Manny's Reviews > A Project to Find the Fundamental Theory of Physics

In his classic How to Cheat at Chess, Bill Hartston explains that it's easy to annotate chess games once you know the result. If a complicated sacrifice worked, praise it as "the logical conclusion to a well-played attack"; if it didn't, dismiss it as "desperation, but White was lost anyway." Hartston says that when you see a chess journalist intently studying the position in an ongoing tournament game, you can be sure that what they're doing is mentally composing those two parallel narratives, so that they'll be able to mail in the right one as soon as the game is finished.

Inspired by Hartston's remarks, I offer you the following two reviews of Stephen Wolfram's new book. I'm hopeful that at least one of them will turn out to be right.

Positive review

People who want to score big often need to do things their own way. Einstein, who changed the face of science for ever, spent his most productive years working as a patent clerk; it gave him time to develop his thoughts at leisure, away from the relentless pressure of the academic world. Wolfram has been no less original. Starting off as a science wunderkind - he got his PhD at 20 - he soon realised that what physics needed most was better computational tools. He started developing the software that eventually turned into the Mathematica package, and founded a successful company to market it.

The more he worked with computation, the more he began to suspect that this was the true substrate of reality. Investigating a huge range of computational systems, he found that even the very simplest ones - in particular, the remarkable 'Rule 30' - could display astonishingly complex behaviour. In his 2002 bestseller, A New Kind of Science, he presented his initial findings, but then returned to growing his company. Now, in this latest book, he completes the Odysseus-like journey that has led him back to the fundamental physics he started from, and shows new ways to address the problems that people have been stuck on since the 70s. Instead of increasingly recalcitrant mathematical frameworks, Wolfram uses a twenty-first century approach and shows how the power of modern computer technology can be harnessed to simulate artificial universes that are starting to reproducing the physical phenomena we see in our own world.

Wolfram's bold hypothesis is that the universe, at its deepest level, consists of a huge network, which is constantly evolving according to a single rule whose precise formulation is, literally, the ultimate answer. His intuition, honed on a lifetime of working with such systems, tells him that this rule is very simple - so simple, in fact, that it can be found by having machines systematically check through the possibilities. He has not yet found it, but he says he's close. Already, he has been able to demonstrate that a large class of these systems follow the equations that define Einstein's theory of general relativity. He has an incomplete but suggestive argument, developed with a student, to show that a slightly more restricted class of systems also display behaviour characteristic of the quantum mechanics we see in our own universe, and in particular reproduce the path integral formula developed by his one-time mentor Richard Feynman.

Wolfram takes a broad-brush approach, and freely admits that important details still need to be resolved. (In particular, it is still not quite clear how to resolve the difficulties posed by Bell's Inequalities). But he thinks his initial successes cannot possibly be accidental, and that we're now within sight of the answer. He is making all his findings and methods generally available, and encourages other people to join in. It's hard not to believe that there will be an enthusiastic response.

Exciting times!

Negative review

People who have screwed up big often feel an urge to justify themselves. Einstein, who wasted the second half of his life on a fruitless search for the Unified Field Theory, ignored horrified entreaties from his friends to stop throwing good money after bad after bad and move on. Similarly, Stephen Wolfram, despite all the evidence to the contrary, is unable to accept that his strategy for investigating the nature of reality might be totally mistaken. Having abandoned a promising career in fundamental physics to start a software company, he is determined to show that software engineering is going to reveal the secrets of the universe. He says himself that his physicist friends beg him in vain to do something else.

Discovering that simple rewriting systems like the overpublicised 'Rule 30' can sometimes create complex structure (a result already well-known, for example, from Conway's 'Game of Life' and the Mandelbrot set), Wolfram decided that the nature of reality had to be in some way related to this finding. In his 2002 self-published screed, A New Kind of Science, Wolfram outlined these ideas and was met with a resounding lack of interest from the scientific community. He went back to working on his company. But he has evidently not had the sense to leave well alone, and is now wheeling out the same old arguments again. He seems to have forgotten how to work with equations. Instead, he shows us interminable pages of output from his simulations, and tries to convince us that they are somehow better.

Wolfram's tired hypothesis, all too familiar from the earlier book, is that the universe, at its deepest level, is some kind of rewriting system. The novelty this time round is that it is, specifically, a certain type of graph rewriting system. A New Kind of Science already made the unsupported claim that systems of this kind can display behaviour describable using the formal apparatus of general relativity. Here, there are a few more details and a link to a recent arXiv paper by a student. There is some extremely handwavy speculation about how quantum mechanics fits into the picture. I do not understand the claimed derivation of the Feynman path integral, which relies on some legerdemain in which energies are related to angles in "multiway causal graph space" (where does the geometry of multiway causal graph space come from?) and the magic formula ∫ exp(iHt)dt emerges from what certainly looks like Wolfram's sleeve. He has still not found any actual rewriting system that displays the properties he keeps promising us must be there.

Bell's Inequalities come up momentarily in the final chapter and are then brushed under the rug, despite the fact that, as Wolfram says, they are the standard objection to hidden variables accounts of quantum mechanics. The best known book on this kind of framework, Nobel laureate Gerard 't Hooft's The Cellular Automaton Interpretation of Quantum Mechanics, is not even mentioned. For some reason, Wolfram thinks his work is interesting, and is making it all publicly available. Knowledgeable people seem less than blown away. You can see some typical comments here.

Depressing times.