Skip to content
Mental Models

Game Theory: Thinking in Moves

The Prisoner's Dilemma: When Rational Players Both Lose

Two players, each acting perfectly rationally, each choosing the move that's best for them — and both ending up worse off than if they'd cooperated. The most important game in the world, and the engine under arms races, price wars and the tragedy of the commons.

15 min Updated Jun 27, 2026

We now arrive at the most famous, most useful, and most quietly tragic result in all of game theory. It’s a game where every player does everything right — reasons clearly, follows their incentives, plays their provably best move — and because of that, they all end up worse off than a pair of cooperating fools would have. It is not a story about stupidity or malice. It’s a story about how individual rationality and collective good can pull in opposite directions, and once you’ve seen it you’ll notice it everywhere: in arms races, overfishing, price wars, doping, traffic, and climate talks. This is the prisoner’s dilemma.

The story

Two suspects — call them Al and Bo — are arrested for a serious crime and held in separate cells, unable to communicate. The prosecutor has enough evidence for a minor charge but needs a confession for the big one, so she offers each the same deal:

  • Stay silent (this is cooperating — with your partner) and the other talks: you take the fall, 10 years, while they walk free.
  • Confess (this is defecting — on your partner) and the other stays silent: you walk free, they get 10 years.
  • Both confess: you each get 5 years — the evidence convicts you both, but your cooperation with the prosecutor shaves a little off.
  • Both stay silent: she can only pin the minor charge on you, so you each get just 1 year.

Here’s the situation as a payoff matrix. To keep “higher is better” honest, we write a sentence of n years as a payoff of −n — so 0 (walk free) is the best outcome and −10 (take the whole fall) is the worst. Test the logic with the steppers.

Payoff matrix

The prisoner's dilemma (payoff = minus the years inside)

Each cell is (Al's payoff, Bo's payoff), where a payoff of −5 means a 5-year sentence. Higher is better: 0 means walking free. Compare each prisoner's two rows/columns one rival-move at a time.

AlBo
Al chooses a row; Bo chooses a column. Each cell lists the row payoff then the column payoff.
Bo
Stay silent (cooperate)Confess (defect)
AlStay silent (cooperate)-1-1-100
 Confess (defect)0-10NE-5-5

A ringed payoff is that player’s best response to the rival’s choice. A cell where both are ringed is a Nash equilibrium.

What the matrix says

Al — dominant strategy: Confess (defect)

Bo — dominant strategy: Confess (defect)

Nash equilibrium (pure): (Confess (defect), Confess (defect))

Confessing is a dominant strategy for both: whatever the partner does, confessing serves a shorter sentence. So both confess and land on (−5, −5) — even though both staying silent (−1, −1) would have been far better for the pair.

Why both rational players confess

Put yourself in Al’s cell and reason it through — exactly the column-by-column dominance check from lesson 2.

  • Suppose Bo stays silent. If I stay silent too, I get 1 year (−1). If I confess, I walk free (0). Confessing is better.
  • Suppose Bo confesses. If I stay silent, I take the full 10 years (−10). If I confess, I get 5 years (−5). Confessing is better.

Confessing wins in both cases — it’s a dominant strategy. So Al confesses. But the game is symmetric, so Bo runs the identical reasoning and confesses too. They both follow their flawless logic straight into (−5, −5): five years each.

Now the gut-punch. Look at the cell they abandoned: if both had stayed silent, it’s (−1, −1)one year each. Both prisoners would have strictly preferred mutual silence to mutual confession. They both knew it. And they still couldn’t get there, because from the silent cell each individually had an incentive to defect and walk free, and each knew the other did too. The rational move and the good outcome were different moves.

Warning:

The exact shape of the tragedy

The mutual-defection cell (−5, −5) is the Nash equilibrium: neither can improve by switching alone (silence while your partner confesses gets you 10 years). The mutual-cooperation cell (−1, −1) is better for both — but it is not an equilibrium, because from it each player can do even better by defecting. The dilemma is precisely this clash: the only stable outcome is the one both players wish they could escape. Individually rational, collectively self-defeating.

In the prisoner's dilemma, mutual cooperation (−1, −1) gives both players a better payoff than mutual defection (−5, −5). So why isn't (−1, −1) the Nash equilibrium?

This is not a story about prisoners

The prison tale is just a vivid wrapper. The structure — two parties, each with a dominant incentive to defect, whose mutual defection leaves them both worse than mutual cooperation would — is everywhere. Whenever you see it, you’re looking at a prisoner’s dilemma wearing a costume.

Situation”Cooperate""Defect”Why both defect → worse for both
Two firms pricingKeep prices highUndercut the rivalBoth undercut → a price war guts both margins
Arms raceDon’t build weaponsBuild weaponsBoth arm → vast cost, no change in relative power
OverfishingFish modestlyCatch all you canBoth overfish → the stock collapses for everyone
Doping in sportStay cleanTake the drugBoth dope → same rankings, ruined health
Climate actionCut emissionsKeep emittingAll emit → shared climate damage

In every row, defecting is individually tempting (undercut and grab share, arm and feel safe, catch more fish today), and in every row, mutual defection is a disaster both sides would have paid to avoid. That shared skeleton is why the prisoner’s dilemma is sometimes called the most important game in the social sciences.

Info:

The prisoner's dilemma is the engine of the tragedy of the commons

If you’ve met the tragedy of the commons — shared pasture overgrazed because each herder gains from one more cow while everyone shares the cost of the ruined field — you’ve met a many-player prisoner’s dilemma. “Graze modestly” is cooperate, “add another cow” is defect, defecting is individually rational, and universal defection destroys the common resource. The commons tragedy is a prisoner’s dilemma scaled from two players to many. Same engine, bigger stage.

Each situation has a 'cooperate' and a 'defect' move. Which ones have the prisoner's-dilemma structure (defecting is individually tempting, but mutual defection is worse for both than mutual cooperation)?

Place each item in the right group.

  • Two startups hoping to pick the same connector standard
  • Rival airlines deciding whether to start a fare war
  • Teammates each deciding whether to slack off on a shared-grade project
  • Two drivers deciding which side of the road to drive on, wanting to match
  • Fishers on a shared lake each choosing how hard to fish
  • Two nations each deciding whether to keep building missiles

”But surely they should just cooperate?”

Every newcomer protests this, so let’s take it seriously. Can’t they just agree to cooperate? Three honest answers:

  1. A promise isn’t binding. Even if Al and Bo swear mutual silence beforehand, once they’re in separate cells each still gains by breaking the promise. Cheap talk doesn’t change the payoffs, so it doesn’t change the equilibrium. (Change the payoffs — say, with a mafia that murders confessors — and you’ve changed the game, which is exactly how real cartels enforce cooperation.)
  2. It’s one-shot. As the matrix stands, the prisoners play once and never meet again, so there’s no future in which betrayal can be punished. This is the crucial crack we’ll pry open next lesson.
  3. The dilemma is real, not a failure of cleverness. Smarter players don’t escape it; they confess faster. That’s what makes it profound — and what makes the escape routes genuinely interesting.

Almost every real “prisoner’s dilemma” that humans actually solve gets solved by changing the game so the payoffs or the structure shift:

  • Repetition — if the same players meet again and again, today’s defection invites tomorrow’s revenge, and cooperation can become rational. (All of next lesson.)
  • Binding contracts & enforcement — an outside referee who punishes defection (courts, regulators, fishing quotas, arms-inspection treaties) rewrites the payoffs.
  • Reputation — when others watch and remember, a defector pays a long-run cost in lost future deals.
  • Changing who decides — merging the two firms, or pooling the fishery under one owner, turns “us vs them” into a single decision-maker with no one to defect against.

None of these deny the dilemma. They each alter the game until cooperation becomes the equilibrium. That’s the practical lesson: when you’re trapped in a dilemma, don’t moralise — re-engineer the payoffs.

Match each idea to its precise role in the dilemma.

Pick a term, then click its definition.

Capture the dilemma in one line.

Pick the right option for each blank, then check.

In a prisoner's dilemma, is a dominant strategy for each player, so the Nash equilibrium is mutual defection — even though mutual would leave both players strictly off.

Recap

Big picture

The prisoner's dilemma in one picture

  • Prisoner's Dilemma
    • The structure
      • Defect is dominant for both → equilibrium is mutual defection
    • The tragedy
      • Mutual cooperation is better for both, but not stable (each can gain by defecting)
    • In the wild
      • Arms races, price wars, overfishing, doping, climate, the commons
    • The escapes
      • Repetition, enforcement, reputation, merging — all *change the game*

When to use it

Reach for the prisoner’s dilemma whenever two or more parties each face a tempting move that, if everyone takes it, leaves all of them worse off — and ask the diagnostic questions: Is defecting dominant? Is mutual cooperation better but unstable? If yes, you’ve found the trap, and you now know moralising won’t fix it. The fix is structural: introduce repetition, an enforcer, reputation, or a way to merge the deciders. The most powerful of those — repetition — is so rich it gets the whole next lesson, where you’ll watch cooperation claw its way back to rationality in a famous computer tournament.

Mark lesson as complete