Review: “A New Kind of Science”

The ancient Chinese game of Go has fairly simple rules. In general, it is much easier to teach someone the rules of Go than those of poker, for example, or of chess. Nonetheless, popular wisdom says that in all the 2,500 years that Go has been in existence, no two games have ever been identical. It’s impossible to know whether this is actually true, but it’s statistically plausible; thus, the game of Go demonstrates that it’s possible for very complex systems to arise from a very simple set of rules.

Makes sense, right? At least, when I say it that way it seems pretty obvious. You probably had some inkling in your mind of the idea that “complex behavior can arise from simple sets of rules” even before I mentioned it to you — didn’t you?

Well, strangely enough, Stephen Wolfram — although a mathematical prodigy who published his first scientific paper at 15, went on to school at Eton, Oxford, and Caltech, and invented the mathematical computation software Mathematica, among other things — did not. When he stumbled upon this idea back in the 1980s (simple rules could produce complexity), he found it not only surprising, but incredible. Astounding. MIND-BOGGLING. So much so that he decided to withdraw from academia, or indeed much of public life, and devote the next 20 years of his life and at least 1,200 pages of text to precisely this phenomenon.

The culmination of this labor is A New Kind of Science, a doorstop of a book that Wolfram claims revolutionizes all forms of scientific research and indeed every kind of human endeavor, from art, to philosophy, to the Meaning of Life itself. I kid you not. And I’ll grant you the favor that Wolfram cannot and allow you to jump off here, if you wish, by giving you my own summary:

This book is complete and utter bullshit.

It’s not so much the information that Wolfram presents that’s the problem. Most of the book is devoted to discussion of cellular automata, a particular kind of computer program the behavior of which my Go example gives only a poor analogy. Wolfram did indeed do some pioneering work with cellular automata in the 1980s, and he knows whereof he speaks.

No, the problem is threefold: A.) Much of what Wolfram presents in this text is not actually new, and much of it was developed by others than Wolfram himself (and he seldom gives credit where it’s due); B.) Wolfram insists that his computational methods are so significant and important that they effectively brush aside all previous scientific methods and indeed mathematics, which Wolfram dismisses as inadequate (despite thousands of years of historical success with those methods); and C.) He makes all these claims with such a pompous, self-congratulatory air that his text is almost impossible to read. (And here I admit that I could not finish it; reading turned to skimming; skimming turned to flipping pages; flipping pages turned to throwing the book down in disgust.)

Take, for example, this passage from chapter 3, which hopes to explain how Wolfram (and Wolfram alone) was able to arrive at the incredible “discoveries” laid on in that chapter: “…one of the problems with very direct experiments is that they often generate huge amounts of raw data. Yet what I have typically found is that if one manages to present this data in the form of pictures then it effectively becomes possibly to analyze very quickly just with one’s eyes. And indeed, in my experience it is typically much easier to recognize unexpected phenomena in this way than by using any kind of automated procedure of data analysis.”

Just on the face of it this paragraph seems intellectually dishonest. Wolfram’s brilliant technique is to “present this data in the form of pictures”? What… you mean like graphs? Anyone who’s made it past elementary algebra without seeing the graph of an equation must be blind.

What Wolfram is in effect proposing, however, actually goes beyond simple graphs of mathematical systems. It’s much crazier than that. Wolfram claims that traditional methods of scientific analysis are needlessly vague and imprecise. Instead, he says, scientists can learn all that they need to know by searching for the precise sets of rules that govern the behavior they’re trying to observe and then reproducing them on the computer. Regular observation-based science can only yield statistical probabilities of outcomes. Nail the right ruleset, however, and you need only run the program a few million times in the computer and you will actually SEE what will happen — not with any kind of statistical analysis, but with your own eyes. All of the universe, Wolfram claims, is merely based on computations of this kind.

He then goes on to give examples of automata that can generate graphs that look curiously like spirals, or leaf forms, or trees, or what-have-you. And this is all well and good — you’ve probably seen such effects before, even if Wolfram claims he hasn’t, in fractal images. There’s even a software program called Painter that uses fractal-based computation to simulate natural art media, including oil painting, pastels, and watercolors.

But that’s just it: Painter is a simulation. Ask any artist whether the experience of using the oil-paint tools in Painter is the same as using actual oil paint, or if it yields the same results, and they will tell you no. The results might be effective enough to convince an observer that the picture was made with real oil paints, but it will only be an illusion. A real oil painting created by the same artist would look substantially different, because the model is imperfect.

But to Wolfram, who has spent the last 20 years sitting in front of his computer, an imperfect model only means that there must be a perfect model out there, waiting to be discovered. To Wolfram, who believes that all of the universe can be represented through computation, the model and the reality are the same thing. Instead of wasting their time with fruitless experimentation, Wolfram says — he’s quick to dismiss just about anybody else’s work as “fruitless” — scientists should sit down at their computers, like he has done, and start looking for the cellular automata-based patterns that unlock the keys to nature.

The idea that man can capture the essence of the universe in a simple calculation, like a genie in a bottle, is an attractive one — but in much the same way that the idea of a perpetual motion machine is attractive. It’s fun, it’s wistful, but it’s not very practical. (Oh wait — did I forget to mention that Wolfram claims to call into question the Second Law of Thermodynamics? p451: “Starting nearly a century ago it came to be widely believed that the Second Law is an almost universal principle. But in reality there is surprisingly little evidence for this.”)

The biggest problem with Wolfram’s automata is that they have virtually no predictive power — and so therefore are virtually useless to scientists. Suppose a scientist discovered an automata that, given a certain input, produced a pattern that looked remarkably like the coastline of Norway. What would it prove? What predictions could we make based on that? Could we know what Norway will look like a hundred, or eighteen billion years from now? Probably not, because as any geologist will tell you, what Norway looks like has precious little to do with what Norway IS. No wonder Wolfram advocates discarding all previous scientific method — since it has little use for his “experiments,” it must itself be useless.

“It has taken me the better part of twenty years to build the intellectual structure that is needed, but I have been amazed by the resuls,” Wolfram writes in the first chapter (if he does say so himself). “For what I have found is that with this new kind of science I have developed it suddenly becomes possible to make progress on a remarkable range of fundamental issues that have never successfully been addressed by any of the existing sciences before.”

Fine. Proof is in the pudding. What progress? What advances have been made based on this material? What science is it revolutionizing, and what great steps forward have humankind taken?

Ultimately, though it’s sad to say it, this book belongs nowhere other than the crank file, along with all the books on secret energy sources, ESP, and UFOs building the pyramids. I’ll never again be able to think of Stephen Wolfram without being reminded of that character in the movie Pi, being driven mad by all those numbers. Somewhere out there, even now, Stephen Wolfram is staring at page after page of graphical output from cellular automata programs — completely and irreversibly batshit insane.

Save yourself the same fate and don’t waste a minute on this overlong, repetitious, hollow, arrogant, self-aggrandizing, dishonest, tedious, intellectual dead-end of a text.