Learning and Stability in Big Uncertain Games
A big game is repeatedly played by a large population of anonymous players. The game fundamentals and the set of players change with time and there is incomplete and imperfect information about the play. This paper argues that simple big games, in which the fundamentals are uncertain but fixed, admit natural myopic Markov perfect equilibria with limited proportions of unpredictable and potentially chaotic periods. These special properties make these equilibria extendable to more general classes of big games. Examples include social adaptation of technology and price stability in production games.
I will also review earlier results on learning and stability in long and in large games that serve as foundation for the current research.
Joint work with Eran Shmaya