Is the universe a computer?

A hidden logic lies behind just about everything we see, according to maverick prodigy Stephen Wolfram. NOTE: Click Print button at foot of page if formatting on this page looks awry (and cancel the print dialoque box, if necessary).

Stephen Wolfram: New Darwin or thwarted prodigy?

Some 40 years ago infant graphics systems began revealing things no-one had ever seen before: beautiful, infinitely intricate fractals generated by repeatedly solving very simple equations, using the result of one calculation as the starting point for the next. The calculations were trivial but these massive iterations were practically impossible before the advent of computing. They were reminiscent of many natural growth processes, like the way cells reproduce and change to become an organism. Mapped onto a computer display, they soon began to throw up images eerily close to natural forms.

     It has become clear that we are computers of a kind, albeit little like the ones on our desks, and that our evolution and growth has involved some form of information processing. Yet there has been remarkably little discussion outside universities, about how our knowledge of computing should change the way we view ourselves and the universe. It is as if we feel demeaned by the idea.
     This has all changed in the past few weeks, with the publication of a wonderfully eccentric doorstopper of a book called A New Kind of Science by Eton-educated former child prodigy Stephen Wolfram. The book says we should start thinking of just about everything in terms of very simple programs. It has become an instant best-seller, generating fury, scorn and praise in almost equal measure.
     Wolfram was born in London in 1959, published his first scientific paper at the age of 15 and received a PhD in theoretical physics from the California Institute of Technology at the age of 20, after leaving Oxford without taking his degree. He was one of the early workers in complexity theory, but changed direction in 1986 to found a company called Wolfram Research, based around a product called Mathematica, which has become one of the standard software toolboxes for mathematicians.

Early passes of a simple one-dimensional automaton (top) with (middle) the rule that produced it. The lower pattern, generated from an equally simple rule, has regions that are not intuitively predictable. Click here to see how this pattern continues and for a fuller explanation

Wolfram is an acknowledged world expert on cellular automata, programs that are about as basic as you can get: start with a dot, use a simple rule to decide where and whether to place another next to it, and then iterate the process to produce a pattern (see left).
     Ankos, as Wolfram's book has quickly become dubbed, serves on one level as an introduction to automata and the forms they generate. Its 1,200-odd pages pack in more than 10 years of work, and I have to say that this article is based on my humble understanding after a very quick reading - it contains many interesting-looking ideas that I have yet to wrap my brain round.
     Wolfram describes in great detail how what appears to be complex natural forms can stem from very simple procedures.

Four shapes generated by tweaking two parameters of a 3D automaton - and, below them, the shapes of real shells. Click here for larger image (171Kb) and more examples.

That this is happening in nature is clear from the pictures right, taken from Ankos, which show real shells with shapes identical to forms generated by different tweaks of a 3D automaton. The natural structures literally have a logic to them.
     You might think these ideas are a perfect complement to those of Darwin, filling in the gaps in how natural selection works: all evolution had to do to get those different shell shapes was to hit on these same tweaks. Similarly, the evolution from a single light-sensitive cell to a complex eye is comprehensible in terms of reproducing the cell in a simply generated pattern.
     But Wolfram says evolution is merely choosing from a limited number of possible forms and that he has revealed an organising principle which is at least as important. Critics point out elementary mistakes in his description of modern Darwinism.
     These may owe something to the fact that Wolfram rejected the usual method of expounding original scientific ideas in formal academic papers submitted for peer review to weed out howlers. He felt this would be to constricting. Instead, he wrote Ankos in flat but effective journalese for a wide readership, allowing himself complete freedom to speculate. His ideas are indeed too broad in scope to be restricted to specialists.
     But many researchers are enraged at how he hypes his ideas like a marketing man, apparently oblivious to work done by others, and claiming as unique personal insights concepts that have been chewed over by many for decades. They dismiss Ankos as the burblings of a man who never lived up to his early promise. But there is raw jealousy in some of the more vitriolic web postings - some researchers refuse to even read the book.
     Wolfram, if his supporters are right, has caught the early 21st Century zeitgeist in the way Darwin pulled together Victorian ideas about evolution, though he fails to come up with an encapsulating idea as simple and powerful as natural selection - or if he has, it has yet to be demonstrated.
     Some of his most provocative ideas relate to maths, physics and even religion, and stem from the observation that some cellular automata generate patterns that are what he calls 'effectively random'.
     You'd expect simple rules to generate simple patterns, he says, yet here is complexity deriving from what amounts to basic arithmetic. He identifies three types of randomness, or unpredictability: environmental, like the buffeting of a cork by turbulent waters; extreme sensitivity to starting conditions, like the proverbial butterfly setting off a storm; and intrinsic, like that shown by cellular automata.
     He argues that the equations of conventional mathematics can never encompass this intrinsic complexity, even though it may derive from very simple rules. And just about all physical systems, according to Wolfram, can be seen as automata. The implication is that there are hidden logical layers in the behaviour of molecules, atoms and sub-atomic particles, not to mention stars, galaxies and, well, the universe, that could explain some of their mysteries.
     Looking at the patterns produced by simple automata, it seems that similar effects must be inherent in the growths and aggregations of the physical universe, which at all scales involves distinct units obeying clear rules. But whether the logic goes beyond patterning is another matter. Genetic algorithms in digital computers have demonstrated how simple programs can evolve into complex useful code, but evolution requires a selection mechanism which is not at all obvious in the inanimate universe. For those who doubt the potential of simple rules, Wolfram points out that the almost infinite capabilities of digital computers stem from a small set of basic machine instructions.
     He says that the way current science views complexity is like looking at the patterns produced by automata rather than at the rules that generated them. Equations may allow you to predict behaviour for practical purposes, but when applied to the real world they always involve some approximation.
     Physicists are still looking for an equation or set of equations to unify all they know about the physical universe. Wolfram suggests they will never succeed because they are looking for the wrong thing. They should be seeking a program. This is an enchanting notion, but hard to prove or disprove.
     The problem is that many automata are not reversible: the pattern they produce does not contain enough information to reveal the starting condition or growth rules. Wolfram uses this to 'explain' the directionality of time and entropy; but it also means that the only way to find the root logic of a system is to try all possible programs until you get one that fits. It is unclear how useful this approach can be.
     Wolfram suggests treating expanding space as a cellular network, something that grows from a set of basic rules like a standard automaton, but includes information links between event nodes. At one point, he postulates that the universe is a giant Turing Machine: that is, a computer.
     Oddly quiet in the debate so far have been the creationists; sooner or later one of them is sure to jump up and claim that Wolfram's cosmic logic implies that there must have been a programmer. It does nothing of the kind of course: you get back to the old circularities about who programmed the programmer. But Ankos, like Darwinism, can be seen as consistent with both theism and atheism.
     Wolfram throws out such a torrent of ideas that some are bound to be wrong, but there is surely something in what he says. At the very least he gives a glimpse of how a thinking human can stem from a strip of DNA containing roughly the same amount of information as the software with which this article was written. Some things you want to believe: there is a curious comfort in the idea that the universe may be smarter than we thought.
     ? A New Kind of Science, published by Wolfram Media. List price £40, but can be found for less than £30. A CD is imminent. Mathematica routines to generate automata, and much more iformation on the subject, are available at Wolfram Science.