From: Vladimir Nesov (robotact@gmail.com)
Date: Sun Mar 16 2008 - 02:27:01 MDT
On Sun, Mar 16, 2008 at 1:54 AM, Jeff L Jones <jeff@spoonless.net> wrote:
> Exactly. I think the only reason John Clark and Vladimir Nesov are
> coming to idiotic conclusions here is because they think that
> anticipation is subjective or tied in with consciousness or identity.
> While they can come up with whatever bizarre definition for identity
> or consciousness they like, it really doesn't tell you anything
> meaningful, because they are only talking about something subjective
> (viewing someone as an "adjective" for instance). You can believe
> whatever you want about identity, since at heart it's not really a
> meaningful concept, and neither is subjective anticipation. But you
> *cannot* believe whatever you want about objective anticipation.
> Anticipating the wrong thing is going to cause most of your copies to
> behave the wrong way.
>
I think that answer is 50% because what happens to observers doesn't
matter, as long as they have the same information about this question,
and information that reasoning engine has is what allows it to
calculate probabilities. It's precisely attempt to do the math, to
formulate the problem so as to assess the right number to base actions
of agents on, that changed my mind about the answer. Common intuition
in this case seems to be wrong. My error was to immediately accept the
answer I intuited, 100/101, and then to rationalize the reason for
that answer to be right.
'Subjectivity' doesn't matter either. When you observe the
environment, the body that you happen to inhabit is just one more
feature of said environment, it is only relevant as a functioning
reasoning engine that also must come in contact with the territory to
obtain information. If the question doesn't concern the body and
information already resides in it, it doesn't matter how this body is
organized or transformed. It might be a 100-headed borg, it might be a
single person. If information available to me about the world says
that I'm ignorant about coin flip, I asses probability of it coming up
heads as 50%.
If my copy that might have observed "heads" is destroyed with
probability 90% instead of being copied 100 times, it reformulates the
question. Now the question is not just about the coin "if the coin
came up heads", it's "if the coin that you will observe came up heads"
that includes parts about the coin, and parts about agent body. In
this case what happened to the body does matter, it's a part of the
territory that question addresses. If body is destroyed with
probability 90% in event of coin coming up "heads", only 55% of
original probability mass corresponds to situations where there is a
body. The fact that I'm doing the reasoning tells me that a surviving
body is on the territory, so the world is in those 55% or original
probability mass. Given this fact, I conclude that probability of coin
coming up "heads" is 1/11.
-- Vladimir Nesov robotact@gmail.com
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT