RE: Subjective counterfactuals

hal@rain.org
Mon, 5 Apr 1999 12:17:54 -0700

Billy Brown, <bbrown@conemsco.com>, writes:
> If this simulation is based on a human (or even animal) mind, the two states
> will almost never be the same. Most of what happens in the mind is
> probabilistic, not mechanistic - so if you simulate the same 100ms of
> processing a million times, you will get a million slightly different end
> states.
>
> Now, I don't know whether this is a necessary property of a conscious mind
> or not, but it seems entirely plausible that it might be. If it is, then
> you need a Turing machine with a random number generator to run a mind as
> software, and the whole system has very different properties than an
> algorithmic program.

In that case, you could equip your TM with a deterministic random number generator. Cryptographic RNGs can be created very practically which are thought to be indistinguishable from true random numbers using even astronomical amounts of computing power. These results are not yet proven, but if you can distinguish the output of the "Blum Blum Shub" RNG from randomness, you can easily factor large numbers, which is thought to be impossible.

If this actually makes a difference in whether consciousness is created or not it would imply that consciousness is non-computational in nature. It would also imply the existence of zombies (beings which behave just like conscious beings but aren't actually conscious), since the resulting TM would act conscious (or else it somehow acts as an oracle for factoring huge numbers, which would certainly be unexpected).

A a deterministic RNG is just another computer program, so if a thus-augmented TM produced consciousness, then the replay paradoxes still apply. It does not appear that an appeal to randomness resolves the difficulties.

Hal