Observer Review: A New Kind Of Science By Stephen Wolfram Essay, Research Paper
Life, the universe and a game of chequersA New Kind of ScienceStephen WolframWolfram Media Inc ?40, pp1,197The tradition of scientists presenting their grand ideas in popular books rather than scientific papers has given us Darwin’s Origin of Species, EO Wilson’s sociobiology and Jim Lovelock’s Gaia hypothesis. The attraction is that the author can expound the central idea at length without the burden of nitpicking detail that accumulates when one has to please one’s peers. The disadvantage is that the devil is often in those details: it is easier to pull the wool over a lay reader’s eyes.Stephen Wolfram’s long-awaited new book demonstrates these pros and cons in spades. Few science books have had a grander scope: he sets out to solve it all, from mathematics to evolution to perception to a theory of everything, with an analysis of free will for good measure. Yet in the end what we get is a thousand pages of free-floating speculation.His central thesis is that a theory of life might not be encapsulated in a single über-equation but might consist of an algorithm: a set of rules, like the lines of a computer programme, which, when enacted again and again, allow the universe to unfold in all its variety. This is an immensely stimulating idea, and although he has not a shred of direct evidence to support it, it is enough to allow this often infuriating book to emerge in credit.Wolfram is that rare thing, an independent scientist, and since the early Nineties he has been reclusive, working through the nights on his ‘new kind of science’. His grand idea stems from the study of cellular automata (CA): a kind of chequers game in which the addition and removal of pieces are dictated by simple rules about the configuration of neighbouring pieces. For example, a rule might specify that we place a new piece on a blank grid square if there are two pieces on neighbouring squares, or remove one if it has more than four neighbours.There are endless permutations of such rules. Wolfram’s pivotal discovery was that some CA produce patterns that are not just complex but random. The only way to know what arrangement the algorithm will generate after thousands of iterations is to run it. The behaviours he has found have convinced him that the universe itself might be a CA, playing out its rules on an incredibly fine-grained network of nodes that define space and time.In explaining this notion Wolfram is inventive and ingenious. But to put the idea to the test, we are limited to the ‘everyday’ systems that Wolfram claims as instances of CA in action. Time and again he holds up some puzzle from everyday life, such as how snowflakes grow or where the second law of thermodynamics comes from, then hints darkly that no one has really made any progress on this problem, before producing a CA that appears to perform the right trick.This is simply dishonest. There are already good explanations of these things based on just the equation-rich theories that Wolfram calumniates. Models of snowflake growth, for instance, are not ‘rather unsuccessful’, but quite the contrary.On the question of how living organisms are moulded, the existing theory – natural selection – is too well known to be spirited away like this. Wolfram instead presents so grotesque a misrepresentation of neo-Darwinism that his CA can’t but seem superior.Nowhere in the book does Wolfram provide anything so conventional as scientific hypothesis-testing. All his arguments proceed by analogy. His strategy is to find a CA that looks suggestive of a particular phenomenon, and then to admit a ’suspicion’ that real life works this way too. It is like saying that because ants have evolved a complex social structure, they must have discovered socialism.Yet the book’s key message may be extremely fertile – perhaps even correct. Wolfram should swallow his hubris and start talking to his peers again. If nothing else, they will keep him honest.