In the models of Young (1993a,b), boundedly rational individuals are recurrently matched to play a game, and they play myopic best replies to the recent history of play. It could therefore be an advantage to instead play a myopic best reply to the myopic best reply, something boundedly rational...