Re: UDA last question (was UDA step 9 10).

From: Marchal <>
Date: Fri Jul 6 12:32:53 2001

Joel Dobrzelewski wrote:

>So the Universal Dovetailer simply enumerates all possible states for "me".

Yes. I prefer to say it generates all computations going through that
states(). Bennett has introduced a nice notion of depth from string, which is
grosso modo the runtime to produce it. It is still possible (even probable)
that "me" is deep and "you" are deep so that the occurence in UD* of you
and me can be taken as evidence that we do share a "long" computation.

() I see Hal Finney says the same thing.

>(Whatever "me" is, is unimportant). And somehow, these states are joined by
>their similarity to one another, which allows (virtual) motion from one
>state to the next.

I think the relation of similarity, or proximity bears on the
computations/histories/dreams. But you are correct. The only way
to keep sharing the same deep computation, and thus sharing our
domain of undeterminacy, is that the computation is "linear in our common
ignorance". This will multiply both of us, in some conservative way,
and in extremely "explosive" way.
I guess the first person *plural* has it origin in such self multiplication

>So, if I am currently sitting in the state of "sitting at my desk", I have
>several realities branching off from this one where in the next moments I am
>"rising from my chair", or "still sitting at my desk", or "petting a white

In case comp is true, I'm afraid necessarily so.
Note that the existence of QM
gives confirming evidence that there is some branching. And perhaps
you can guess that comp predicts that any machine looking at her
neighborood sufficiently closely, i.e. below her level of substitution
will discover some "observable weirdness" and other continua of
parallel computations ...
Note that histories can be considered as fusing by difference amnesy.
I really hope to succeed in "rediscovering" the quantum computer
through "machine's introspection".

>But I am left with one nagging question.

I really hope so Joel. My strongest goal with UDA is to show that
computationalism or mechanism does not solve *per se* the mind
body problem. Quite the contrary. "Before UDA" you can believe there
is only a "consciousness problem", "after UDA" you got a "body" problem
too. Understanding comp consists in understanding that matter/space
/time is necessarily not obvious and must be recovered from the
space of all computations as seen as some internal points of views.
I told you that my UD or your MUCA or Schmidhuber's Great Programmer, ...
are not the solution, there are only steps
toward a mathematical *formulation* of the problem.

Of course the formulation gives insight, (the reversal)... and
that's nothing compare to the arithmetisation of that formulation,
weird quantum logics, but that's technical ... (if you know modal logic
search for LASE in the archive).

>1) Where is Time?

Time is *the* first person concept "par excellence". It is linked
with intuition, consciousness, but also construction, and truth.
The arithmetical translation of UDA let me hope that time is captured
by the modal logic known in the literature as S4Grz.

>1a) What governs the trajectory of one's awareness through all his/her
>possible states?

The geometry generated by the notion of proximity on the (maximal ?)
consistent extensions (our alternative "completed" futures).
Or if you prefer, what governs the possible trajectory of awaraness is
Schroedinger Equation. Our problem: extract it from the geometry above.

>1c) How do I get from one state to another?

You 3-don't. You 1-do. This must be related to the geometry above.

>1d) Isn't this "awareness", and its motion path necessarily "outside" those
>states and the Universal Dovetailer?

No. But perhaps you are close. A bizare thing is that if someone look
at a portion of a representation of UD*, he can say there is, in some sense,
 no awareness possible in that portion.
And this remains true for all finite portions of UD*.
The "meaning" appears only when the whole UD* is considered. This is coherent
with the fact that the neighborhood of "first person state" are defined
on that limit.

An image is that each instant, each observer moment perhaps, is defined
by a "trip" from the base of the cone (UD*) to the (non existing) top,
at infinite speed, just because we cannot be aware of the delays, nor
of any initial represention.

>Since I am into cellular automata, I always assumed that time is implemented
>naturally... by the automaton... from one tick to the next - just like a

That's the integers sequence. No?

>Can we really assume UD* exists?

Unless you are finitist you can realise that with comp you cannot escape its
existence. You get UD* once you accept the existence of *all* natural numbers,
and all their describable relations.

But no problem without changing your mind. Abandoning comp by abandoning
arithmetical platonisme is a way like another :-)

>Doesn't it take Time to execute UD?

No. Only a concrete UD, concrete relatively to "your" most probable
histories, will take enormous time.

But tell me, if you postulate that time, how could we explain it.
What would that time be, and where would it come from?

You tell me you don't need time or space with Your MUCA, isn't?

In Platonia it takes neither millimeters nor seconds.

Arithmetical relation are not "temporal",
still less (with Occam) if we can justify or just identifies
the discourses on the "temporal" by the "soud introspecting machines".

Received on Fri Jul 06 2001 - 12:32:53 PDT

This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:07 PST