Nash Equilibrium, Minmax and Entropy
Prof. Sourav Bhattacharya
Economics Group
Indian Institute of Management Calcutta
In two-person zero-sum games, players impose maximum harm and (often) maximum surprise on each other. The first part follows from the familiar result that every Nash equilibrium gives each player her minmax value. The second part and its prerequisites can be made precise under a certain notion of ‘surprise’ – payoff-relevant entropy. We show that for a large class of non-zero-sum games, there is some Nash equilibrium that is endowed with these properties. We identify simple sufficient conditions on the payoff matrices that give rise to such equilibria. The existence of malevolent equilibria even in games of substantial common interest, such as stag hunt, is an antithesis to Adam Smith’s invisible hand. It also provides a surprising upper bound on the costs of strategic uncertainty and coordination failure.


















