Re: Quantum Theory of Immortality - research proposal

From: Hal Finney <hal.domain.name.hidden>
Date: Mon, 16 Nov 1998 18:25:58 -0800

Jacques M. Mallah, <jqm1584.domain.name.hidden>, writes:
> On Mon, 16 Nov 1998, Higgo James wrote:
> > Does the 'many-worlds' interpretation of quantum mechanics
> > imply immortality?
>
> No, it sure doesn't. While in some worlds you survive, the total
> measure associated with all copies of you is clearly reduced by a factor
> equal to what your 'chance of survival' appears to be. For all practical
> purposes, that is the same as if there were, for example, 100 copies of
> you and I go out and kill 50. You are 50% likely to be killed.

I would say, on the other hand, that if you have 100 IDENTICAL copies of
yourself, and kill 50 of them, you are still alive. There was really only
one consciousness, instantiated in parallel across 100 brains.

Consider an AI system running on a computer, and suppose it is conscious
and intelligent. Run exactly the same program on 100 computers, in lock
step. Shut down 50 of them. Does the AI feel anything? I would say not.

Suppose the 100 computers are connected by wires, so that each point on
each internal circuit is electrically connected to the corresponding
points on all the other 99 computers. Then you really have just
one computer, with internal redundancy. Remove 50 of the computers,
decreasing the redundancy by 50%. Does the AI feel anything? I would
say not.

For people who would give different answers in the two cases, let the
wires be variable resistances, so that we can smoothly vary between
the two cases - with perfect resistance we have the first case, with
perfect conduction we have the second case. Now we can vary smoothly
between the cases and get two different answers. It does not seem
reasonable that there should be an objectively definable threshold
resistance which splits the consciousness of the AI.

If someone wants to answer "yes" to both cases, I can construct more,
leading down to a computer with double-width metal traces and transistors,
and ask whether an AI run on such a wide computer would feel different
from an AI running on a thin computer. Surely everyone would agree that
the answer here is "no." Then I can split the traces with a variable
resistance and reproduce the cases above.

Hal Finney
hal.domain.name.hidden
Received on Mon Nov 16 1998 - 17:52:06 PST