Le 23-janv.-07, à 04:36, Stathis Papaioannou a écrit :
>>
>> When you ask your computer to print a document, the computer typically
>> does not search the meaning of the words "print" or "document" in a
>> dictionary. Other more subtile self-reference are handled by the
>> diagonalization technic which makes it possible to cut the infinite
>> regresses. IF and when I come back on the Fi and Wi, I will give you
>> Kleene second recursion theorem which solves all those infinite
>> regress
>> appearing in computer self-reference.
>
> An association has been made between "print" and "document" with
> objects
> in the real world. You can work out what the print command is on an
> unknown
> computer by experimenting with different inputs and observing outputs.
> But
> if the real world is internalised, even if you could work out
> regularities in the
> syntax of an unknown computer (and I don't know if this is necessarily
> possible:
> it might be a military computer with syntax deliberately scrambled
> with a one-time
> pad) you would be unable to work out what it originally meant - what
> the computer
> is thinking. It is like finding an unknown language without a Rosetta
> stone or any
> cultural background which might help you with a translation. This
> reminds me of the
> impossibility of sharing 1st person experience: you can only do so if
> you share some
> 3rd person quality allowing at least some interaction.
First, I do agree we cannot extract semantics from syntax, or behavior
of a program from their codes. There is a well know theorem (Rice
theorem) explaining why, and I could come back on this when I come back
on the Fi and Wi. But so we agree. This does not depend on "cultural
background" unless you define "cultural background" by "most probable
universal neighborhood history".
Now I am not sure this is directly related to the 1-3 distinction.
Also, I have no clue of what you mean by "real world".
>>
>> From Pythagoras to Proclus, "intellectuals" were proud not making
>> that
>> error. Aristotle is in part responsible for having made "appearance"
>> reality, coming back to the (provably wrong assuming comp) common
>> sense
>> in those matters.
>> (of course as you know we have to rely on common sense to go beyond
>> common sense).
>
> OK, but we have to start with some basic observation. It looks like
> objects
> are pulled to the Earth by a force - that is a basic observation, with
> a minimal
> implicit theory.
I agree. I call that sometimes "grandmother physics". Even physicists
use it in everyday life.
> General Relativity explains this differently, but it takes a rather
> complex series of arguments to arrive at GR.
Already Galilee makes grandmother's physics globally wrong. Now
Galilee, Newton and GR works because they does not contradict
grandmother physics, and recast it in frame compatible with larger set
of data.
> You can't call Newton stupid because
> of this.
Not at all. Especially not Newton, who wrote some text showing that he
was aware on the lack of serious metaphysical foundations for his
physics.
> Similarly, your conclusion that there is no separate physical reality
> follows
> from a number of carefully argued steps, and at the start of the chain
> is the fact
> that there does appear to be a physical world... if there did not, we
> would not be
> having this or any other discussion.
I think we agree. I have never doubted the appearances of a physical
reality, even assuming comp.
What I do pretend, is that IF we assume the comp hyp, then the
appearance of a physical reality does not reflect the existence of a
*primary* physical reality. It is simpler to describe the
epistemological consequences (albeit probably looking more
provocative), which is that physics cannot be the fundamental science.
It really means that the laws of physics, not only can be derived from
computer science/number theory, but has to be derived from computer
science number theory if we are asssuming comp. All stable appearances
must emerge through a notion of first person plural observation.
Now such "first person plural observation" can be described in the
language of a Universal Machine, and this gives a way to test the comp
hypothesis.
I am before all an empiricist. True, I'm saying that if comp is true
then the "laws of physics" are in your head (actually in any universal
machine's "head"). So let us test comp by 1) deriving the
"comp-physics" (the physics in the head), 2) let us compare it with the
usual observations. If the empirical data contradict comp: comp is
refuted. If the data are coherent with comp, comp is not refuted. If
comp is correct, the data will never be contradicted, and we will never
know if comp is correct, but may be we will bet on it according to
possible circumstances.
Note that in the UDA I do start, not only from the appearances of a
physical world, but from its primary existence. But this assumption is
eliminated in the course of the reasoning (by the movie graph/maudlin).
> You seemed to be disputing the idea that a serial computation cannot
> be broken
> up arbitrarily into parallel components, or suggesting that they need
> to be glued
> together in some way if they are. This seems to contradict most of the
> teleportation
> thought experiments we have discussed, in which it is sufficient for
> continuity of
> conscious that I vanish at A and a copy with close enough brain be
> created at B:
> there need be no glue, no causal connection (although of course it
> would help to
> make the copy if you had the right information, the result would be
> the same if the
> copy just came about by random processes), no regard for temporal or
> spatial
> displacement.
Let us call S my digital (generalized) brain state here and now, in
Brussels for example.
Let us call S' the state of "consistent extension" of that state "me in
Moscow" or "me in Washington" to take those examples.
What the thought experiments show indeed is that we don't need (and
even cannot use) a causal or physical connection between those states S
and S'. But S' has to be a state "near S" for me having a feeling like
"I am in Washington, and I come from Brussels", so as to say yes to the
doctor and continue to say yes to the doctor. This works if S-S'
belongs to a "normal" computational history, that is a third person
computation having the "right" probability (if that exists, if not comp
is already incorrect). But then, by definition of computation S and S'
have to be mathematically related, even arithmetically related. If not
there would be too much white rabbits at the start.
In UD*, i.e. the entire (countably infinite) dovetailing work of the
Universal Dovetailer UD, we must distinguish third person computations
and all possible first person data, which can indeed, from a first
person perspective (which abstracts all delays in the running of the
UD) contained all arbitrary sequences. But almost all such sequences
are NOT computable. They are never generated by the UD, except by local
dovetaling on the finite initial segment of (all) real numbers, but
this perturbs only the measure (which we are searching) in the limit.
This limit makes a first person sense only in the limit.
Consciousness supervenes on computational histories which are immune
for that "real" unavoidable first person (plural) randomization.
The UDA (UD Argument) illustrates that we have to take into account
only the arithmetical gluing. But we have to take account of that
gluing. If not, everything is a computation, and this would make Church
thesis false (among many things ...).
> If there are more arbitrary sequences than third person computations,
> how
> does it follow that arbitrary sequences are not computations?
Stathis, I must go now, I will answer this later. I will comment the
rest of your post + a commentary on your kind reference to Searle. We
agree on many things. I am afraid we have to dig in slightly more
technical stuff for making the remaining posts understandable. I have
to explain those "arithmetical (first person/third person) glues".
Bruno
http://iridia.ulb.ac.be/~marchal/
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Everything List" group.
To post to this group, send email to everything-list.domain.name.hidden
To unsubscribe from this group, send email to everything-list-unsubscribe.domain.name.hidden
For more options, visit this group at
http://groups.google.com/group/everything-list?hl=en
-~----------~----~----~----~------~----~------~--~---
Received on Tue Jan 23 2007 - 08:35:55 PST