(Extracted from Barking Up the Wrong Tree by Eric Barker)
This is where Robert Axelrod got started. With the Cold War raging between the United States and the USSR, he wanted to explore what it takes to get people to trust and cooperate, what strategy is most effective. So he decided to have a tournament where different computer programs with different strategies play Prisoner’s Dilemma together to see which one racked up the most points. Researchers from psychology, economics, math, sociology, and other disciplines sent in a total of fourteen algorithms plus one program that would behave randomly. One of the programs was insanely nice: it always trusted its opponent even after being screwed over. Another of the programs—named ALL D—was the opposite: it always betrayed its opponent without fail. Other programs rested somewhere in between. Some of the more complex programs played nice for the most part while occasionally trying to sneak in a betrayal to get a leg up. One program called Tester monitored the other player’s moves to see how much it could get away with and then would backpedal if caught betraying its opponent. Which ethical system reigned supreme in the end? Shockingly, the simplest program submitted won the tournament. It was only two lines of code. And it’s something we’re all familiar with: tit for tat. All TFT did was cooperate on the first Prisoner’s Dilemma round, then in every subsequent round, it did whatever the opponent did previously—that is, if on the previous round the opponent cooperated, it cooperated on the next round; if the opponent betrayed, it betrayed on the next round. This simple program decimated the competition. So Axelrod ran the tournament again. He reached out to even more experts and this time had sixty-two entries. Some algorithms were more complex and some were variants on TFT. Who won? Simple ol’ tit for tat. Again.