1. It’s a form of political theory that harkens back to classical authors from Hobbes to Rousseau (with echoes of Plato’s Crito and other ancient works). That is, it makes certain assumptions about the preferences and goals of “players”–usually individuals or states–and then asks what must happen when they interact. This is the same method that led Hobbes to believe that individuals, motivated by the goal of minimizing pain, would kill one another absent a state. Hobbes’ conclusions were rejected by other theorists, but his method remains alive in modern game theory. There is a rival tradition of political theory that treats people as deeply embedded in cultural contexts. For Hegel, Nietzsche, Dewey, Foucault, Habermas, and others, the important question is how and why culture has changed, not how individuals will act under specified theoretical conditions. Some results of game theory seem to generalize across all existing cultures–which wouldn’t have surprised Hobbes or Locke.
2. Since game theory starts with players and models their interaction, it can handle markets, wars, and votes equally well. Schelling’s work is typical in that it doesn’t fit within the borders of his own field (economics), but could equally belong to political science or–in the case of his famous model of racial “tipping points”–sociology. There is something impressive about a theory that explains human behavior without arbitrary limits.
3. Some people assume that the “players” in game theory are selfish. That is not true. A game-theoretical model can work very well to explain behavior driven by any motives. Usually, altruism makes human interactions turn out better, and then games are uninteresting–but not always. Consider, for example, the bad outcomes that can result when X and Y are picking a restaurant, and X only wants to eat at Y’s favorite place, and Y only wants to go where X wants to go. They may withhold information about their own preferences, causing a big mess, even though their motives are selfless.
4. If game theory has a limitation, it is not an assumption of selfishness but rather a presumption that the players have preferences and identities prior to interacting. For instance, if the players are the USA and USSR (as in Schelling’s classic work), then their identities are those of the two nations and their goals are assumed to be security, or domination, or whatever. However, a person’s identity as a representative of the USA or the USSR is not just given; it is forged as a consequence of social and historical change, and it can fall apart. Soviet officials were supposed to bear the identity of “international Communists”; they really identified with the USSR or narrowly with their individual security interests; and then suddenly around 1990 most began to see themselves as Russians or even Europeans, but not as Soviets. This was a massive political change.
Even given players with fixed identities, it is not obvious that they will want any particular goals (such as security, pleasure, dominance, honor or salvation). We may start wanting one thing and persuade ourselves to value something different. It’s not clear that these processes of identity-formation and preference-setting can themselves be modeled as games. When we deliberate about who we are and what we want, the reasoning is not strategic in the same way. However, this is not a criticism of game-theory, simply an argument that it belongs in a broader context.