An Empirical Study of Free-Riding Behavior in the Maze P2P File-Sharing System

Maze is a P2P file-sharing system with an active and large user base. It is developed, deployed and operated by an academic research team. As such, it offers ample opportunities to conduct experiments to under-stand user behavior. Embedded in Maze is a set of incentive policies designed to encourage sharing and contribution. This paper presents an in-depth analysis of the effectiveness of the incentive policies and how users react to them. We found that in general the policies have been effective. But they also encourage the more selfish users to cheat by whitewashing their ac-counts as a variation of Sybil attack. We examine multiple factors that may contribute to the free-riding behavior. Our conclusions are that upload speed, NAT and amount of shared files are not the problems, and selfish behavior is demonstrated more by shorter online time. Since free-riders are also avid consumers of popular files, we suggest a two-pronged approach to reduce free-riding further: mechanisms to direct queries to sources that would otherwise be free-riders, and policies to encourage users make their resources more available.