Thought Toys · Strategy & computation · Exhibit 10

The evolution of trust

Two strangers can each either cooperate or cheat. Cheating always pays more in the moment — so why does trust exist at all? Fill a world with simple players — some always kind, some always rotten, some who just copy whatever you did last — and let the winners breed. When players meet only once, the cheats devour everyone. Let them meet again and again, and something better claws its way to the top.

What you're seeing

Every coloured band is one strategy's share of the population, stacked to fill the whole height. Each generation, every player meets every kind of player and plays the dilemma for the chosen number of rounds; whoever earns the most points has more offspring next generation, whoever earns least dies back. So the bands are evolution in fast-forward — the strategies that thrive widen, the ones that flop pinch shut.

Start with Rounds per meeting at 1 and press play. With only a single encounter there's no future to protect, so being nice is pure loss — the red Cheater band swallows everything. That's the bleak baseline: in a one-shot world, betrayal is simply the better move. Now slide rounds up to ten or more and reset. The picture flips. Cheaters still feast at first on the naive Cooperators, but once the easy victims are gone they're left facing players who remember — the green Copycat (cooperate first, then do whatever you just did) and the amber Grudger (kind until crossed, then never again). Against those, cheating gets punished right back, and the cheats starve out. Trust wins not by being soft, but by being repeated and answerable.

That's the whole surprising lesson: cooperation doesn't need anyone to be a saint. It just needs the same players to keep meeting, to remember, and to return like for like. Nudge the Mistakes slider and watch how a little noise — fingers slip, signals get crossed — shakes the outcome, since now even the well-meaning occasionally wrong each other.

They cooperateThey cheat
You cooperate+2 / +2−1 / +3
You cheat+3 / −10 / 0
The rule, exactly. The repeated prisoner's dilemma with the payoffs above (per round, your score first). Five strategies compete: Copycat (tit-for-tat: cooperate, then echo the opponent's last move), Cooperator (always cooperate), Cheater (always defect), Grudger (cooperate until the opponent ever defects, then defect forever), and Random. Each generation, a strategy's fitness is its average score against the current population mix; shares then update by the replicator rule (a strategy's share grows in proportion to how far above the pack its score sits). "Mistakes" flips a chosen move with the set probability. (Checked offline before shipping: the per-round match scores equal the matrix exactly; with 1 round the population collapses to all-Cheater, and with many rounds the Cheaters go extinct while Copycat and Grudger take over.) Counter-example, verified in node: cooperation is not always favoured — in a single-round game defection scores higher, so trust can never get started.

Also in Strategy & computation: Sorting algorithms →

All 23 in Strategy & computation
  1. 10The evolution of trust — you are here
  2. 23Sorting algorithms
  3. 24PageRank & the random surfer
  4. 25Huffman coding
  5. 26Dijkstra's shortest path
  6. 27Nash equilibria
  7. 33The learning-rate cliff
  8. 36A* pathfinding
  9. 37Braess's paradox
  10. 44Diffie–Hellman key exchange
  11. 45Preferential attachment
  12. 46Aliasing & the Nyquist limit
  13. 47The secretary problem
  14. 49Freeze too fast, stay stuck
  15. 51Cross one line, and its territory closes
  16. 56Catch one error, miss the next
  17. 57Why more processors stop helping
  18. 58Why a busy line explodes
  19. 59The set that's only sure when it says no
  20. 60The fit that memorizes instead of learns
  21. 61When the wire breaks, pick one
  22. 67Better at both, and still better off trading
  23. 69Everyone was consistent. The vote wasn't.

← Back to the cabinet  ·  Thought Toys — Exhibit 10.