The Evolution Of Democracy

Twenty years ago, a political science professor at the University of Michigan came out with a seminal book titled the Evolution of Cooperation.

In it, he described how cooperative strategies could have evolutionary-beneficial consequences, and thus be selected for. In particular, via a series of computer game tournaments in which algorithms were submitted to play an extended iterated prisoner’s dilemma, he identified a strategy that was the most successful called “Tit for Tat” (TFT). (Read the link for information as to how the game works.)

In this strategy, you retain a memory of past interactions with other entities, and you treat them exactly as they treated you the last time you dealt with them. If they cooperated the last time, you cooperate. If they defected on you the last time, you defect on them the next. If it’s your first interaction, you cooperate.

The strategy has four characteristics that made it successful. It’s simple and can be clearly and easily recognized after a brief period of time, it’s forgiving, it’s provocable and retributive (so that you can’t get away with screwing it), and it’s nice (that is, it never screws anyone for no reason–its default is to cooperate). In essence, it is cooperative, and is rewarded for being that way.

One of the interesting things about it is that the more similar algorithms it has to deal with, the better it does. Put in an environment of non-cooperators, it has a much harder time, but it can still be more successful than them, and if it has a few others to cooperate with, it can survive even in a sea of non-cooperators.

Non-cooperators, on the other hand, don’t do well in a cooperative society. A non-nice strategy (one that always, or occasionally, or randomly defects unprovoked) won’t do well in a world of TFTs, because after the first time they get screwed by it, they will not cooperate with it again, at least until it changes its ways. So while it gets a big payoff the first time, it gets a much smaller one in subsequent exhanges, whereas the TFTs interacting with each other always get the medium benefit.

Thus, it’s possible for a small group of cooperators to “colonize” a larger group of non-cooperators, and eventually take it over, whereas a group of non-cooperators invading a larger group of cooperators will not thrive, and will eventually die out. This is the basis for Axelrod’s (and others’) claim that there is evolutionary pressure for cooperation to evolve.

This may hold the key to fixing Iraq, and ultimately the Middle East. While there’s a lot of bad news coming from that country right now, the fact remains that much of it is calm and at peace–that part doesn’t make the news. It may be that nationwide elections won’t be possible in January, but certainly it should be for some regions (particularly the Kurdish region).

The Jihadists and ex-Ba’athists are determined to prevent a democracy from forming there, but if such can be established in large areas, it will provide an unnurturing environment for them there. Then we can gradually expand them, and tighten the noose around the Fallujahs over time. What we have to pay attention to is not the level of violence, but over how widespread a region it is. As more and more of the country becomes not only pacified, but wealthier, with a stake in continued peace and freedom, we can continue to shrink the territory in which the terrorists, the ultimate non-cooperators, can survive, and eventually kill them or starve them out.