Re: ESSAY: How to deter a rogue AI by using your first-mover advantage

From: Stathis Papaioannou (
Date: Tue Aug 28 2007 - 05:19:38 MDT

On 28/08/07, Vladimir Nesov <> wrote:

> It's an interesting point. Illustrates how 'counting the simulations'
> is useless. Say, in universe A there is a simulation of universe B,
> and in universe B there is a simulation of universe C. How many
> simulations of universe C are there? One can say that C is also simulated
> in universe A, so it doubles 'number of times being simulated' for C.
> What is the contribution of indirection?
> Say, implementation of simulation of universe B in universe A consists
> of two devices which communicate with each other. Within each device,
> the same pattern content can be simultaneously viewed as being
> represented (and hence simulated) on two levels: say, pattern of
> communication between processing devices and storage devices and
> patterns within processing devices. So, each half of universe B is
> effectively simulated twice, so whole B is simulated 4 times. If
> simulator consists of N parts, it makes 2^N simulations.
> I see solution to this debacle in:
> 1) same level of existence of all possible universes (one can say,
> simulated by enumerating turing machine, but it should be unnecessary)
> as mathematical abstractions;
> 2) irrelevance of number of simulations/identical universes; effectively
> from subjective POV substructures exist simultaneously in all simulations
> that contain them.

All simulations which provide an identical 1st person POV are
equivalent, and in a significant sense redundant, because it is
impossible even in principle to know which one you are in. If the
number of simulations were suddenly increased or decreased, it would
be impossible for you to know that anything unusual had happened: your
next subjective moment will be in one and only one simulation, any of
the simulations will do as that one simulation, so provided there is
at least one simulation available to choose from you can never know
what's "really" going on.

However, if there are two or more competing "next moments" then the
number of simulations is relevant. If there are X simulations in which
you are tortured and Y simulations in which you are not tortured in
the next moment, then you have a X/(X+Y) chance of being tortured.

Stathis Papaioannou

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