In 1950, a 22-year-old mathematician named John Nash walked into his Princeton adviser’s office with a 28-page PhD thesis. It introduced a concept that would, within a generation, become the backbone of modern economics, political science, evolutionary biology, and the design of everything from spectrum auctions to kidney-exchange networks. Nash called it an “equilibrium point.” Everyone else would eventually call it the Nash equilibrium.
The deeper subject Nash was working in — game theory — was barely two decades old at the time. It had been codified in 1944 by John von Neumann and Oskar Morgenstern in a massive book titled Theory of Games and Economic Behavior, but mostly applied to a narrow class of problems. Nash’s thesis extended the framework to cover essentially every competitive situation involving rational actors. It is difficult to overstate what that extension made possible.
This article is a tour of the central ideas of game theory — what they mean, where they came from, and why a branch of mathematics that sounds like it’s about parlour games has turned out to be one of the most consequential intellectual exports of the twentieth century.
What a game actually is
In game theory, a “game” is any situation with the following structure:
- There are multiple players, each of whom can choose from a set of strategies.
- Each player has a payoff that depends not only on their own strategy but on what the other players choose.
- Each player is trying to maximise their own payoff, knowing the others are doing the same.
That’s it. The word “game” is unfortunate because it suggests poker or chess, but the framework applies to any situation of interdependent decision-making. Setting a price when competitors are setting theirs. Deciding whether to cooperate with a colleague. Bidding in an auction. Choosing whether to trust someone. Voting. War. All of these fit the template.
The central mathematical question is: what should a rational player do? The answer turns out to depend sensitively on the structure of the game.
The prisoner’s dilemma
The most famous game in the subject is the prisoner’s dilemma. Two suspects have been arrested and are being interrogated separately. The prosecutor offers each the same deal: if you confess and your partner doesn’t, you go free and they get 10 years. If you both confess, you both get 5 years. If neither of you confesses, you both get 1 year on a lesser charge.
Laid out as a payoff table (smaller numbers are better):
| B stays silent | B confesses | |
|---|---|---|
| A stays silent | A: 1, B: 1 | A: 10, B: 0 |
| A confesses | A: 0, B: 10 | A: 5, B: 5 |
What should A do? If B stays silent, A is better off confessing (0 years vs 1). If B confesses, A is also better off confessing (5 years vs 10). No matter what B does, confessing is strictly better for A. The same logic applies to B. So both confess — and both get 5 years, when they could have gotten 1 each by cooperating.
This is the disturbing lesson of the prisoner’s dilemma: rational self-interest can produce outcomes that are worse for everyone than an achievable alternative. It doesn’t matter how intelligent the players are. The logic of the game drives them to a bad equilibrium.
The prisoner’s dilemma is not a trick. It is a faithful model of real situations: arms races (both sides should disarm; neither dares), pollution (no individual firm should invest in clean technology if competitors don’t), overfishing (no fisherman should catch fewer fish voluntarily). Recognizing the structure is often the first step to designing institutions that can change the payoffs — that’s what regulation, treaties, and enforcement mechanisms are for.
Nash equilibrium
The prisoner’s dilemma has a unique equilibrium: both confess. Every player is doing the best they can given what the others are doing. Neither player regrets their choice after seeing what the other chose. That’s the Nash equilibrium condition.
Formally: a strategy profile is a Nash equilibrium if no player can improve their payoff by unilaterally changing their strategy, assuming the others keep theirs fixed.
Nash’s 1950 theorem is that in any finite game with finitely many players and strategies, at least one Nash equilibrium exists — possibly involving randomisation (mixed strategies). This was a purely mathematical result, proved using Kakutani’s fixed-point theorem from topology. It looked, at the time, like a technical contribution to a narrow subject.
Its consequence, however, was that every strategic interaction has some rational prediction. If you can model a situation as a finite game, the Nash equilibrium tells you where rational play has to end up (or at least, where it can’t leave without someone deviating profitably). That made game theory a general-purpose tool rather than a bag of tricks for specific puzzles.
Multiple equilibria are possible. In a game of “drive on the left or right,” both “everyone drives left” and “everyone drives right” are equilibria; no single driver benefits from switching unilaterally. Which equilibrium actually obtains depends on history, coordination, and social convention — not on the mathematics. Game theory tells you the space of stable outcomes; it doesn’t always tell you which one will occur.
Mixed strategies and bluffing
Some games have no equilibrium in pure strategies. Consider rock-paper-scissors. Whatever you pick, your opponent has a response that beats it, so they’d switch. Whatever they pick, you’d switch too. There’s no stable pure strategy for either player.
The resolution is mixed strategies: you play each option with some probability. In rock-paper-scissors, the unique Nash equilibrium is to play each move with probability 1/3. At this point, neither player can improve by switching, because the opponent’s random play makes every pure strategy equally good.
This is why poker players and military strategists actually randomize their behaviour. Being predictable is a losing strategy in any game where the opponent is adapting. A mixed strategy is not a failure to commit; it is the optimal response to strategic interdependence.
Nash’s theorem says such equilibria always exist if you’re allowed to randomize. That sweeps up the entire class of zero-sum games that von Neumann had studied, plus enormously more.
Where game theory gets applied
The reach of game theory goes well beyond textbooks. A few concrete examples.
Auction design. When the US government auctioned off spectrum for cellular networks in the 1990s, they hired game theorists (notably Paul Milgrom and Robert Wilson) to design the auction format. The wrong format would have let bidders collude or would have left billions on the table. The right format — simultaneous multi-round ascending-bid auctions with specific rules to prevent signalling — raised tens of billions and has been copied worldwide. Milgrom and Wilson won the 2020 Nobel Prize in Economics for this body of work.
Market design. Kidney-exchange networks, which match patient–donor pairs across the country to enable chains of compatible transplants, are designed using matching theory — a branch of game theory pioneered by Lloyd Shapley and Alvin Roth, who also won a Nobel Prize. Centralized school-choice systems in New York, Boston, and elsewhere use similar mechanisms.
Evolutionary biology. John Maynard Smith and George Price adapted game theory to evolution in the 1970s, introducing the concept of an evolutionarily stable strategy (ESS). This is a strategy that, once adopted by most of a population, cannot be invaded by a rare alternative. It has become the standard framework for studying why animals behave the way they do — why males of some species fight and others display, why some flowers self-pollinate and others don’t.
Computer science. Algorithmic game theory, a younger field, studies computation in strategic settings. How should ride-sharing platforms price rides so that drivers and riders cooperate? How should search engines run ad auctions so that advertisers bid truthfully? These are active research areas with direct commercial implications.
International relations. The stability of mutually assured destruction during the Cold War was a game-theoretic argument — Thomas Schelling, who later won the Nobel Prize, was a pioneer. So was the design of arms-control treaties, where credible commitment and verification mechanisms matter more than declared intentions.
The limits
Game theory is not a universal predictor of human behaviour. It assumes players are rational, that they know the structure of the game, and that they know other players are rational too — assumptions that are often approximately true and sometimes very wrong.
Behavioural game theory, developed since the 1980s by economists like Colin Camerer and Ernst Fehr, studies how actual human play deviates from Nash predictions. People punish unfair behaviour at cost to themselves. They cooperate in one-shot prisoner’s dilemmas more often than strict rationality allows. They have limited ability to reason many steps ahead.
These deviations aren’t failures of game theory so much as corrections to it. The mathematical framework still provides the baseline; behavioural research describes how human cognition systematically departs from that baseline. Together they produce a more accurate picture of how strategic interactions actually unfold.
What the subject teaches
The most useful thing game theory gives anyone who studies it is a habit of mind. When you see a contested situation, ask: what are the strategies? What are the payoffs? Who is choosing what, given what they expect others to choose? What equilibrium does the incentive structure drive toward?
That line of questioning doesn’t always produce quantitative answers. But it reliably produces better analyses than the alternative, which is to treat decisions as if they happen in isolation.
The mathematical apparatus — payoff matrices, equilibrium concepts, evolutionary dynamics — is the formal tool. The underlying idea is simpler: that most interesting situations involve interdependence, and that the right way to think about interdependence is to take seriously that everyone else is also thinking about it.
John Nash died in 2015, in a traffic accident, returning from Oslo where he had just received the Abel Prize. The 28-page thesis he wrote at 22 had, by then, influenced every social science and reshaped significant parts of mathematics, biology, and computer science. Game theory is, in many ways, the twentieth century’s most surprising intellectual export: a framework from pure mathematics that turned out to describe how almost every competitive system in the world actually works.
Frequently asked
Is game theory actually used in practice?
Extensively. Auction design (Google Ads, spectrum auctions), market design (kidney exchange, school choice), antitrust analysis, diplomatic negotiation, and evolutionary biology all use game-theoretic tools. The 2020 Nobel Prize in Economics went to Milgrom and Wilson for auction theory, which is essentially applied game theory.
Did John Nash prove something most people can understand?
The statement, yes: every finite game with a finite number of players has at least one equilibrium in mixed strategies. The proof is harder — it uses a fixed-point theorem from topology. The statement is taught in undergraduate economics; the proof is graduate-level mathematics.