Tit for Tat: The Simple Strategy That Beat Every Supercomputer Trick

An adaptation of Robert Sapolsky's 2024 lecture on the evolution of behavior. How a tournament of 200 experts crowned the simplest possible strategy as the best way to win at cooperation.

July 24, 20262 min read8 / 10

Once game theorists agreed that an uncertain future makes cooperation possible, one question remained: what is actually the best way to play? Two researchers decided to find out by testing everything at once.

A Tournament of 200 Strategies

In the early 1970s, the scientist W.D. Hamilton, the same one behind Hamilton's rule for kin selection, teamed up with political scientist Robert Axelrod. They invited around 200 people, economists, chess champions, war strategists, and more, to each submit their best strategy for playing the Prisoner's Dilemma repeatedly against an unknown number of future rounds.

Some submissions were extremely complex mathematical formulas. The strategy that won, by a wide margin, was the simplest one submitted. It was called tit for tat.

How Tit for Tat Works

The rule could not be simpler: cooperate on the very first round, and after that, just copy whatever the other player did to you in the previous round.

  • If they cooperated last time, you cooperate this time.
  • If they betrayed you last time, you betray them back once.
  • If they go back to cooperating, you go back to cooperating too.

Tit for tat won for four specific reasons, each one measurable in the tournament results.

  • It starts nice. Opening with cooperation gives every interaction a chance to go well from the start.
  • It punishes betrayal. A cheater gets an immediate, proportionate response, so cheating stops paying off quickly.
  • It forgives. As soon as the other player goes back to cooperating, tit for tat does too, instead of holding a permanent grudge.
  • It's simple and predictable. Other players can quickly learn exactly what to expect from it, which itself encourages cooperation.

None of the complicated strategies could beat a rule this basic, because complexity was never the advantage. Being clear, fair, and quick to forgive was.

This looked like the final answer, until researchers noticed a flaw hiding inside real-world relationships: what happens when one player thinks they were betrayed, but it was actually just a mistake?

Further Reading

Keep reading

The Wildebeest Myth That Fooled America

3 min read

Why You Share 98% of Your DNA With a Chimp, but Only Half With Your Mom

2 min read

I Would Die for Two Brothers or Eight Cousins

3 min read
PreviousThe Prisoner's Dilemma and the Shadow of the FutureNext When a Mistake Starts a War
Tit for Tat: The Simple Strategy That Beat Every Supercomputer Trick | Durgesh Rai