UNIT – III

CASCADING BEHAVIOR

IN NETWORKS
 Syllabus
 Power Laws, Rich-Get-Richer Phenomenon,
Diffusion, Cascading Behavior, Cascades and
Clusters, Role of Weak Ties.
 Popularity as a Network
Phenomenon
 We have been studying situations in which a person’s behavior
or decisions depend on the choices made by other people –
either because the person’s rewards are dependent on what
other people do or because the choices of other people convey
information that is useful in the decision-making process.
 Popularity is a phenomenon characterized by extreme
imbalances: while almost everyone goes through life known
only to people in their immediate social circles, a few people
achieve wider visibility, and a very, very few attain global
name recognition.
 Analogous things could be said of books, movies, or almost
anything that commands an audience.
 Popularity as a Network
Phenomenon
 Web as a concrete domain to measure popularity very
accurately.
 It is easy to take a snapshot of the full Web and simply count
the number of links to high-profile Web sites such as Google,
Amazon, or Wikipedia.
 The full set of links pointing to a given Web page as the in-
links to the page.
 The number of in-links to a Web page is a measure of the
page’s popularity.
 Power Laws
 Power law means that the vast majority of nodes have very
few connections, while a few important nodes (we call them
Hubs) have a huge number of connections.
 In studies over many different Web snapshots, taken at
different points in the Web’s history, the recurring finding is
that the fraction of Web pages that have k in-links is
approximately proportional to 1/k2.
 The crucial point is that 1/k2 decreases much more slowly as
k increases.
 A function that decreases as k to some fixed power, such as
1/k2 in the present case, is called a power law.
 Power Laws
 Similar power laws arising in measures of popularity in
many other domains as well:
 For example, the fraction of telephone numbers that receive
k calls per day is roughly proportional to 1/k2.
 The fraction of books that are bought by k people is roughly
proportional to 1/k3.
 The fraction of scientific papers that receive k citations in
total is roughly proportional to 1/k3.
 Power Laws
 There’s a simple method that provides at least a quick test for
whether a data set exhibits a power-law distribution.
 Let f (k) be the fraction of items that have value k, and suppose
you want to know whether the equation f (k) = a/k c
approximately holds, for some exponent c and constant of
proportionality, a.
 Then, if we write this as f (k) = ak−c and take the logarithms of
both sides of this equation, we get
log f (k) = log a − c log k.
 This says that if we have a power-law relationship, and we plot
log f (k) as a function of log k, then we should see a straight
line: −c will be the slope, and log a will be the y-intercept.
 Power Laws
 Such a “log-log” plot thus provides a quick way to see if
one’s data set exhibits an approximate power-law: it is easy
to see if one has an approximately straight line, and one can
read off the exponent from the slope.
 Rich-Get-Richer Models
 In social networks, there is a phenomenon called Rich get Richer
also known as Preferential Attachment.
 In Preferential Attachment, a person who is already rich gets more
and more and a person who is having less gets less.
 For example, assume there are some students in a class and every
student is friend with some students which is called its degree i.e a
degree of a student Is the number of friends it has.
 Now the student with a higher degree is rich and the student with a
low degree is poor.
 Now suppose there comes a new student in the class and he/she has
to make m friends, so he/she will select students with a higher degree
and become friends with them which increases the degree of rich.
 This is called Rich getting Richer or Preferential Attachment.
 Rich-Get-Richer Models
 Ideas from the analysis of information cascades and
network effects provide the basis for a very natural
mechanism to generate power laws.
 Power laws arise from the feedback introduced by
correlated decisions across a population.
 We will build our model of power law not from the
internals of each person’s decision-making process
but from the observable consequences of decision-
making in the presence of cascades.
 Rich-Get-Richer Models
 Based on this idea, here is a simple model for the creation of links among Web
pages.
1. Pages are created in order, and named 1,2,3,...,N.
2. When page j is created, it produces a link to an earlier Web page by choosing
between actions (a) and (b) below according to the following probabilistic rule
(controlled by a single number p between 0 and 1):
(a) With probability p, page j chooses a page i uniformly at random from among all
earlier pages, and creates a link to this page i.
(b) With probability1−p, page j instead chooses a page i uniformly at random from
among all earlier pages and creates a link to the page that i points to.
(c) This describes the creation of a single link from page j; one can repeat this
process to create multiple, independently generated links from page j. (However, to
keep things simple, we will suppose that each page creates just one outbound link.)
 Rich-Get-Richer Models
 Part 2(b) of this process is the key: after finding a random
earlier page i in the population, the author of page j does
not link to i, but instead copies the decision made by the
author of page i – linking to the same page that i did.
 The main result about this model is that if we run it for
many pages, the fraction of pages with k in-links will be
distributed approximately according to a power law
1/k c, where the value of the exponent c depends on the
choice of p .
 Rich-Get-Richer Models
 The copying mechanism in 2(b) is really an
implementation of the “rich-get-richer” dynamics:
when you copy the decision of a random earlier
page, the probability that you end up linking to
some page l is directly proportional to the total
number of pages that currently link to l.
 Rich-Get-Richer Models
 we call this a “rich-get-richer” rule because the
probability that page l experiences an increase in
popularity is directly proportional to l’s current
popularity.
 This phenomenon is also known as preferential
attachment , in the sense that links are formed
“preferentially” to pages that already have high
popularity.
 A page’s popularity grows at a rate proportional to its
current value, and hence exponentially with time.
 Diffusion in Networks
 The distribution and transmission of culture and social
systems across geographic areas is called as “Diffusion”.
 Types of diffusion
1) Demic Diffusion: Diffusion through migration of a
population that had already adopted the innovation is termed
as Demic Diffusion.
2) Cultural Diffusion: Diffusion through presumed imitation
or adoption of what must have been considered a superior
system is termed as Cultural Diffusion.
 Diffusion in Networks
 Biological Diffusion: Diffusion and spread of diseases through
direct contact is termed as Biological Diffusion. Epidemiology
is the study of biological diffusion
 The diffusion process involves three main elements :
 Sender:A sender is responsible for initiating the diffusion
process.
 Receiver: A receiver receives the diffusion information from
the sender. Commonly, the number of receivers is higher than
the number of senders.
 Medium: This is the channel through which the diffusion
information is sent from the sender to the receiver.
 This can be TV, newspaper, social media ,social ties, air (in the
case of a disease spreading process), etc.
 Diffusion in Networks
 There are two distinct kinds of reasons why
imitating the behavior of others can be beneficial:
 Informational effects, based on the fact that the
choices made by others can provide indirect
information about what they know, and
 Direct-benefit effects, in which there are direct
payoffs from copying the decisions of others.
 The Diffusion of Innovations
 New behaviors, practices, opinions, conventions,
and technologies spread from person to person
through a social network, as people influence their
friends to adopt new ideas.
 Our understanding of how this process works in
sociology known as the diffusion of innovations
 The Diffusion of Innovations
 The person-to-person influence was due primarily to
informational effects: as people observed the decisions of
their network neighbors, they obtained indirect
information that led them to try the innovation as well.
 In Ryan and Gross’s study, they interviewed farmers to
determine how and when they decided to begin using
hybrid seed corn; they found that while most of the
farmers in their study first learned about hybrid seed corn
from salesmen, most were first convinced to try using it
based on the experience of neighbors in their community.
 The Diffusion of Innovations
 Other important studies in the diffusion of innovations
focused on settings in which decisions about adoption
were driven primarily by direct-benefit effects rather than
informational ones.
 A long line of diffusion research on communication
technologies has explored such direct-benefit effects; the
spread of technologies such as the telephone, the fax
machine, and e-mail has depended on the incentives
people have to communicate with friends who have
already adopted the technology .
 Modeling Diffusion through a
Network
 Network models based on direct-benefit effects because we
have certain social network neighbors like friends,
acquaintances, or colleagues.
 The benefits of adopting a new behavior increase as more
and more of these neighbors adopt it.
 For example, you may find it easier to collaborate with co-
workers if you are using compatible technologies.
 Similarly, you may find it easier to engage in social
interaction – all else being equal – with people whose
beliefs and opinions are similar to yours.
 Modeling Diffusion through a
Network
 A Networked Coordination Game
 In an underlying social network, we will study a situation in which each
node has a choice between two possible behaviors, labeled A and B.
 If nodes v and w are linked by an edge, then there is an incentive for
them to have their behaviors match.
 We represent this using a game in which v and w are the players and A
and B are the possible strategies.
 The payoffs are defined as follows:
• if v and w both adopt behavior A, they each get a payoff of a>0;
• if they both adopt B, they each get a payoff of b>0; and
• if they adopt opposite behaviors, they each get a payoff of 0.
 Modeling Diffusion through a
Network
 We can write this in terms of a payoff matrix.
 Modeling Diffusion through a
Network
 v’s choice of strategy will be based on the choices made by all of
its neighbors, taken together.
 Suppose that some of its neighbors adopt A, and some adopt B;
what should v do in order to maximize its payoff?
 This clearly depends on the relative number of neighbors doing
each, and on the relation between the payoff values a and b.
 With a little bit of algebra, we can make up a decision rule for v
quite easily, as follows.
 Suppose that a p fraction of v’s neighbors have behavior A, and a
(1−p) fraction have behavior B; that is, if v has d neighbors, then
pd adopt A and (1−p)d adopt B.
 Modeling Diffusion through a
Network
 So if v chooses A, it gets a payoff of pda, and if it
chooses B, it gets a payoff of (1−p)db.
 Thus, A is the better choice if
pda ≥(1−p)db,
 or, rearranging terms, if.
pda ≥(1−p)db
pda ≥db−pdb
pda +pdb ≥db
pd(a+b) ≥db
p ≥b/(a+b)
 Modeling Diffusion through a
Network

Node v must choose between behavior A and behavior B, based on what its neighbors are doing.
 Modeling Diffusion through a
Network
 Use q to denote this expression on the right-hand side.
 This inequality describes a very simple threshold rule: it
says that if a fraction of at least q = b/(a +b) of your
neighbors follow behavior A, then you should, too.
 When q is small, then A is the much more enticing behavior,
and it only takes a small fraction of your neighbors engaging
in A for you to do so as well.
 However, if q is large, then the opposite holds: B is the
attractive behavior, and you need a lot of your friends to
engage in A before you switch to A.
 Cascading Behavior
 In any network, there are two obvious equilibria to the network
wide coordination game: one in which everyone adopts A, and
another in which everyone adopts B.
 Suppose that everyone in the network is initially using B as a
default behavior.
 Then a small set of “initial adopters” all decide to use A.
 We will assume that the initial adopters have switched to A for
some reason outside the definition of the coordination game–they
have somehow switched due to a belief in A’s superiority, rather
than by following payoffs – but we’ll assume that all other nodes
continue to evaluate their payoffs using the coordination game.
 Cascading Behavior
 An initial set of nodes adopts A while everyone else
adopts B.
 Time then runs forward in unit steps; in each step,
each node uses the threshold rule to decide whether
to switch from B to A.
 The process stops either when every node has
switched to A or when we reach a step where no
node wants to switch, at which point things have
stabilized on coexistence between A and B.
 Cascading Behavior
 Let’s consider an example of this process using the
social network in this Figure
 Suppose that the coordination game is set up so that
a =3 and b =2; that is, the payoff to nodes
interacting using behavior A is 3 /2 times what it is
with behavior B.
 Using the threshold formula, we see that nodes will
switch from B to A if at least a fraction q
=2/(2+3)=2/5 of their neighbours are using A.
Cascading Behavior
 Cascading Behavior
 Now, suppose that nodes v and w form the set of initial adopters of
behavior A, while everyone else uses B.
 where dark circles denote nodes adopting A and lighter circles denote
nodes adopting B.
 Then after one step, in which each of the other nodes evaluates its
behavior using the threshold rule, nodes r and t will switch to A for each
of them, 2 /3 > 2/ 5 of their neighbors are now using A.
 Nodes s and u do not switch, on the other hand, because for each of
them, only 1 /3 < 2 /5 of their neighbors are using A.
 In the next step, however, nodes s and u each have 2/ 3 > 2 /5 of their
neighbors using A, and so they switch. The process now comes to an end,
with everyone in the network using A.
 Cascading Behavior
 Consider the social network in Figure
 Cascading Behavior
 Suppose that in the A-B coordination game we have a =3 and b =2,
leading to a threshold of q = 2 /5.
 If we start from nodes 7 and 8 as initial adopters, then in the next three
steps we will first see (respectively) nodes 5 and 10 switch to A, then
nodes 4 and 9, and then node 6. At this point, no further nodes will be
willing to switch.
 We call this chain reaction of switches to A a cascade of adoptions of A.
 We have to distinguish between two fundamental possibilities:
(i) that the cascade runs for a while but stops while there are still nodes
using B.
(ii) that there is a complete cascade, in which every node in the network
switches to A.
Cascading Behavior
 Cascades and Clusters
 When people are connected in networks to each
other then they can influence each other's behavior
and decisions. This is called Cascading Behavior in
Networks.
 A group of similar things or people positioned or
occurring closely together is called a Cluster.
 Cascades and Clusters
 The property of densely connected communities is that when you
belong to one, many of your friends also tend to belong.
 We can take this as the basis of a concrete definition, as
A cluster of density p is a set of nodes such that each node in the
set has at least a fraction p of its network neighbors in the set.
 The set of nodes a, b, c, d forms a cluster of density 2/ 3 in the
network .The sets e, f, g, h and i, j, k, l each form clusters of
density 2/ 3 as well.
 Each node in a cluster have a prescribed fraction of its friends
residing in the cluster as well, implying some level of internal
“cohesion.”
 On the other hand, it does not imply that any two particular nodes
in the same cluster necessarily have much in common.
Cascades and Clusters
 Cascades and Clusters
 For example, in any network, the set of all nodes is
always a cluster of density 1 because all your
network neighbors reside in the network. .
 If you have two clusters of density p, then the union
of these two clusters is also a cluster of density p.
 The Relationship Between Clusters
and Cascades
 A cascade comes to a stop when it runs into a dense
cluster, and, that this is the only thing that causes
cascades to stop.
 Clusters block the spread of cascades.
 Clusters are the natural obstacles to cascades.
 The Figure shows the cluster structure of a network
which tell us something about the success or failure
of a cascade.
 In this figure, we see two communities, each of
density 2 /3.
The Relationship Between Clusters
and Cascades
 The Relationship Between Clusters
and Cascades
 Consider a set of initial adopters of behavior A, with a
threshold of q for nodes in the remaining network to adopt
behavior A.
(i) Clusters Are Obstacles to Cascades
 If the remaining network contains a cluster of density
greater than 1 − q, then the set of initial adopters will not
cause a complete cascade.
(ii) Clusters Are the Only Obstacles to Cascades
 Moreover, whenever a set of initial adopters does not
cause a complete cascade with threshold q, the remaining
network must contain a cluster of density greater than 1 −
q.
 The Relationship Between Clusters
and Cascades
 In this figure, clusters of density greater than 1− 2/ 5
= 3 /5 block the spread of A at threshold 2 /5.
 Clusters Are Obstacles to Cascades
 The Relationship Between Clusters
and Cascades
 Clusters Are the Only Obstacles to Cascades
 Role of Weak Ties
 Weak Ties also focuses on holes in the network.
 Our acquaintances (weak ties) are less likely to be socially
involved with one another than our close friends ("strong
ties").
 Weak ties serve as a bridge between groups of people that
are closely knit with each other, but not with the other
groups.
 Through these bridges, information can be fed from one
group to another.
 It facilitates the flow of information from distant parts of a
network.
 It helps to integrate social systems.
Role of Weak Ties