So this guy John Conway invents this(not the first, not the concept, Stanislaw Ulam was doing this in the 1940s, for crying out loud!) cellular automaton, and people spend 3 decades watching squiggly graphics. What's that to do with the Universe?

Life (the game, not The Game) is just a remarkably simple cellular automaton, with the property that it can simulate a Turing machine. And anything which can simulate a Turing machine can simulate any computation -- that's the Church-Turing thesis. This is definitely neat: just some simple rules are enough to define computation.

Of course, since the 1930s logicians have been coming up with loads of these Turing complete systems, none of them very complex. E.g. Conway also came up with FracTran, which embeds any program inside some rules for manipulating rational numbers. The point is that universal machines for what we term "computation" are really very simple. John von Neumann was probably the first to write it down, even if he probably wasn't the first to think it. And that was back in the 1940s. Note: pfft correctly points out that I use "universal" here and in the sequel in the sense of "universal for Turing machines", or "universal for CAs", or "universal for Life", or "universal for recursive functions", or your favourite Turing complete model.

History repeats itself, first as tragedy, then as farce. In the 1970s, people realised that apart from being a neat footnote in mathematical logic, Conway's "Life" afforded the opportunity of drawing pretty pictures on a computer! Interest surged, and many modules are available for Life. Universal machines based on Life have grown ever smaller. Oh, and in the same field Hilbert's tenth problem was solved, but how important could that be? Yuri Matiyasevich's name is impossible to write down, and no pretty pictures were available.

Enter the 3rd millennium. We're now repeating stuff from the 1970s. Fredkin said that the Universe could be a cellular automaton, because cellular automata were powerful enough to model it (if we smooth over minor problems of continuous vs. discrete systems, and the like). Wolfram (in the very aptly-named A New Kind of Science) says it might be, or even must be. For the same reason.

Note the word UNIVERSAL in "universal machine". It doesn't refer to the famous Universal Studios Tour, it refers to an annoying property of those damned machines. Universal machines are computers, aka fools: What one fool can do, any other fool can also do. And what one universal machine can do, any other universal machine can also do. If the Universe can be modeled by the game of Life, then it can be modeled equally well by any other universal machine. So if we want to claim that the Universe IS a cellular automaton, we'd better have a much better argument to back us up than just that cellular automata are powerful enough to simulate the Universe. After all, so are polynomials (see the work of the Russian bloke with the hard name beginning with M, and the American linguist logician and philosopher with a girl's name, plus some other gals and guys).

It's impossible to distinguish between different universal machines. Arguing whether the Universe runs on your favourite universal machine or on mine is, if at all possible, an even sillier exercise than arguing whether C or Java is "the best" programming language. Both can do anything. So...

Why Life and not FracTran?

Let me be the first to claim that The Universe is a FracTran Program. That way, when someone claims it in the April 2031 issue of Scientific American, it won't be for the first time.

Oh, and there's the small business of having to solve nondeterminism in quantum mechanics before we can do this. And unifiying the general theory of relativity with it. A piece of cake, that.


Footprints sez "...you clearly ignored the question, which is the title of the node: WHAT IF...?", which is true. This often happens to me while ranting, in that I am too busy wiping the foam off my mouth to answer the question. So here's the answer.

It wouldn't make one bit of difference. It COULDN'T make one bit of difference. The whole point is to claim that at some unimaginably minute scale, our Universe is a cellular automaton. But -- since it's just a universal cellular automaton -- it could equally well have been any other universal machine with appropriate initial conditions, and we'd be none the wiser.