Re: Simulation argument in the NY Times

From: Gwern Branwen (
Date: Wed Aug 15 2007 - 23:38:26 MDT

On 0, Norman Noman <> scribbled:
> On 8/15/07, Henry Wolf VII <[1]> wrote:
> The problem with those assumptions is that you only have access to one
> game. There are still plenty of 2D games available. If you only have
> access to that one 2D game, you would assume that there were only two
> dimensions and time. This ignores that there may be many more that we
> can't yet perceive. Looking back fifteen years, you would assume that
> the universe only has sixteen colors.
> Most 2D games still have the APPEARANCE of a third dimension, and most 3D
> games have action confined to a surface. It's fairly clear in both cases
> that they are made to be played on a 2D screen, and thus by 3D people.
> Our one instance in this "universe" is nowhere near enough to
> extrapolate accurately to an outside universe. If this is a simulation,
> the beings that created it would be post-singularity level beings. Due
> to the nature of the singularity, their existence is currently
> impossible to describe.
> It seems to me the very fact that they are "post-singularity level beings"
> is the beginning of a description. If indeed they are, then many more
> predictions might be made, because while a post-singularity entity could
> be and do almost anything, what it WOULD do depends on the original
> optimization target of whoever orchestrated the singularity, and whether
> something went wrong and that target became corrupted.

Seems to me that even using 'beings' is stretching matters. I've been thinking about this and trying to apply the anthropic principle.

If we are not in a simulation, then the universe seems to be computable in general (per our observations). If we are in a simulation, then whatever universe is outside the simulation must at least allow for the simulation to run (otherwise we would not be making observations of computable phenomena), and so at least allows of computable phenomena. This is not to say that the outside universe must be computable in general as well! I'm not sure what a universe that mixed incomputable with computable stuff would be like - stranger than you or I could imagine, probably - and so this isn't really a useful argument.

But I don't think this is useless speculation. I've noticed a lot of suggestion that the simulations could be recursive, but one could argue that it's unlikely that there will be nested total simulations of the universe. If simulating an entire universe requires properties stemming from being uncomputable (say being able to carry out an infinite amount of computation in a finite time?), then obviously any simulation wouldn't be able to conduct its own total simulation of a universe. Or if they did, inhabitants of *that* simulation would discover limits to the simulation set by the conventional and finite resources available to the 1st simulation's inhabitants.

rusers Secert 888 SAW toffee Becker sigint or LRTS charges

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:58 MDT