Results in the Emergence of Social Equilibria: Some Computational Experiments

摘要

Disequilibrium states and non-optimal choices are common facts of social behavior that contradict the conclusions of classical economic theory. This encourages research on models in which outcomes exhibit those properties. In these models -in which uncertainty is a fundamental feature- the social behavior is an emergent property of the interactions among components of the system. In this work we present a framework for the simulation of the emergence of social behavior and the analysis of the mathematical properties of a set of agents engaged in interactions mediated by a protocol that regulates the interactions and the allocation of resources. The formal-ization allows to predict the existence of an equilibrium and the simulation confirms this for a particular but paradigmatic case. The simulation is based on an implementation of the iterated Prisoner's Dilemma. Prior experimentation (for example, Axelrod's tournaments) showed that tit-for-tat is the most adequate solution of this models. Our framework, based on certain realistic assumption, shows instead that tit-for-tat is unstable: a population of tit-for-tatters mimics a population of irrational agents if it becomes infected by even a low percentage of them.