Re: first person indeterminacy vs predictability
On 08 Mar 2011, at 09:23, Digital Physics wrote:
As you suggested, I tried to check the archive to make sense of your
replies, but I utterly failed. The archive seems to be full of
unexplained terminology as well. In your opinion, which previous
messages provide justifications of your claims on finite random
strings and white rabbit hallucinations?
Which claims?
Do you agree that the one bit string "1" can be considered as random,
in the first person view (like when you measure a (up+down) electron
in a {up, down} apparatus? I am not saying much more than that, except
that I situate such first person randomness in the classical situation
of self-duplication. And I did answer about the rabbit hallucination.
Its depth comes from the human processing of the hallucination, not
from just a video-game rabbit description.
Have you consult the sane04(*) paper? It is probably simpler than
finding what I said in this list. You were supposed to read a lot, I
realize. I cannot take exerts out of context . So it is far simpler, I
think, to start from sane04. The older 'escribe' archive were easier
to search in, but I have problem myself with the Google group archive.
You have intervened in a already long conversation, and you are
certainly welcome, but your question was out of the context of the
discussion, and it is not clear for me what you already understand or
not. Please read sane04, so I can figure out what is your precise
point if you are not satisfied with my explanation above. It is the
base of the list discussion since many years. Indeed the whole
discussion, is mostly based on my work which has developed intuitions
similar to Tegmark and Schmidhuber (but published much earlier) and
which shows both their defects and the (theological) price of their
amelioration with respect of the mind-body question.
The sane paper is a not too bad complete version of my contribution. I
give a precise version of a very weak form of computationalism, from
which I show that the mind-body problem is reduced to a problem of
justifying the appearance of the physical laws in the relative
number's or machine's mind.
The point is theological. It means that if we take seriously into
account the computationalist hypothesis, then Plato's theology is
correct, and Aristotle's theology (used by atheists and christians,
among others) is not correct. In that frame, physical reality does no
more describe the fundamental ontology, but appears to be the border
(or projection, shadow, ...) of (arithmetical) truth as seen from an
internal number-theoretically definable perspective. My contribution
is divided into a not too involved technical part (UDA), and much more
technical part (AUDA, or the Löbian machine's interview). The
technical AUDA makes the comp hypothesis, together with the classical
theory of knowledge, testable. I use the term 'theology' in the
original greek sense of 'theory of everything', or 'theory of truth'.
I can explain this only if I we are able to share some context.
In case you know the french you can consult, on my web page, the
original thesis, and a long detailed version. Tegmark and Schmidhuber
does not really address the mind-body problem. They assumed implicitly
some 'identity thesis' which just cannot work when assuming the
computationalist hypothesis. You might also consult Russell Standish's
book 'theory of nothing" which introduces well the kind of debate we
have on the everything-list.
If you have question we can proceed steps by steps in short mail.
I have no idea of your background, and the comp mind-body problem is
highly interdisciplinary, and different people have different
problems, and need sometimes different explanations. You might, but
are not obliged, to present yourself, mister or miss digital physics.
Bruno Marchal
(*) http://iridia.ulb.ac.be/~marchal/publications/SANE2004MARCHALAbstract.html
http://iridia.ulb.ac.be/~marchal
From: marc...@ulb.ac.be
To: everything-list@googlegroups.com
Subject: Re: first person indeterminacy vs predictability
Date: Mon, 7 Mar 2011 18:19:33 +0100
On 07 Mar 2011, at 17:26, Digital Physics wrote:
I agree that white rabbits have programs much shorter than those of
random structures.
It depends. Very short programs can generate all random structures.
You mean the short program that computes the entire set! But this is
irrelevant here: to predict a concrete individual history, we must
consider the probability of the program that computes this concrete
individual history, and nothing else. The description of the entire
set is much shorter than the description of most of its individual
elements. But it is useless as it has no predictive power.
Schmidhuber has a lots of papers on this:
http://www.idsia.ch/~juergen/computeruniverse.html
Th
RE: first person indeterminacy vs predictability
As you suggested, I tried to check the archive to make sense of your replies,
but I utterly failed. The archive seems to be full of unexplained terminology
as well. In your opinion, which previous messages provide justifications of
your claims on finite random strings and white rabbit hallucinations?
From: marc...@ulb.ac.be
To: everything-list@googlegroups.com
Subject: Re: first person indeterminacy vs predictability
Date: Mon, 7 Mar 2011 18:19:33 +0100
On 07 Mar 2011, at 17:26, Digital Physics wrote:
I agree that white rabbits have programs much shorter than those of random
structures.
It depends. Very short programs can generate all random structures.
You mean the short program that computes the entire set! But this is irrelevant
here: to predict a concrete individual history, we must consider the
probability of the program that computes this concrete individual history, and
nothing else. The description of the entire set is much shorter than the
description of most of its individual elements. But it is useless as it has no
predictive power. Schmidhuber has a lots of papers on this:
http://www.idsia.ch/~juergen/computeruniverse.html
The entire set of random string is useful to illustrate the first person
indeterminacy, and that was its role in my reply to Andrew and 1Z. So your
remark is unfounded.We have discussed this a lot with Juergen on this list. To
keep its position he was obliged to assume that finite strings can never be
said random, even form a first person point of view. You might take a look in
the archive. To sum up, Schmidhuber missed the first person indeterminacy.You
have to understand that the point here consists not in solving the mind-body
problem, but in formulating it in the computationalist theory of the mind.
White rabbits have intrinsically very deep (in Bennett's sense) programs.
No, because many programs making white rabbits for video games are both short
and fast, that is, those rabbits are not deep in Bennett's sense.
But a sustaining white rabbit human hallucination is another matter. And this
is what we have to take into account in the "measure problem" when we are
confronted with the universal dovetailing.
But you also claim that "most will consider their histories ...
Chaitin-incompressible".
In the case of you being duplicated in W and M iteratively. Not in
case of you in the UD's work.
This seems very unclear. What's the difference?
It is the difference between a counting algorithm, and a universal algorithm.
You might identify numbers and programs, and in that case the difference is the
difference between a list of programs, and a list of the executions of the
programs. If you have read enough in the archive or in my paper to understand
the first person indeterminacy notion, you might understand that, from the
first person points of view, such a distinction does matter.
This means long programs and no predictability at all, contradicting
daily experience.
Not at all. If you agree with Everett, and send a beam of particles
prepared in the state (up + down) on a "{up, down}-mirror", you see
the splitting of the beam. If you label the left and right electrons
by W and M, you can bet the strings will be incompressible,
sure, this still makes sense
and this is a quantum analog of iterated self-duplication. This gives an hint
for the vanishing of the WR: computable histories about the
substitution level, and randomness below. That justifies in part the
quantum appearance from the digitalness of the mind (not of matter).
Well, to me this sounds a bit like jargon used to hide the lack of substance.
I meant "computable histories *above* the substitution level", and "randomness
below". More precisely the randomness pertains on the set of all computations
going through my current relative states. This is a consequence of the UD
Argument. I refer you to my sane04
paper:http://iridia.ulb.ac.be/~marchal/publications/SANE2004MARCHALAbstract.html
Or can you explain this clearly? Excuse me for skipping the remainder of this
message.
I suggest you read the paper sane04(*). If you have a (real precise, not
philosophical) problem, just ask a precise question. We were discussing the
seventh step of the UD Argument. It would already be easier if you can
acknowledge the understanding of the first six steps. Note that the skipped
message was alluding to the more technical part of the work, where the measure
one is given by a variant of Gödel-Löb self-reference logics, which I name
"arithmetical hypostases", because I have used them to provide an arithmetical
interpretation of Plotinus theology, including his notion of matter. The whole
result is that comp, with the classical theory of knowledge, is an empirically
testable theory.
Bruno Marchal
(*)http://iridia.ulb.ac.be/~marchal/publications/SANE2004MARCHALAbstract.html
http:
Re: first person indeterminacy vs predictability
On 07 Mar 2011, at 17:26, Digital Physics wrote:
I agree that white rabbits have programs much shorter than those
of random structures.
It depends. Very short programs can generate all random structures.
You mean the short program that computes the entire set! But this is
irrelevant here: to predict a concrete individual history, we must
consider the probability of the program that computes this concrete
individual history, and nothing else. The description of the entire
set is much shorter than the description of most of its individual
elements. But it is useless as it has no predictive power.
Schmidhuber has a lots of papers on this:
http://www.idsia.ch/~juergen/computeruniverse.html
The entire set of random string is useful to illustrate the first
person indeterminacy, and that was its role in my reply to Andrew and
1Z. So your remark is unfounded.
We have discussed this a lot with Juergen on this list. To keep its
position he was obliged to assume that finite strings can never be
said random, even form a first person point of view. You might take a
look in the archive. To sum up, Schmidhuber missed the first person
indeterminacy.
You have to understand that the point here consists not in solving the
mind-body problem, but in formulating it in the computationalist
theory of the mind.
White rabbits have intrinsically very deep (in Bennett's sense)
programs.
No, because many programs making white rabbits for video games are
both short and fast, that is, those rabbits are not deep in
Bennett's sense.
But a sustaining white rabbit human hallucination is another matter.
And this is what we have to take into account in the "measure problem"
when we are confronted with the universal dovetailing.
But you also claim that "most will consider their histories ...
Chaitin-incompressible".
In the case of you being duplicated in W and M iteratively. Not in
case of you in the UD's work.
This seems very unclear. What's the difference?
It is the difference between a counting algorithm, and a universal
algorithm. You might identify numbers and programs, and in that case
the difference is the difference between a list of programs, and a
list of the executions of the programs. If you have read enough in the
archive or in my paper to understand the first person indeterminacy
notion, you might understand that, from the first person points of
view, such a distinction does matter.
This means long programs and no predictability at all, contradicting
daily experience.
Not at all. If you agree with Everett, and send a beam of particles
prepared in the state (up + down) on a "{up, down}-mirror", you see
the splitting of the beam. If you label the left and right electrons
by W and M, you can bet the strings will be incompressible,
sure, this still makes sense
and this is a quantum analog of iterated self-duplication. This
gives an hint
for the vanishing of the WR: computable histories about the
substitution level, and randomness below. That justifies in part the
quantum appearance from the digitalness of the mind (not of matter).
Well, to me this sounds a bit like jargon used to hide the lack of
substance.
I meant "computable histories *above* the substitution level", and
"randomness below". More precisely the randomness pertains on the set
of all computations going through my current relative states. This is
a consequence of the UD Argument. I refer you to my sane04 paper:
http://iridia.ulb.ac.be/~marchal/publications/SANE2004MARCHALAbstract.html
Or can you explain this clearly? Excuse me for skipping the
remainder of this message.
I suggest you read the paper sane04(*). If you have a (real precise,
not philosophical) problem, just ask a precise question. We were
discussing the seventh step of the UD Argument. It would already be
easier if you can acknowledge the understanding of the first six
steps. Note that the skipped message was alluding to the more
technical part of the work, where the measure one is given by a
variant of Gödel-Löb self-reference logics, which I name "arithmetical
hypostases", because I have used them to provide an arithmetical
interpretation of Plotinus theology, including his notion of matter.
The whole result is that comp, with the classical theory of knowledge,
is an empirically testable theory.
Bruno Marchal
(*)
http://iridia.ulb.ac.be/~marchal/publications/SANE2004MARCHALAbstract.html
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.
RE: first person indeterminacy vs predictability
> > I agree that white rabbits have programs much shorter than those of random
> > structures.
> It depends. Very short programs can generate all random structures.
You mean the short program that computes the entire set! But this is irrelevant
here: to predict a concrete individual history, we must consider the
probability of the program that computes this concrete individual history, and
nothing else. The description of the entire set is much shorter than the
description of most of its individual elements. But it is useless as it has no
predictive power. Schmidhuber has a lots of papers on this:
http://www.idsia.ch/~juergen/computeruniverse.html
> White rabbits have intrinsically very deep (in Bennett's sense) programs.
No, because many programs making white rabbits for video games are both short
and fast, that is, those rabbits are not deep in Bennett's sense.
> > But you also claim that "most will consider their histories ...
> > Chaitin-incompressible".
>
> In the case of you being duplicated in W and M iteratively. Not in
> case of you in the UD's work.
This seems very unclear. What's the difference?
> > This means long programs and no predictability at all, contradicting
> > daily experience.
>
> Not at all. If you agree with Everett, and send a beam of particles
> prepared in the state (up + down) on a "{up, down}-mirror", you see
> the splitting of the beam. If you label the left and right electrons
> by W and M, you can bet the strings will be incompressible,
sure, this still makes sense
> and this is a quantum analog of iterated self-duplication. This gives an hint
> for the vanishing of the WR: computable histories about the
> substitution level, and randomness below. That justifies in part the
> quantum appearance from the digitalness of the mind (not of matter).
Well, to me this sounds a bit like jargon used to hide the lack of substance.
Or can you explain this clearly? Excuse me for skipping the remainder of this
message.
--
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.
Re: first person indeterminacy vs predictability
On 07 Mar 2011, at 15:26, Digital Physics wrote:
You write "white rabbits (flying crocodiles) are not random
structures. They are aberrant
consistent extensions, a bit like in our nocturnal dreams." I agree
that white rabbits have programs much shorter than those of random
structures.
It depends. Very short programs can generate all random structures.
White rabbits have intrinsically very deep (in Bennett's sense)
programs. They are relatively costly. But technically this is not
enough for eliminating them from the first person appearance, unless
we use the self-referential logics.
But you also claim that "most will consider their histories ...
Chaitin-incompressible".
In the case of you being duplicated in W and M iteratively. Not in
case of you in the UD's work.
This means long programs and no predictability at all, contradicting
daily experience.
Not at all. If you agree with Everett, and send a beam of particles
prepared in the state (up + down) on a "{up, down}-mirror", you see
the splitting of the beam. If you label the left and right electrons
by W and M, you can bet the strings will be incompressible, and this
is a quantum analog of iterated self-duplication. This gives an hint
for the vanishing of the WR: computable histories about the
substitution level, and randomness below. That justifies in part the
quantum appearance from the digitalness of the mind (not of matter).
Then you say "but computer science and mathematical logic shows that
it is not easy either to prove that comp and first person
indeterminacy implies [flying rabbits]". I don't understand - it has
been shown it's not easy to prove that? How has it been shown it's
not easy to prove that?
That is actually rather obvious, if you know just a bit of computer
science. To get all the computational histories, you need Church
thesis and the enumeration of all partial computable function. By the
padding theorem, this is a highly redundant and fractal (and complex)
structure, and by the theorem of Rice, the set of codes corresponding
to any non trivial functions is not recursive (making our substitution
level) unknowable. So it is rather highly complex to derive the
possibility of white rabbits from that. In this list we discuss
alternate manner to approach that measure problem.
And you say: "There is no reason for making all relative histories
equally likely." But then what's the alternative?
To study the math of the universal dovetailing, and of what machine
can say about themselves and about they consistent extension
relatively to it.
Accepting the comp theory, together with the classical theory of
knowledge, although we don't have the measure, we can extract the
logic obeyed by the particular case of the "measure one". I have
succeeded in showing that it obeys already a quantum-like logic. This
needs a bit of advanced computer science/mathematical logic. See my
paper for details and references.
I have to say that I am a bit astonished that some people seems to
have difficulties to grasp that once we assume comp, theoretical
computer science becomes *the* key tool to progress on the fundamental
question. The beam example above suggests empirically that we are
physically duplicated in the iterative way. But obviously we are not
just duplicated iteratively, we are also obeying computational laws,
and arithmetical laws, etc. If that was not the case, comp would imply
white noise and would fall immediately in Russell's Occam catastrophe.
But, thanks to God, universal numbers does not put only mess in
Platonia, they generate also a lot of order.
-- Bruno Marchal
From: marc...@ulb.ac.be
To: everything-list@googlegroups.com
Subject: Re: first person indeterminacy vs predictability
Date: Mon, 7 Mar 2011 14:58:15 +0100
On 07 Mar 2011, at 10:47, Digital Physics wrote:But if most
histories are equally likely, and most of them are random and
unpredictable and weird in the sense that suddenly crocodiles fly
by, then why can we predict rather reliably that none of those weird
histories will happen?
From: marc...@ulb.ac.be
To: everything-list@googlegroups.com
Subject: Re: first person indeterminacy
Date: Sun, 6 Mar 2011 19:47:20 +0100
You can also consider the iteration of self-duplication. If you
iterate 64 times, there will be 2^64 versions of you. First person
indeterminacy is the fact that most of the 2^64 versions of you will
agree that they were unable to predict in advance what was the next
outcome at each iteration. Most will consider that their histories
(like:
"WMMMWWMWWWWMMWMMWM ..." (length 64)
are random, even Chaitin-incompressible.
Nobody said that the histories are generated by the iterated self-
duplication. The iterated self-duplication is used here only to
u
RE: first person indeterminacy vs predictability
You write "white rabbits (flying crocodiles) are not random structures. They are aberrant consistent extensions, a bit like in our nocturnal dreams." I agree that white rabbits have programs much shorter than those of random structures. But you also claim that "most will consider their histories ... Chaitin-incompressible". This means long programs and no predictability at all, contradicting daily experience. Then you say "but computer science and mathematical logic shows that it is not easy either to prove that comp and first person indeterminacy implies [flying rabbits]". I don't understand - it has been shown it's not easy to prove that? How has it been shown it's not easy to prove that? And you say: "There is no reason for making all relative histories equally likely." But then what's the alternative? From: marc...@ulb.ac.be To: everything-list@googlegroups.com Subject: Re: first person indeterminacy vs predictability Date: Mon, 7 Mar 2011 14:58:15 +0100 On 07 Mar 2011, at 10:47, Digital Physics wrote:But if most histories are equally likely, and most of them are random and unpredictable and weird in the sense that suddenly crocodiles fly by, then why can we predict rather reliably that none of those weird histories will happen? > From: marc...@ulb.ac.be > To: everything-list@googlegroups.com > Subject: Re: first person indeterminacy > Date: Sun, 6 Mar 2011 19:47:20 +0100 > You can also consider the iteration of self-duplication. If you > iterate 64 times, there will be 2^64 versions of you. First person > indeterminacy is the fact that most of the 2^64 versions of you will > agree that they were unable to predict in advance what was the next > outcome at each iteration. Most will consider that their histories > (like: > "WMMMWWMWWWWMMWMMWM ..." (length 64) > are random, even Chaitin-incompressible. Nobody said that the histories are generated by the iterated self-duplication. The iterated self-duplication is used here only to understand what is the first person indeterminacy in a very simple context (the context of pure iterated self-duplication). Assuming comp, the 3-histories(*) are generated by the UD, which is a non trivial mathematical object, and 1-histories(*) appears in the relative 1-person way by a highly complex mixing of computable histories and oracles (which can be handled mathematically with the logics of self-reference). There is no reason for making all relative histories equally likely. It is not easy to prevent white rabbits and flying crocodile, but computer science and mathematical logic shows that it is not easy either to prove that comp and first person indeterminacy implies them. And if we prove comp implies them, then observation and induction makes comp false or very non plausible. "Note also that, as Russell Standish recalled recently, white rabbits (flying crocodiles) are not random structures. They are aberrant consistent extensions, a bit like in our nocturnal dreams." Bruno (*) the suffix 1 and 3, in 1-x and 3-x, means x as seen by the first person or the third person respectively, as defined for example in the sane04 paper:http://iridia.ulb.ac.be/~marchal/publications/SANE2004MARCHALAbstract.html 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.
Re: first person indeterminacy vs predictability
On 07 Mar 2011, at 10:47, Digital Physics wrote: But if most histories are equally likely, and most of them are random and unpredictable and weird in the sense that suddenly crocodiles fly by, then why can we predict rather reliably that none of those weird histories will happen? > From: marc...@ulb.ac.be > To: everything-list@googlegroups.com > Subject: Re: first person indeterminacy > Date: Sun, 6 Mar 2011 19:47:20 +0100 > You can also consider the iteration of self-duplication. If you > iterate 64 times, there will be 2^64 versions of you. First person > indeterminacy is the fact that most of the 2^64 versions of you will > agree that they were unable to predict in advance what was the next > outcome at each iteration. Most will consider that their histories > (like: > "WMMMWWMWWWWMMWMMWM ..." (length 64) > are random, even Chaitin-incompressible. Nobody said that the histories are generated by the iterated self- duplication. The iterated self-duplication is used here only to understand what is the first person indeterminacy in a very simple context (the context of pure iterated self-duplication). Assuming comp, the 3-histories(*) are generated by the UD, which is a non trivial mathematical object, and 1-histories(*) appears in the relative 1-person way by a highly complex mixing of computable histories and oracles (which can be handled mathematically with the logics of self-reference). There is no reason for making all relative histories equally likely. It is not easy to prevent white rabbits and flying crocodile, but computer science and mathematical logic shows that it is not easy either to prove that comp and first person indeterminacy implies them. And if we prove comp implies them, then observation and induction makes comp false or very non plausible. Note also that, as Russell Standish recalled recently, white rabbits (flying crocodiles) are not random structures. They are aberrant consistent extensions, a bit like in our nocturnal dreams. Bruno (*) the suffix 1 and 3, in 1-x and 3-x, means x as seen by the first person or the third person respectively, as defined for example in the sane04 paper: http://iridia.ulb.ac.be/~marchal/publications/SANE2004MARCHALAbstract.html 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.
RE: first person indeterminacy vs predictability
But if most histories are equally likely, and most of them are random and unpredictable and weird in the sense that suddenly crocodiles fly by, then why can we predict rather reliably that none of those weird histories will happen? > From: marc...@ulb.ac.be > To: everything-list@googlegroups.com > Subject: Re: first person indeterminacy > Date: Sun, 6 Mar 2011 19:47:20 +0100 > You can also consider the iteration of self-duplication. If you > iterate 64 times, there will be 2^64 versions of you. First person > indeterminacy is the fact that most of the 2^64 versions of you will > agree that they were unable to predict in advance what was the next > outcome at each iteration. Most will consider that their histories > (like: > "WMMMWWMWWWWMMWMMWM ..." (length 64) > are random, even Chaitin-incompressible. -- 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.
