Evolutionary Design by Computers [With Cd Rom] [Hardcover]

by Richard Dawkins

What might Charles Darwin have done if he had had a computer at his disposal? What, indeed, would he make of this book? Of course he'd first have to get to grips with the astonishing power of the digital computer as an amplifier of human intelligence, and this might be initially shattering to an elderly Victorian gentleman. But having penetrated that barrier, I suspect that he'd have found here a cathartic rounding off of his 'one long argument': an elegant closure.

In Darwin's own time, the idea he had to offer was just as shattering to his contemporaries. In first chapter of The Origin of Species, he wisely softened up his readers with a weakened, almost homeopathic, dose of the medicine. Not natural selection - not the real thing - but a familiar and non-threatening analogue of natural selection. Artificial selection helped the Victorian mind in something like the same way computer models and 'artificial life' help ours. You want to design a new breed of pigeon with a long neck, in a way analogous to nature's way? The obvious engineering solution might seem to be direct mechanical manipulation: hang weights from the neck to stretch it, or put increasing numbers of rings round the neck like a fashionable tribeswoman? But this is not the Darwinian way. Instead, you wait for spontaneous genetic variation and breed from the longest necked individuals in successive generations. This is the principle by which we have engineered faster horses, tamer wolves reshaped into poodles and weimeraners. This how we have rejigged cows into milk machines, hens into mass production egg machines. A human designer can imagine a desired animal or plant shape, and set about achieving that goal generations later. This is evolutionary design of domestic animals and plants.

Darwin's original readers understood this process. The sticking point for them was when he took the designer out of the equation, replacing the deliberate human will of the breeder by unconscious natural selection: the survival of the fittest in the struggle for life. A hundred and forty years on, we have got used to natural selection. Today, our problem is opposite, and Darwin would have enjoyed the irony. Deliberate, creative human design, which Darwin used as his starting point because it is apparently so easy to understand, is today our mystery. How did the simple, mechanical, automatic process of survival of the fittest generate a brain powerful enough to engage in this novel - by evolutionary standards - kind of design process? For that matter, how did it produce a brain capable of painting pictures, of writing music or proving Fermat's last theorem? When an Eddison designs the electric light, or a Whittle the jet engine, what goes on inside the brain? Is it a kind of Darwinian survival of a minority of good ideas from a population of wild inspirations well up inside the head? If so, what is the population of varying entities that are selected, and by what criteria do they survive? Are they memes, as suggested by Derek Gatherer in Chapter 3. The meme was originally defined as a unit of cultural inheritance, but is the transfer of memes from brain to brain just the tip of the iceberg? Does most of the struggle for survival among memes, most of the cut and thrust of population memetics, go on inside individual brains? Do only the winners of an internal struggle see the light of day and enter the wider competition to jump to another brain?

Darwin would surely have been fascinated by such questions, not least in the context of computer art. As a masterwork takes shape, is each brushstroke - or some larger unit of creativity? - the winner of a Darwinian struggle for existence inside the artist's head? If so, should we define the survival criterion as 'taste', or is that still too anthropomorphic and empty? Is it rather that the artist's brain, before he even begins the painting, is already swarming with memes, and new ones gain a foothold only if they are compatible, in complicated ways, with the existing population? If so, the analogy with genetic Darwinian selection is closer than many people realise, for the most important part of the 'environment' in which genes survive is the other genes of the gene pool, as encountered in the cells of successive generations of bodies within the species. In any case, the possibility of using Darwinian selection explicitly in making computer art is a fascinating one, as we learn from some of its leading practitioners, in Section 3.

Most of all, Darwin would enjoy this book for its repeated demonstrations of the emergent, creative power of his principle of selection, the nonrandom survival of randomly varying manufacturing instructions, with or without something equivalent to sexual recombination. Here, Darwin would surely feel, is the experimental demonstration he really needed; a fully transparent manifestation of natural selection, not artificial selection, albeit in the artificial environment of the computer. Karl Sims's creatures, evolved to swim, walk, jump or follow, are sufficiently similar to real life to strike an instant chord with Darwin. But so, in their different ways, are the results of many others among these chapters. It only remains, Darwin might feel, to release such virtual creatures into an artificial ecosystem to prey on each other, run away from each other, coevolve in lockstep like antelopes and big cats. Easier said than done, of course. Meanwhile, I think we can be confident that Darwin would love this book.