te: Wed, 27 Oct 1993 14:08:15 -0400 From: Bryan Douglas Caplan Subject: Game theory and evolution To: libernet-d@Dartmouth.EDU I've recently been reading a book called Demanding the Impossible: A History of Anarchism by Peter Marshall, and it has sparked a few thoughts on game theory, evolution, and their political ramifications. Many people interpret Darwin's theory of evolution as basically a Hobbesian theory. Everything is struggling to survive, and mercilessly crushes everything that stands in its path. Darwin's view was not this simple, but he did delay the publication of the Origin of Species because he could not explain the suicidal behavior of bees who die when they sting to defend their hive. If we put this over-simplified version of Darwin's theory into the terms of game theory, we would say that always-defect is the evolutionarily strategic way to play the game. Applying the theory to human society, we would likely draw the same conclsions that Hobbes did: namely, a powerful state is necessary to check the destructive effects of the "evolutionary game." Now the 19th-century anarcho-communist Kropotkin wrote a critique of this interpretation of Darwin. The critique was called Mutual Aid; it argued that cooperation was a much more important factor in human evolution than conflict. Species that learned to cooperate would tend to survive much better than those stuck in perpetual conflict. In terms of game theory, Kropotkin basically thought that we are evolutionarily programmed to play always-cooperate. He drew the political consequence that with the abolition of the state, our natural tendency to cooperate would re-appear and intensify. In consequence, everyone would happily take part in a order of voluntary communism. The conclusions of socio-biology and modern game theory make an interesting contrast to both theories. Socio-biology explains Darwin's bee-paradox thusly: the basic unit of evolution is not the individual but the gene. If one bee can save the lives of other bees WHO SHARE HIS GENES, he can promote his genes ever though he perishes himself. On this view, we have (as Kropotkin argued) an innate desire to help others; but in keeping with more standard evolutionary theory, this desire to help is limited to a fairly narrow group of other people who share our genes. But what about the prevalence of cooperation not based on genetic ties? This is where modern game theory comes into play. Cooperation between people who share no genes couldn't be based on universal altruism; otherwise, the all-defect strategy could always take advantage of the all-cooperate strategy. Rather, cooperation in the absence of genetic ties would have to be based on tit-for-tat, on reciprocation. Strategically speaking, BOTH all-defect and all-cooperate encourage the other player to play all-defect. But tit-for-tat, in contrast, encourages other players to play tit-for-tat. This makes it an evolutionarily dominant strategy, since it makes cooperation correspond to narrow self-interest. Now I think that this evolution/game theory has interesting political implications as well. It tends to argue that, unlike the (oversimplified) Darwinian theory, a powerful state is not necessary to create cooperation. But it also argues against Kropotkin's theory that voluntary communism could work well, since his system is based on universal altruism rather thyan on reciprocation. Instead, the third theory tends to support libertarianism , since it shows how cooperation can arise without the state, but only if the system preserves individual incentives and individual responsibility. Evolution does promote cooperation, but only cooperation based on reciprocation, not unilateral giving.