# [extropy-chat] Probability of identity - solution?

Russell Wallace russell.wallace at gmail.com
Fri Oct 13 04:37:15 UTC 2006

```(For anyone who hasn't read my earlier post: suppose you're copied into 2
copies, A and B, then B is copied into 999, should you subjectively expect
to have a 1/2 probability of finding "yourself" as A, as intuition and
causal logic would suggest, or 1/1000, as measure accounting would suggest?)

I think I may have the solution now.

It depends on what you choose as your reference class. "I" after all really
refers to oneself _at the present time_. In practice for various reasons
evolution not least among them we tend to extent that to one's future self
also, but that's a rather arbitrary choice. Most of the time that doesn't
matter because it's the obvious one (though note that Ethos can be taken as
a form of extension of the reference class). The paradox can be taken as an
exception to "most of the time".

Specifically: if you choose to define your reference class by causal logic,
then you get the causal logic conclusion. If you choose to define your
reference class by measure accounting, then you get the measure accounting
conclusion. If you want to know which "really" defines you - then the answer
is, you'll "really" have died a second from now anyway, because "yourself"
then will not be the same entity as "you" now (scare quotes because we're
voiding the warranty on the words in question, but you get the idea). So
decide what you care about, and aim for that. (This meshes nicely with the
way the measure accounting viewpoint is consistent with morality - it gives
the same policy decisions one would get if the 999 B's were randomly
selected from your fellow humans rather than being copies of you.)

And this is the kind of explanation one was looking for - in the absence of
an undercutting defeater for one viewpoint against the other, one would
philosophically expect different results to be explained by different
premises. I think this one is satisfactory, though I'm open to
counterarguments.
-------------- next part --------------
An HTML attachment was scrubbed...