6 December 2019, Computational Social Choice Seminar, Mehmet Ismail

Abstract

Applications of the maximin criterion extend beyond economics to statistics, computer science, politics, and operations research. However, the maximin criterion—be it von Neumann's, Wald's, or Rawls'—draws fierce criticism due to its extremely pessimistic stance. I propose a novel concept, dubbed the optimin criterion, which is based on (Pareto) optimizing the worst-case payoffs of tacit agreements. The optimin criterion generalizes and unifies results in various fields: It not only coincides with (i) Wald's statistical decision-making criterion when Nature is antagonistic, but also generalizes (ii) Nash equilibrium in n-person constant-sum games, (iii) the core in cooperative games, (iv) stable matchings in matching models, and (v) competitive equilibrium in the Arrow-Debreu economy. Moreover, every Nash equilibrium satisfies the optimin criterion in an auxiliary game. Finally, the optimin criterion is the first parameter-free concept that can selectively explain the puzzle of cooperation in games, including the repeated prisoner's dilemma, the traveler's dilemma, the centipede game, and the repeated public goods game.

For more information on the Computational Social Choice Seminar, please consult https://staff.science.uva.nl/u.endriss/seminar/.