Saturday 27 April 2013

Why We're Probably Real, or at Least Simulated

     I have always been interested in games and simulations. Most recently I've been playing around with developing a set of self-imposed rules to simulate the growth of villages and cities in Minecraft, but I've been at it since before I got that Apple ][ for my sixteenth birthday. Naturally, when I first encountered Conway's Game of Life, I was intrigued with the possibility that our universe itself might be a vast cellular automata system, in which the various subatomic particles were essentially multidimensional equivalents of Life's glider guns and other emergent phenomena.
     Well, last week, a friend forwarded me a link to this article on the question of whether or not we're living in a big computer simulation of reality, and ways we might be able to detect if this was so, essentially by looking for artifacts of the limitations of whatever computing device is running the simulation.
      I haven't read the original academic sources, and most of the math would probably be over my head anyway, but my impression is that what they're trying to do is similar to something I described once in an article about conducting science in a roleplaying game. Imagine you are a character in a game, where certain aspects of the gameworld are simulated on a roll of two six-sided dice. If you kept careful records of these sorts of events, over time you might find that their probabilities tended to converge towards certain multiples of 1/36, those corresponding to the odds of a given sum. You might find this an odd coincidence, and while you might not immediately conclude that it was actually a literal roll of the dice that determined the outcomes in your world, it would be evidence for some such explanation. Maybe God really does play dice with the universe.

     So it seems to me like a splendid idea for scientists to look for telltale patterns of computational artifacts in our universe, and it would certainly be a momentous discovery to find one. Still, it seems to me it would be amazingly difficult to imagine just what sort of a machine could even be running this simulation, much less make meaningful guesses as what kind of computational artifacts to look for. It's probably running on something a little more powerful than my old Apple ][, but not necessarily. See, in principle, any Turing machine could run any computable simulation, given sufficient time and memory. Sure, it would take a long time, but not for us inside the simulation. We would have no way to measure the passage of time between "frames" of the movie, trapped as we are within those frames. Looking at the text on this page now, there is no way for you to tell how many seconds, hours or days passed between my typing the letters of this word.

     But then, why does it actually need to be computed in the first place? We often speak of mathematical facts as having a kind of independent existence, regardless of whether or not any human happens to know them. For example, there is a fact of the matter as to what the googolth digit of pi is, even though no one is ever likely to calculate it.
     Now consider the universe of a computer simulation. It, like pi, is really just a vast computational problem. All of the macroscopic phenomena we might experience within it (including the capacity to experience and reflect on phenomena) would be simply emergent properties, like the glider guns and more complex structures of Life, possibly detectable to someone watching the monitor, so to speak, but possibly just lost in the noise, but nonetheless intrinsically present in the data itself.
     So whether or not our universe simulation is ever actually executed, there still would be a fact of the matter as to what you're thinking about right now, locked within the complete mathematical description of this universe. So again, why even postulate some computer simulation, when we could simply be an emergent property of an unimaginably vast number? That may be all the existence we have or need. We may have no more "real" solid existence than the googolth digit of pi.

     I believe there is an answer. For one thing, if we are just numbers, then we are surely not alone. All mathematically possible universes share the same degree of existence. Of course, this means that infinitely many parallel universes "exist", but infinitely many more chaotic messes of data with no intelligible emergent properties to speak of. The fact that we find ourselves in one of the vanishingly small fraction of number sets that includes the emergent property of consciousness is kind of astonishing, but it shouldn't be, thanks to the anthropic principle; while it's improbable that a universe picked at random will include someone wondering about the improbability of consciousness is, it's a certainty that someone wondering about the improbability of consciousness will be in such a universe.
     But there are infinitely many possible wondering consciousnesses, and like the infinitely many universes, most of them may be pretty much random. That is, the subjective experience of a consciousness may just be there, but with little more coherence than that. Most will be nothing like the subjective experience we have here, of a universe that appears to follow coherent laws of physics, where principles of causality seem to hold sway, where there may be mysteries and apparent paradoxes but they fairly consistently tend to be resolved within the rules.
     So consider: of all the infinitely many crazy chaotic universes in which conscious could be randomly encoded, we seem to be in one where there are impressively consistent regularities. The odds of being in such a universe are submicroscopically tiny, and yet here we are. The anthropic principle doesn't account for why these regularities should exist wherever we find ourselves. So it's reasonable to look or some sort of explanation for the apparent order we perceive in our universe.
     One such explanation is that these emergent properties of planets, solar systems, trees and consciousness really do need to be instantiated in some way in order to become real. In other words, there has to be some sort of physical, mechanical encoding of the data, which by obeying its mechanical or computational rules, generates an orderliness to the pattern.

     And that's why I'm satisfied that we must either be in a "real" universe, or at least running as an emergent property of a simulation on a machine in some "real" universe. Not that anyone was really worrying too much about whether or not we were just a matrix of numbers.