Evolving networks for social optima in the "weakest link game"

Abstract

Previous models have applied evolving networks based on node-level "copy and rewire" rules to simple two player games (e.g. the Prisoner's Dilemma). It was found that such models tended to evolve toward socially optimal behavior. Here we apply a similar technique to a more tricky co-ordination game (the weakest link game) requiring interactions from several players (nodes) that may play several strategies. We define a variant of the game with several equilibria-each offering increasing social benefit. We found that the evolving network functions to select and spread more optimal equilibria while resisting invasion by lower ones. Hence the network acts as a kind of "social ratchet" selecting for increasing social benefit. Such networks have applications in peer-to-peer computing and may have implications for understanding social systems. © Springer Science+Business Media, LLC 2008.