Research On Social Incentive Mechanisms Based On Iterated Prisoner's Dilemma

Cooperation is the eternal topic of human development,and also the key solution of many social dilemmas in modern society.Human society has always been troubled with the tragedy of the commons and prisoner's dilemma resulting from the conflicts between individual interests and collective interests.How to lead people out of difficulties,solve problems and promote cooperation is the fundamental goal of the design and application of social incentive mechanisms.That means we should design proper social incentive mechanisms to make individual's rational choice can also achieve collective rationality and obtain cooperation dividends.Two typical social incentive mechanisms has been proved the mechanisms based on peer pressure are more effective than those traditional methods.One is Pigouvian tax with peer-information mechanism,which means individuals with cooperation behaviors receive rewards and their opponents are noticed with this information,thus making defecting peers feel pressure to cooperate.The other is peer incentive mechanism,which means individuals receive rewards for their opponents' cooperation behaviors and peer pressure.Besides,cooperation promotion is also the topic of evolutionary game theory on complex networks.In this paper,we apply the framework of evolutionary game theory to model and analysis peer-pressure-based social incentive mechanisms.We proposal an iterated prisoner's dilemma with external incentives.The model can be constructed by different parameters of different social incentive mechanisms.Through the study of replicator dynamics in the well-mixed condition and the Fermi dynamics in the lattice network,we can find that the cooperator ratio is determined by the relationship between incentive compensation and the advantage of defectors to cooperators.In well-mixed condition,if the incentive compensation exceeds the advantage of defectors to cooperators,the final population is all cooperative;otherwise it is in cooperation/defection coexist state.The research on lattice network is basically the same as that in well-mixed condition.In addition,we construct a model of iterated prisoner's dilemma with adjustable external influence range.This model can not only construct different social incentive mechanisms with parameters combination,but further distinguish the globality of externalities and the influence domain between peers through the adjustment of influence range.By uniformed parameters,we find that under well-mixed condition,if the incentive compensation from the incentive mechanism exceeds the advantage of defectors to cooperators,the population is in cooperatordominated state;otherwise,the population is in cooperation/defection coexist state.Simulation on the lattice holds basically same conclusion and the difference may result from the resistance cluster made by same strategy holders.What's more,we compared Pigouvian tax with peerinformation and peer incentive on real on-line social networks.When the degree distribution follows the power law distribution,peer incentive performs better than Pigouvian tax with peerinformation in general,but the best performance is under Pigouvian tax with peer-information in some conditions.