Measure, and Time

From: Lee Corbin <lcorbin.domain.name.hidden>
Date: Mon, 16 May 2005 18:54:06 -0700

Brian wrote

> > > [Peter]
> > > And if Platonia exists, it already contains every string that
> > > could be output by the UD -- so what do you need a UD for?
> >
> > [Lee]
> > It'll probably be retorted that it all has to do with
> > measure, but there are lots of machines (I mean TMs) in
> > Platonia, and lots and lots of them are a lot shorter
> > (and so by any measure more plentiful) than the old UD.
>
> That's true, there are lots of shorter programs than the dovetailer.
> But how many of those short programs support conscious observers?
> Let's suppose you wanted to write a program that supports a conscious
> observer. It's going to be a very long, complicated program right?
> Very informationally complex. Well, I'm lazy. I would just write a
> program that generated all possible programs then ran them. I would
> then know that on some subset of those programs I've got conscious
> observers. My program is considerably simpler than your program for
> generating a conscious observer, it just takes a lot longer to
> run.

Okay, thanks for that idea. But there is still the problem of
measure to you, no? (I know Bruno has other ideas, and has
explained them, and I think that there are a lot of ways people
have spoken of regarding how to assign measure. Where are you?)

That is, the UD will devote less than 1/10^10^10^10 of its resources
to "running" the first conscious observer on the list of *all* TMs.
On the other hand, I could believe that the rarity of a program
DEVOTED TO RUNNING ME or (more regrettably, some other conscious
entity) might be less than that. I.e., maybe only one in 10^^3.
But more sensibly: d'you think that Schmidhuber or anyone has made
a very good case why the UD would give me more runtime (i.e.
observer-moments) than a specialized devoted program (which must
also exist in Platonia)?

> So it takes much less information to compute everything than it
> does to compute one particular universe. I'm guessing the UD in its
> most compact form is the simplest program for computing a conscious
> observer! (know of any others?). And it is simpler than most of those
> bitstrings appearing in its computational history.

Yes.

> > [Lee]
> > Well, I think that David Deutsch's version includes all our
> > latest and best physical theories, which still includes the
> > Schrödinger equation and other time-based foundations. I'm
> > guessing that the universes (I mean *slices*) are real; that
> > is, have a higher ontological status than completely unrelated
> > snapshots. But that would be just a guess.
> >
> > As for me, the whole multiverse as well as any slice strongly
> > exhibits a dependence on time.
>
> I might be misunderstanding, but neither the multiverse as a whole nor
> any slice of it is time-dependent.

The laws of physics link snapshots of places with higher time
clock readings with those with lower readings. I think that in
FoR, slices *are* spacetimes, and are objectively linked together
just as the pieces in a crossword puzzle are even though it's
still in the can.

Thus the linkage is dictated by the laws of physics. And almost
all of those have this little parameter t in them. It seems to
me that this might afford time a special role.

But I'm still busy re-reading the chapter on Time; I know that
I once agreed with every word, and thought that I understood it.
But I'm a lot less sure now.

Lee
Received on Mon May 16 2005 - 21:55:53 PDT