Nash equilibrium: the idea that no one wants to move

A Nash equilibrium is a situation in which no player can do better by changing their own choice alone. Each person is already doing the best they can, given what everyone else is doing. The arrangement need not be fair, efficient, or pleasant. It is simply stable: if any single player departs from it unilaterally, that player ends up worse off. Stability, not desirability, is what the concept describes.

The idea is due to John Nash, who set it down in 1950 in a paper that ran barely a page in the Proceedings of the National Academy of Sciences. Its reach was far larger than its length. The theory of games that John von Neumann and Oskar Morgenstern had assembled in 1944 offered a complete solution only for two-person, zero-sum conflicts, where one player's gain is exactly the other's loss. Most economic and social life does not take that form. There are many participants, their interests are partly aligned and partly opposed, and they can all gain or all lose together. Nash's contribution was to define equilibrium for any finite game, whatever the number of players or the structure of payoffs, and to prove that at least one such equilibrium always exists, provided players are allowed to mix — to choose among their options at random with fixed probabilities. The existence proof rested on a fixed-point theorem; its consequence was that the concept is never empty. Every finite game has at least one resting point.

The economist Roger Myerson later judged this one of the great intellectual advances of the twentieth century, and compared its effect on economics to the discovery of the double helix in biology. The comparison is large, and it is deserved. Before Nash, strategic situations were a collection of special cases. After him, they shared a common language.

The machinery is worth stating precisely, because most confusion comes from skipping it. Each player has a set of available strategies. A best response is the strategy that yields the highest payoff once the others' choices are held fixed. A Nash equilibrium is a combination of strategies in which every player is at the same time playing a best response to all the others. No one, acting alone, can improve their position. That is the whole of the definition. Everything else follows from it.

Two consequences are routinely missed. The first is that equilibrium carries no promise that the outcome is good. The prisoner's dilemma is the standard illustration: two suspects each reason their way to informing on the other, and both end up worse than if they had stayed silent. That outcome is the game's only equilibrium, and it is the bad one. Mutual betrayal is stable precisely because, given that the other will betray, betraying is the best available response. Stability and desirability are different properties, and the equilibrium concept guarantees only the former. A good deal of economic and political dysfunction has exactly this shape: an arrangement that no individual can escape alone, even though all would prefer a different one.

The second consequence is that a game can have more than one equilibrium. In the coordination problem economists call the Battle of the Sexes, two people would rather be together than apart but disagree about where to go; there are two equilibria, and the payoffs by themselves cannot say which will hold. This is not a defect in the theory to be corrected. It is an accurate reflection of how coordination works. When several stable arrangements exist, something outside the payoffs — habit, precedent, a salient option that draws the eye — settles which one holds. Driving on the left and driving on the right are both equilibria; which one a country adopts is a matter of history, not of mathematics. The same dynamic governs competition between technologies: once enough buyers coordinate on one option, it can lock in and persist even when a better alternative exists, held in place by history and network position rather than merit.

Several distinctions keep the idea honest. A Nash equilibrium is not the same as a dominant strategy, a move that is best whatever anyone else does. Every dominant-strategy equilibrium is a Nash equilibrium, but the reverse is false, and most interesting games contain no dominant strategy at all. Equilibrium is also not a prediction. It is a consistency condition: a description of what could persist if it arose, not a forecast of what people will do. When Merrill Flood and Melvin Dresher first ran the prisoner's dilemma as an experiment at the RAND Corporation in 1950 — Nash's adviser Albert Tucker supplied the story of the two prisoners afterwards — they found their players cooperating far more often than the equilibrium implied. The distance between the stable point and the observed behaviour is not an embarrassment for the theory. It is the starting point of behavioural game theory, which asks why real people depart from equilibrium and how they learn their way toward it or away from it.

The reach of the concept comes from how little it assumes. Wherever two or more parties must each act on a guess about what the others will do, an equilibrium analysis applies: firms setting prices, states honouring or breaking a treaty, drivers choosing routes that then become congested, bidders shading their offers in an auction. Because each outcome depends on how rivals respond, organisations increasingly rehearse these interactions in simulation, proposing a move and letting a credible opponent counter it before any commitment is made in the market. Governments have used the same machinery deliberately, designing the rules of spectrum auctions so that the self-interested behaviour of bidders produces an efficient allocation rather than a collusive one. The equilibrium does not tell any of these actors what is right or fair. It tells them where the situation can come to rest, and therefore what would have to change for it to rest somewhere better.

That last point is the practical one. An equilibrium is a description of stability, and stability is what makes bad outcomes durable. To improve an outcome, wishing for a better one is not enough, because the better arrangement is not stable on its own terms. What has to change is the game itself: the payoffs, the information available, the order of moves, the capacity to make a binding commitment. Alter the structure so that the preferable arrangement becomes the one no player wants to abandon, and it will hold of its own accord. This is why the most consequential strategic work is so often done before the game begins: in the design of the rules, not in the playing of the hand.