Selfish behavior can often lead to suboptimal outcome for all participants, a phenomenon illustrated by classical examples in game theory, such as the prisoner's dilemma. Over the last decade we have developed a good understanding of how to quantify the impact of strategic user behavior on overall performance in some concrete games. In this talk, we will consider online auctions from this perspective. A key property of this environments is that players typically participate in multiple auctions, have valuations that are complex functions of multiple outcomes, and are using learning strategies to deal with an uncertain environment. In this talk we show how to provide robust guarantees for the performance of many simple auctions even in such complex environments.
Theory Lunch
Composable Mechanisms, Learning, and Price of Anarchy in Auctions
January 14, 2015