The Science of Discworld II

complex system viewpoint, evolution isn’t like that at all. Organisms may sometimes compete directly – two birds tugging at the same worm, for instance. Or two baby birds in the nest, where direct competition can be fierce and fatal. But mostly the competition is indirect – so indirect that ‘compete’ just isn’t the right word. Each individual bird either survives, or not, against the background of everything else, including the other birds. Birds A and B do not go head-to-head. They compete against each other only in the sense that we choose to compare how A does with how B does, and declare one of them to be more successful.
    It’s like two teenagers taking driving tests. Maybe one of them is inthe UK and the other is in the USA. If one passes and the other fails, then we can declare the one who has passed to be the ‘winner’. But the two teenagers don’t even know they are competing, for the very good reason that they’re not. The success or failure of one has no effect on the success or failure of the other. Nevertheless, one gets to drive a car, and the other doesn’t.
    The driving-test system works that way, and it doesn’t matter that the American test is easier to pass than the British one (as we can attest from personal experience). Evolutionary ‘competition’ mostly works like the driving test, but with the added complication that just occasionally it really is more like a tennis match.
    From this point of view, evolution is a complex system, with organisms as entities. Which organisms survive to reproduce, and which do not, are system-level properties. They depend as much on context (American driving test versus British) as on the internal features of the individuals. The survival of a species is an emergent feature of the whole system, and no simple short-cut computation can predict it. In particular, computations based on the frequencies of genes in the gene-pool can’t predict it, and the alleged explanation of altruism by gene-frequencies is unconvincing.
    Why, then, does altruism arise? An intriguing answer was given by Randolph Nesse in the magazine Science and Spirit in 1999. In a word, his answer is ‘overcommitment’. And it is a refreshing and much-needed alternative to bean-counting.
    We have said more than once that humans are time-binders. We run our lives not just on what is happening now, but on what we think will happen in the future. This makes it possible for us to commit ourselves to a future action. ‘If you fall sick, I will look after you.’ ‘If an enemy attacks you, I will come to your aid.’ Commitment strategies change the face of ‘competition’ completely. An example is the strategy of ‘mutual assured destruction’ as a deterrent for nuclear war: ‘If you attack me with nuclear weapons, I will use mine to destroy your country completely.’ Even if one country has many more nuclear weapons, which on a bean-counting basis means that it will ‘win’, the commitment strategy means that it can’t.
    If two people, tribes or nations make a pact, and agree to commitsupport to each other, then they are both strengthened, and their survival prospects increase. (Provided it’s a sensible pact. We leave you to invent scenarios where what we’ve just said is wrong.) Ah, yes, that’s all very well, but can you trust the other to keep to the agreement? We have evolved some quite effective methods for deciding whether or not to trust someone. At the simplest level, we watch what they do and compare it to what they say. We can also try to find out how they have behaved in similar circumstances before. As long as we can get such decisions right most of the time, they offer a substantial survival advantage. They improve how well we do, against the background of everything else. Comparison with others is irrelevant.
    From a bean-counter’s point of view, the ‘correct’ strategy in such circumstances is to count how many beans you gain by committing yourself, compare that to how many you gain by cheating, and see which pile of beans is biggest. From Nesse’s point of view, that approach doesn’t amount to a hill of beans. The whole calculation can be sidestepped, at a stroke, by the strategy of overcommitment. ‘Stuff the beans: I guarantee that I will commit myself to you, no matter what . And you can trust me, because I will prove