Mediators

July 17, 2007
Moshe Tennenholtz | Technion

Mediators are reliable parties that can act on behalf of agents that give them the right of play. Unllike mechanism designers, mediators can not design new games, but are designed to lead to desired behaviors in given games. We introduce a theory of action mediators and show that they significantly enrich the set of situations where we can obtain stability against correlated deviations by coalitions. Moreover, we introduce the study of routing mediators, in which the above mediators may possess information also about the behavior of agents who do not give the mediator the right of play. We study the relationships between the power of different routing mediators in establishing correlated strong equilibrium. We show a natural class of routing mediators that allow to implement fair efficient outcomes as a correlated super-strong equilibrium in a wide class of games. Finally, we introduce the use of mediators in pre-Bayesian games, and apply it in the context of position auctions. If time permits, we will also discuss the connections between mediators and the study of program equilibrium.

Based on joint work with Itai Ashlagi, Dov Monderer, and Ola Rozenfeld.

Speaker Details

Moshe Tennenholtz is a professor with the faculty of Industrial Engineering and Management at the Technion, where he holds the Sondheimer Technion Academic Chair. During 1999-2002 he has been a visiting professor at Stanford CS department, where he has also been a research associate during the years 1991-1993. Moshe received his B.Sc. in Mathematics from Tel-Aviv University (1986), and his M.Sc. and Ph.D. (1987, 1991) from the Department of Applied Mathematics and Computer Science in the Weizmann Institute. His area of research lies in the interface between Artificial Intelligence and Game Theory. Among his contributions, in joint work with colleagues and students, he introduced the theories of artificial social systems, co-learning, distributed games, and non-cooperative computing, the axiomatic approach to ranking systems, as well as the study of program equilibrium and learning equilibrium.Until recently Moshe served as the editor-in-chief of the Journal of Artificial Intelligence Research (JAIR). He is also an associate editor of Games and Economic Behavior (GEB), the international journal of autonomous agents and multi-agent systems, and an editorial board member of the AI magazine, and of the Journal of Machine Learning Research (JMLR).