is only vulnerable to exogenous shocks (see
Vergne and Durand 2010).
Given this theoretical background, we set out to develop a
configurational perspective by using a multimethod research
design (Mingers 2001). We first conducted a 4-year in-depth
case study (Gerring 2007; George and Bennett 2005) of an
airline in order to identify mechanisms with the power to
Henfridsson & Bygstad/Digital Infrastructure Evolution
cause digital infrastructure evolution. We then conducted a
case survey (Larsson 1993) to analyze the causal paths
through which these mechanisms are combined to produce
successful digital infrastructure evolution. The next section
outlines the in-depth case study.
The Norwegian Case Study
Norwegian Corp is an international airline carrier that pio-
neered the low-price airline market in Scandinavia. Enabled
by the European deregulation of the airline industry, the
firm’s strong growth started in 2002 as it established a
national network of destinations. Some 9 years later, in 2011,
Norwegian operated a total of 303 routes to 118 destinations
in Europe and the Middle East, and carried 15.7 million
passengers. The company had 2,500 employees and revenues
were $1.9 billion. Over this nine-year period, Norwegian has
MIS Quarterly Vol. 37 No. 3/September 2013 913
Henfridsson & Bygstad/Digital Infrastructure Evolution
successfully used information technology in new ways, culti-
vating the digital infrastructure that was the focal point of our
in-depth research.
Methods
We selected the case for our in-depth research based on two
criteria. First, long-term involvement and access was deemed
important, since studies of evolution benefit from rich longi-
tudinal data. Such data also resonates well with critical
realism’s theorizing process, commonly referred to as
retroduction, that is, taking an empirical observation and
hypothesizing a mechanism that might explain that particular
outcome (Danermark et al. 2002; Sayer 1992). Second,
among cases to which we had access, we used a so-called
extreme technique of case selection by engaging in intense
data collection and analysis of a digital infrastructure per-
ceived to be unusually successful. Extreme cases correspond
“to a case that is considered to be prototypical or paradigmatic
of some phenomenon of interest” (Gerring 2007, p. 101) and
is useful for theory generation because extremes or ideal types
typically define theoretical concepts. Compared to a repre-
sentative case selection technique, we anticipated that the
selection of the airline Norwegian would offer access to ideal
types “formed by the one-sided accentuation of one or more
points of view” (Weber, 1949, p. 90). In this regard, the case
study generated conceptual constructs that manifest theorizing
through idealization (Lopreato and Alston 1970; Ohlsson and
Lehtinen 1997).
Data Collection: We did not begin our study at Norwegian
with the intention of studying the mechanisms of digital infra-
structure evolution (see Plowman et al. 2007). It started as an
inquiry into service innovation in a fast-growing firm. As the
study unfolded, however, we discovered how Norwegian’s
success in terms of becoming an important player on the
Scandinavian aviation market was intimately related to its
evolving infrastructure. We therefore gradually appreciated
the process as a paradigmatic example of digital infrastructure
success and reoriented the study to inquire into the underlying
generative mechanisms of this evolution.
We used three methods to collect our data: interviewing,
participant observation, and document analysis. First, we
conducted 31 semi-structured interviews with a total of 19
respondents (we interviewed 6 respondents more than once).
Some interviews were tape-recorded and transcribed ver-
batim, while others relied on field notes. We interviewed the
CIO, two top managers, three business line managers, two
business analysts, two IT managers, two systems developers,
two project managers, one booking assistant, one consultant,
914 MIS Quarterly Vol. 37 No. 3/September 2013
one vendor representative, and two customer representatives.
Second, participant observation was another important source
of data. In addition to system demonstrations and observa-
tions of direct system use, we spent approximately 12 hours
observing 7 meetings related to Norwegian IT and strategy
projects. In addition to these discrete events, we conducted
ongoing debriefings after meetings and numerous informal
interviews with Norwegian employees. Finally, our study
included a significant volume of archival data, including busi-
ness plans, project plans, joint venture contracts, and IT archi-
tecture documents. We also had access to 18,846 Facebook
postings of how Norwegian communicated with their cus-
tomers during the European airspace “ash crisis” in 2010.
Although many of the documents we collected were confi-
dential and could not be used directly in this research, the
material served to confirm or disconfirm interpretations made
throughout the data analysis process.
Data Analysis: We used a four-step approach (see Table 2)
for analyzing the data collected at Norwegian. First, we used
an open coding procedure to discover key events. While
many events were deemed important a priori (e.g., business
decisions or contract agreements), others emerged from data
analysis (e.g., the importance of the IT architecture and the
use of Facebook). This coding procedure helped us establish
a timeline of key events that occurred over time (see
Figure 3).
Then, in the second step, we identified the objects of the case
(see Danermark et al. 2002). We then used Hedström and
Swedberg’s (1998) work on mechanisms for typifying the
identified objects as being potential elements in macro–micro,
socio-technical action, or micro–macro mechanisms (see
DeLanda 2006). We visualized the objects of the Norwegian
case as a data display (Miles and Huberman 1994; see
Table 3). The third step was retroduction, that is, the identifi-
cation of key mechanisms among candidate mechanisms. We
started by analyzing the interplay of objects, in particular the
interplay between social and technical objects, which allowed
for the identification of socio-technical mechanisms. With the
infrastructure as point of departure, we tried to identify
macro–micro mechanisms, that is, enabling socio-technical
features of the digital infrastructure. Transcripts from the
cases were analyzed in order to identify these enabling condi-
tions. These were tracked through organizational behavior
(socio-technical action) and the self-feeding outcome, that is,
how the emergent behavior at the micro-level increased the
capabilities of the infrastructure (micro– macro). At this point
in the analysis, we had identified six candidate mechanisms.
They were detailed and assessed through backward-chaining
(Pettigrew 1985), going from outcomes to causes (mech-
anisms), and through forward-chaining, going from causes to
 Henfridsson & Bygstad/Digital Infrastructure Evolution

Table 2. In-Depth Data Analysis

Steps
1. Coding key events
2. Identification and typifying
components
3. Retroduction of
mechanisms
4. Establish contexts and
outcomes of mechanisms
Tasks Outputs
a. Identify key events in the data material A chronology of key events of the case
b. Establish a time line of the key events (Figure 3)
a. Identify networks of social and A set of components and related data
technical components (Table 3)
b. Use Hedström and Swedberg (1998)
for typifying components as macro–
micro, socio-technical action, or micro–
macro
c. Display components and related data
a. Investigate interplay between micro Three digital infrastructure mechanisms
and macro elements to explain (adoption, innovation, and scaling) including
outcomes definitions and measures (Table 4)
b. Identify and analyze candidate mech-
anisms. Assess explanatory power of
each.
c. Define mechanisms and develop mea-
sures to be used in the case survey
a. Analyze mechanisms to confirm A typology of contextual conditions of
contextual conditions and outcomes mechanisms over two dimensions: control
b. Develop contextual typology and technical architecture

National network of
destinations Bank Call
established Norwegian Norwegian

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Service-oriented Low-price In-flight

Internet Mobile
architecture calendar broadband
bank portal
introduced established services

Facebook used
during the ash
Internet Digital customer crises
bookings communication
dominated

Figure 3. Chronology of Key Events

outcomes. In this way, we assessed the explanatory power of
each one of them, in relation to the empirical evidence, and
finally arrived at three (see Table 4) that consistently could
explain the sequence of events over time.
Finally, in the fourth step, we analyzed the three selected
mechanisms to establish contextual conditions and outcomes.
Whether a mechanism is actualized or not is contingent
(Pawson and Tilley 1997). In our analysis at Norwegian, the
decentralized control and loosely coupled architecture of the
digital infrastructure stood out as important conditions of the
mechanisms observed. Given that these empirical results
resonated well with our initial assumptions grounded in prior
literature, we decided that mode of control and technical
architecture would be important conditions to be investigated
further in the case survey.

MIS Quarterly Vol. 37 No. 3/September 2013 915

Henfridsson & Bygstad/Digital Infrastructure Evolution

Table 3. Events and Objects Overview

Key Events Objects Data
Establishing a service Enterprise Service Bus (ESB), booking • CIO and two architects designed the SOA
oriented architecture systems, IT architects, data center • open source software for ESB
(SOA) • run at outsourcing data center
Internet bookings Internet portal, booking systems, • Internet portal
Amadeus, users • integration with Amadeus, allowed for direct booking
• user-printed tickets
Low-price calendar SOA, business developers, IT staff • SOA-enabled low-price calendar
• new user pattern triggered
Digital customer SOA, marketing department, customers • successful portal and email services
communication • 85% of customer communication was electronic in
dominating 2006
Bank Norwegian Business developers, SOA, banking • the bank was innovated on experiences from the
established systems airline solutions
• SOA allowed for short time-to-market
Call Norwegian Mobile portal, SOA, vendors, GSM • the telecom services extended services to airline
established network customers
• mobile phones is perceived as the most important
user platform
Using Facebook during Facebook, customers, crisis team • fast response to the “ash crisis” in 2010
the “ash crises” • Facebook used for problem solving

Case Findings We knew exactly what we needed; a service oriented

In 2002, a virtually unknown airline, Norwegian Air Shuttle,
decided to challenge the dominant airline on the Scandinavian
market, SAS. The deregulation of the European airline mar-
ket paved the way for Norwegian’s ambition to outperform
the incumbent airline through a low-price strategy.
As a new entrant without legacy, Norwegian started with a IT
solution only consisting of the most necessary systems for
making an airline work, such as a basic booking system and
a simple back-office solution. As the company expanded,
however, it was soon realized that a professional approach to
IT was imperative. In less than a year, the unknown airline
flew more than 300,000 passengers and, given plans to grow
internationally, the IT solution soon turned out to be ill-
dimensioned.
Accordingly, Norwegian hired a CIO and two IT architects
with extensive experience from the airline business. Starting
in 2003, this team took a strategic outlook on IT governance.
The team envisioned an IT infrastructure that would need to
scale over time to be at par with Norwegian’s growth ambi-
tions. Based on his earlier experiences of battling legacy
systems when pursuing change ambitions, the CIO recalled:
architecture, enabling easy communication and reuse
of components across different technologies. We
settled for a simple, but fast open-source enterprise
service bus, which could be scaled up in order to
handle transaction growth. Since we had no legacy
systems we could implement the solutions within
months, including the integration with the Amadeus
system.
The new architecture was mainly designed in 2004, and, as
outlined below, it gradually expanded into a large-scale
infrastructure over the following years. Figure 3 provides a
chronology of this infrastructure evolution. Reflecting an
entrepreneurial culture with little bureaucratic control, each
new service was typically conceptualized in a meeting with
business developers and IT seniors, and then developed and
deployed during a short project. “You see,” commented a
business director, “we do not have ‘IT projects’ in Nor-
wegian, only business projects, with a clear economic objec-
tive.” As an example of the agile approach, usability issues
were typically not seriously addressed in early releases.
However, after the launch of a new service, use patterns were
systematically monitored, and used as a basis for quick
adjustments in an iterative fashion.

916 MIS Quarterly Vol. 37 No. 3/September 2013


Internetworking with customers: Since travel agent ser-
vices were considered too expensive for a low-cost airline,
Norwegian’s first challenge was to enable Internet bookings.
In 2004, the airline launched its Internet portal to by-pass
travel agents. This launch was enabled by a solution that
allowed customers to print their boarding cards at home,
making the traditional paper ticket obsolete. The printout
tickets included a bar code, which was scanned at the gate.
This strategy turned out to be successful, not least because
customers quickly adopted this new type of travel processing.
In fact, in 2006, when the airline flew a total of 5.1 million
passengers, 90 percent of all customer interaction, including
e-mail and web marketing, on-line sales, booking, and check-
in, was digital.
Capitalizing on the service-oriented architecture, the Internet
portal gave birth to another innovation, the so-called low-
price calendar. The low-price calendar provided an overview
of the cheapest flights to any chosen destination. It helped
customers adjusting their travel to dates when traveling was
good value for money. Addressing a problem perceived by
customers in the low-price segment, the new innovation was
an outstanding success, substantially increasing the number of
bookings after its inception in 2005. The low-price calendar
was later copied by many other airlines, including Nor-
wegian’s competitor, SAS.
Branching out: In 2007, the company, somewhat sur-
prisingly, decided to enter banking by establishing a bank
called Bank Norwegian. Motivating this boundary-crossing
initiative, Norwegian’s CEO, stated, “Today we have one of
the most visited web pages in Norway, with 2 to 3 million
visitors each month. We aim at coupling this traffic with
bank services.” In addition to the large volume of potential
customers, this radical diversification could be traced to the
architecture of digital infrastructure, which allowed quick
integration with its banking partner’s systems. As the Direc-
tor of Business Development commented,
We had established a very flexible IT architecture,
and we realized at the time that it would be possible
to innovate new services on this. First we were just
brainstorming rather freely: how could a combina-
tion of brand and technology generate new business?
The establishment of the Internet bank took a mere 6 months
and, in 2008, it served 50,000 customers.
Stimulated by the success in banking, the company launched
a telecom company, Call Norwegian. Initially, Call Nor-
wegian involved a mobile portal to enable easy airline
booking, and to offer airport Wi-Fi hotspots and GSM
network services. The Mobile Portal Director commented,
Henfridsson & Bygstad/Digital Infrastructure Evolution
We focused on how to make money on new ser-
vices, analyzing which services we should provide
ourselves, which we should buy and how they
should be integrated. At the same time we are very
concerned about our architecture. A chief ambition
is to maintain it as “clean” as possible. We don’t
really go for cutting-edge solutions. Rather, we
combine known and stable components in new
ways.
Between 2009 and 2011, the mobile solution was further
enhanced to include cell phone-enabled bar code tickets, a
wide-range GSM and mobile broadband services, and on-
flight broadband services (as the first airline in Europe). As
the CIO commented, “We strongly believe that the mobile
phone will become our key customer platform.”
This belief proved relevant earlier than anticipated. In April
2010, a volcano in Iceland, the Eyjafjallajøkull, erupted and
an enormous, shifting ash cloud covered parts of Europe for
about 10 days. Most of the North and Central European air-
space was closed, and hundreds of thousands of air passen-
gers were grounded. Angry passengers started to contact
airline and travel agent call centers, which quickly collapsed.
As an improvised response, Norwegian quickly established a
large-scale Facebook-enabled customer communications and
problem-solving operation, which actually addressed most of
the problems. Facebook was not only used for information
purposes, but also to negotiate customer rerouting and new
tickets. The team leader commented afterward, “Frankly, we
do not know where this Facebook thing will take us, but we
certainly realized that our customers preferred this commu-
nication channel in this urgent situation.”
Looking back at this evolution, Norwegian was characterized
by modularity in both organization and technology. The busi-
ness units of Norwegian were loosely connected, allowing for
modular innovation, while still being able to draw on business
and technical resources of other units. As an example, the IT
architect of Norwegian strongly believed that successful infra-
structure evolution depended on balancing increasing variety,
on the one hand, and modularization, on the other. If increas-
ing variety (resulting from continuous innovation) was not
balanced with continuous modularization of the architecture,
it would lead to chaos. Indeed, the IT architect reasoned, “My
first and top priority is protecting the integrity of the SOA
structure, no matter how important a project deadline is.”
Mechanisms at Norwegian
The remarkable success of Norwegian’s digital infrastructure
was characterized by growth along three dimensions: (1) ser-
MIS Quarterly Vol. 37 No. 3/September 2013 917
Henfridsson & Bygstad/Digital Infrastructure Evolution
vices, such as the web portals, the low-price calendar, the
bank, and mobile services; (2) users, from a few thousand in
2002 to several million in 2011; and (3) stakeholders, as
Norwegian aligned a number of networks to its infrastructure
over time. Our fieldwork revealed at least two conditions that
provided a powerful environment for digital infrastructure
evolution. First, the enabling service-oriented architecture,
which allowed the addition, replacement, or change of com-
ponents with relative ease. Second, the entrepreneurial and
open culture that supported an evolution that was not totally
dependent on top management directives. While these condi-
tions triggered the evolution, our data analysis showed that
there were three mechanisms behind the successful evolution
observed at Norwegian.
The innovation mechanism: As evidenced by the testi-
monies of our respondents, there existed a profound optimism
that Norwegian would be able to expand its business propo-
sition beyond its original scope. The malleability of the
service-oriented architecture created a space of possibilities
that served as a melting pot for innovation. In fact, this space
of possibilities spawned a pace of service innovation that, at
the time, was largely unheard of in the institutionalized airline
industry.
The infrastructure malleability fostered a creative process
involving IT personnel, business managers, and external
vendors, which allowed them to start recombining infrastruc-
ture resources such as technical elements, business routines,
vendors, and user groups. This recombination resulted in new
ideas for services, which in many cases materialized at
Norwegian. These services were typically designed in rela-
tively short, intensive projects, and then launched. In the
Norwegian case, it had to do with patterns of assembling
different components into new services (Figure 4).
We refer to this mechanism as the innovation mechanism, that
is, a self-reinforcing process by which new products and ser-
vices are created as infrastructure malleability spawns recom-
bination of resources.
The adoption mechanism: Norwegian managed to attract
customers to purchase tickets online. Although the web-
based infrastructure initially was relatively unsophisticated,
it was easy to use and relied on resources readily available to
customers. For instance, customers quickly embraced the
laser printing of tickets, available at their work or home
office. Indeed, the growth was so strong that by 2006—only
two years after the introduction of the services—online buyers
of tickets were a majority of customers (85 percent). This
convinced top management that the Internet strategy was
paying off, and that the airline should further exploit the
momentum of growth. Accordingly, with more users adopting
918 MIS Quarterly Vol. 37 No. 3/September 2013
the infrastructure services, more resources were allocated to
improve and extend the infrastructure. For instance, resources
were spent on maintaining rapid response time in spite of
increased website traffic, which, in turn, allowed for more
services offered.
As the number of users of the digital infrastructure increased,
revenues increased, which, in turn, attracted more resources
to the infrastructure for providing more and improved ser-
vices. We refer to this mechanism as the adoption mechanism
(Figure 5), that is, a self-reinforcing process by which more
users adopt the infrastructure as more resources invested
increase the usefulness of the infrastructure.
As indicated above, “new services” is the outcome of the
innovation mechanisms and “more services are offered” is the
starting point of the adoption mechanism. It makes analytical
sense to keep the two mechanisms apart, rather than viewing
them as parts of a single, more complex mechanism. Essen-
tially, the two mechanisms involve different sets of actors.
While business and technology developers drive the innova-
tion mechanism, users drive the adoption mechanism.
The scaling mechanism: The infrastructure at Norwegian
did not only succeed by expanding its services and user base.
It also increased its scope by including the partner solutions
of its emerging network of stakeholders. Norwegian’s busi-
ness infrastructure was unusually open, which lowered the
barriers for outside actors to integrate with the airline’s
infrastructure. For instance, early on, Norwegian standard-
ized its central business bus to achieve compatibility with the
European Amadeus system, opening up an entire array of
possible partners. Furthermore, in 2005, Norwegian provided
full access (through an application programming interface
referred to as the Norwegian Application Interface) to its ser-
vices to agencies and search engines. As a result, Norwegian
started to attract partners such as travel agencies, other air-
lines, hotel chains, and car rental companies. Partner solu-
tions were integrated with the web portal, which, as a result,
increased the reach of the infrastructure.
We refer to this as the scaling mechanism (Figure 6), that is,
a self-reinforcing process by which an infrastructure expands
its reach as it attracts new partners by creating incentives for
collaboration.
Table 4 summarizes the definitions of the mechanisms gener-
ated from our in-depth case study research at Norwegian. The
case survey, presented in the next section, uses these con-
ceptualizations as a starting-point for analyzing how mech-
anisms interact in infrastructure evolution. Looking back at
the Norwegian case study, there were clear indications that
the three mechanisms interacted. The innovation mechanism
 Henfridsson & Bygstad/Digital Infrastructure Evolution

Digital Infrastructure

Technical New
malleability services

Recombination

Figure 4. The Innovation Mechanism

Digital Infrastructure

More services More resources

offered invested

More users adopt

Figure 5. The Adoption Mechanism

Digital Infrastructure

Partners Reach
attracted expanded

Partner solutions
added

Figure 6. The Scaling Mechanism

Table 4. Mechanisms
Mechanism Definition
Innovation A self-reinforcing process by which new products and services are created as infrastructure
malleability spawns recombination of resources
Adoption A self-reinforcing process by which more users adopt the infrastructure as more resources
invested increase the usefulness of the infrastructure
Scaling A self-reinforcing process by which an infrastructure expands its reach as it attracts new
partners by offering incentives for collaboration

MIS Quarterly Vol. 37 No. 3/September 2013 919

Henfridsson & Bygstad/Digital Infrastructure Evolution
interacted with the adoption mechanism, in the sense that it
provided the new services that attracted more users. At the
same time, the adoption mechanisms provided the financial
resources necessary to maintain the innovation mechanism.
We also observed that the scaling mechanism increased the
space of innovation by adding more elements to the
infrastructure. This type of contingencies between mech-
anisms is explored in more depth in the following sections.
The Case Survey
Our in-depth case study research at the unusually successful
Norwegian company yielded three mechanisms of digital
infrastructure evolution. We chose the company as an
extreme case, since deriving ideal types is typically seen as a
useful starting-point for theory generation (Gerring 2007).
However, to substantiate this theoretical inquiry, we decided
to survey whether (1) these mechanisms were activated and
(2) resulted in a successful outcome, in other cases as well.
Taking a critical realist stance, we hypothesized the existence
of equifinality in digital infrastructure evolution. In other
words, we anticipated that there would be a set of different
causal paths by which the mechanisms could be combined to
produce successful outcomes. To investigate these issues fur-
ther, we conducted a case survey (Larsson 1993) of 41 cases
of infrastructure evolution reported in the scholarly literature
to analyze its space of configurations. Configurational
thinking has emerged as powerful way of studying “any
multidimensional constellation of conceptually distinct char-
acteristics that commonly occur together” (Meyer et al. 1993,
p. 1175). Among critical realists, Pawson and Tilley’s (1997)
thinking epitomizes the promise of configurational analysis
and, as outlined in the theoretical background above and fur-
ther elaborated in the description of methodological choices
below, their work serves as a backdrop to our analysis.
Methods
A case survey involves systematic collection and coding of
case studies, where preference is given to the case charac-
teristics rather than the original authors’ analysis and conclu-
sions (Yin and Heald 1975). Bridging nomothetic surveys
and ideographic case studies, the case survey method is an
inexpensive way of learning from many rich case studies
(Larsson 1993) and is ideal for configurational theorizing
(Fiss 2007).
Case Selection and Data Collection: Similar to Rivard and
Lapointe’s (2012) recent study of IT implementers’ responses
920 MIS Quarterly Vol. 37 No. 3/September 2013
to user resistance, we (1) collected a large sample of digital
infrastructure studies from scholarly sources, (2) refined the
initial sample using inclusion and exclusion criteria (Yin and
Heald 1975), and (3) coded the cases using the definitions of
the mechanisms identified in the in-depth study.
The initial sample, including journal articles, peer-reviewed
conference papers, book chapters, and working papers,
covered well over 60 cases. In our further investigation, the
inclusion criteria were (1) that the case documented the
evolution of a digital infrastructure, and (2) that the case
narrative was long enough to provide sufficiently rich data.
The main exclusion criterion was that the case was not rich
enough to make it possible to determine whether mechanisms
were actualized or not.
All in all, we included 41 cases in our sample (see Appendix).
The research database included cases from articles published
in information systems journals (e.g., MIS Quarterly, Infor-
mation Systems Research, Journal of AIS, and Information &
Organization), medicine (e.g., Methods of Information in
Medicine), development studies (e.g., Information Technology
for Development), as well as science and technology studies
(e.g., Science, Technology & Human Values). It also included
cases from articles published in conference proceedings (e.g.,
the International Conference on Information Systems and the
International Federation for Information Processing). In addi-
tion, cases were selected from book chapters and working
papers. The cases covered a variety of settings including
banking, healthcare, natural resources, pharmaceutics, public
sector, and telecom.
Coding: We used Pawson and Tilley’s (1997) configura-
tional framework for designing a coding scheme focusing on
three elements: context, mechanisms, and outcome. First, we
drew on our in-depth case study and the literature review to
distinguish technical architecture and organizational control
as two key contextual conditions. In the in-depth case study,
we discovered that a loosely coupled architecture and decen-
tralized control were important conditions at Norwegian for
activating the power of the identified mechanisms. Based on
this insight, we returned to our database of related literature
and studies to (1) assess what other possible architectures
were present in digital infrastructure evolution and (2) assess
what other modes of control were present. In the architecture
case, our assessment yielded several examples of infrastruc-
tures that were based on tight integration (see Hanseth and
Braa 2000). We decided to label this category tightly coupled
architecture. In the case of control as contextual condition,
previous studies strike a difference between top-down imple-
mentation and bottom-up cultivation of infrastructures. We
decided to label this distinction as centralized versus
decentralized control (see Kirsch 2004; Yoo et al. 2010).

Then, the three mechanisms (and their definitions, see
Table 4) generated from the in-depth study were used as a
basis for our coding of the studies included in the case survey.
The coding of mechanisms required careful assessment of
whether the mechanism in question in fact was actualized.
For example, most (if not all) digital infrastructure projects
have the ambition of user adoption, and therefore involve
management interventions designed to achieve this. How-
ever, the sheer presence of management intervention to
achieve user adoption was not considered sufficient to code
the adoption mechanism as actualized. Rather, this would
require evidence that user adoption was self-reinforcing, as
described in Figure 1. As an example, consider that a man-
agement team may stimulate user adoption (action-formation)
through a commercial campaign in the hope that it would
increase user adoption of the infrastructure. However, if we,
in this hypothetical example, could not verify that this
management intervention paid off in a self-reinforcing process
that generated resources (transformation) that maintained user
adoption beyond the end of the campaign, we reasoned that
the mechanism was not actualized. Similarly, the other two
mechanisms were coded positively only to the extent that they
were self-reinforcing. In this context, it should be empha-
sized that, in specific cases where mechanisms were coded as
unactualized, the claim of actual presence of such unactual-
ized mechanisms builds on an assumption of coherence across
cases of infrastructure evolution.
Finally, we coded outcomes based on the relative success of
the evolution process. Starting from the assessment of the
author(s), we critically examined if the infrastructure had
reached a state where it displayed a capacity to endure, and
adapt to environmental changes. In particular, we interpreted
if (1) the infrastructure filled a relevant role in its business
ecosystem, and (2) the extent to which the infrastructure’s
affordances were possible to escape for its stakeholders. If in
doubt, we supported our interpretations by disciplined imagi-
nation of possible scenarios through which individual
stakeholders would be able to reverse the infrastructure’s state
of success or failure.
Three people independently coded the 41 studies included in
the sample. In addition to the two authors of this paper, a
Master’s student served as a coder. In case of disagreements,
the coders reread the case and discussed the coding until 100
percent agreement was achieved. The inter-reliability of the
coding was approximately 85 percent in the first round of
independent coding.
Data Analysis: We first examined the relationship between
the eight possible combinations of the three mechanisms
(adoption, innovation, and scaling (AIS), AI, AS, IS, A, I, S,
Henfridsson & Bygstad/Digital Infrastructure Evolution
and none) and their outcomes (Table 5). This revealed clearly
that the absence of activated mechanisms was strongly asso-
ciated with a negative outcome, and that two configurations
were highly successful. Second, in order to understand the
causal paths of the two successful configurations (AS and
AIS), we conducted a qualitative investigation of all cases
associated with these configurations. The first step involved
comparative analysis between the cases in the same configu-
ration. This analysis revealed how the mechanisms inter-
acted. The second step involved a comparison of cases across
the two successful configurations. This helped us understand
why both AS and AIS produced a successful outcome. Third,
we analyzed contextual conditions and their relation to the
two configurations. One interesting observation was that the
activation of mechanisms in one configuration (AS) appeared
to be less dependent on the two contextual conditions advo-
cated in the literature than the other configuration.
Case Survey Results
Table 6 outlines the descriptive statistics of the 41 cases of
infrastructure evolution included in our sample. It shows the
frequency distribution of the cases across mechanism configu-
rations and evolution outcome. Since the three mechanisms
(adoption (A), innovation (I), scaling (S)) can be combined in
eight different ways (none, A, I, S, AI, AS, IS, AIS), there are
eight logically possible configurations of which a particular
case can be a member (see Appendix for coding details of
each case). Among the cases in the sample, the IS subset was
the only mechanism configuration that remained unactualized.
Furthermore, we identified 11 cases (26.8 percent) where
none of the 3 mechanisms were actualized, all of which
resulted in unsuccessful digital infrastructure evolution. At
the other end of the spectrum, we identified 12 cases (29.3
percent) as members of the AIS subset and 7 cases (17.1
percent) as members of the AS subset. These 19 cases were
coded as successes, and will be the primary focus for our
further analysis.
Our in-depth study at Norwegian established three individual
mechanisms: adoption, innovation, and scaling. It also sug-
gested that the actualization of all three of them in the same
case would effectively contribute to a successful outcome of
digital infrastructure evolution.9 The descriptive statistics of
our case survey support this suggestion by indicating a strong
correlation between the AIS configuration and a successful
9
As highlighted by one of the anonymous reviewers, this is not to suggest a
closed system. Consistent with critical realism assumptions, there might
always be other mechanisms “lurking around the corner.”
MIS Quarterly Vol. 37 No. 3/September 2013 921
Henfridsson & Bygstad/Digital Infrastructure Evolution

Table 5. Case Survey Data Analysis

Steps
1. Descriptive statistics
2. Qualitative analysis
3. Analysis of contextual
conditions
Tasks
a. Examine relationships between combinations of
mechanisms and outcomes
b. Identify highly successful and highly
unsuccessful combinations of mechanisms
a. Comparative analysis of cases within the two
successful combinations (AS and AIS)
b. Comparative analysis between the two
successful configurations
c. Examination of alternative explanations
a. Analyze the effect of decentralized control on
the two combinations
b. Analyze the effect of loosely-coupled
architecture on the two combinations
c. Assess explanatory power of the configurations
Outputs
Frequency distribution of the cases
across combinations of the three
identified mechanisms (Table 6)
A systematic assessment of causal
relationships in the successful cases.
Two configurations of successful
infrastructure evolution (Figure 7)
Completion of two successful causal
paths in digital infrastructure evolution

Table 6. Descriptive Statistics

Mechanism Unsuccessful Successful
Combination N (%) Infrastructure Infrastructure Total
NONE 11 (26.8%) 11 0 11 (100%)
A 3 (7.3%) 2 1 3 (100%)
I 4 (9.7%) 2 2 4 (100%)
S 1 (2.4%) 1 0 1 (100%)
AI 3 (7.3%) 1 2 3 (100%)
AS 7 (17.1%) 0 7 7 (100%)
IS 0 (0%) 0 0 0 (100%)
AIS 12 (29.3%) 0 12 12 (100%)
Total: 41 (100%) 17 (41.5%) 24 (58.5%)

outcome. Beyond this confirmation of the Norwegian case
study results, we can make at least two additional obser-
vations. First, the actualization of a single mechanism is
insufficient for leading to a successful outcome. For instance,
our case survey suggests that the innovation mechanism alone
is insufficient for a successful outcome. Although there were
a few exceptions, our results show that the innovation mech-
anism is contingent on the adoption and scaling mechanisms.
These findings resonate well with critical realism’s assump-
tion of contingent causality, suggesting that the outcome of a
mechanism depends on other mechanisms (Elder-Vass 2010;
Sayer 1992). Second, the results also suggest that the actuali-
zation of the innovation mechanism is not a necessary con-
dition for success if the adoption and scaling mechanisms are
interacting in the same evolution process. This is interesting,
since our analysis of the Norwegian case suggested it as a
vital mechanism for adoption and scaling in the AIS configu-
ration. This insight indicates that the natures of the AIS and
AS configurations are qualitatively different and warrant
further attention.
In view of these initial insights of our case survey results, we
further investigated how the mechanisms interacted in the two
successful configurations (AIS and AS). In the Norwegian in-
depth case study (as an example of the AIS configuration), we
noted how the three mechanisms interacted. In what follows,
we investigated whether the same type of interaction existed
in the 12 AIS cases included in the survey case. We also
investigated the seven cases with the AS configuration to
further understand how the adoption and scaling mechanisms
alone can interact to produce successful infrastructure evolu-
tion. In doing so, we also turned to the contextual conditions
of architecture and mode of control for both configurations to
understand the differences in causal paths. Manifesting our

922 MIS Quarterly Vol. 37 No. 3/September 2013


Figure 7. Configurations of Digital Infrastructure Evolution
configurational perspective, Figure 7 portrays the causal paths
that we identified through our analysis, including the number
of cases that followed each path.
Adoption-Innovation-Scaling: Our examination of the AIS
cases (Case #6, The Internet; Case #10, Environmental Health
in the French Public Health Administration; Case #12, Local
Danish Electronic Patient Record initiative; Case #22, The
SWIFT Network; Case #23, Criminal Case Management in
Finland; Case #27, Pharmaceutics; Case #29,Gateways versus
Standards; Case #30, Internet IPv6; Case #32, Telecom; Case
#33, Broadband Mobile Services in South Korea; Case #39,
U.S. Petroleum Company; Case #41, and U.S. Retail Com-
pany)10 included in the case survey revealed a similar
interplay of mechanisms as we found in the Norwegian case
study. The innovation and adoption mechanisms fed on each
other, which created fertile ground for the scaling mechanism
as combinatorial possibilities (innovation) increased and the
provision of more users (adoption) leveraged the scope of the
infrastructure. In the Local Danish Patient Record initiative
(Case #12, Aanestad and Blegind Jensen 2011), as one of
10
See Appendix for sources and coding.
Henfridsson & Bygstad/Digital Infrastructure Evolution
three examples11 summarized in Table 7, the innovative way
of using existing systems to achieve standardized extraction
of patient data (SEP) triggered adoption from county hos-
pitals, which, in turn, created the resources for establishing a
national SEP project that could support the initiative’s scale
and scope. Starting as a local initiative supported by two
county hospitals, the SEP infrastructure counted 4.3 million
Danes as users some 9 years after its inception. Similar inter-
action between mechanisms existed in the seemingly quite
different SWIFT network case (Case #22, Scott and
Zachariadis 2010), in which 68 banks in 11 countries joined
forces to create an infrastructure for financial transactions that
developed “from an efficiency initiative driven by a closed
‘society’ of banks to a network innovation of world-class
standing” (p. 2). The widespread adoption by banks of
SWIFT in the 1970s created incentives for innovating the
infrastructure further, where speed, costs, volume, security,
and uniform formats were benefits that were targeted in a
series of improvements of the infrastructure over the years.
11
Our selection of illustrative examples of the AIS configuration was done
with emphasis on making sense to readers without too much contextual
background.
MIS Quarterly Vol. 37 No. 3/September 2013 923
Henfridsson & Bygstad/Digital Infrastructure Evolution

Table 7. Content Analysis of Adoption–Innovation–Scaling Configuration

12. Local Danish Elec-
tronic Patients Records
initiative: The SEP project
in Denmark was a local
initiative in 2000 to share
patient data, but expanded
into a national solution
during a nine-year period.
22. International Financial
Communication Network:
SWIFT is an infrastructure
for financial transactions,
now used in 200 countries.
Starting in 1971 as a stan-
dard, SWIFT went live in
1977 and expanded into a
global infrastructure.
23. Criminal Case
Management in Finland:
The Criminal Case Manage-
ment system in Finland was
introduced in 1992, and
developed into a national
integrated infrastructure.
Contextual
conditions Mechanisms Outcome Reference
Architecture: Innovation: Started as a simple “Our empirical material Aanestad and
Modularized, problem-solving initiative, using describes two Danish Blegind Jensen
many systems existing systems in new ways. This initiatives, where a (2011)
involved was cleverly expanded into a national national project failed to
patient database. deliver interoperable
Control: Electronic Patient Record
Distributed Adoption: Starting with pilot in two systems while a small,
counties, gradually expanded into a local solution grew and
national solution. Patients were also now offers a nation-wide
allowed access. solution for sharing patient
record information” (p.
Scaling: First accessing one EPR 161).
and PAS, later including all such
systems in Denmark into the solution.
Architecture: Innovation: Building on Telex solu- SWIFT developed into Scott and
SWIFT is a net- tions, financial actors and technology one of the world’s key Zachariadis
work, with a vendors designed and developed the infra-structures, proving (2010)
common standard network with continuous innovation. fast, seamless and secure
implemented in a Packet switching was introduced in exchange of financial
large number of 1979, and SWIFT 2 was launched in transactions.
financial systems 1983.
“Over time, what began as
Control: Adoption: Initiated by 68 banks in 11 a closed society…grew
Distributed, countries. As services expanded, into an industry cooper-
SWIFT is owned even more banks and countries ative supporting an enthu-
by member banks joined. siastic community of prac-
tice and transformed into
Scaling: The expansion was charac- an unexpected network
terized by negotiations and adapta- phenomenon” (p. 2).
tions. Moving from a closed society,
SWIFT became a global network.
Architecture: Innovation: The Sakari solution Sakari was considered a Fabri (2009)
Modular, helped transforming the whole legal success in Finland.
expanded into criminal case process, and was “It is recognized that it has
service oriented extended with new services annually. helped make criminal
architecture proceedings quicker and
Adoption: Courts, police, prose- more accurate, () and the
Control: cutors and prisons were gradually system has also helped to
Centralized (but enrolled as new services were create a useful exchange
managed by integrated. of information and
representatives of practices among the
user institutions) Scaling: Linking into other structures different organizations and
was a key strategy. actors involved” (p. 123).

The financial infrastructure scaled considerably in terms of its
coverage and scope, and has since long become the most
comprehensive one in the global bank sector. Regarding the
Criminal Case Management system in Finland (Case #23,
Fabri 2009), innovation in the civil procedure rules (e.g.,
elimination of original signature requirement) boosted new
civil case management applications. The new infrastructure
was rapidly adopted by debt collecting companies, while
lawyers’ use progressed well but more slowly. Stimulated by
this success, the infrastructure increased its reach by ex-
panding into criminal cases, which usually is considered more
difficult since more stakeholders are involved.
In view of the similarities in interaction between the mech-
anisms in the AIS configuration cases, we further explored the
contextual conditions that actualized the AIS configuration.
As shown in Figure 7, and in line with the extant infrastruc-
ture literature, there was a high correlation between loosely

924 MIS Quarterly Vol. 37 No. 3/September 2013


coupled architecture and decentralized control and the AIS
configuration. One plausible explanation is that the innova-
tion mechanism is a less predictable process than adoption
and scaling, and that the innovation mechanism is dependent
on a “space of possibilities” (Bygstad 2010; Davenport and
Short 1992) that centralized control and tightly coupled IT
architectures cannot trigger.
Adoption-Scaling: Compared to the AIS cases, the examina-
tion of the AS cases (Case #1, Health IS in Developing
Countries; Case #9, Legal IS in Austria; Case #15, Power
Systems; Case #21, Health IS in India; Case #28, OSI Versus
IP standards; Case #34, University Software; and Case #36,
E-Government in Germany)12 showed significantly different
results. The results suggested that as long as the adoption and
scaling mechanisms were actualized, the powers of the inno-
vation mechanism did not have to be actualized. In view of
our analysis of the AIS configuration, where we noted that the
innovation mechanism helped in actualizing the adoption
mechanism, this result requires further analysis.
The Legal IS in Austria (Case #9, Bernroider and Koch 2009),
as one example summarized in Table 8, was a result of in-
house development work carried out by the Federal Chan-
cellery, sometimes done with the support of external systems
development firms. The system was initially adopted for
increasing efficiency. Over time, however, more users
including the ministries and eventually citizens were adopting
the infrastructure. Once citizens came into the picture, a web-
based version was developed in parallel to the system used in-
house to facilitate user access. The reach of the infrastructure
increased as citizens were provided comprehensive law infor-
mation through this web-based system. Later on, the infra-
structure was further extended in reach by digitally supporting
the process for producing judicial material from inception to
Internet publication. This coincided with the Council of
Ministers’ decision to adopt the infrastructure in all ministries.
Similar interaction between the adoption and scaling mech-
anisms could be observed in the Health in India (Case #21,
Sahay and Walsham 2006) and e-Government in Germany
(Case #36, Pipek and Wulf 2009) cases.
Given this interaction between adoption and scaling, which
was qualitatively different from the mechanisms interaction
in the AIS cases, we turned to the contexts of the cases. As
illustrated in Figure 7, three (Legal IS in Austria; Health IS in
India; and e-Government in Germany; see Table 8) out of the
seven AS cases were characterized by centralized control and
a tightly coupled architecture. This breaks with some of the
12
See Appendix for source and coding.
Henfridsson & Bygstad/Digital Infrastructure Evolution
seminal works in the infrastructure literature, which argue that
loosely coupled architecture and distributed control are
important elements in making evolution processes successful.
It appears that the adoption and scaling mechanisms reinforce
each other in a way that leads to successful evolution, and that
central control and tightly coupled architecture sometimes are
important conditions in making this happen. For instance, in
the Health IS in India case, centralized control supported the
scaling of the infrastructure beyond the pilot to align health
centers. Increasing the reach, in turn, the health IS solution
became more attractive to adopt. However, as a concluding
remark, it should be emphasized that our investigation of con-
textual conditions of the AS configuration does not provide a
clear-cut picture. Although being beyond the scope of this
study, this observation indicates that there exist other con-
textual conditions that interact with these mechanisms,
suggesting an opportunity for further research.
Discussion
The adoption of critical realism has helped us to explore a
number of issues that challenge the way we think about digital
infrastructure. In particular, we highlight the existence of
three generative mechanisms (adoption, innovation, and
scaling) that serve as causal powers in digital infrastructure
evolution. In addition, we develop a configurational perspec-
tive that suggests multiple causal paths of such evolution. In
what follows, we outline implications for infrastructure
research and discuss the usefulness of critical realism in infor-
mation systems research. In addition, we highlight the study
limitations and future issues of research.
Implications
Our work synthesizes and conceptualizes earlier insight on
self-reinforcing mechanisms (Grindley 1995; Katz and
Shapiro 1985; Sydow et al. 2009), innovation (Arthur 2009;
Schumpeter 1980), and scaling (Hanseth and Lyytinen 2010;
Pollock et al. 2007; Sahay and Walsham 2006) as generative
mechanisms. This is useful, since the causal powers of each
of the mechanisms have been largely masked in prior
research. Our critical realist conception of digital infrastruc-
ture recognizes that mechanisms act transfactually (see
Fleetwood 2009) and that the actualization of their powers is
contingent on other mechanisms. In this regard, our research
indicates that successful digital infrastructure evolution can-
not be explained by merely attending to a single mechanism.
Among the 41 cases included in our sample, there were only
3 cases where an individual mechanism coincided with a suc-
cessful outcome (see Case #18, Hanseth and Aanestad 2003;
MIS Quarterly Vol. 37 No. 3/September 2013 925
Henfridsson & Bygstad/Digital Infrastructure Evolution

Table 8. Content Analysis of Adoption-Scaling Configuration

Contextual
Conditions Mechanisms Outcome Reference
9. Legal IS in Austria: Architecture: Tightly Adoption: First, only internal The LIS was used by 4.5 million Bernroider and
The LIS system is a coupled (first main- users, then professional users per month, and was con- Koch (2009)
legal repository and frame, then Internet users, and finally the public. sidered successful, both in terms
supports the workflow version) of internal efficiency and public
for law making in Scaling: The Internet version access. It ensured support for the
Austria. Developed Control: Centralized was launched in 1996, and law process. It provided trans-
since 1972. (managed in-house in extended with eLaw access. parency and services to citizens.
ministry)
21. Health IS in India: Architecture: Tightly Adoption: Starting with a The HISP solution was in full use Sahay and
The design, develop- coupled (but adapt- simple solution and pilot in Andhra Pradesh. The Walsham (2006)
ment, and implemen- able) users, more services were infrastructure was scaled to
tation of a Health added, which increased the support a whole state in a low-
Information System in Control: Centralized user base. resource setting. Adoption,
Southern India. (through HISP project although challenging, was steadily
and state authorities) Scaling: Starting with a pilot, increasing.
the solution was scaled in two
steps, first to 46 health cen-
ters, then to the whole state.
36. e-Government in Architecture: Tightly Adoption: Comprehensive The solution was considered Pipek and Wulf
Germany: The case is coupled workflow requirements elicitation successful, with a consistent (2009)
a solution to support system process with users, started growth of services and users.
work processes con- growing use and adaptation. Transition to Internet technology
necting the state Control: Centralized, was problematic, but solved.
government located in cooperation Scaling: A solution that first
the state's capital with was proprietary, and later
the German was expanded to Internet
Bundesrat. technology. This facilitated
the alignment of new
stakeholders.

Case #31, Rolland and Monteiro 2002; Case #20, Scott and
Walsham 1998). In all of the other 21 successful cases, the
outcome was traced to a configuration of two or three mech-
anisms. Although there is accumulating evidence on what
drives infrastructure evolution in the literature, the generative
mechanisms reported in this paper offer the analytical dis-
tinctiveness needed not only to zoom in on specific aspects of
infrastructure but also to zoom out to understand its
contingencies.
In addition, while the literature is full of testimonies that
express the complex relationships between elements that
make up an infrastructure (Braa et al. 2007; Ciborra et al.
2000; Ciborra and Failla 2000; Hanseth et al. 2006), there
have been few, if any, attempts to formulate perspectives that
allow the simultaneous study of multiple causes. Concurring
with its critical realist underpinnings, our configurational
perspective recognizes both the inherent complexity of infra-
structures and the need of an analytical lens for making sense
of them. As an example, consider that we observed that
successful infrastructure evolution does not depend on the
innovation mechanism as long as the powers of the adoption
and scaling mechanisms are actualized. Both the AS and AIS
configurations lead to successful evolution. Such equifinality
is accommodated by our configurational perspective, which
allows investigation of specific causal paths as well as the
multiplicity characterizing digital infrastructure.
Moreover, previous work on digital infrastructure has sug-
gested that centralized control is detrimental to the outcome
of the evolution process (Ciborra et al. 2000). Our research
provides a less polarized view, where we found significant
empirical support for the prevalent stance in the case of the
AIS configuration, while our evidence for the AS configura-
tion tells another story. The same applies to architecture:
while a loosely coupled architecture is found to be a valuable
trigger for the AIS configuration, a tightly coupled architec-
ture does not impede the AS configuration from a successful
outcome. In other words, it seems that there is no specific
relationship between mode of control and architecture in
successful cases of combining the adoption and scaling
mechanisms.

926 MIS Quarterly Vol. 37 No. 3/September 2013


We believe that these findings have important implications for
practitioners who face the challenge of managing the evolu-
tion of digital infrastructures. Understanding the difference
between the two configurations is seminal for the choice of
management strategy. The AIS configuration describes the
causal path of the really ambitious undertakings, where inter-
action between the three mechanisms is crucial, and where the
innovation mechanism should be targeted as the driver.
Typical examples of such projects are the Norwegian (our in-
depth case study) and the International Financial Communi-
cation Network (Case #22, Scott and Zachariadis 2010) cases.
In such cases, managers should note that the dynamics of the
configuration require a loosely coupled architecture and
decentralized control in order to create the space of possi-
bilities necessary for actualizing the innovation mechanism.
Also, tight scheduling is detrimental, because the configura-
tion feeds on exploration until momentum is established.
Our study also suggests that the AS configuration involves
lower stakes than AIS because the interplay between the
adoption and scaling mechanisms is relatively straight-
forward. It allows for a wider choice of management inter-
ventions. Seeking to actualize this configuration, managers
should have confidence in traditional project management
techniques, and observe that both loosely coupled and tightly
coupled architectures may be effective.
The Usefulness of Critical Realism
in Information Systems
The promises of critical realism as a philosophical tradition
for information systems research have been voiced for some
time now (Mingers 2004, 2010, Smith 2010). However, to
date, there is a paucity of empirical research, underpinned by
critical realist assumptions, that makes substantive contribu-
tions to established streams of information systems research.
While there exist a few exceptions (Bygstad 2010; Smith
2010; Volkoff et al. 2007), current writings on critical realism
in our discipline mainly provide useful conceptual guidance.
Our research is an early example to challenge some of the
conventional wisdom in the extant literature of a substantive
area of information systems research.
The delay between the initial recognition of a philosophical
tradition and its wider adoption in the community is under-
standable. As an applied discipline, information systems is
largely motivated by its members’ capacity to generate
knowledge with a capacity to, directly or indirectly, contribute
to the corresponding community of practitioners (Lyytinen
1999). In this regard, recognition of new developments and
directions in the philosophy of science is a necessary but not
Henfridsson & Bygstad/Digital Infrastructure Evolution
sufficient step for paving the way for a philosophical tradition
in the discipline. For instance, it took some 10 years until the
early writings of Boland, Orlikowski, and Walsham on inter-
pretive research were appropriated in the field as a broadly
recognized basis for conducting high-quality research. One
important element of this consolidation was examples of
empirical research that could demonstrate its value in gener-
ating new knowledge deemed relevant by the community. In
doing this, they then served as sources of inspiration for
research to come.
In a similar vein, we view this article as an intermediate step
between the pioneering writings on critical realism and its
adoption in wider circles in the community. In particular, we
view the research design, combining the in-depth case study
and case survey methods, and the adoption of configurational
thinking as a promising direction for leveraging qualities of
critical realism in our discipline. For instance, as the field is
beginning to deal with complex phenomena such as digital
innovation (Yoo et al. 2010), digital infrastructure (Tilson et
al. 2010), and platforms (Tiwana et al. 2010), conceptions of
causality that recognize contingency and multiple paths are
needed (see El Sawy et al. 2010). Our configurational per-
spective offers an approach to accommodate such a view on
causality that can be reused in other settings. As indicated in
the implications section, we offer a number of contributions
to extant infrastructure literature that can be traced back to our
adoption of critical realism for information systems purposes.
Limitations and Issues for Future Research
Future studies could address several limitations in our work.
First, as noted by Rivard and Lapointe (2012), the use of
secondary data in case survey research certainly introduces
some limits to what can be inferred from the cases included in
the sample. The cases were originally written for a different
purpose, meaning that it would be unrealistic to imagine that
we would be able to create a situated understanding of the
particulars associated with each of the cases. However, our
use of inclusion and exclusion criteria for deciding which
cases to include in the sample restricted our inquiry to studies
that reported sufficiently rich accounts of the case setting. In
order to offer scholars of infrastructure the opportunity to
challenge our coding, we also included an overview of the
coding in the Appendix. This makes it possible for original
contributors to assess our coding, and it enables use of our
data material in future research. Second, the use of the case
survey method necessarily quantifies qualitative data in a way
that risks draining ideographic accounts of their richness. We
compensated for this loss of situated understanding by con-
sidering 41 cases in the same study in a relatively ideographic
MIS Quarterly Vol. 37 No. 3/September 2013 927
Henfridsson & Bygstad/Digital Infrastructure Evolution
manner. Third, mechanisms are causal structures that gener-
ate observable events, and we offer an understanding of three
mechanisms, as well as their combinations and contextual
conditions, that lead to successful infrastructures. We ac-
knowledge that the granularity of our analysis of causality is
at a relatively high level, suggesting the existence of nested
causal paths in digital infrastructure evolution left unad-
dressed in this study. Thus, we do not claim that we have
discovered all of the mechanisms relevant for infrastructure
evolution. For instance, while identifying mechanisms of
successful evolution of digital infrastructures, we have not ad-
dressed negative self-reinforcing mechanisms. One example
of such a mechanism is described by Hanseth et al. (2006) as
reflexive standardization. Generally increasing the com-
plexity of an infrastructure, this mechanism has been con-
firmed in recent cases, such as the European eCustoms case
(Case #25, Henningsson and Henriksen 2011). Future
research that adopts critical realism for studying negative self-
reinforcing mechanisms would be worthwhile. Finally,
focusing on successful configurations, our research primarily
deals with the equifinality of infrastructure evolution. While
our case survey findings also indicate multi-finality, we did
not further analyze why particular configurations result in
different outcomes (e.g., A, I, and AI). Our understanding of
digital infrastructure would benefit from future research on
this issue.
Conclusion
This paper proposes an alternative understanding of digital
infrastructure evolution, which emphasizes the relevance of
more closely examining its generative mechanisms. The
paper details and illustrates a critical realist approach to
digital infrastructure where configurational thinking serves as
a vehicle for understanding the combinations of mechanisms
that lead to successful evolution. We suggest that this ap-
proach may serve as a foundation for informing our under-
standing, as well as future studies, of digital infrastructure and
its inner workings.
Our perspective provides insights into the contributions and
limitations of previous understandings of digital infrastruc-
ture. To date, the four streams of digital infrastructure
research identified in our literature review have paved the
way for establishing an area of research that recognizes the
arrays of systems and technologies that confront today’s
managers and CIOs, rather than the conventional inquiry in
information systems within the confines of the single system.
However, as our research shows, conventional wisdom in the
area falls short when it comes to articulating the multiple
928 MIS Quarterly Vol. 37 No. 3/September 2013
paths by which successful digital infrastructure evolution
comes about. In the extant literature, there is a tendency to
offer partial explanations, rather than focusing attention on
the complete set of key mechanisms and their interaction.
This is problematic since it tends to inhibit a comprehensive
understanding of how and why digital infrastructures evolve
the way they do across cases.
In response, our critical realist approach accommodates both
the interpretivist and positivist assumptions manifested in the
extant infrastructure literature. The plausibility of this accom-
modation is demonstrated in the paper by offering theoretical
implications that challenge some of the hegemony in infra-
structure studies. It may appear somewhat ironic that our
perspective draws on many of the findings already available
in the area, yet brings new light on established truths by intro-
ducing new philosophical assumptions. We attribute this
result to the power of a philosophical tradition, critical
realism, which offers a set of new pillars for approaching
substantive areas in information systems. We hope that our
example will inspire others to adopt critical realism in their
attempts to advance areas of information systems research
where the findings in prior literature need to be released from
some of their long-established convictions.
Acknowledgments
We thank Carsten Sørensen, Olga Volkoff, Youngjin Yoo, the two
anonymous reviewers, the associate editor, and the senior editors for
constructive and insightful comments on earlier versions of this
manuscript. Special thanks to Ole Hanseth for intellectual support
in designing and crafting this research. We are also thankful for the
feedback received when presenting our work to researchers in the
global infrastructure group at the Department of Informatics,
University of Oslo. Finally, we thank Hrafnhildur Jonasdottir for
assistance in coding the sample of digital infrastructure cases used
in this research.
References
Aanestad, M., and Blegind Jensen, T. 2011. “Building Nation-
Wide Information Infrastructures in Healthcare Through Nodular
Implementation Strategies,” Journal of Strategic Information
Systems (20), pp. 161-176.
Archer, M. S., Bhaskar, R., Collier, A., Lawson, T., and Norrie, A.
1998. Critical Realism: Essential Readings, London: Rout-
ledge.
Arthur, W. B. 2009. The Nature of Technology: What It Is and
How It Evolves, New York: Free Press.
Beckert, J. 1999. “Agency, Entrepreneurs, and Institutional
Change: The Role of Strategic Choice and Institutionalized
Practices,” Organization Studies (20:5), pp. 777-799.

Bernroider, E. W. N., and Koch, S. 2009. “Aligning ICT and Legal
Frameworks in Austria’s e-Bureaucracy: From Mainframe to the
Internet,” in ICT and Innovation in the Public Sector—European
Studies in the Making of E-Government, F. Contini and G. F.
Lanzara (eds.), Basingstoke, UK: Palgrave Macmillan, pp.
147-173.
Bharadwaj, A. 2000. “A Resource-Based Perspective on Informa-
tion Technology Capability and Firm Performance: An Empi-
rical Investigation,” MIS Quarterly (24:1), pp. 169-196.
Bhaskar, R. A. 1979. The Possibility of Naturalism, London:
Routledge.
Bhaskar, R. A. 1997 [1975]. A Realist Theory of Science, London:
Verso.
Bhaskar, R.A. 1998. “General Introduction,” in Critical Realism:
Essential Readings, M. S. Archer, R. Bhaskar, A. Collier, T.
Lawson, and A. Norrie (eds.), London: Routledge, pp. ix-xxiv.
Braa, J., Hanseth, O., Heywood, A., Mohammed, W., and Shaw, V.
2007. “Developing Health Information Systems in Developing
Countries: The Flexible Standards Strategy,” MIS Quarterly
(31:2), pp. 381-402.
Broadbent, M., and Weill, P. 1997. “Management by Maxim: How
Business and IT Managers Can Create IT Infrastructures,” Sloan
Management Review, Spring, pp. 77-92.
Broadbent, M., Weill, P., and St.Clair, D. 1999. “The Implications
of Information Technology Infrastructure for Business Process
Redesign,” MIS Quarterly (23:2), pp. 159-182.
Bunge, M. 2004. “How Does it Work? The Search for Explanatory
Mechanisms.” Philosophy of the Social Sciences, (34:2), pp.
182-210.
Bygstad, B. 2008. “Information Infrastructure as Organization: A
Critical Realist View,” in Proceedings of the 29th International
Conference on Information Systems, Paris, December 14-17,
Paper 190.
Bygstad, B. 2010. “Generative Mechanisms for Innovation in
Information Infrastructures,” Information and Organization
(20:3-4), pp. 156-168.
Callon, M. 1986. “Some Elements of a Sociology of Translation:
Domestication of the Scallops and the Fishermen,” in Power,
Action and Belief: A New Sociology of Knowledge, J. Law, (ed.),
London: Routledge & Kegan, pp. 196-233.
Child, J. 1972. “Organizational Structure, Environment and Perfor-
mance: The Role of Strategic Choice,” Sociology (6:1), pp. 1-22.
Child, J. 1997. “Strategic Choice in the Analysis of Action, Struc-
ture, Organizations and Environment: Retrospect and Prospect,”
Organization Studies (18:1), pp. 43-76.
Ciborra, C. 2000. “From Alignment to Loose Coupling: From
Mednet to WWW.Roche.Com,” in From Control to Drift—The
Dynamics of Corporate Information Infrastructures, C. U.
Ciborra, K. Braa, A. Cordella, B. Dahlbom, A. Failla, O.
Hanseth, V. Hepsø, J. Ljungberg, E. Monteiro, and K. A. Simon
(eds.), Oxford: Oxford University Press, pp. 193-211.
Ciborra, C. U., Braa, K., Cordella, A., Dahlbom, B., Failla, A.,
Hanseth, O., Hepsø, V., Ljungberg, J., Monteiro, E., and Simon,
K. A. (eds.). 2000. From Control to Drift—The Dynamics of
Corporate Information Infrastructures, Oxford: Oxford Univer-
sity Press.
Ciborra, C., and Failla, A. 2000. “Infrastructure as a Process: The
Case of CRM at IBM,” in From Control to Drift—The Dynamics
Henfridsson & Bygstad/Digital Infrastructure Evolution
of Corporate Information Infrastructures, C. U. Ciborra, K. Braa,
A. Cordella, B. Dahlbom, A. Failla, O. Hanseth, V. Hepsø, J.
Ljungberg, E. Monteiro, and K. A. Simon (eds.), Oxford: Oxford
University Press, pp. 105-124.
Danermark, B., Ekstrom, M., Jakobsen, L., and Karlsson, J. C.
2002. Explaining Society. Critical Realism in the Social
Sciences, London: Routledge.
Davenport, T. H., and Short, J. E. 1990. “The New Industrial Engi-
neering: Information Technology and Business Process Reengi-
neering.” Sloan Management Review (31:4), pp. 11-27.
DeLanda, M. 2006. A New Philosophy of Society, London:
Continuum.
DeLone, W. H., and McLean, E. R. 1992. “Information Systems
Success: The Quest for the Dependent Variable,” Information
Systems Research (3:1), pp. 60-95.
DeLone, W. H., and McLean, E. R. 2003. “The Delone and Mclean
Model of Information Systems Success: A Ten-Year Update,”
Journal of Management Information Systems (19:4), pp. 9-30.
Demetriou, C. 2009. “The Realist Approach to Explanatory Mech-
anisms in Social Science: More than a Heuristic?” Philosophy of
the Social Sciences (39), pp. 440-462.
Edwards, P. N., Jackson, S. J., Bowker, G. C., and Williams, R.
2009. “Introduction: An Agenda for Infrastructure Studies,”
Journal of the Association for Information Systems (10:5), pp.
364-374.
Elder-Vass, D. 2010. The Causal Power of Social Structures:
Emergence, Structure and Agency, Cambridge, UK: Cambridge
University Press.
El Sawy, O. A., Malhotra, A., Park, Y., and Pavlou, P.A. 2010.
“Seeking the Configurations of Digital Ecodynamics: It Takes
Three to Tango,” Information Systems Research (21:4), pp.
835-848.
Engeström, Y. 1987. Learning by Expanding, Helsinki: Orienta-
Konsultit Oy.
Fabri, M. 2009. “E-Justice in Finland and in Italy: Enabling
Versus Constraining Models,” in ICT and Innovation in the
Public Sector: European Studies in the Making of E-
Government, F. Contini and G. F. Lanzara (eds.), Basingstoke,
UK: Palgrave Macmillan, pp. 115-146.
Fiss, P. C. 2007. “A Set-Theoretic Approach to Organizational
Configurations,” Academy of Management Review (32:4), pp.
1180-1198.
Fleetwood, S. 2009. “The Ontology of Things, Properties and
Powers,” Journal of Critical Realism (8:3), pp. 343-366.
Fleetwood, S. 2011. “Powers and Tendencies Revisited,” Journal
of Critical Realism (10:1), pp. 80-99.
George, A. L., and Bennett, A. 2005. Case Studies and Theory
Development in the Social Sciences, Cambridge, MA: MIT
Press.
Gerring, J. 2007. Case Study Research: Principles and Practices,
New York: Cambridge University Press.
Glennan, S. 2009. “Mechanisms, Causes, and the Layered Model
of the World,” Philosophy and Phenomenological Research
(81:2), pp. 362-381.
Grindley, P. 1995. Standards, Strategy, and Policy: Cases and
Stories, New York: Oxford University Press.
Hagel III, J., Seely Brown, J., and Davison, L. 2010. The Power of
Pull: How Small Moves, Smartly Made, Can Set Big Things in
Motion, New York: Basic Books.
MIS Quarterly Vol. 37 No. 3/September 2013 929
Henfridsson & Bygstad/Digital Infrastructure Evolution
Hanseth, O., and Aanestad, M. 2003. “Design as Bootstrapping:
On the Evolution of ICT Networks in Health Care,” Methods of
Information in Medicine (42), pp. 385-391.
Hanseth, O., and Braa, K. 2000. “ Who’s in Control: Designers,
Mangers—or Technology? Infrastructures at Norsk Hydro,” in
From Control to Drift—The Dynamics of Corporate Information
Infrastructures, C. U. Ciborra, K. Braa, A. Cordella, B. Dahlbom,
A. Failla, O. Hanseth, V. Hepsø, J. Ljungberg, E. Monteiro, and
K. A. Simon (eds.), Oxford, UK: Oxford University Press, pp.
125-147.
Hanseth, O., Jacucci, E., Grisot, M., and Aanestad, M. 2006.
“Reflexive Standardization: Side Effects and Complexity in
Standard Making,” MIS Quarterly (30:2), pp. 563-581.
Hanseth, O., and Lyytinen, K. 2010. “Design Theory for Dynamic
Complexity in Information Infrastructures: The Case of Building
Internet,” Journal of Information Technology (25), pp. 1-19.
Hanseth, O., and Monteiro, E. 1997. “Inscribing Behaviour in
Information Infrastructure Standards,” Accounting, Management
& Information Technology (7:4), pp. 183-211.
Hanseth, O., Monteiro, E., and Hatling, M. 1996. “Developing
Information Infrastructure: The Tension Between Standardi-
zation and Flexibility,” Science, Technology, and Human Values
(11:4), pp. 407-426.
Hedström, P. 2008. “Studying Mechanisms to Strengthen Causal
Inferences in Quantitative Research,” in The Oxford Handbook
of Political Methodology, J. M. Box-Steffensmeier, H. E. Henry
E. Brady, and D. Collier, (eds.), Oxford, UK: Oxford University
Press, pp. 319-336.
Hedström, P., and Swedberg, R. 1998. “Social Mechanisms: An
Introductory Essay,” in Social Mechanisms: An Analytical
Approach to Social Theory, P. Hedström and R. Swedberg (eds.),
Cambridge, UK: Cambridge University Press, pp. 1-31.
Henningsson, S., and Henriksen, H. Z. 2011. “Inscription of
Behaviour and Flexible Interpretation in Information Infrastruc-
tures: The Case of European e-Customs,” Journal of Strategic
Information Systems (20:4), pp. 355-372.
Holland, J. H. 1995. Hidden Order: How Adaptation Builds
Complexity, Reading, MA: Addison-Wesley.
Katz, M. L., and Shapiro, C. 1985. “Network Externalities, Compe-
tition and Compatibility,” American Economic Review (75), pp.
424-440.
Kirsch, L. J. 2004. “Deploying Common Systems Globally: The
Dynamics of Control,” Information Systems Research (15:4), pp.
374-395.
Larsson, R. 1993. “Case Survey Methodology: Quantitative
Analysis of Patterns Across Case Studies,” Academy of
Management Journal (36:6), pp. 1515-1546.
Latour, B. 1987. Science in Action: How to Follow Scientists and
Engineers Through Society, Cambridge, MA: Harvard Univer-
sity Press.
Lave, J., and Wenger, E. 1991. Situated Learning: Legitimate
Peripheral Participation, Cambridge, UK: Cambridge Univer-
sity Press.
Lee, A. S. 2004. “Thinking About Social Theory and Philosophy
for Information Systems,” in Social Theory and Philosophy for
Information Systems, J. Mingers and L. Willcocks (eds.),
Chichester, UK: Wiley, pp. 1-26.
Lopreato, J., and Alston, L. 1970. “Ideal Types and the Idealization
Strategy,” American Sociological Review (35:1), pp. 88-96.
930 MIS Quarterly Vol. 37 No. 3/September 2013
Lyytinen, K. 1999. “Empirical Research in Information Systems:
On the Relevance of Practice in Thinking of IS Research,” MIS
Quarterly (23:1), pp. 25-27.
Malhotra, A., Gosain, S., and El Sawy, O. A. 2005. “Absorptive
Capacity Configurations in Supply Chains: Gearing for Partner-
Enabled Market Knowledge Creation,” MIS Quarterly (29:1), pp.
145-187.
Merton, R. 1967. On Theoretical Sociology: Five Essays, Old and
New, New York: The Free Press.
Meyer, A. D., Tsui, A. S., and Hinings, C. R. 1993. “Con-
figurational Approaches to Organizational Analysis,” Academy
of Management Journal (36:6), pp. 1175-1195.
Miles, M. B., and Huberman, A. M. 1994. Qualitative Data
Analysis, Thousand Oaks: Sage Publications.
Mingers, J. 2001. “Combining Is Research Methods: Towards a
Pluralist Methodology,” Information Systems Research (12:3),
pp. 240-259.
Mingers, J. 2004. “Re-Establishing the Real: Critical Realism and
Information Systems Research,” in Social Theory and Philosophy
for Information Systems, J. Mingers and L. Willcocks (eds.),
Chichester, UK: Wiley, pp. 372-406.
Mingers, J. 2010. “The Contribution of Systemic Thought to
Critical Realism,” Journal of Critical Realism (10:3), pp.
303-330.
Mol, A., and Law, J. (eds.). 2002. Complexities: Social Studies of
Knowledge Practices, Durham, NC: Duke University Press.
Ohlsson, S., and Lehtinen, E. 1997. “Abstraction and the
Acquisition of Complex Ideas,” Journal of Educational Research
(27:1), pp. 37-48.
Orlikowski, W. J., and Baroudi, J. J. 1991. “Studying Information
Technology in Organizations: Research Approaches and
Assumptions,” Information Systems Research (2:1), pp. 1-28.
Pawson, R., and Tilley, N. 1997. Realistic Evaluation, London:
Sage Publications.
Pawson, R., and Tilley, N. 2009. “Realistic Evaluation,” in
Evidence-Based Practice: Modernising the Knowledge Base of
Social Work?, H-U. Otto, A. Polutta, H. Ziegler (eds.),
Leverkusen, Germany: Barbara Budrich, pp. 151-180.
Pettigrew, A. M. 1985. “Contextualist Research and the Study of
Organizational Change Processes,” in Research Methods in
Information Systems, E. Mumford, R. Hirschheim, G. Fitgerald,
and A. T. Wood-Harper (eds.), Amsterdam: North-Holland, pp.
53-78.
Pipek, V., and Wulf, V. 2009. “Infrastructuring: Toward an Inte-
grated Perspective on the Design and Use of Information
Technology,” Journal of Association for Information Systems
(10:5), pp. 447-473.
Plowman, D. A., Baker, L. T., Beck, T. E., Kulkarni, M., Solansky,
S. T., and Travis, D. V. 2007. “Radical Change Accidentally:
The Emergence and Amplification of Small Change,” Academy
of Management Journal (50:3), pp. 515-543.
Pollock, N., Williams, R., and D’Adderio, L. 2007. “Global Soft-
ware and its Provenance: Generification Work in the Production
of Organizational Software Packages,” Social Studies of Science
(37:2), pp. 254-280.
Ragin, C. C. 2006. “Set Relations in Social Research: Evaluating
Their Consistency and Coverage,” Political Analysis (14:3), pp.
291-310.

Ragin, C. C. 2008. Redesigning Social Inquiry: Fuzzy Sets and
Beyond, Chicago: The University of Chicago Press.
Rai, A., Patnayakuni, R., and Seth, N. 2006. “Firm Performance
Impacts of Digitally Enabled Supply Chain Integration Capa-
bilities,” MIS Quarterly (30: 2), pp. 225-246.
Rihoux, B., and Ragin, C. C. (eds.). 2009. Configurational Com-
parative Methods: Qualitative Comparative Analysis (QCA) and
Related Techniques, Thousand Oaks, CA: SAGE Publications.
Rivard, S., and Lapointe, L. 2012. “Information Technology
Implementers’ Responses to User Resistance: Nature and
Effects,” MIS Quarterly (36:3), pp. 897-920.
Rolland, K., and Monteiro, E. 2002. “Balancing the Local and the
Global in Infrastructural Information Systems,” Information
Society (18:2), pp. 87-100.
Sahay, S., and Walsham, G. 2006. “Scaling of Health Information
Systems in India: Challenges and Approaches,” Information
Technology for Development (12:3), pp. 165-200.
Sayer, A. 1992. Method in Social Science. A Realist Approach,
New York: Routledge.
Schumpeter, J. 1980. Theory of Economic Development, London:
Transaction Publishers.
Scott, S., Van Reenen, J., and Zachariadis, M. 2010. “The Impact
of the Diffusion of a Financial Innovation on Company Per-
formance: An Analysis of SWIFT Adoption,” Discussion Paper
No 992, London School of Economics.
Smith, M. L. 2010. “Testable Theory Development for Small-N
Studies: Critical Realism and Middle-Range Theory,” Inter-
national Journal of Information Technologies and Systems
Approach (3:1), pp. 41-56.
Star, S. L., and Ruhleder, K. 1996. “Steps Toward an Ecology of
Infrastructure: Design and Access for Large Information
Spaces,” Information Systems Research (7:1), pp. 111-134.
Sydow, J., Schreyögg, G., and Koch, J. 2009. “Organizational Path
Dependence: Opening the Black Box,” Academy of Management
Review (34:4), pp. 689-709.
Tanriverdi, H., Konana, P., and Ge, L. 2007. “The Choice of
Sourcing Mechanisms for Business Processes,” Information
Systems Research (18:3), pp. 280-299.
Tilson, D., Lyytinen, K., and Sorensen, C. 2010. “Digital Infra-
structures: The Missing IS Research Agenda,” Information
Systems Research (21:4), pp. 748-759.
Tiwana, A., Konsynski, B., and Bush, A. A. 2010. “Platform
Evolution: Coevolution of Platform Architecture, Governance,
and Environmental Dynamics,” Information Systems Research
(21:4), pp. 685-687.
Urry, J. 2003. Global Complexity, Cambridge, UK: Polity Press.
Vaast, E., and Walsham, G. 2009. “Trans-Situated Learning:
Supporting a Network of Practice with an Information Infra-
structure,” Information Systems Research (20:4), pp. 547-564.
Vergne, J.-P., and Durand, R. 2010. “The Missing Link Between
the Theory and Empirics of Path Dependence: Conceptual
Clarification, Testability Issue, and Methodological Implica-
tions,” Journal of Management Studies (47:4), pp. 736-759.
Volkoff, O., Strong, D. M., and Elmes, M. B. 2007. “Technological
Embeddedness and Organizational Change,” Organization
Science (18:5), pp. 832-848.
Henfridsson & Bygstad/Digital Infrastructure Evolution
Walsham, G. 1995. “The Emergence of Interpretivism in IS
Research,” Information Systems Research (6:4), pp. 376-394.
Weber, M. 1949. The Methodology of the Social Sciences,
Glencoe, IL: Free Press.
Webster, J., and Watson, R. T. 2002. “Analyzing the Past to
Prepare for the Future: Writing a Literature Review,” MIS
Quarterly (26:2), pp. xiii-xxiii.
Wynn, D., and Williams, C. K. 2012. “Principles for Conducting
Critical Realist Case Study Research in Information Systems,”
MIS Quarterly (36:3), pp. 787-810.
Yin, R. K., and Heald, L. A. 1975. “Using the Case Survey Method
to Analyze Policy Studies,” Administrative Science Quarterly
(20:3), pp. 371-381.
Yoo, Y., Henfridsson, O., and Lyytinen, K. 2010. “The New
Organizing Logic of Digital Innovation: An Agenda for Infor-
mation Systems Research,” Information Systems Research (21:4),
pp. 724-735.
Yoo, Y., Lyytinen, K., and Yang, H. 2005. “The Role of Standards
in Innovation and Diffusion of Broadband Mobile Services: The
Case of South Korea,” Journal of Strategic Information Systems
(14), pp. 323-353.
Zammuto, R. F., Griffith, T. L., Majchrzak, A., Dougherty, D. J.,
and Faraj, S. 2007. “Information Technology and the Changing
Fabric of Organization,” Organization Science (18:5), pp.
749-762.
Zittrain, J. L. 2006. “The Generative Internet,” Harvard Law
Review (119), pp. 1974-2040.
About the Authors
Ola Henfridsson is a professor of Information Systems and
Management at Warwick Business School, University of Warwick.
He is also an adjunct professor at the Department of Applied
Information Technology, Chalmers University of Technology. His
research interests include digital innovation and technology
management. Ola’s research has been published in Information
Systems Research, MIS Quarterly, and other journals in the infor-
mation systems discipline. He is a senior editor of Journal of
Information Technology and a former senior editor of MIS
Quarterly. He also serves on the editorial boards of Information
Technology and People and Journal of the Association for
Information Systems.
Bendik Bygstad is a sociologist who is currently a professor at the
Norwegian School of Information Technology, and an adjunct
professor at the University of Oslo. His main research interests are
IT-based service innovation and the relationship of IS and
organizational change. He is also interested in IS research methods,
in particular the philosophical and methodological implications of
critical realism. He has published articles in journals such as
Information Systems Journal, Journal of Information Technology,
and International Journal of Project Management.
MIS Quarterly Vol. 37 No. 3/September 2013 931
932 MIS Quarterly Vol. 37 No. 3/September 2013
 SPECIAL ISSUE: CRITICAL REALISM IN IS RESEARCH

THE GENERATIVE MECHANISMS OF DIGITAL

INFRASTRUCTURE EVOLUTION
Ola Henfridsson
Warwick Business School, The University of Warwick,
Coventry CV4 7AL UNITED KINGDOM {ola.henfridsson@wbs.ac.uk}

Bendik Bygstad
Norwegian School of IT, Schweigaards gt. 14, 0185 Oslo NORWAY and
Department of Informatics, University of Oslo, Oslo NORWAY {bendik.bygstad@nith.no}

Appendix
Sample Cases

Contextual Conditions
Architecture: tightly coupled (0); loosely coupled (1)
Control: centralized (0); decentralized (1)

Mechanisms
Adoption (A): unactualized (0); actualized (1)
Innovation (I): unactualized (0); actualized (1)
Scaling (S): unactualized (0); actualized (1)

Outcome: unsuccessful (0); successful (1)

Comb: combination of mechanisms

Contextual Out-
No Case Conditions Mechanisms come Comb Reference
Arc Con A I S
1 Health Information Systems 1 1 1 0 1 1 AS Braa, J., Hanseth, O., Heywood, A., Mohammed,
Project HISP: A successful W., and Shaw, V. 2007. “Developing Health
standardization strategy in low- Information Systems in Developing Countries:
resource countries, based on The Flexible Standards Strategy,” MIS Quarterly
flexible and simple solutions. (31:2), pp. 381-402.
Continuously from 1992-2007.
2 National Hospital: A case of 0 0 0 0 0 0 – Hanseth, O., Jacucci, E., Grisot, M., and
increasing complexity of Aanestad, M. 2006. “Reflexive Standardization:
requirements, leading to Side Effects and Complexity in Standard
paralysis. Making,” MIS Quarterly (30:2), pp.563-581.

MIS Quarterly Vol. 37 No. 3—Appendix/September 2013 A1

Henfridsson & Bygstad/Digital Infrastructure Evolution
No Case
3 Norsk Hydro: A case of an
expanding corporate standard in
the 1990s, leading to broad
adoption, but difficult to scale
4 IBM: An innovative CRM project,
with scaling problems.
5 EDI: An ambitious project in
health, but failing to align a
complex network of actors and
technology.
6 Internet: Describes how the
dynamics of bootstrapping and
adaptation explains the success
of the Internet.
7 Genome project: An ambitious
scientific community project,
which fails to establish a
sustainable solution.
8 Statoil: An innovative project of
knowledge management, which
fails to trigger internal dynamics.
9 Legal systems: An expanding
legal infrastructure in Austria,
growing organically from 1972.
10 Environmental Health in the
French Public Health
Administration: Analyzes a
successfully distributed network
of practice, 2000 to 2005,
supported by an emerging
information infrastructure.
A2 MIS Quarterly Vol. 37 No. 3—Appendix/September 2013
Contextual Out-
Conditions Mechanisms come Comb Reference
Arc Con A I S
0 0 1 1 0 1 AI Hanseth, O., and Braa, K. 2000. “Who’s in
Control: Designers, Managers or Technology?
Infrastructures at Norsk Hydro,” in From Control
to Drift—The Dynamics of Corporate Information
Infrastructures, C. U. Ciborra, K. Braa, A.
Cordella, B. Dahlbom, A. Failla, O. Hanseth, V.
Hepsø, J. Ljungberg, E. Monteiro, and K. A.
Simon (eds.), Oxford: Oxford University Press,
pp. 125-147.
0 0 1 1 0 0 AI Ciborra, C., and Failla, A. 2000. “Infrastructure
as a Process: The Case of CRM at IBM,” From
Control to Drift—The Dynamics of Corporate
Information Infrastructures, C. U. Ciborra, K.
Braa, A. Cordella, B. Dahlbom, A. Failla, O.
Hanseth, V. Hepsø, J. Ljungberg, E. Monteiro,
and K. A. Simon (eds.), Oxford: Oxford
University Press, pp. 105-124.
1 1 0 0 0 0 – Monteiro, E., and Hanseth, O. 1995. “Social
Shaping of Information Infrastructure: On Being
Specific about the Technology,” in Information
Technology and Changes in Organizational
Work, W. J. Orlikowski, G. Walsham, M. R.
Jones, and J. I. I DeGross (eds.), London:
Chapman & Hall, pp. 325 -343.
1 1 1 1 1 1 AIS Hanseth, O., and Lyytinen, K. 2010. “Design
Theory for Dynamic Complexity in Information
Infrastructures: The Case of Building Internet,”
Journal of Information Technology (25:1),
pp.1-19.
0 1 0 0 0 0 – Star, S. L., and Ruhleder, K. 1996. “Steps
Toward an Ecology of Infrastructure: Design
and Access for Large Information Spaces,”
Information Systems Research (71), pp.111-134.
0 0 0 0 0 0 – Hepsø, V., Monteiro, E., and Rolland, K. 2009.
“Ecologies of eInfrastructures,” Journal of the
AIS (10:5), pp.430-446.
0 0 1 0 1 1 AS Koch, S., and Bernroider, E. 2008. “Aligning
ICT and Legal Frameworks in Austria’s e-
Bureaucracy: From Mainframe to the Internet,”
in ICT and Innovation in the Public Sector:
European Studies in the Making of E-
Government, F. Contini and G. F. Lanzara
(eds.), Basingstoke, UK: Palgrave Macmillan,
pp. 147-173.
1 1 1 1 1 1 AIS Vaast, E., and Walsham, G. 2009. “Trans-
situated Learning: Supporting a Network of
Practice with an Information Infrastructure,”
Information Systems Research (20:4), pp.
547-564.

No Case
11 French Rail: Aiming to transfer
an airline booking system to a
railway context. Fails because of
“translation” problems.
12 Local Danish Electronic
Patient Record initiative: A
local initiative, which surprisingly
develops and scales into a
national Danish Electronic
Patient Record solution.
13 Health: A national Danish
Electronic Patient Record
standardization initiative, which
never gets off the ground.
14 GIS in India: An attempt to
introduce GIS technology into an
Indial local administration. Fails
because of “translation”
problems.
15 Power systems: En epic
description of how the US
electric grid and companies
expanded as networks of power.
16 Health Broadband Networks:
A telemedicine solution at the
National Hospital is successfully
innovated and adopted by health
personnel.
17 Health: An EDI initiative gets
mired in standardization issues,
and never comes off the ground.
18 Telemedicine: A successful
case of telemedicine in
ambulances, but mainly as a pilot
project.
19 EDIFACT standard: A national
standardization initiative fails
because of technical and
organizational complexity.
20 Banking: A study of an
innovative decision support
system, with limited adoption and
scaling.
21 Health IS: A successful
adoption and scaling of a health
information system in an Indian
state.
Henfridsson & Bygstad/Digital Infrastructure Evolution
Contextual Out-
Conditions Mechanisms come Comb Reference
Arc Con A I S
0 0 0 0 0 0 – Mitev, N. 2000. “Toward Social Constructionist
Understandings of IS Success and Failure:
Introducing a New Computerized Reservation
System,” in Proceedings of the 21st International
Conference of Information Systems, Brisbane,
Australia, pp. 84-93.
1 1 1 1 1 1 AIS Aanestad, M., and Blegind Jensen, T. 2011.
“Building Nation-Wide Information Infrastructures
in Healthcare Through Modular Implementation
Strategies,” Journal of Strategic Information
Systems (20:2), pp.161-176.
0 0 0 0 0 0 – Aanestad, M., and Blegind Jensen, T. 2011.
“Building Nation-Wide Information Infrastructures
in Healthcare Through Modular Implementation
Strategies,” Journal of Strategic Information
Systems (20:2), pp.161-176.
0 1 0 0 0 0 – Sahay, S., and Walsham, G. 1996.
“Implementation of GIS in India: Organizational
Issues and Implications,” International Journal of
Geographical Information Systems (10:4), pp.
385-404.
0 1 1 0 1 1 AS Hughes, T. P. 1987. “The Evolution of Large
Technical Systems,” in The Social Construction
of Technological Systems, W. E. Bijker, T. P.
Hughes, and T. Pinch (eds.), Cambridge, MA:
MIT Press, pp. 51-82.
1 0 1 1 0 1 AI Hanseth, O., and Aanestad, M. 2003. “Design
as Bootstrapping. On the Evolution of ICT
Networks in Health Care,” Methods of Informa-
tion in Medicine (42), pp.385-391.
1 1 0 0 0 0 – Hanseth, O., and Aanestad, M. 2003. “Design
as Bootstrapping. On the Evolution of ICT
Networks in Health Care,” Methods of Informa-
tion in Medicine (42), pp. 385-391.
0 0 1 0 0 1 A Hanseth, O., and Aanestad, M. 2003. “Design
as Bootstrapping. On the Evolution of ICT
Networks in Health Care,” Methods of Informa-
tion in Medicine (42), pp. 385-391.
1 1 0 0 0 0 – Hanseth, O., and Monteiro, E. 1997. “Inscribing
Behaviour in Information Infrastructure
Standards,” Accounting, Management and
Information Systems (7:4), pp. 183-211.
0 0 0 1 0 1 I Scott, S. V., and Walsham, G. 1998. “Shifting
Boundaries and New Technologies: A Case
Study in the UK Banking Sector,” in
Proceedings of the 19th International Conference
on Information Systems, Helsinki, Finland,
pp.177-187.
0 0 1 0 1 1 AS Sahay, S., and Walsham, G. 2006 “Scaling of
Health Information Systems in India: Challenges
and Approaches,” Information Technology for
Development (12:3), pp. 165-200.
MIS Quarterly Vol. 37 No. 3—Appendix/September 2013 A3
Henfridsson & Bygstad/Digital Infrastructure Evolution
No Case
22 The SWIFT Network: A
successful standards innovation
in early 1970s, and the gradual
expansion into a global financial
network.
23 Law: A Criminal Case
Management system in Finland
was developed and expanded
gradually from 1992 into a
successful legal network.
24 Law: The Civil Trial Online
project is aimed at improving the
workflow between courts and
lawyers in Italy, had almost no
results after 6 years, due to
growing complexity.
25 e-Customs in Europe: An
ambitious EU project to integrate
customs had some local
successes, but has problems in
scaling and adoption.
26 Health: The NHS summary
care record project is
characterized by a number of
problems, and fails to establish a
sustainable development.
27 Pharmaceutics: The evolution
of an intranet, from corporate
asset to local adaptation in a
loosely coupled architecture.
28 OSI vs. IP standards:
Compares the development and
adoption of the OSI and IP
standards, explain-ing the
success of IP as the successful
balancing between flexibility and
standardization
29 Gateways vs. standards:
Analyzes a “standards war” in
Scandinavia in the 1980s,
concluding with the importance
of gateways.
30 Internet IPv6: The case
investigates the efforts in the
early 90s to address the IP
address shortage. Aligning the
various actor-networks and
protecting the installed base
proved successful.
A4 MIS Quarterly Vol. 37 No. 3—Appendix/September 2013
Contextual Out-
Conditions Mechanisms come Comb Reference
Arc Con A I S
1 1 1 1 1 1 AIS Scott, S., and Zachariadis, M. 2010. “A
Historical Analysis of Core Financial Services
Infrastructure: Society for Worldwide Interbank
Financial Telecommunication (S.W.I.F.T.),”
Working Paper Series, No 182., London School
of Economics and Political Science.
1 0 1 1 1 1 AIS Fabri, M. 2009. “E-Justice in Finland and in
Italy: Enabling Versus Constraining Models,” in
ICT and Innovation in the Public Sector: Euro-
pean Studies in the Making of E-Government, F.
Contini and G. F. Lanzara (eds.), Basingstoke,
UK: Palgrave Macmillan, pp. 115-146.
1 0 0 0 0 0 – Fabri, M. 2009. “E-Justice in Finland and in
Italy: Enabling Versus Constraining Models,” in
ICT and Innovation in the Public Sector: Euro-
pean Studies in the Making of E-Government, F.
Contini and G. F. Lanzara (eds.), Basingstoke,
UK: Palgrave Macmillan, pp. 115-146.
0 0 1 0 0 0 A Henningsson, S., and Henriksen, H. Z. 2011.
“Inscription of Behaviour and Flexible Interpreta-
tion in Information Infrastructures: The Case of
European e-Customs,” Journal of Strategic
Information Systems (20:4), pp. 355-372.
0 0 0 0 0 0 – Greenhalgh, T., Stramer, K., Bratan, T., Byrne,
E., Mohammad, Y., Russell, J. 2008. “Introduc-
tion of Shared Electronic Records: Multi-Site
Case Study Using Diffusion of Innovation
Theory,” British Medical Journal (337: a1786;
doi: 10.1136/bmj.a1786).
1 1 1 1 1 1 AIS Ciborra, C. U. 2000. “From Alignment to Loose
Coupling: From Mednet to WWW.Roche.Com,”
in From Control to Drift—The Dynamics of
Corporate Information Infrastructures, C. U.
Ciborra, K. Braa, A. Cordella, B. Dahlbom, A.
Failla, O. Hanseth, V. Hepsø, J. Ljungberg, E.
Monteiro, and K. A. Simon (eds.), Oxford:
Oxford University Press, pp. 193-211.
1 1 1 0 1 1 AS Hanseth, O., Monteiro, E., and Hatling, M. 1996.
“Developing Information Infrastructure: The
Tension Between Standardization and
Flexibility,” Science, Technology, and Human
Values (11:4), pp. 407-426.
1 1 1 1 1 1 AIS Hanseth, O. 2001. “Gateways—Just as
Important as Standards: How the Internet Won
the ‘Religious War’ about Standards in
Scandinavia,” Knowledge, Technology and
Policy (14:3), pp. 71-89.
1 1 1 1 1 1 AIS Monteiro, E. 1998. “Scaling Information Infra-
structure: The Case of the Next Generation IP in
Internet,” The Information Society (143), pp.
229-245.
 Henfridsson & Bygstad/Digital Infrastructure Evolution

Contextual Out-
No Case Conditions Mechanisms come Comb Reference
Arc Con A I S
31 Maritime Classification 1 1 0 1 0 1 I Rolland, K., and Monteiro, E. 2002. “Balancing
Company: Balancing local the Local and the Global in Infrastructural Infor-
contexts and corporate mation Systems,” The Information Society
standards, adoption was (18:2), pp. 87-100.
successful, but scaling
problematic.
32 Telecom: The case explores the 1 1 1 1 1 1 AIS Nielsen, P., and Aanestad, M. 2006. “Control
balance between central control Devolution as Information Infrastructure Design
and local autonomy. Strategy: A Case Study of a Content Service
Relinquishing control led to Platform for Mobile Phones in Norway,” Journal
innovative and successful of Information Technology (21), pp. 185-194.
infrastructure.
33 Broadband Mobile Services in 1 1 1 1 1 1 AIS Yoo, Y., Lyytinen, K., and Yang, H. 2005. “The
South Korea: The case explains Role of Standards in Innovation and Diffusion of
the rapid diffusion of broadband Broadband Mobile Services: The Case of South
mobile services in Korea. Korea,” Journal of Strategic Information Systems
(14), pp. 323-353.
34 University software: SAP 0 1 1 0 1 1 AS Pollock, N., Williams, R., and D’Adderio, L.
module/Uni module: The 2007. “Global Software and its Provenance:
successful generification and Generification Work in the Production of
adaptation of two university Organizational Software Packages,” Social
software packages. Studies of Science (37:2), pp. 254-280.
35 Health: A case on an innovative 0 1 0 1 0 0 I Constantinides, P., and Barrett, M. 2006.
regional health network in “Large-Scale ICT Innovation, Power, and
Greece, with adoption and Organizational Change: The Case of a Regional
scaling problems, because failing Health Information Network,” Journal of Applied
to take part in a power network. Behavioral Science (41:1), pp. 76-90.
36 eGovernment: A project aimed 0 0 1 0 1 1 AS Pipek, V., and Wulf, V. 2009. “Infrastructuring:
at improving workflow between a Toward an Integrated Perspective on the Design
state and federal level, with and Use of Information Technology,” Journal of
sustainable growth. Association for Information Systems (10:5), pp.
447-473.
37 Health: A large NHS project in 0 0 0 1 0 0 I Greenhalgh, T. 2010. “Adoption, Non-Adoption,
the UK fails because of too and Abandonment of a Personal Electronic
technical focus and no clear Health Record: Case Study of HealthSpace,”
adoption strategies. British Medical Journal (341: c5814; doi:
10.1136/bmj.c5814)
38 US petroleum company 0 0 0 0 1 0 S Broadbent, M., Weill, P., and St.Clair, D. 1999.
CostCo: Infrastructure project “The Implications of Information Technology
with limited effect because of Infrastructure for Business Process Redesign,”
strong focus on cost reduction. MIS Quarterly (23:2), pp. 159-182.
39 US petroleum company: 0 0 1 1 1 1 AIS Broadbent, M., Weill, P., and St. Clair, D. 1999.
Innovative infrastructure project “The Implications of Information Technology
with redesigned business Infrastructure for Business Process Redesign,”
processes and sustaining MIS Quarterly (23:2), pp. 159-182.
growth.
40 US retail company: Innovative 0 0 1 0 0 0 A Broadbent, M., Weill, P., and St. Clair, D. 1999.
project with adoption and scaling “The Implications of Information Technology
problems. Infrastructure for Business Process Redesign,”
MIS Quarterly (23:2), pp. 159-182.
41 US retail company: Opportunity 0 0 1 1 1 1 AIS Broadbent, M., Weill, P., and St. Clair, D. 1999.
oriented project with redesigned “The Implications of Information Technology
processes and successful Infrastructure for Business Process Redesign,”
adoption and scaling. MIS Quarterly (23:2), pp. 159-182.

MIS Quarterly Vol. 37 No. 3—Appendix/September 2013 A5