From: Eugen Leitl (eugen@leitl.org)
Date: Sun Jun 16 2002 - 10:23:38 MDT
On Wed, 12 Jun 2002, Lee Corbin wrote:
> > I'm one of the [two] runs.
>
> I think that this is wrong. You are both of the runs. Duplicates
> are selves. I would say that you have no physical basis for saying
> that you are one of the runs and not the other.
This only applies if you're looking at this from the inside. An external
observer whose window of observation is not horizonted by the pressing of
the button event sees that one trajectory goes on, and the other
trajectory suddenly ends. There's clearly a difference between those two
outcomes, it's a fork (one prong of it length zero though). Not only that:
if you decide to terminate, and nothing happens, you know you're the first
run (because the first run runs always through regardless of what you
choose). You can never observe your nonexistance after the execution of
the decision. You cannot tell which branch of the run you are after after
you decide to not terminate. The OS can't tell you two different things,
because it would fork your state.
> > The entity may or may not listen.
>
> By hypothesis, the entity does not listen. We are talking
Life is uncertain. I can't know it in advance, and I refuse accepting
absolutes resulting on consequence of such magnitude on what I think I
know. Hey, I might just have a bad day, and overlook something trivial. I
think I'll be playing safe here, if if that costs me some game money (the
amount of resources available to me within the course of the simulation),
so be it. This is somewhat influenced by the amount of time I have to
decide, and the amount of game money I'm promised. (If there's *a lot* of
it, and there's years of time for me to mull over it, and I eventually
figure out something I think is safe the outcome may change).
> about runs (program executions) that from the Operating
> System's point of view are entirely deterministic. Your
> calculation (i.e. your experience) determines only one
> bit of output that the OS uses: namely whether or not
> to terminate the second execution mid-way through.
Because both runs are deterministic, only one outcome (to press or not to
press) is possible. That information is available at the first run. If
you're the second run, and you vote to terminate, you're doomed already,
regardless of what you do.
There's no way to tell, which of the runs is me. The probability is 50%,
right? The odds here seem to be lots worse than russian roulette, albeit
minus the blown out brains splattering the bitscape.
No, I think won't terminate.
> Quite true. If billions are offered, I'll consider
> aborting my second run. As you say, the total number of
Notice that this is game money, unless there are many beings caught up
with you in the simulation. Is the fate of these beings dependent on your
decision? Or are you just the one asked, and their worldlines continue
undisturbed, whether you're the one still part of the sim or no? This
looks more and more construed by the microsecond.
> copies of me executing (currently or in the past or future)
> can also logically affect my preferences. Also, as you say,
> (A) you could be surrounded by mere portrayals of people that
> the OS puts on for your benefit, but who are not real and do
> not have experiences, or (B) there could be other people
> present (though it's much more complicated and takes some
> work to make sure inconsistencies don't arise).
Actually, because the runs are deterministic those others must be
deterministic, too. It makes me wonder about the degree of fidelity of a
zombie I know intimately, yet that is not a person. I don't think a zombie
would pass a Turing test without being a real person. If they are not
zombies, the implications of your decisions have been just amplified
n-fold (for the number of beings boxed in the sim with you).
> Okay, as you will: they're not "separate people". But part of
> my hypothesis is that there is exactly zero chance that they
> can diverge; they're quadruples in a (formal) Turing Machine,
> and this TM either runs one-and-a-half times, or two full times.
> Your choice.
If the system is deterministic, there's exactly zero choice there. Though
it doesn't help you to know this one bit, without knowing what your choice
is, and the outcome of it in above model.
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:14:49 MST