From: Stathis Papaioannou (stathisp@gmail.com)
Date: Wed Mar 12 2008 - 01:22:29 MDT
On 12/03/2008, Lee Corbin <lcorbin@rawbw.com> wrote:
> > What I had in mind was that Sj is used as input to produce its
> > successor Sm, from which the rest of Tuesday follows
> > deterministically. If Sj is reliably transferred from the first to the
> > second computer that would be the same as if the computation
> > unfolded in the one intact machine, no?
>
> Right.
>
> > But if the transfer were unreliable it would only be by chance
> > that the "right" sequence of states was set into motion giving
> > rise to the Tuesday experience.
>
> Yes. To me, that's as if we perfectly emulate a person's 48hr
> life quite a few times without incident. Then in some run of that
> single computer a God-particle level cosmic ray strikes at
> midnight on Monday, and the rest of the computation takes
> off in an entirely random and different direction, probably no
> longer emulating the same person at all
>
> To you, is that equivalent also?
No, the cosmic ray would cause a different computation. But if instead
the cosmic ray set off a Monday computation that otherwise wouldn't
have happened, that would be the same as a deliberate Monday
computation from the point of view of the internal observers.
What you refer to as the significance of a causal link in computation
is equivalent to what is sometimes called the requirement that the
computational system "handle counterfactuals". For example, if some
event on Monday had been different, then should be taken into account
by the computation and events on Tuesday would also have been
different; whereas if Tuesday just happened due to a random event this
sort of sensitivity to Monday would be lacking. The idea is that if
the computational system doesn't handle counterfactuals it is not only
useless for computation, but also that it doesn't give rise to
consciousness. But while it is obvious that a computation that doesn't
handle counterfactuals is useless to an external observer, it isn't
obvious that it would lack an internal observer. You would have to
invoke seemingly magical effects to explain how the computation knows
or cares that it hasn't been "properly" implemented when there is
nothing in the physical states underpinning it which indicates this.
> >> What is the appeal to you and Schmidhuber and the rest of
> >> the gang to all that dust comprising 0.99999999999 of your
> >> experience? That would seem to me to induce fatalism.
> >
> > Determinism, yes. Is that the same as fatalism?
>
>
> No, I mean something like this: Suppose that I hear that an
> extremely large asteroid by sheer luck is barreling down on
> the Earth at thousands of kilometers per second. I see this
> as a big, big hit on my runtime. (We're all going to die.)
> Why, in the Tegmark example, there is no other Lee Corbin
> for about 10^10^29 lightyears from here. So for me personally,
> not to mention all the people I love, this (to me) is very bad
> news.
>
> How is it for you? If 0.99999999999 of you (say in the
> visible universe 10^42 ly across), is unaffected by the asteroid,
> why worry?
You're assuming that you somehow know that you are the "real" Lee
Corbin, as opposed to all those simulations in Platonia. But if the
theory you scoff at is for the sake of argument right, there is no
reason to posit a separate, concrete universe at all. And you can
still hope that the computational branches where the asteroid misses
the Earth are of higher measure than the ones where it hits.
-- Stathis Papaioannou
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT