Social networks and gossip Jeroen Bruggeman 1. Cooperation? 2.

Social networks and gossip

Jeroen Bruggeman

1

Cooperation?

2

Cooperation• Reputation: infomation [knowledge] that

one individual has about another

• Reputation: through observation and for humans - with language - also through gossip

• Gossip requires network

• Reciprocity as special kind of reputation-based cooperation

3

Reputations• Different reputations for different kinds

of actions and skills

• Let’s distinguish reputations for cooperation - as in game theory - from prestige for remainder actions and skills (gossip for both)

• Meta-reputations for contributing to transmission of reputations

4

Gossip in theory• Who are trustworthy candidates to

cooperate with? Whom to avoid?

• Gossip as (evaluative) statements about non-present persons

• Gossip-based reputations can foster cooperation, even for public goods (Panchanathan & Boyd 2004)

• Boundary condition: actions observable

5

Gossip in actuality• Emotional aspects: irresistable to hear,

gratifying to do• (Tacit) references to norms• Context dependent who can say what

about whom• Diffusion in network, but locally sticky

around gossipee• Conformist bias: (nearly) consensus

within groups; not necessarily across6

Gossip in actuality

• Negative gossip does not necessarily create trust, but requires it: risk of retaliation when gossiper is exposed

• Signaling group loyalty is self-serving without risk of sanctions: speak positively about group members who live up to group norms and negatively about norm violators

7

Gossip in actuality• Combined with self-presentation • Not always honest: strategic actions to

make oneself look favorable compared with others (social comparison theory)

• Meta-gossip about reliability and goals gossiper, e.g. excessive or unreliable gossip mongers are less trusted

• Tactics: “the art of gossiping while not appearing to” N.Besnier

8

Can network topology reduce noise (error & manipulation) and increase trust?

Appropriate definition of social cohesion to distinguish ‘better’ parts from ‘worse’ parts 9

Social cohesion as k-connectivity(Douglas White, Frank Harary 2001)

• (sub)group bonding as strong as minimal number of (sub)group members, k, who hold group together

• redundancy of information channels helps to reduce noise: minimal number of independent paths, m, that connect any pair of members

• Theorem (Menger 1927): m = k10

To be done: public goods experiment with imputed noise

3-connected 1-connected, with same size, density, degree distribution, degree centralization, path distance

Gossip tuples(X,Y,) 11

Complications

• Through meta-reputations, people will trust some info-sources more than others - no adoption of average gossip but weighted average

• In large groups, k-connectivity expected to have non-monotonic effect on cooperation.

• Acutal networks change, probably also with non-monotonic effect

12