Re: all of me or one of me

From: Wei Dai <weidai.domain.name.hidden>
Date: Thu, 1 Apr 1999 22:43:24 -0800

On Thu, Apr 01, 1999 at 06:36:31PM -0800, hal.domain.name.hidden wrote:
> Wouldn't the situation most relevant to the question of identity be one
> where the two people in the room are both copies of you? Where you can
> know that each would do exactly the same thing?

The reason I didn't use a scenario with copies is because the copies might
care about each other as much as they care about themselves, which would
complicate the situation. In my scenario it's clearer that the two players
are not altrustic toward each other. I think we can abstract away the
problem that they won't have exactly identical memories and experiences
and assume that the temporary amnesia treatment actually makes them
completely identical.

> If you knew that each would do the same thing, I think you would push A.
> As I recall, if Hofstadter set up different scenarios to try to get
> people to push A. The other fellow is your identical twin, etc. When he
> finally got to where the "other player" was just the player in a mirror,
> then finally people would push A. With two instances of the same person,
> I think it would be as certain as myself in the mirror.

It is not necessary to invoke "superrationality" in order to get the
result that both players should press A. If each player cannot tell which
player he is in the game, conventional rationality in the form of decision
theory actually tells him to press A. Unlike Hofstadter, who thinks the
players SHOULD press A, I think they SHOULDN'T, and therefore decision
theory must be incomplete. So basicly I'm making a similar point as
Hofstadter, but in the opposite direction.
Received on Thu Apr 01 1999 - 22:44:23 PST