Re: Separate Copies Contribute Separately to One's Runtime

From: Mikko Rauhala (mjrauhal@cc.helsinki.fi)
Date: Sat Mar 08 2008 - 13:57:38 MST


la, 2008-03-08 kello 10:19 -0800, Lee Corbin kirjoitti:
> I claim that isomorphic copies separated enough in spacetime
> to constitute separate physical processes are separate subjects.

This is rather nonsensical to me.

> Suppose that there is an absolutely identical you being held
> in an Earth-like simulated city near Sirius. It ought to matter
> to you that one of them is going to shortly die.

You're using faulty terminology to gain psychological advantage for your
faulty assertion. ("One of them is going to shortly die" implies death
of a self in this context, which really isn't the case here; see further
down.)

Anyway, certainly it matters in that there'd be one less instance of me
(in this universe, anyway), and therefore less redundancy and
fault-tolerance. (Were one to go more abstract, also some of my
potential futures might be weeded out by that, but that happens all the
time anyways without it being deadly.)

An sich I see no reason to fret over one instance terminating.

> Yes, right now you are "both", i.e., you can't tell whether you're
> on Earth or near Sirius, (and I always claim that you actually are
> in both places at the same time).

Good, you're somewhat sane so far. However this puts at a rather odd
light your previous statement about "one of them shortly dying". Rather,
one would cease to be (instantiated) in one of the places one previously
was. This does not sync up with "death" in any way or form (to stretch a
bit, unless moving my arse to the kitchen kills me in the living room;
sure, you can say so, but it isn't a very practical definition nor does
it reflect the subjective experience). I suggest that this is where you
should reconcile your views a bit.

> But suppose that one atom of the Earth version is missing. Does that
> really matter? Surely there are not suddenly two people where just
> a second ago there was just one. Likewise, taking away more atoms,
> one of them will have to be damaged a great deal to cease being you.

One atom missing would presumably (chaos and all) diverge the copies
given some time. I don't claim to know _exactly_ where the line goes as
to when two similar processes have the same subjective existence and
when they don't. We can already see clearly where the line isn't crossed
(multiple identical digital error-resistant instances) and where it is
(noticable difference of output given same sensory input), the rest is
for future study.

Anyways, I generally suggest discussing digital error-resistant
processes and going by rather direct mathematical comparisons of the
instances; otherwise we just get into wanking over something we just
don't know how to properly quantify (at this point, anyway), and that
doesn't really lead anywhere productive.

Incidentally and slightly relevantly, as far as value judgements go, I
am of the opinion that if there are several highly similar but
nevertheless noticably different selves, it's as such a lesser loss if
one of them kicks the bucket than if they were a diverse group. (For
instance, also everyone transforming into orgasmatrons of one basic
architecture would be quite a loss, even if they didn't merge all the
way.)

This is also relevant for the concept of "local death" used by Egan in
at least Schild's Ladder; even a separate instance of you dying can be
shrugged off as little if anything more than a slight case of amnesia,
if you have a timely backup. Here one's own judgement should take
precedence; for instance, let's have a freak accident instantiate an
earlier snapshot while a more experienced one is still in play. Is it
suicide for it to terminate? I don't really think so, but regardless it
should be that instance's call. (I also reserve the right to make a
different call than I think I'd make in that situation :] )

Anyway, back to business.

> To return to Eliezer's hypothesis---sorry---instantly after the coin flip,
> the one who got heads is duplicated all over the place. If we stop
> right there, then *subjectively* you have to figure your odds at
> only 1:100 of seeing tails since

No, because there is only one subject. It is just in many places at the same
time, to use your own terminology :]

> and here is why: For clarity, consider that ten minutes later,
> 100 out of the 101 suddenly cease to exist. Then this is repeated,
> and again ten minutes later instantaneous disintegration comes to
> a random 100 of the 101, and this is repeated again and again.

Here you manage to demonstrate that if you terminate the heads-seeing
self with a probability of 1/101 and the tails-seeing self with a
probability of 100/101 you'll most often end up with a heads-seeing one.

I agree. :)

-- 
Mikko Rauhala   - mjr@iki.fi     - <URL:http://www.iki.fi/mjr/>
Transhumanist   - WTA member     - <URL:http://www.transhumanism.org/>
Singularitarian - SIAI supporter - <URL:http://www.intelligence.org/>


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT