Re: New website: The Simulation Argument

From: Bryan Moss (
Date: Mon Dec 03 2001 - 20:54:05 MST

Eliezer S. Yudkowsky wrote:

> > There's even a hint of an argument against simulations
> > here: if your physics allows you to state the exact
> > configuration of some part of the universe at a particular
> > time, then there's the chance that you're living in a
> > simulation, therefore you can't know that your physics is
> > correct -- PARADOX!
> No, not paradox. It simply means that you go on simulating
> the ancestors without being certain that your simulation
> uses the real laws of physics.

True. It might, I think, be expedient of physicists to make
"substrate dependence" a basic postulate of any new
fundamental physics. It wouldn't shock me given the current
crackpot-esque state of modern physics.

> Or if you can't get back the exact world-state, then you run
> a series of successive approximations.

How? Do we create an approximation, let it run to the present
moment, and then compare...? Seems unlikely. Even with the
precise, correct "world-state" there is no way of knowing it
is indeed the precise, correct "world-state." Retrodiction is
a bit difficult to confirm. I'd argue that we can rule out
any scientific uses of "ancestor-simulation," although that
really gets us nowhere what with all those potentially
mischievous posthumans running around. (Boundless possibility
rears its ugly head yet again.)

> Or for that matter the bizarre purpose behind the simulation
> could simply require some generic ancestorlike situations,
> not the reduplication of specific past history.

Perhaps (and "bizarre" is the word). This, of course, assumes
that there *are* "generic ancestor-like situation," which
there may very well not be at this level of "granularity." As
always, the whole thing might be designed to fool me (and
maybe you and a few others), for some perverse reason, and
pulls all sorts of dastardly tricks to keep us in check, but
such speculation is an exercise in the pointless.


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT