If we do this again, I’m almost tempted to submit a really perverse example which does the opposite of tit for tat. So, defect on the first turn, and cooperate on turn n iff they defected on n-1. I’m pretty sure this would go really badly.
Hmm, I notice also that no one tried naive always cooperate. Would be interesting to see how well that does.
I thought of this too, but it really should play C then D to start so it cooperates against itself—otherwise, it will defect against itself.
Given the payoff matrix for this tournament, it isn’t a very good strategy. However, if the payoff matrix rewarded D-C more strongly, then a pair of (tit-for-tat, tit-for-tat’s complement) would do very well.
If we do this again, I’m almost tempted to submit a really perverse example which does the opposite of tit for tat. So, defect on the first turn, and cooperate on turn n iff they defected on n-1. I’m pretty sure this would go really badly.
Hmm, I notice also that no one tried naive always cooperate. Would be interesting to see how well that does.
I thought of this too, but it really should play C then D to start so it cooperates against itself—otherwise, it will defect against itself.
Given the payoff matrix for this tournament, it isn’t a very good strategy. However, if the payoff matrix rewarded D-C more strongly, then a pair of (tit-for-tat, tit-for-tat’s complement) would do very well.
Tested that, in the round-robin it is somewhere about the average, in the evolutionary dies out (albeit more slowly that the defectors).