← All Games

Evolution of Trust

Six strategies compete in repeated Prisoner's Dilemma matches. After each tournament, successful strategies reproduce proportionally. Watch which behaviors survive — and why.

Generation
0
Leading Strategy
Status
Setup
Starting Population — Presets
Strategy Populations (Total: 100)
Population — 100 agents

How does cooperation evolve?

In a one-shot game, betrayal is always the "rational" choice — it pays more regardless of what your partner does. But in repeated interactions, the calculus changes: future rounds punish past betrayal. Robert Axelrod's 1980 computer tournaments showed that Tit-for-Tat (Copycat) won against all submitted strategies by being nice (never defects first), retaliatory (punishes betrayal immediately), forgiving (resumes cooperation after retaliation), and clear (easy to read, no hidden games).


Payoffs used in this simulation (per round):

They Cooperate
They Defect
You Cooperate
+2 / +2
−1 / +3
You Defect
+3 / −1
0 / 0

Key insight: Always Defect thrives in a naive world full of cooperators, but collapses once retaliating strategies like Copycat and Grudger become common. Copycat performs well across diverse populations. Detective exploits nice strategies but is checked by retaliators. The simulation reflects real-world dynamics: the environment a strategy lives in determines its success.