Why cooperation is hard

Two people would each do better if they cooperated than if they did not, and each does better individually by defecting whatever the other chooses. That is the whole trap, it is stated in four numbers, and versions of it run through evolution, ecology and economics without anyone needing to understand it.

The four numbers

Each of two players independently chooses to cooperate or defect, and the payoffs look like this.

What changes when it happens again

In one round, defection wins and there is nothing to discuss. The interesting version is when the same two players meet repeatedly and each can remember what the other did, because now today's choice has consequences tomorrow.

Robert Axelrod ran a famous tournament in 1980, inviting strategies from game theorists and running them against each other. The winner was the simplest entry submitted: cooperate on the first move, then copy whatever the opponent did last time. Tit for tat, four lines of description, beat every elaborate strategy in the field.

Axelrod's account of why it won generalises well beyond the tournament. It was nice, in that it never defected first. It was retaliatory, so exploiting it was not free. It was forgiving, returning to cooperation the moment the other side did, which stopped a single mistake becoming a permanent feud. And it was clear, so opponents could work out how it behaved and adapt to it.

The catch, discovered later, is that tit for tat is fragile to noise. If a cooperative move is misread as a defection, two tit-for-tat players fall into alternating retaliation and never recover. Strategies that occasionally forgive an apparent defection do better in any world where signals are imperfect, which every real world is.

Where biology runs into this

Nothing in nature reasons about payoffs, and the structure applies anyway, because selection does the reasoning. If defecting leaves more descendants, defecting spreads, and no intention is required at any point.

This is why cooperation is a genuine puzzle rather than an obvious good. A cell that ignores the body's constraints and reproduces faster is the definition of a defector, and it is roughly what cancer is. A mutualist that takes the benefit and gives nothing gets the benefit for free. The question is never why organisms cooperate, it is why defectors have not taken over.

The answers are structural rather than moral, and each removes the incentive rather than policing it: repeated interaction with the same partner makes today's defection tomorrow's cost, shared reproductive fate means damaging your partner damages you, and partner choice lets the exploited simply leave. Those are the same three mechanisms that hold multicellular bodies together, set out in how individuals became groups, and the same ones behind stable mutualism in living together.

Competition without a dilemma

It would be easy to describe this world as one where organisms compete, and to let that imply something game-theoretic. It does not.

Competition here is arithmetic. Resources deplete where they are eaten and regrow over time, so an organism that feeds leaves less for whatever arrives next. That is a shared finite pool, not a strategic interaction: nothing chooses between cooperating and defecting, nothing can identify the other party, and nothing can remember a previous encounter with a specific individual.

The requirements for a dilemma are all absent. There is no channel identifying another organism as an organism, so there is no partner to recognise. There is no memory of individuals, so repeated interaction cannot be detected even if it occurred. And there is no action whose payoff depends on another organism's simultaneous choice.