|
Most digital physics proposals seem to be unfalsifiable, but there are exceptions. For example, statistical Physics tells us that a system can be full of complex, random fluctuations, but it can still be 'stable' if those fluctuations cancel each other out. Boltzmann hypothesised that we could just be a random fluctuation in some otherwise stable system (perhaps a vast cloud of gas). This hypothesis is falsifiable, since large fluctuations are far less likely than small fluctuations. It's very unlikely that the room I'm in, including me, is a fluctuation in a gas cloud, but the anthropic principle says that I wouldn't notice all those small fluctuations which don't produce me. If I am a fluctuation then I can predict that it's incredibly likely that the fluctuation is limited to this room, as anything larger would be far less likely. However, when I step outside I find a whole city, which falsifies the hypothesis. Likewise we can observe a whole planet, solar system, galaxy, local group, cosmic web, etc. which goes exactly against the predictions of the fluctuation hypothesis. However, from a digital physics perspective we can get the opposite result. Let's hypothesise that we're running on a giant Turing Machine and our program is a random fluctuation on its tape; ie. a random series of bit flips on an otherwise empty tape. Since small fluctuations are vastly more likely than large fluctuations, we would expect to be part of a small program rather than a large one. Again, the anthropic principle says that I'll never observe tiny programs that don't produce me, but what can I say about tiny programs which do produce me? Well, randomness and asymmetry is hard to produce using a computer program: it must be encoded as part of the program, since it can't be produced spontaneously. Hence I would predict small programs to have less randomness and asymmetry than large programs, so I would predict a symmetric and uniform Universe, which is largely what we see. One falsifiable prediction of such a digital physics theory is that quantum phenomena are actually pseudorandom, ie. deterministic and predictable, since the only way to encode unpredictable values in a program is to write them out bit-for-bit in the source. A pseudorandom number generator would require far fewer bits, and hence is more likely; also, a smaller pseudorandom number generator is more likely than a large one. If we find that quantum phenomena cannot be predicted by any short program, we can falsify this hypothesis. |