Incentives for cooperation in peer-to-peer systems

Peer-to-peer (P2P) systems have emerged as a widely deployed alternative to traditional client-server architectures for the distribution of information goods. The fundamental premise of P2P systems is that of voluntary resource sharing among individual peers. However, the inherent tension between individual rationality and collective welfare produces a misalignment of incentives which threatens to degrade the performance of P2P systems.; The design of protocols for distributed systems comprising of parties with diverse and selfish interests has spawned a great deal of recent research at the boundary between microeconomics, game theory and computer science. While traditional system design assumes obedient participants, the assumption of rational behavior brings new types of failures that cannot be addressed by traditional engineering tools, but incentive mechanisms such as penalties, reputation schemes and contracts.; In this dissertation we develop a framework for understanding the technical and economic characteristics of P2P systems and their implications for economics-informed design of P2P systems. Employing methods ranging from analysis and simulations to design and characterization, we study the incentives and disincentives that shape user behavior in P2P systems, and propose a diverse set of approaches that seek to alleviate the challenges inherent in incentive schemes relative to the unique characteristics of P2P systems.; In particular, we demonstrate the role that shared history plays in promoting cooperation in systems with large populations, few repeat transactions and highly dynamic memberships; we propose a maxflow-based algorithm for shared history to mitigate against collusive behavior; we study the implications of invisible actions and show that contracts can be designed to elicit cooperative behavior in equilibrium; we quantify the tradeoff between cooperating with and defecting on strangers in the presence of whitewashers, made possible by the availability of cheap pseudonyms (low cost identities), and show that cooperation may be sustained by adjusting one's behavior toward strangers based on their observed cooperativeness and/or by imposing a large identity cost; we find that cheap pseudonyms incur significant social loss only if the turnover rate is high and the societal generosity is intermediate; and, finally, we provide insights for quantifying performance-related disincentives in file-sharing systems and propose the acknowledgement prioritization technique as a means of eliminating existent disincentives.