Co-operatively solving the prisoner’s dilemma

Listening to a back edition of Seth Godin’s Akimbo podcast “Game Theory & the Infinite Game”, it struck me that a co-operative is, by design, an attempt to collectively escape the prisoner’s dilemma.

To recap the parameters of the ‘game’, two people are caught red-handed for a crime. They are going to serve a year’s jail-time for this. But they are wanted for a bigger crime too. So they are each offered a way out of jail-time, rat on your partner for the bigger crime and you will go free but your partner will get 50 years. The trouble is that if they both sing, then they will both get 5 years.

The expectation is that the rational selfish response will be to rat on the other. In practice, humans have demonstrated a systemic bias to co-operate – we seem to understand that the pursuit of single-minded self-interest can lead to a bad outcome for all.

So now I have added The Evolution of Cooperation by Robert Axelrod to my reading list (here for the abridged version). This book reports on a Prisoner’s Dilemma tournament where he invited game theory experts to submit programs that would be paired off against each other to see which did best over repeated interactions.

“Amazingly enough, the winner was the simplest of all candidates submitted.
This was a strategy of simple reciprocity which cooperates on the
first move and then does whatever the other player did on the previous
move.”

He called it the Tit-for-Tat strategy.

By analysing the top-scoring strategies, Axelrod stated several conditions necessary for a strategy to be successful:

• Nice – Almost all of the top-scoring strategies were nice, that is, it will not defect before its opponent does.
• Retaliating – A successful strategy must not be a blind optimist. It must sometimes retaliate to avoid being exploited by others..
• Forgiving – Though players will retaliate, a successful strategy will once again fall back to cooperating if the opponent does not continue to defect. This stops long runs of revenge and counter-revenge.
• Non-envious – The best strategies did not strive to score more than the opponent.

Whether the players trust each other or not is less important in the long run than whether the conditions are ripe for them to build a stable pattern of cooperation with each other. The successful strategy learns through trial-and-error that it is better to co-operate than not.

Some lessons here for co-operative design!