Date of Original Version
Abstract or Description
Peer-to-peer networks have emerged as a popular alternative to traditional client-server architectures for the distribution of information goods. Recent academic studies have observed high levels of free-riding in various peer-to-peer networks, leading some to predict the imminent collapse of communities hosted on these networks.
Our research develops static and dynamic analytic models to analyze the behavior of peer-topeer networks in the presence of free-riding. In contrast to previous predictions we find that freeriding is sustainable in equilibrium and in some cases occurs as part of the socially optimal outcome. We also show that network performance can be improved by grouping peers into communities that provide similar value and by providing external incentives to peers to share. With regard to the latter finding, we show that without external incentives, the level of free-riding in peer-to-peer networks will be higher than the socially optimal level, but that quality of service tied to the contribution of content can be used as a lever to induce users to share and thereby achieve the socially optimal network outcome. Our results are robust to consideration of a dynamic game where content is duplicated in the network over time