Saibal Mitra wrote:
>Maybe I didn't understand you correctly, but in
>
>http://www.escribe.com/science/theory/m1726.html
>
>you make the point that if a person is simulated, it doesn't matter how
>often that person is simulated, as long as the simulations are identical.
Well, I make that point inside a reasoning where I just conclude that it
is difficult to rely on intuition when you evaluate ``probabilities" in
self-multiplication experiences. That is why I try to extract the
probabilities or the ``credibilities" from the interview of the
self-referentially correct universal machine (and from the interview
of its "guardian angel" which exists through Solovay generalisation
of Godel and Lob incompleteness theorems).
>Assuming you are right, I intended to show that without any form of
>(quantum) suicide, you can change the probabilities of observing rare
>events. Maybe the above example was not so clear, so let's look at a slight
>variant of this.
>
> Suppose you want to observe some rare phenomena, such as proton-decay. You
>then place a computer next to the experimental set-up. The computer is
>simulating your entire life over and over again (the biological version of
>you is annihilated).The computer receives information about the experiment,
>but this will not influence the simulation unless evidence for proton decay
>is found. In that case you are made aware of the positive result.
>
>Now, all the simulations where you don't receive news about the experiment
>are equivalent and should (according to you) be counted once. The
>simulations where you do receive information are all different, because a
>positive result could come at any time during your simulated life.
>
>Therefore you would have to conclude that the probability of observing
>proton-decay is almost equal to 1.
You could aswell cryogenise me, and wake me up when the proton has
decayed.
But I am neutral or ignorant about the way we should or should'nt take
into
account the multiplied emulations of a layer of my subjective life. I
have
only positive evidence that in case the multiple lives bifurcate, and so
when they are relatively distinguishable, in that case indeed we should
take into account the multiple continuations for our expectancies.
My strategy is to modelize the first person sensible observer's certainty
of
an atomical proposition p by ``provable(p) and consistent(p)" where
`provable' is
the (arithmetical) godelian provability predicate. And consistent(p) is
an abbreviation of `not provable not(p)'. And I modelize UD* by the set
of
verifiable arithmetical propositions. This leads to sort of quantum
logics,
from which I expect the possibility of recovering quantum probabilities.
Bruno
Received on Sat Sep 30 2000 - 08:15:01 PDT