RE: New website: The Simulation Argument

From: Nick Bostrom (
Date: Mon Dec 03 2001 - 16:56:34 MST

Emlyn wrote:

>I imagine that you mean the following:
>We can, however, conclude that one na´ve transhumanist dogma (that humanity
>will achieve a post-human status and run ancestor simulations) is false,
>unless we are now living in such a simulation.
>Does that make sense?


>Oh, and by the way, no I didn't think I was running in a sim, until recently
>maybe. I came across this argument about nested simulations recently (or
>fully digested it, anyway), and it's intrigued me, I must admit... possibly
>the best argument for the existence of "god" (in the broadest possible
>sense) that I've ever come across. That's not fun given that I am an atheist

Moving to a better-considered could be fun!

(I'm a non-religious agnostic, if anybody wondered.)

>One piece of the argument which troubles me is this:
>"Since the experiences that an observer has if she is living in a simulation
>are indistinguishable from those she has if she is living in unmediated
>physical reality, it follows from a very weak form of the principle of
>indifference that the probability of her living in a simulation equals the
>fraction of observers that live in simulations[5]. "
>This seems to take it for granted that an entity living in the "real world"
>(at the root) and an entity living inside any level of sim are equivalent in
>terms of what I might call "being", for wont of a better term. ie: there is
>no significant difference between a simulated being and an actual being.
>While this may be the case, it does assume a strong materialist view of
>consciousness. This must be the case for the principle of indifference to
>apply. That's probably ok, but maybe it needs to be stated explicitly?

Yes, I call it substrate-independence. I don't argue for this assumption in
the paper but I note that it enjoys fairly wide support (among those doing
the philosophy of mind) and it is generally assumed in transhumanist
discussions about uploading etc.

>A couple of other points:
>"Reality may thus contain many levels. Even if it is necessary for the
>hierarchy to bottom out at some stage - the metaphysical status of this
>claim is somewhat obscure - there may be room for a large number of levels
>of reality, and the number could be increasing over time. "
>Speaking about change over time in the overall hierarchy of simulations is a
>bit dangerous. While it might be true that there is a loose "arrow of time"
>across the whole set of simulations/universes (simuverses?), with the
>hierarchy defining dependencies, it is still the case that time, as part of
>the notion of space/time, is local to each given simulation... if it is
>sensibly defined in all. To give a very basic example, it is entirely
>possible that a given sim might be run backwards, if the universes' creators
>had reversed the physical laws, such that the local experience of time in
>the inner sim was reversed when compared to the outer sim.

The above remark presumes the case where the hierarchy bottoms out, and the
time-direction referred to is then simply the direction of time's arrow in
the physical basement universe at this bottom level.

>I would have thought that, if you are arguing that the probability that we
>are in a sim approximates 1,

No, I don't argue that. (I'd actually put the probability to less than 30%.)

> you are likewise arguing that the number of
>sims approximates infinity (or some slightly more sensible statement), and
>does so right now (in whatever reference you draw "now" from).

If there were a rapidly increasing

>"(One consideration that counts against the multi-level hypothesis is that
>the computational costs for the basement-level simulators would be very
>great. Simulating even a single posthuman civilization might be too
>expensive. If so, then we should expect our simulation to be terminated when
>we are about to reach the posthuman stage.)"
>This makes major assumptions about the physics of computation outside of our
>universe, which are entirely ungrounded in my opinion. It may be that it is
>more common for infinite computing capacity to be possible in sims than not.

It counts against the multi-level hypothesis because we don't know that
computation is (nearly) free in the basement-universe.

>Alternatively, it does not take into account the possibility that the sim
>would be rigged to slow down, relative to the outer sim, to make up for lack
>of concurrent execution resources. As the relative time frames became
>infinitely divergent (the inner sim approaching stasis with respect to the
>outer sim), the inner sim would approach infinite computational ability
>*with respect to its own time frame".

The simulators may, however, decide to shut it down when the pace in their
world of our simulation gets too slow so that they don't see anything
interesting happening anymore.

>Thanks for a thought provoking paper, Nick.

Thanks, Emlyn!

Nick Bostrom
Department of Philosophy, Yale University
New Haven, CT 06520 | Phone: (203) 432-1663 | Fax: (203) 432-7950

This archive was generated by hypermail 2b30 : Sat May 11 2002 - 17:44:24 MDT