From: Hal Finney (hal@finney.org)
Date: Tue Jun 18 2002 - 12:27:14 MDT
Hal wrote:
> > If computer programs and their outputs, even their execution histories,
> > have Platonic realism, then it is easier to see how we could be living
> > in such a reality.
>
> > We know that if we were running as a simulation on a computer
> > which existed in the physical world, it would not be possible
> > to tell what sort of reality embeds the computer program.
> > Maybe Platonic reality is enough.
Lee replied:
> Here's where you've made an *enormous* jump. Each of us
> individually knows that (following Descartes) he or she
> is experiencing at the present moment. We logically,
> sensibly, and rationally infer that others are too, but
> then, it could always turn out that your brain is in a
> jar somewhere, or some vast cool entity is running a
> simulation of the 21st century where you're emulated
> but everyone else is merely portrayed.
>
> If Platonic reality were enough, then it would completely
> wipe out all our reasons for doing anything. As Greg Egan
> hinted at in his absurd Theory of Dust (in Permutation
> City), your life will continue (has continued--is continuing)
> in timeless platonic space regardless of what happens in
> the actual execution. So why save a little girl from Nazis?
> Why strive to build a friendly AI? Whether you fail or
> succeed, and it's all happening in platonic space anyway
> then who gives a shit?
I'd say two things about this. First, even if this were true about
the pointlessness of our activities, it would have no bearing on the
truth of the proposition that Platonic reality of computer programs
is enough to explain our experiences. Just because you may find the
outcome distasteful is not an objection to the logic of the argument.
OTOH in this case I did not so much present an argument as an invitation
to consider the possibility that Platonic existence suffices, so perhaps
emotional reactions are relevant.
But second, I don't think you have to reach such a bleak conclusion.
Let me suggest a couple of ways to see it differently.
One way to see the supposed pointlessness of human action is that in
this model, the "multiverse" (the totality of all realities created
by all programs) is in a sense deterministic. Whatever you do here in
this universe is matched by other universes where you do other things.
The same thing is true, as a matter of fact, in the Many Worlds
Interpretation (MWI) of quantum mechanics. So it might seem that even
admitting the actual existence of alternative worlds is enough to call
into question the meaningfulness of our actions.
I think the answer is to adopt a position of "compatibilism", where
the workings of our free will are *identified* with the deterministic
physical processes which underly our brain activities. In a sense, our
actions do change the universe, in that they change the probabilities of
the various outcomes. The doctrine of compatibilism has an extensive
philosophical literature which I won't try to summarize here, but you
can do a search on that term online to get some pointers.
In the context of a multiverse model like the MWI, compatibilism means
that our actions can change the probabilities of the various multiverse
worlds. In effect, in seeking to stop harm to others, we reduce the
fraction of worlds in the multiverse where that harm occurs.
For the full Platonic multiverse model, in the case where we think of
the multiverse as being generated by all possible computer programs,
we first need to understand that there are both pragmatic and technical
reasons to believe that not all worlds have equal "measure". This means
that if you selected a random conscious observer from the multiverse,
residents of some worlds would be more likely to be selected than others.
One reason to believe this is that our own world seems to be quite special
and orderly, perhaps generated by a relatively simple computer program.
(Stephen Wolfram suggests in his new book that 5 lines of Mathematica
code might do it.) If we're going to accept multiverse theories at all,
we need to explain why our universe is as lawful as it is, and that seems
to require a non-uniform probability distribution across universes.
As it happens, there is a plausible reason to expect this as well.
If we assume that all programs are equally probable, it turns out that
many programs produce exactly the same universes. Basically, if you
have a program which is inherently short, then it doesn't matter if there
is junk after the end of the program, as the junk doesn't get executed.
So if you look at all the million-bit programs, then if there is a program
which only uses its first thousand bits, there are 2^1000 effectively
identical copies of that program, corresponding to all the possible
ignored values of the second thousand bits. Hence a universe created
by a thousand bit program might be expected inherently to have 2^1000
times more measure than a universe created by a million bit program.
This means that universes created by short programs would plausibly be
much more likely than universes created by long programs, which offers
an explanation of why our universe seems to be creatable by a relatively
short program.
Given this perspective, our activities can in effect change the measure
of universes in the multiverse by making them more or less likely, just
as in the MWI. When we intervene to help another person, we reduce the
measure of universes where that person is suffering. We make them less
probable; we make it less likely that a random person selected from the
multiverse would be suffering in that way. We reduce the total amount
of suffering in the vast multiverse.
These are positive results and are arguably just as meaningful as when
we take steps to improve conditions in our own universe. Therefore the
existence of a multiverse via Platonic reality of computer programs is
compatible with the meaningfulness of human action.
Hal
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:14:53 MST