Re: New website: The Simulation Argument

From: Nick Bostrom (
Date: Sun Dec 02 2001 - 11:56:49 MST

Eliezer wrote:

>I can't say that I recall ancestral simulations playing any
>significant part in most transhumanist scenarios. Personally I always
>thought it would be a horrible thing to do even if possible and
>permitted. A Friendly SI outcome does not allow for nonconsensual
>simulations, and most ancestral simulations would presumably fall into
>that category.
>Of course, to some extent this argument is disingenuous, since the
>Bayesian argument requires only a very small probability of an SI that
>runs lots of ancestor simulations for the simulated ancestor worlds to
>outnumber real ones. I'm just saying that "posthumans running ancestor
>simulations" is not necessarily the most probable projected outcome, in
>either my own model or transhumanism in general.

That's good, but as you say, the scenarios that are called into question
are not only those where a civilization has large probability of some day
running ancestor-simulation, but also scenarios where a civilization has
any significant chance (even a very tiny one) of doing that ( - all
assuming we are not in a simulation, of course). So this rules out all
those scenarios we hear about people with human-like motives setting off
into space and colonizing planets; it also rules out all scenarios where
civilization develops along a "noisy" in a rather haphazard fashion. For
example, I think most of the scenarios that arch-transhumanist Anders
Sandberg is exploring in his role playing games are now refuted.

Some Singleton-hypotheses are among the survivors, including many versions
of Friendly AI, so the paper could be taken to add some credence to these

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 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT