Robust Mechanisms
Joint Friday Theory Seminar and RISCS Center Talk
Speaker: Jing Chen, MIT
Friday, May 7, 3:00 PM, Room 135
Abstract: Mechanism design aims at leveraging the players’ knowledge and rationality, so as to make it in their interest to obtain the outcomes the mechanism designer wants. Originally developed for economic applications, mechanism design may hold the key to build decentralized and complex system, that are held together by strong individual incentives. Currently, however, mechanism design is very fragile, and cannot deliver engineering-grade solutions, due to classical problems (such as equilibrium selection) and less classical ones (such as collusion, complexity, and privacy). We thus are working to make mechanism design robust and ready for engineering applications of today.
In this talk, we take a first step in our program by proving that total robustness can be achieved in the classical setting where the players have complete information about each other. Namely, we prove that extremely robust mechanisms exist for the classical problem of maximizing revenue (in a general setting of “quasi-linear utilities”).
Joint work with Avinatan Hassidim and Silvio Micali.