A case for holistic incentive design

Introduction As distributed systems that span multiple administrative domains proliferate, robust protocols increasingly need to incorporate the incentives of multiple stakeholders into their design. A significant challenge in designing incentive aware systems is layering: where are incentives required? who are the principals? Peer-to-peer systems (P2P) push these challenges to the limit. Users are transitory, act independently, and strategize to maximize unknown utility functions. Moreover, users themselves are not the only players; ISPs and IT administrators impose limits on bandwidth hungry services. Content providers actively police P2P networks for copyrighted material, often introducing polluted files that merit system-level countermeasures [6]. In this paper, we argue for a holistic approach to incentive design in a P2P environment. P2P systems offer a rich base for experimentation regarding incentives. Their users act as a set of “fruit flies” for incentives research; their willingness to jump from service to service has resulted in the evolution of a diverse set of P2P protocols, implementations, and research insights. In this context, a holistic incentive design is one which is both complete and self-contained. By complete, we refer to addressing the challenges of operating on end hosts, e.g. rapid churn, potential for Sybil attacks, etc. By self-contained, we mean that all primitives required by the system are provided by the system, e.g. no out-of-band authentication mechanisms. In the remainder of this paper, we expand on the notion of holistic incentive design, exploring a common set of issues that future P2P incentive strategies should take into account. We argue that existing work on incentive design has been crucial in bringing the community to a turning point; lessons learned can now be synthesized into a common set of holistic design criteria for incentive aware distributed systems.