Session 8: Cascading behavior in

networks
IDS 564, Prof. Ali Tafti

1
2
Figure 19.2: Node v must choose between behavior A and behavior B, based on
what its neighbors are doing.

3
Figure 19.2: Node v must choose between behavior A and behavior B, based on
what its neighbors are doing.

A is a better choice if:

“We’ll use q to denote this expression on the right-hand side. This inequality describes
a very simple threshold rule: it says that if at least a q = b/(a+b) fraction of your
neighbors follow behavior A, then you should too.” (p. 567 EK) 4
5
6
7
 Complete cascade
• Consider a set of initial adopters who start with a new
behavior A, while every other node starts with behavior B.
Nodes then repeatedly evaluate the decision to switch from B
to A using a threshold of q.

• If the resulting cascade of adoptions of A eventually causes

every node to switch from B to A, then we say that the set of
initial adopters causes a complete cascade at threshold q.

8
 Cascading behavior and viral marketing
• Tightly-knit communities in the network can hinder the spread of
an innovation

• Two different norms A and B, separated by the boundary

between adjacent communities

• “Viral marketing”: Network positioning of the initial adopters

determines the spread of the innovation; as does the relative
payoff of a and b.
– in the previous example, if the marketer focuses on nodes 12 and 13,
then cascading to A starts up again.
– focusing on 11 or 14 would not be as fruitful.

9
Population level vs. network-level cascading

10
CASCADES AND CLUSTERS

11
12
 Cluster density
• We say that a cluster of density p is a set of nodes such
that each node in the set has at least a p fraction of its
network neighbors in the set.

• Does not imply that any two particular nodes in the same
cluster necessarily have much in common.
– In any network, the set of all nodes is always a cluster of
density 1
– If you have two clusters of density p, then the union of these
two clusters (i.e. the set of nodes that lie in at least one of
them) is also a cluster of density p.

13
Claim: 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) 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) 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.

• e.g. in Fig. 19.7, clusters of density greater than (1 – 2/5) = 3/5 block the
spread of A at threshold q = 2/5.

14
15
16
Bryce Ryan and Neal C. Gross. The diffusion of hybrid seed corn in two Iowa communities. Rural
Sociology, 8:15–24, 1943.
17
18
Suppose that w and x are the initial adopters of a technology spreading with
threshold ½. Then, everyone else in their tightly-knit six-node community will
adopt this tech, but u and v will not.

19
 Double-edge aspect of local bridges
• Powerful ways to convey awareness

• Weak at transmitting behaviors that are risky or

costly to adopt
– where you need to see a higher threshold of
neighbors doing it too.

• Explains why many social movements tend to

build support locally and relatively slowly.
20
Heterogenous thresholds

21
 Heterogenous thresholds

payoff from choosing A: pdav

payoff from choosing B: (1-p)dbv

threshold: p >= bv/(av + bv)

22
Watts and Dodds (2007): To
understand spread of behaviors,
we need to consider the
presence of people who are
easily influenced.
Duncan J. Watts and Peter S. Dodds. Networks, influence, and
public opinion formation. Journal of Consumer Research,
34(4):441–458, 2007.

23
Cascade capacity of an infinite network—in which each node has a finite
number of neighbors—is the largest value of the threshold q for which some
finite set of early adopters can cause a complete cascade.

If q ≤ 1/2 , then nodes u and v will switch to A, after which nodes w

and x will switch, and the switches will simply propagate all the
way down the path.

The cascade capacity of this infinite path is exactly ½.

28
Infinite grid has cascade capacity of 3/8.
If q ≤ 3/8 , then behavior A moves to the neighbors of shaded nodes.

If 3/8 ≤ q ≤ 1/2, then A fails to spread even though it is the better

technology.

29
 Limits on cascade capacity
• If threshold q > 1/2, then “interface” between adopters of A
and B decreases at every step. This means that diffusion will
stop at some point if q > 1/2 ; the spread will eventually stop.

30