[extropy-chat] Probability of identity - solution?

Lee Corbin lcorbin at rawbw.com
Mon Oct 16 01:56:24 UTC 2006

Robin writes

>>What is wrong in duplication (or forking) experiments is to suppose that
>>there is only a probability that the bad outcome will occur. We *know*
>>that both outcomes occur. Therefore probability is an atrocious way to
>>approach the problem. Probability can be used, as I say, only for
>>planning purposes in that one might have to do something awkward
>>in the present circumstance while a lot of copies of one are being made.
> You seem to be rejecting the concept of indexical uncertainty, which seems
> to me to be valid and central to these situations.   Even when you know
> all of the physical details of a universe, you can still be uncertain 
> about which creature you are in such a universe.

Yes, especially if you are addressing a case in which there are real differences:
In some experiments, subjects remain identical over spatial distances. But in
either case, one may wonder if he is A, the duplicate that was to be created
near Alpha Centauri, or B, the duplicate that was to be created near
Betelgeuse.  However!  That is not crucial to your identity:  your identity
is independent of spatial location.   (Let me be clear: Robin Hanson lives
equally well both near Alpha Centauri and Betelgeuse.)

> An agent who has forked but has not yet found out which forked agent
> he is can sensibly have uncertainty about his identity.

He *can* be uncertain about his location.  About what else can he
be uncertain?

And are you defending a statement such as "with probability .6 I will be
at A and with probability .4 I'll be at B"?  I have said that I will defend
such a statement only for *planning* purposes---you need to be wearing
a raincoat if you know it's raining at one of the places.  But the truth is
that there is a 100% probability that you will be at A.


More information about the extropy-chat mailing list