On 29 Aug 2005, at 03:56, chales1.domain.name.hidden wrote:
> Before I disappear back into the eeeeeeeuw enteric neuroscience PhD
> life….again…
>
> Firstly:
> In http://iridia.ulb.ac.be/~marchal/publications/
> SANE2004MARCHAL.pdf is:
> Definition:
> “Fundamental Physics: I define it by the correct-by-definition
> discourse about observable and verifiable anticipation of possible
> relatively evolving quantities and/or qualities.”
>
> Do you see the assumption here? It assumes that models derived from
> experiential qualities (=empiricism) necessarily capture those
> processes of the natural world that generate the experiential
> qualities used to be empirical.
Why? No. I don't make that assumption at all. The definition of
physics I gave is very general. It does neither assume comp nor
assume the existence of an empirical world. The definition is
agnostic. It assumes the notion of stable observable and sharable
histories, but is neutral on the nature of them. See also my answer
to Tom (daddycaylor).
Precisely, models derives from experiential qualities does not
necessarily capture the processes of some natural world (if that
exists) generating the qualities, only if the "model" is correct, and
that is why I put it in the definition. It looks trivial? It is
certainly, at first sight. But the whole paper will show that once
comp is assumed, the definition is no more trivial.
> This is simply a belief. Yes the universe may behave in the way
> characterised –as you say – by ‘definition’. They are tautologous.
Well, no. The argument that I gave shows the contrary.
> To assume that these are ‘fundamental’ is to make the whole
> definition an oxymoron! The ‘fundamental laws’ that responsible for
> experiential qualities and the laws obtained by USING those
> experiential qualities to do observation cannot logically be
> claimed to be the same set! No useful outcome can be acquired from
> this thinking.
You must take the definition literally and abstract yourself from
your own naturalist hypotheses. All what I say, is that if "H phi = E
phi" is correctly related to what we can always observe, then it is a
physical law. It is very near a tautology, but with comp, strangely
enough perhaps, it will appeared to be everything but tautological.
But then you must follow the reasoning.
To be sure I am not so glad of that definition because people tend to
take it in some non literal sense.
Another definition works: physics is what is true in all "observer-
moment" when they are seen by betting machine, or still better:
"physics" is what is invariant for all observable point of views. The
defect is that the terms "observable" are defined in the course of
the reasoning.
>
> Secondly:
> To make any sense at all of the words ‘first-person’ you have to
> have some sort of definition of what that is. I don’t think there
> is any real definition of that anywhere in QM or UD or YD or any of
> the other models discussed here.
I should stop here because I can deduce you have not read even the
short papers. The notion of first person is completely defined both
during the UDA and in the interview of the Lobian machine. The UDA
definition is equivalent to the definition of "subjective" by
Everett. And the arithmetical definition (in the AUDA) is taken from
Theaetetus. The definition use in UDA is based on fold-psychology,
and in AUDA: on self-reference logic (making my work 100% third
person falsifiable).
> There is an assumption that the idea of a ‘third person’
> perspective and ‘third person experience’. These I think are both
> an oxymoron like military intelligence. J This is another
> assumption deriving from the idea that first person perspective is
> an any way captured by QM or a COMP machine.
I explictly prove that the first person notion of a machine cannot
been described in any third person term by the machine itself. I can
do that consistently for the class of sound machine thanks to the
axiomatizability of the gap which exists between truth on machine and
truth provable by the machine. It is the gap captured by the
difference between the modal logics G* and G (but UDA gives already
an intuitive picture of the gap).
> Not shown, not justified…
I explain in detail the necessary hypothetical part of comp, once
comp is true. This has never been done in *any* theory so far. I know
it is a subtle matter, and I have been convinced of the consistency
of that procedure thanks to the taking into account of incompleteness
(which has eventually been axiomatized by the G and G* logics, by
Solovay).
>
>
> Third and finally
> COMP can be found to be false as follows:
> In http://iridia.ulb.ac.be/~marchal/publications/
> SANE2004MARCHAL.pdf there is, on page 10, the following statement:
> “It can be seen as a manner to emulate digital parallelism in a
> linear sequential way.”
>
> According to your own document the UD is an enormous serial proof
> machine. It _assumes_ that the very act of being such a machine
> somehow preserves all the characteristics of what it is describing.
> This is simply untrue.
>
> Proof:
> Two turing machines A and B working in lockstep actually do 3
> ‘computations’ or ‘proofs’: machine A proof Ta, machine B proof Tb
> and C) the ‘relative’ proof Tab (and Tba) between machines A and
> B....Equally true, but not literally proven by the two machines.
> These are virtual theorems that precisely fit the classification of
> Gödellian ‘unprovable truths’ (which, I hereby claim priority as
> discoverer). Russel has my paper on them.
>
> The UD has Ta in it.
> The UD has Tb in it.
> The UD does not have Tab or Tba
Why ? The UD dovetails on all interactions between all machines. You
are not giving evidence that you did grasped what is a dovetailer
(universal or not).
> Furthermore:
> In order that Tba/Tab be excluded in the UD you have to assume they
> are not relevant to subjective experience.
I don't excluded them. If you think so, prove it. The UD excludes
only actual use of actual infinities.
> If Tba and Tab are recognised at all they are certinly not included
> in the UD by defintion of the serial execution. Nor are they proven
> eliminable from the UD (=unimportant).
>
> Tba and Tab are IMPLICIT proofs = virtual theorems that are only
> present in inherently parallel systems.
Why do you think the Universal Dovetailer do dovetail? The UD
dovetails on all inherently parallel systems once digital. If not
digital, we go out of my comp working hypothesis. We can do that, but
then this would not been a critic of my theorem.
> Ergo it is simply an erroneous basis to start any model of the
> natural world
What you call "natural world" *is* a model (in your sense of the
world). If not, tell me what do you mean by that.
> and especially erroneous if it has anything to say about subjective
> experience, which, if it is anything at all is NOT about what a
> thing IS,
Oh! I agree, here. I guess you mean IS in some 3 person viewpoint.
> but is about what a thing IS NOT.
... and much more!!!
> When you explicitly define any ‘thing’ you implicitly define NOT
> THING!
Hal Ruhl says something like that, but even after asking for an
explanation I have not yet figure what that could mean.
Which "NOT" ? It should be stronger than the actually very strong
classical negation.
> Any computationalist approach formulates seriality and disposes of
> all the virtual theorems! No amount of wishful computing will
> reinstate them! You have to BE PARALLEL to get them.
But the universal dovetailer is *the* master in implementing all
digital parallel processes, even some infinite Garden of Eden. Only
parallelism + explicit call to actual infinities would cause trouble
to comp. That the UD does not miss anything thing computable
(including the inherently parallel procedures) is a non trivial
consequence of Church Thesis.
>
> Alternatively you can prove conclusively prove that they are not in
> the universe….woops…you will then prove all Feynman diagrams to be
> false…(virtual theorems = virtual matter) …and….I can find a
> mechanism whereby the human brain makes very good use of virtual
> theorems.
You are mystifying parallelism. I guess you know Johnson Laird's work
in the comp philosophy of mind. He made that moves but this has not
yet explain anything "physical" or "psychological". Beside, it
contradicts my working hypothesis, which you have not shown false or
inconsistent.
>
> So “what is it like to be a UD?”
> It’s like being a whopping great fat Turing machine = like being a
> tape reader and tape: probably nothing at all, but certainly
> different from whatever the UD thinks it is emulating!
Why do you want the UD ever thinks. Gosh, please, read my papers. The
UD is certainly not a person, or an infinitely dumb and shallow one.
> What’s it like to be a computer based on silicon? Like being a hot
> rock? What’s it like to be a quantum computer? Like being a
> cryogenically cold rock. :)
Remember I don't assume rock and silicon. I try to explain them as we
are obliged when we postulate the comp hyp.
>
> Finally: This is development of Progoginian thinking. The two
> people on this list who seem to be getting nearest to it are
> Stephen Paul King and Hal Ruhl.
>
> Can we please move on? The universe is absolutely bristling with
> virtual theorems. There are no theories of everything here… but
> merely ‘theories of not much at all’!
>
> Experiential qualities are not in the UD.
Right! But totally unrelated with the demonstration I gave. Only
Lobian machine can have experential qualities. The UD is definitely
not a Lobian machine. Actually it is more a Robinsonian Machine (but
that's is for later ifr you ask me to explain).
> All human knowledge is symbolically grounded in experiential
> qualities (a la Harnad).
Agreed! And explained in term of Lobian machine if you accept
Theaetetus' definitions of knowledge.
>
> Ergo ... Bye bye COMP.
You are quick, to say the last. Give me your hypotheses please, and
show me how precisely they are in contradiction with comp. Don't say
"my first person experience", because "machine's first person
experiences" also are not describable in third person term by the
machine. To abandon comp effectively you must refer to some third
person process which would be not emulable by a Turing machine. That
exists. Nielsen has given a nice example: e^iOMEGAt, with OMEGA being
Chaitin's number.
You really need that? How could you build an experimental device
capable or recognizing e^iOMEGAt ?
If you follow the UDA argument, you will understand that if you
assume a "natural primitive world", then comp is false, and you can
search for actual infinities. With comp, it would be a waste of time,
because a sound machine cannot recognize something as being provably
more complex than eself. She can only bet on things like that, in
some circumstances.
And in some (lucky?) situations machines can even correctly bet on
those uncomputable things.
Bruno
http://iridia.ulb.ac.be/~marchal/
Received on Mon Aug 29 2005 - 10:38:08 PDT