[ExI] Midazolam, Memory Erasure, and Identity

[Lee Corbin]

Eliezer writes


> Lee Corbin wrote:
> 
>> For example, suppose that teleportation became the norm, only
>> it was "teleportation with delay".  That is, the original lived on for
>> another sixty seconds while it was confirmed that his remote 
>> duplicate had been successfully incarnated.   He would see his
>> newly created duplicate over closed circuit TV alive and well
>> at the remote destination, and his remote duplicate would see
>> him.
>> 
>> Now over time, the commuter who employed this teleportation
>> with delay would become completely accustomed to it working
>> just fine.  Each time he used it, he would appear at the destination
>> and see his original on the TV screen and wave.  But after a few
>> months of this, he would note the increasing astonishment apparent
>> on the original's face.  Why?  Because the original would have
>> remembered being on the *receiving* end numerous times, having
>> never experienced being on the to-be-disintegrated end. 
>> 
>> Even I would be astonished, I couldn't help it.  But *I* would not
>> be alarmed, just rather amused.  Because I have internalized that
>> tomorrow I will still wake up in the same bed, no matter what
>> trifling incidents happen to me today.   In fact, I would realize
>> that *I* was really there at the remote destination!  It would
>> simply be that the instance *here* was not collecting memories
>> of what was happening to me *there*.  Perfectly natural, given
>> the circumstances.
> 
> Suppose we permit the quantum theory of immortality, so that, in a 
> tiny fraction of worlds, a successor to the original-location commuter 
> would survive; the disintegrator would malfunction.  By QTI it would 
> feel like the line of subjective experience definitely continued there 

Yes, not only would it feel like (your nice phrase) "the line of subjective
experience" continued there, it would in fact continue there. Were the
miraculously rescued instance of the guy of the typical frame of mind
he'd exclaim "Wow!  What a close call!  This time instead of the true-me
getting teleported and the false-me getting disintegrated, it happened
the other way around!  Now I have to hunt down that varmit what
was teleported and work out the legal arrangements since there are
two of us."

> (and it still annoys me that I can think of no non-subjective 
> experiment which distinguishes between QTI and its theoretical 
> alternatives).
> 
> Would this change your mind about the whole thing being a case of 
> short-term memory loss?  Would you say that in this case they become 
> different people - different subjective lines?

In my conception of what a *person* is, the answer is no. They're
still the same person and would remain so for many years until enough
difference had accumulated to make them clearly different people
(much as in the same way you are not the four-year-old you once
"were").

The software analogy is perfect: people are like programs, and instances
are like, well, running instances.  And if some hacker makes some small
change to Linux, it's still Linux.  It is, that is, until so many hacks have
been made that it's really not Linux anymore.

> If so, what difference does some tiny infinitesimal fraction of worlds 
> make?  Why not say they are two people to begin with, and one line ends?
> 
> I just can't get rid of the apparently basic nature of the question 
> "What seems to happen next?"

I've always called that the horrid "Anticipation Dilemma".  At one point
in 1986 I couldn't even code because of it:  I'd pace the halls at work
trying to figure out exactly why I didn't have to anticipate with great
relish the dinner I'd had the night before!

(I wrote up a version of this and posted it here on Tuesday, April 10, 2007
at 11:26pm. For some reason, I see that it never made it into the Archives,
though I certainly got sent a copy.)

> I acknowledge this as probably indicating some kind of very basic
> flaw in my understanding,

I don't think so. It's just that *anticipation* cannot, so far as I know,
be put on an entirely consistent rational basis.

I will re-post the essay I wrote to the list in April.  Let me know if
you don't get it.

Lee