Great post Jason! One more aspect about the Axelrod and Hamilton experiment I would like to highlight in this context is that the TFT algos, and almost all the top 10 winning algos first chose to co-operate with their competitor, they also took into account what their competitor did in the last move. Whenever history is involved, and there are going to be future transactions, co-operation always wins out. Not only because it enables the transaction, but also it synergizes the transactions after as well! This will become apparent if you take a look at the pay-off matrix in that experiment. It was paying less to co-operate, but the summation of co-operating games over 50 odd transactions was always more than win-lose kind of transactions.

In real life, I think this synergy is the relationship between two parties. When one party does not co-operate, it impacts all future transaction with that party.