Reference request: stability of cooperation in prisoner's dilemma and other normal form games

Question

While reading some stuff on normal form games, I came across a claim that whilst cooperation in prisoner's dilemma can lead to better result, the cooperation is not "stable".

I have never came across any discussion of stability of cooperative strategy in normal form games. Can someone refer me to some material that discuss the stability of cooperative solutions? Thanks!

Answer 1

I think a lot of writings on game theory will discuss the stability of games, particularly in the context of the Nash equilibrium. In the case of the Prisoner's Dilemma, the instability is because:

1. The "ideal" result, which gives the greatest net benefit to both players, is for them to co-operate.
2. If you know the other player is going to co-operate, you benefit (at their expense) by betraying them.
3. If you know the other player is going to betray, you still benefit (again, at their expense) by betraying them.
4. Therefore, it's in your selfish best interests to betray.
5. So two selfish agents playing the game will betray each other, despite this being a worse outcome for both of them compared to co-operating.

So the "everybody wins" scenario is unstable because as soon as the players become selfish they wind up in the lose-lose situation.