Are You Living in A Computer Simulation?
The comments below address this argument, which is forthcoming in Philosophical Quarterly. They were sent in an email to some people originally, but I thought others might be interested as well. Incidentally, I used to be on a couple of mailing lists where this guy and most of the people he lists in the acknowledgements used to discuss these kinds of things. It’s kind of an ad hominem, but it’s worth noting from the outset that most of them have an almost religious faith in their ability to get themselves made immortal via some technological means within their lifetime. Check web searches on “extropian” or “transhumanism” for more. I ended up unsubscribing both because I started to see holes in their logic and got tired of their almost unbounded optimism for the future. Anyway, regards the article, here are my thoughts.
1. The assumption of substrate-independence, while perhaps uncontroversial in some circles, is being shouted down by new approachments to cognitive science that emphasise embodiment as a crucial aspect of intelligence. These developments are sometimes referred to as second generation cognitive science since they reject the functionalist/symbolist assumptions of the thinkers who founded the field on the computer analogy. Anything by Andy Clark is a good introduction.
2. Even if it were possible to implement minds as simulations (which I doubt given the point above), simulating their world is an entirely different proposition. Daniel Dennett, and I can’t remember where but it’s in one of his more recent books (and not a journal article) argues for instance than given our current knowledge of the physical laws that constrain computation we couldn’t in fact simulate a world for a conscious subject with agency (and the agency is important, since it introduces the problem of dynamically feeding back the right information to the agent based on their actions), due simply to the fact that the kind of processing involved in generating a completely interactive world for a human is just astronomical to the point of unimaginable (he does the numbers). Nick talks about this in the paper but glosses over it with a comment about the massive power of our future computers…
3. When he says that the simulation need not behave like a human, does he mean the computer in which the simulation is run need not behave like a human? Well of course not. But the humans in the simulation need to pass Turing tests, because the claim is that we may be in a simulation, and we as a matter of course do pass Turing tests.
4. When he talks about re-running the simulation if there are mistakes, it makes me wonder what conception of time he is using. Or rather, what conception of consciousness. It seems strange that consciousness under the simulation model could be re-wound. This isn’t an argument of course, just an intuition that there might be an argument here. My own thoughts are that consciousness isn’t sufficiently well understood to be able to make the assumptions necessary for accepting the conclusions of the article, though.
5. The argument that such computers are in-principle possible is not an argument that we could build them, or more importantly write the software for them. Is it plausible that such a simulation is an achievable software project? Once again not an argument, just something to think about.
6. I think it’s funny that he gives the impression that if a meteor were about to hit earth, he could be found adjusting the probabilities in his model rather than, say, participating in the orgies out on the streets or whatever most people would be doing.
7. A more straightforwardly philosophical critique is the following. Why assume that it is future humans who developed the simulation of our world? If we are in a simulation, then the concept of our future selves is irrelevant metaphysically, since what “really” exists is outside of our entire simulation. We may as well assume that it is God running the simulation (he mentions something like this towards the end, but from the angle that we should consider the posthumans that are running our simulations as God. But I say, why consider the God(s) who created our simulations as posthumans?). Or we might assume that the simulation is an accident that happened when a bunch of Platonic-silicon fell off a Platonic-cliff in the Platonic-”Real World” and made a Platonic-computer. That is, if we concede the possibility of a simulation we must also concede there is nothing inside the simulation that could possibly tell us anything about what is outside the simulation (if we are rigorous that is, which I don’t think the article is, with the way it imports knowledge of what is inside our simulation to what is outside the simulation). That is, I don’t think the argument gets beyond Descartes, and I think that it falls to all of the arguments against Descartes. Actually I feel sufficiently confident about that that I might take it on as a summer project to write up the critique and mail it to Philosophical Quarterly. Another way of framing the fallacy is that the paper is assuming that by learning about our future, we thereby learn about our past.
8. The article does drift towards drivel as it moves along, in my opinion; it is filled with a lot of speculative nonsense that belongs in a science fiction novel, not a philosophy paper. Much of it could reduced to his comment “Reality may thus contain many levels.” and then the caveat that we can not in principle ever have knowledge of levels below our own, if such a concept even stands investigation. I did think his ethical speculations were interesting though. As an aside, the Doomsday Argument (Is the end of the world nigh? John Leslie. The Philosophical Quarterly Jan 1990 v40 n158 p65) he refers to looks interesting. Unfortunately there are no electronic versions available through the library here but I’m going to check it out when I’m next there…