From: Hal Finney (hal@rain.org)
Date: Wed Feb 12 1997 - 10:20:42 MST
From: Anders Sandberg <nv91-asa@nada.kth.se>
> A bit like what happens in _Permutation City_, but remember what Durham
> did.
This is a version of that scenario that I could accept. I don't remember
if this is exactly how it was done in the book, but it was something
like this. I am an upload, so it is pretty easy to copy myself.
I make a copy which is intended to live for only a moment. The copy is
distinguished from me in that it starts out in a world where there is
a sign saying "you are the copy". It sees that for less than a second,
then it is instantly terminated.
It would take some courage for me to set up this situation, and some
people would be unwilling to do it. The copy is going to *die*.
However since it will only have a fraction of a second of different
experiences from the other instance, not much is lost. I have convinced
myself that when I run the experiment and push the button, if I see the
sign appear saying "you are the copy", I am willing to accept that my
existence will terminate a moment later, knowing that another version
of me will go on. I'll squeeze my eyes shut, cringe, and wait for the
inevitable, knowing that it will be quick and painless.
According to the Moravec/Egan theory, when I run this experiment, if
I do find myself as the copy, I will not find my existence terminated.
After a few seconds, I will open my eyes disbelievingly. The sign is
still there saying that I am the copy, but my program is still running.
Something has gone wrong. Maybe there was a glitch in the program which
was supposed to terminate me, maybe the other instance of me changed
his mind at the last minute and cancelled the termination, maybe, as
in Egan's story, I learn that the whole thing was an insane fantasy and
now I can begin the long struggle back towards sanity.
What has happened is this. There are an infinite number of possible
universes, but only a finite number of possible mental states for me.
That means that for any mental state, there are an infinite number of
instances of myself experiencing that state. When something happens
which has a random component (as everything does), some of those instances
experience it one way and some experience it another.
If something happens to kill some of them, others will survive, because
although their mental states were the same, some other aspects of the
universe may have been different; or because the random event which
killed some did not happen that way in other universes. No matter
what happens, some, an infinite number in fact, will survive.
It follows that everyone who has ever died in fact still lives. To their
surprise, something happened which allowed their existence to continue.
I don't really know how the cavemen have managed to live for thousands of
years; maybe it turned out that aliens were watching them and swooped down
in the moment of death to rescue them, or maybe they were resurrected
out of a Tipler simulation. Generally we can expect to witness the least
improbable event which allows us to stay alive.
This leads to the well known test of the theory, which is to try killing
yourself and see if it works. But this has the drawback that if the
theory is wrong, you are dead. That's why I like the uploading scenario,
since that is one circumstance in which I might be willing to try
experiencing death.
Of course, if I run the experiment and end up as the original, then I
don't know the answer. I see the copy die. Did it go on living in
some other universe? There is no way to know. I could run the experiment
multiple times, so that the chances that I will end up as the copy sometime
become larger (in some sense). But there will always be an instance which
has never experienced being the copy, so it's not clear whether to think
of this as helping answer the question.
It's also worth noting that Everett's many-worlds interpretation of QM
makes roughly the same predictions, so you don't have to swallow all
the baggage of platonic materialism. You can still hope that reports
of your death will be "greatly exaggerated."
Hal
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:44:10 MST