Inter-Market Competition of Reputation Infrastructure: An Agent Based Approach
How do you evaluate the relative merits of reputation systems? In this paper from the winter of 2011, I take the approach of letting little computer programs decide which e-commerce market they would prefer.
E-commerce markets are prone to cheating between participants because of the anonymity of Internet interactions and the lack of governmental jurisdictional authority to prosecute cheaters. For this reason, there has been much research and development into mechanisms that prevent cheating. Sets of mechanisms that seek to reduce or eliminate cheating in online markets are known as reputation systems or reputation infrastructure. Previous work on reputation systems has considered markets as existing alone. In reality, however, markets exist alongside other markets, so this paper builds on previous work to examine the dynamics of a world in which there are multiple markets differentiated by reputation infrastructure. In the model presented in this paper, agents with varying strategies are matched to play Prisoner’s Dilemma, and then can switch to a different market if they choose. Experiments are performed to examine general behavioral dynamics and the effects of varying the model parameters.