Learnability of a class of Non-atomic Games arising on the Internet

In this paper we study a large class of non-atomic games arising from interactions on the Internet, such as many users sharing a network link, researchers accessing a database or web server, subscribers to a network services provider trying to gain access to the modem pool and many more. We provide simple and easily computable conditions under which these games are learnable by several models of learning, such as adaptive and sophisticated learning (Milgrom and Roberts 1991), calibrated learning (Foster and Vohra 1996), and reasonable learning (Friedman and Shenker 1997). Thus, using these methods one can evaluate the stability of such games on the Internet or in other settings in which players must learn over time.