On 12 Dec 2011, at 04:39, Joseph Knight wrote:
On Sat, Dec 10, 2011 at 6:39 AM, Bruno Marchal <marc...@ulb.ac.be>
wrote:
On 09 Dec 2011, at 19:47, Joseph Knight wrote:
On Fri, Dec 9, 2011 at 3:55 AM, Bruno Marchal <marc...@ulb.ac.be>
wrote:
The heap argument was already done when I was working on the
thesis, and I answered it by the stroboscopic argument, which he
did understand without problem at that time. Such an argument is
also answered by Chalmers fading qualia paper, and would introduce
zombie in the mechanist picture. We can go through all of this if
you are interested, but it would be simpler to study the MGA
argument first, for example here:
http://old.nabble.com/MGA-1-td20566948.html
There are many other errors in Delahaye's PDF, like saying that
there is no uniform measure on N (but there are just non sigma-
additive measures), and also that remark is without purpose because
the measure bears on infinite histories, like the iterated self-
duplication experience, which is part of the UD's work, already
illustrates.
All along its critics, he confuses truth and validity, practical
and in principle, deduction and speculation, science and
continental philosophy. He also adds assumptions, and talk like if
I was defending the truth of comp, which I never did (that mistake
is not unfrequent, and is made by people who does not take the time
to read the argument, usually).
I proposed him, in 2004, to make a public talk at Lille, so that he
can make his objection publicly, but he did not answer. I have to
insist to get those PDF. I did not expect him to make them public
before I answered them, though, and the tone used does not invite
me to answer them with serenity. He has not convinced me, nor
anyone else, that he takes himself his argument seriously.
The only remark which can perhaps be taken seriously about MGA is
the same as the one by Jacques Mallah on this list: the idea that a
physically inactive material piece of machine could have a physical
activity relevant for a particular computation, that is the idea
that comp does not entail what I call "the 323 principle". But as
Stathis Papaioannou said, this does introduce a magic (non Turing
emulable) role for matter in the computation, and that's against
the comp hypothesis. No one seems to take the idea that comp does
not entail 323 seriously in this list, but I am willing to clarify
this.
Could you elaborate on the 323 principle?
With pleasure. Asap.
It sounds like a qualm that I also have had, to an extent, with the
MGA and also with Tim Maudlin's argument against supervenience --
the notion of "inertness" or "physical inactivity" seems to be
fairly vague.
I will explain why you can deduce something precise despite the
vagueness of that notion. In fact that vagueness is more a problem
fro a materialist than an immaterialist in fine.
How so?
With comp, if you want introduce a physical supervenience thesis, the
physical activity can only mean "physical activity relevant with the
computation". So we can say that a physical piece of a computer is
inert relatively to a set S of computations in case the computations
in S are exactly executed with and without the physical piece.
Digitalness makes the notion of exactness here sense full. If someone
says that such a piece of matter has still some physical activity
involved in the computation, it can only mean that we have not chosen
the right level of implementation of those computations. If a
materialist can convince someone that such a piece, which has no role
for the computation in S, has some role, for S bearing a first person
perspective, then we can no more say "yes" to a doctor, in virtue of
building a device which will emulate correctly the computations in S
(assuming some of them corresponds to a conscious experience).
Indeed, it is not yet entirely clear for me if comp implies 323
*logically*, due to the ambiguity of the "qua computatio". In the
worst case, I can put 323 in the defining hypothesis of comp, but
most of my student, and the reaction on this in the everything list
suggests it is not necessary. It just shows how far some people are
taken to avoid the conclusion by making matter and mind quite
magical.
I think it is better to study the UDA1-7, before MGA, and if you
want I can answer publicly the remarks by Delahaye, both on UDA and
MGA.
I feel quite confident with both the UDA and the MGA (It took me a
little while).
Nice.
I read sane04, and quite a few old Everything discussions,
including the link you gave for the MGA as well as the other posts
for MGA 2 and 3.
I might send him a mail so that he can participate. Note that the
two PDF does not address the mathematical and main part of the
thesis (AUDA).
So ask any question, and if Delahaye's texts suggest some one to
you, that is all good for our discussion here.
My main question here would be: when Delahaye says you can't
(necessarily) have probabilities for indeterministic events, is
that true?
Simplifying things a little bit I do agree with that statement.
There are many ways to handle indeterminacies and uncertainty.
Probability measure are just a particular case. But UDA does not
rely at all on probability. All what matters to understand that
physics become a branch of arithmetic/computer science is that
whatever means you can use to quantify the first person
indeterminacy, those quantification will not change when you
introduce the delays of reconstitution, the shift real/virtual, etc.
Formally, the math excludes already probability in favor of
credibility measure. But for the simplicity of the explanation, I
use often probability for some precise protocol. The p = 1/2 for
simple duplication is reasonable from the numerical identity of
reconstituted observers. We have a symmetry which cannot be hoped
for any coins!
Credibility measure? What's that?
Let T be a finite set, and 2^T its power set. A belief function b is a
function from 2^T to [0, 1] quite comparable to a probability
function. We have for example that b({ }) = 0, and b(T) = 1. The main
difference is we don't ask for the Poincare identity, we ask only for
an inequality:
if Ai are subsets of T, we ask that
b(A1 U A2 U ... U An) bigger or equal than Sum_i b(Ai) - Sum_i < j
b(Ai inter Aj) + ...
For probability we ask an equality. I see that in english they use
"belief", but in french we use "croyance" (credibility).
This leads to a variant of probability calculus which can handle
better the notion of ignorance than the bayesian approach.
You might Google on "Dempster Shafer theory of evidence".
Modal logics can be used to formalize the "certainty" case. Alechina,
in Amsterdam, has shown that the normal modal logic K + the formula D
(Bp -> Dp) formalizes this "certainty" completely, and we get
something similar in the relevant variants of the self-reference
logics. (B = [ ], D = < >), except that we loose the modal
necessitation rule, like for the quantum logics.
This is just one example of a calculus of uncertainty (apart from
probability). There are theories of possibility, of plausibility, etc.
Dempster-Shafer theory of evidence got some success in criminal
inquests, medical diagnostic, finding location of secret submarines,
debugging and inductive inference. It is particularly useful when we
don't want to ascribe a uniform probability in case of ignorance,
which is the case when we ignore the content of the set T, that is
when we have only partial knowledge on the elementary results of the
random experiences. It works also better on vague or fuzzy sets. I am
not an expert in that field, but my late director thesis, Philippe
Smets, the founder of IRIDIA, was working on an extension of such a
belief function theory (or theory of evidence).
How would it affect the first few steps of the UDA if there were no
defined probability for arriving in, say, Washington vs Moscow?
Well, in that case, there are probability measure. In the infinite
self-duplication, you can even use the usual gaussian. But even if
there were no such distribution, the result remains unchanged:
physics becomes a calculus of first person uncertainty with or
without probability. As I said, only the invariance of that
uncertainty calculus matter for the proof of the reversal.
Tell me if this answer your question.
That seems to make sense. Thanks
OK. Ask any question in case you want grasp completely, or who knows,
refute, the UDA argument. Please, for the step 8, MGA, use the most
recent version which exists only on this list:
http://old.nabble.com/MGA-1-td20566948.html
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@googlegroups.com.
To unsubscribe from this group, send email to
everything-list+unsubscr...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/everything-list?hl=en.