The price of anarchy in social dilemmas: Traditional research paradigms and new network applications

Abstract

Research on social dilemmas has largely been concerned with whether, and under what conditions, selfish decisions by autonomous individuals jointly result in socially inefficient outcomes. By contrast, considerably less emphasis has been placed on the extent of the inefficiency in those outcomes relative to the social optimum, and how the extent of inefficiency in theory compares with what is observed in experiments or practice. In this expository article, we introduce and subsequently extend the price of anarchy (PoA), an index that originated in studies on communication in computer science, and illustrate how it can be used to characterize the extent of inefficiency in social dilemmas. A second purpose of our article is to introduce a class of social dilemmas that occur when individuals selfishly choose routes in networks, and illustrate how the concept of PoA can be helpful in studying them.