Game theory

Tit-for-Tat

Why does copying your opponent's last move beat every clever strategy?

Tit-for-tat is a strategy for the repeated prisoner's dilemma that cooperates on the first move and thereafter copies whatever the opponent did on the previous move, rewarding cooperation and punishing defection immediately.

TL;DR

  • Tit-for-tat cooperates on the first round of a repeated prisoner's dilemma, then simply repeats whatever its opponent played in the previous round.
  • Robert Axelrod's 1980 computer tournaments showed tit-for-tat winning against far more complex strategies despite being one of the shortest programs submitted.
  • Its success rests on four traits: it is nice (never defects first), retaliatory (punishes defection at once), forgiving (returns to cooperation the moment the opponent does), and clear (easy for opponents to read and adapt to).
  • Cooperation becomes rational only when the game is repeated indefinitely, because the threat of future retaliation can outweigh the one-time gain from defecting.
  • Tit-for-tat is not an equilibrium of the one-shot game, but it makes mutual cooperation self-enforcing in the repeated game once the shadow of the future is long enough.

When to use it

  • Use tit-for-tat when you face the same counterpart repeatedly and want to encourage cooperation without leaving yourself open to exploitation.
  • Use it when your actions are observable, so the other side can clearly link your response to their own previous move.
  • Use it when the relationship is likely to continue indefinitely, because credible future retaliation is what deters present defection.

When it fails

  • Tit-for-tat fails when moves are noisy or misread, because a single accidental defection can trigger an endless echo of mutual retaliation between two tit-for-tat players.
  • It fails in one-shot or fixed-length games with a known end, where backward induction unravels cooperation and defection dominates every round.
  • It fails against opponents who never respond to punishment, since it keeps matching their defection instead of cutting its losses and moving on.

Worked example

In 1980 the political scientist Robert Axelrod invited game theorists to submit computer programs to play a two-hundred-round prisoner's dilemma against one another. The winner, submitted by Anatol Rapoport, was tit-for-tat: a four-line program that cooperated first and then echoed its opponent. It beat longer, cleverer entries because it never started a feud it could not end, and it collected steady rewards from the many opponents willing to cooperate back.

The same logic shows up in real relationships. During World War I, opposing units in the trenches settled into an informal live-and-let-live truce: each side held its fire so long as the other did, and answered any deliberate shelling with immediate retaliation. Neither side trusted the other, yet cooperation held because both knew they would face the same enemy across no-man's-land tomorrow, and a defection today would be punished then.

Tit-for-tat's weakness is noise. If one player's cooperation is misread as a defection, two tit-for-tat players can fall into an unbreakable cycle of mutual retaliation, each punishing the other for a punishment. This is why later tournaments favoured a generous variant that occasionally forgives a defection without responding, restoring cooperation after an accidental slip while still deterring deliberate exploitation.

Payoffs are points per round (higher is better); each cell reads as (Player A, Player B). Player A chooses the row, Player B chooses the column. These are Axelrod's temptation (5), reward (3), punishment (1), and sucker (0) payoffs, satisfying T > R > P > S and 2R > T + S so that steady cooperation beats taking turns exploiting each other.
Player B
Player ACooperateDefect
Cooperate3 / 30 / 5
Defect5 / 01 / 1
The one-shot Nash equilibrium is (Defect, Defect) at (1, 1), yet (Cooperate, Cooperate) at (3, 3) Pareto-dominates it; repeating the game lets tit-for-tat make the (3, 3) outcome self-enforcing through the threat of retaliation.

Related concepts

  • Prisoner's Dilemma — the one-shot game that tit-for-tat is played over and over
  • Nash Equilibrium — tit-for-tat can make mutual cooperation a Nash equilibrium of the repeated game
  • Dominant Strategy — defection is the dominant strategy in the one-shot game that repetition lets tit-for-tat overcome

FAQ

What is tit-for-tat in simple terms?

Tit-for-tat is a simple rule for repeated games: start by cooperating, then do whatever your opponent did last time. If they cooperated, you cooperate; if they cheated, you cheat back once, then forgive as soon as they return to cooperating. It rewards good behaviour and punishes betrayal immediately without holding a lasting grudge.

What is an example of tit-for-tat?

A classic example is a price war between two rival firms. If one holds prices high, the other matches and both profit; if one cuts prices to steal customers, the rival immediately cuts too, erasing the gain. Because they compete month after month, each firm learns that undercutting invites swift retaliation, so high-price cooperation tends to hold.

Tit-for-tat vs grim trigger, what's the difference?

Both punish defection, but grim trigger punishes forever while tit-for-tat forgives. Grim trigger cooperates until the opponent defects once, then defects for the rest of the game with no way back. Tit-for-tat retaliates for only a single round and returns to cooperation the moment the opponent does, making it far more robust to mistakes.

Why does tit-for-tat sustain cooperation?

Tit-for-tat sustains cooperation because it makes defection unprofitable over time. The one-time gain from cheating is wiped out by the opponent's retaliation on the next round, so as long as the game is likely to continue, the future cost outweighs the present temptation. Being nice, retaliatory, forgiving, and clear lets partners safely trust each other.

How do you know when tit-for-tat beats always defecting?

Tit-for-tat beats always-defecting when the game is repeated often enough. Formally, mutual cooperation holds when the discount factor or continuation probability exceeds the threshold (T−R)/(T−P), which weighs the temptation to defect against the value of ongoing cooperation. With Axelrod's payoffs of 5, 3, 1, and 0, that threshold is 0.5.