Re: The Copy Paradox

wolfkin@ldl.net
Thu, 13 Nov 1997 22:42:14 +0000


> Date: Thu, 13 Nov 1997 13:51:31 -0800
> From: Hal Finney <hal@rain.org>

> Wolfkin, wolfkin@ldl.net, writes:
> > I still see a difference between stopping and starting the *same*
> > program (which is held in memory until restart),
>
> Whoa. You want the program to be HELD IN MEMORY? What happens if it
> gets swapped to disk? Do you think silicon transistors are better at
> retaining identity than the magnetic disk surface?

Of course not. The problem (for me) is the question of whether the
pattern which is me (we agree there) is changed significantly during
the process, EVEN IF only temporarily. An identical pattern will end
up being started again in a few minutes, but THIS one will be copied
onto disk and then destroyed. The difference is that the one held in
memory is EXACTLY the same when restarted, while one that's been
copied to disk is an identical simulacrum.

> Suppose you were running as a program, on a time sharing system, and
> you're cool about it. "No problem," you say, "since I'm being held
> in memory between run slices." "Hate to tell you this," I say, "but
> actually this OS swaps you to disk between runs. You've been swapped
> in and out a dozen times during the last three seconds." What do
> you say? How do you react? Do you believe that you've been constantly
> dying?

Several people will have died, but none of them is me. I may even
fail to be very mournful of those people, but I will certainly take
steps (if possible) to put a stop to MY impending death.

> > and starting
> > *another* program while stopping the first permanently. We often
> > think of two copies of a program as being 'the same' when in fact
> > they have features that differ (location, frex).
>
> Now suppose you're running on a network. "No problem," you say, "since
> I'm 'the same' program each time because I have the same location."
> "Hate to tell you this," I say, "but that network is shared with a few
> hundred other programs, and the set of processors you're running on
> is constantly changing. Sometimes you're using these processors here,
> sometimes those over there, sometimes it even pulls in some units from a
> loaner system at the local data bank. Your location has shifted 100
> times since we've been talking." What do you say? How do you react?
> Do you consider that you've been constantly dying with each shift in
> location?

Same as above. A person might be able to cope with this (certainly
some of you would have no problem with it), but I think the only way
that I could deal with it would be to give up caring about my
personal survival altogether. That is, by making the same shift in
priorities that almost everyone eventually does, caring more about
the continuing existance of their children, their works of art, or
ideas, than about identity survival.

> > I would say that such a person has participated only once. :) The
> > fact that he has memories which are taken from previous people
> > doesn't mean that he *is* those people.
>
> Yes, yes, one of the ground rules is that we're avoiding saying that
> A *is* B. All I meant was that he has memories of having gone through
> this experience multiple times.

Let me make myself more clear. Assume for a moment that we become
able to transfer memories from person to person, but not whole
personalities (yet). Say someone, is implanted with the memories of
ten or twelve skydivers who were all killed when their chutes failed
to open. Should he assume, then, that he is immune to sudden high-
speed contact with the ground? It matters very much whether we
assume he is or is not the same person as the ones who fell.

> > Assuming that someone who may exist in the future will think that
> > they are me doesn't really help me accept the idea that *I* should be
> > killed after the copy is made.
>
> Suppose you knew someone who was afraid to fall asleep, because he was
> convinced that every time he fell asleep he died, and a new version of
> him awoke in the morning - someone who had all the same memories, but
> a new person nonetheless. How would you try to persuade him that he was
> mistaken?

I do know someone (or used to) like that. Me. When I first was
convinced that the pattern of my brain was me, this bothered me quite
a bit. How large a change is necessary to produce a new person? I
have never found an answer to that, and finally lack of sleep led me
to put it out of my mind. :)

Wolfkin.