P1: qVa Trim: 6.125in×9.25in Top: 0.5in Gutter: 0.75in
CUUS2079-07 CUUS2079-Zafarani 978 1 107 01885 3 January 13, 2014 17:17

186 Information Diffusion in Social Media

R

RB

R B R B

RB

{R}

{R,R}

{R,R,R} {R,R,R} {B,B,B}

Herding

{B,B,B}

{B,B}

{R,R}

or

{R,B}

{B,B}

or

{B,R}

{B}

B

{}

Figure 7.3. Urn Experiment. Rectangles represent student predictions written on the

blackboard, and edge values represent what the students observe. Rectangles are filled

with the most likely majority, computed from conditional probabilities.

7.1.2 Intervention

As herding converges to a consensus over time, it is interesting how one

can intervene with this process. In general, intervention is possible by pro-

viding private information to individuals that was not previously available.

Consider an urn experiment where individuals decide on majority red over

time. Either (1) a private message to individuals informing them that the

urn is majority blue or (2) writing the observations next to predictions on

the board stops the herding and changes decisions.

7.2 Information Cascades

In social media, individuals commonly repost content posted by others

in the network. This content is often received via immediate neighbors

(friends). Aninformation cascadeoccurs as information propagates through

friends.

Formally, an information cascade is defined as a piece of information or

decision being cascaded among a set of individuals, where (1) individuals