Re: Infinite universe

From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Tue Apr 29 2003 - 16:31:38 MDT


Psy-Kosh wrote:
>
> It would seem to me that if Moravec/Level 4/Dust Theory/"whatever we
> want to call it at the moment" cosmology is valid then no action can
> have any effect except in one's own subjective reality, right? ie, if
> this model is valid, then wouldn't the best we could to would be to
> minimize the probability of it occuring in our own experiance, but
> having absolutely no effect on the totality of it?
>
> ie, any "action" I would take to try to reduce it would correspond to
> some state N.
>
> But all states N already exist. Be taking said action simply means
> that I am state N.
>
> State N is "always" itself though. ie, it isn't like "state N is not
> state N, and then it is." So no change actually occurs even though my
> subjective experiance is that there is change.
>
> What may be true is that the states that correspond to an experance
> of having state N part of a history leading up to it have a lower
> ratio of "evil states" compared to the ratio of total "bad states" to
> all states.

Think of objective reality as being embodied in the web of conditional
probabilities <a|b>. These are all that exist on the quantum level. Why
shouldn't they be all that exist on the subjective level as well?
Similarly, our actions can quite easily affect the relative probabilities
- ratios - of <a1|b> and <a2|b>. In others' lives, as well as our own.
Even if A1 and A2 and B are all supposedly "infinite". Even if they all
supposedly "exist".

You're not trying to make something "exist" or "not exist". You're trying
to shift probability from the bad outcome <a2|b> to the good outcome
<a1|b>. You're not trying to make A2 "not exist" or A1 "exist". They
both exist. You're trying to shift their relative measures, change the
flow of conditional probability from one to the other.

-- 
Eliezer S. Yudkowsky                          http://intelligence.org/
Research Fellow, Singularity Institute for Artificial Intelligence


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:42 MDT