Thursday, February 27, 2014

Are We Living In a Simulation?

Dr. Nick Bostrom of Oxford says: probably. The basic argument is thus: Suppose that it is possible to make simulations rich enough to fool its inhabitants into thinking they are real. Then such simulations are probably common, given the ease of using computational power. So we are probably living in one rather than in a "real" world.

But really such arguments are specious. Bostrom tries to rigorize his argument by postulating that posthuman civilizations run "ancestor-simulations", but they might just as well run some other form of simulation, perhaps one in which only one observer (you, dear reader, of course!) is self aware, and everyone else is simulated just enough to continue the illusion. In which case his formula is significantly off. So his postulate #2 ("any posthuman civilization is extremely unlikely to run a significant number of simulations of their evolutionary history (or variations thereof)") is in fact quite likely to be true - or maybe it is, how could we know? But once that one is true, the others' truth values become unknown. In fact, in any scenario in which #2 is not followed precisely, Bostrom's math stops working.

This is often the problem with anthropic-principle arguments, and this one is no exception. You might be living in a simulation, but this paper is not a convincing argument.

1 comment:

  1. Like solipsism, his idea is essentially impossible to disprove. And, whether it is true or not should make little difference to us -- if it's true, we wouldn't know it, and would not have control over what happens. We can't do anything except continue to live as if this was the real world....which it probably is.

    At first I thought he meant that, as living biological beings, we were living in a simulation, but he is instead proposing that we are sentient computer programs. This is harder to argue against. The first, living being put into a simulation, require additional steps of memory replacement, and additionally require the simulation builders to be either immoral or amoral.

    However, sentient programs could be created in the simulation, rather than put into it, and so could easily have manufactured memories that would not conflict with the simulated world around them. And creating sentient computer programs would be at least less immoral in degree than say, abducting sentient beings and wiring them into a simulation against their will.

    Overall though, I think he's watched the Matrix one too many times.

    As for running a "significant number of simulations of their evolutionary history (or variations thereof)", I think that is actually very likely, although not necessarily for historical or scientific purposes. I think with such computational power, they'd be running lots of them for entertainment, and they would not need to be historically or scientifically accurate. As evidence for this argument, I point to.....Skyrim, and all the other hundred of video games out there, being run thousands or millions of times each day. Contrast that with the number of scientific simulations.