On 31 Jan 2011, at 12:44, Andrew Soltau wrote:

On 27/01/11 17:44, Bruno Marchal wrote:

On 25 Jan 2011, at 18:24, Andrew Soltau wrote:

On 24/01/11 21:35, Bruno Marchal wrote:

Thanks for all this. I will do some reading and then go through the points again. And get back to you.

You are welcome. Ask any question.



I have been trying to decipher your response to

> However, structures of information are instantiated in the physical.

OK, but this cannot work if DM is correct, by MGA. That's the whole point. There is no "physical reality" available. It is not obvious to understand this. The UDA+MGA explains this, and the AUDA (the Löbian interview, or Abstract Universal Dovetailer Argument) provides a path to extract physics, and the logic explains why the theory splits into quanta and qualia. Quanta appear as sharable qualia.

I have read your paper The Origin of Physical Laws and Sensations, but am still at a loss. I confess I find the blizzard of acronyms difficult to follow. (In particular it would help me greatly if we referred to the Computationalist Theory of Mind as CTM, as do wikipedia and Standford philosophy website, rather than COMP)

Comp is just an abbreviation of computationalism. It is synonymous with CTM, DM (digital mechanism), or simply here Mechanism, MEC, ...) etc. I change the wording when people add special meaning to the term. Sometimes comp means the precise theory "yes doctor + Church thesis", but it can indeed be shown equivalent with CTM. Some people use CTM having in mind the idea that the computation has to be physically instantiated, but then it is the point of the paper to show this does not work. Also "yes doctor" is really the assertion of the existence of a level where I am Turing emulable. Quickly we can understand that such a level cannot be found by any machine, but they can bet on them. It does not matter that you need to emulate the entire galactic quantum field to get your experience. In that sense comp is much weaker than the implicit intent of most version of CTM, closer to high level and neurophilosophy.

eg Is DU the same as UD? Or is DU the infinte trace of the universal dovetailer, as seems to be suggested by diagram 7?

UD is the english for the french DU. Sorry for that typo.
I use UD* for the infinite trace of the UD.
MGA is the movie graph argument (same consequences as Maudlin's argument). UDA = Universal Dovetailer Argument. In sane04 I add the MGA as a last step of UDA. But in all other publications I put the MGA, before UDA. UDA and MGA were originally introduced to remind people that science has not yet decided between Plato and Aristotle, and to provide motivation for mathematical definition of belief, knowledge, observation and feeling in the case of ideally correct universal (Löbian) machine. A Löbian machine is a universal machine with proving abilities, and "knowing" in a technical sense that she is universal. My work is a work on Gödel's theorem (and Löb, Solovay, Kleene, etc.) in relation with physics, reality, dreams, etc. By Aristotle, I mean (to simplify) the idea that physical reality is primary, or that physics is the fundamental science. By Plato, I mean (to simplify again) the idea that physical reality is the border, or the shadow, or the projection, or the creation, of a non physical vaster reality (be it mathematical, theological, computer science theoretical, arithmetical ...). MEC makes it arithmetical, because it becomes absolutely undecidable. It makes it also theological when listening to what the machine say and stay mute about, or say with interrogation mark.

Obviously it is trivial to show that the physical universe is redundant,

It is not trivial. It took me 30 years to make about ten person understanding the entire thing. It is the whole point of the proof. It shows the falsity of physicalism. I have come on this list to explain that Tegmark's idea that the physical universe is a mathematical object among others cannot work, assuming CTM, due to the first person indeterminacy. I think that you are still using the identity thesis in the philosophy of mind. Tegmark is still guilty, if you want, of a form of physicalism, by assuming that the physical universe might be a mathematical structure among another. Physical is undefined, and mechanism, when taken enough seriously, leads to the idea that the coupling consciousness/realities is a purely arithmetical phenomenon. The only way to show that the physical universe is redundant consists in showing how the physical laws appear to be believed in absence of physical universe(s). This makes physics no more a fundamental science, but a science which has to be explained from another science. With MEC it can be shown that the other science is arithmetic, or any first order logical specification of a universal (in the Turing sense) theory (that is a theory in which you can represent the partial computable functions). The internal epistemologies are handled by modal variant of self-reference.

At some level it is trivial: if you agree that mind is immaterial, you might understand that it is easier to explain the illusion of matter to a mind than the reality of the mind to matter.

MGA is the movie graph argument, first published in 1988. See my URL for the exact references.
UDA is the Universal Dovetailer Argument, first published in 1991.

The main part of the work is AUDA (Arithmetical Universal Dovetailer Argument). It is an abstract, but arithmetically concrete, version of the UDA that universal machine finds by themselves when looking inward.

MGA and UDA are mainly tools for explaining that science has not yet solve the mind body problem, and that science has not yet decide between Plato and Aristotle theology (in the greek sense of the word).

but the move to show that it is disproven I do not follow.

The contradiction is epistemological. The disproof has to use some (weak) form of Occam razor. You are right.

Essentially, I do not follow your argument that "I. The Universal Dovetailer Argument shows why comp necessarily *forces* a reversal between physics and machine psychology"

The UDA argument contains 8 steps. Just tell me where you have a problem. From a theoretician of the mind perspective it can be seen as a reduction of the mind body problem to a new body problem, itself reduced into a mathematical problem in computer science. The first seven steps of the UDA already entails that CTM entails indeterminacy, non locality, and even non-clonability (although this does not appear in that paper). The eighth step, more or less equivalent (and older) than Maudlin shows directly the incompatibility between weak materialism (MAT) (MAT = the doctrine that there is a primary physical universe, or epistemologically, the doctrine that physics is the fundamental science) and mechanism (MEC, comp, CTM, ...).

You quote Maudlin's “Computation and Consciousness,” The Journal of Philosophy, pp 407-432, as having more complete arguments.

Maudlin's argument provides more information, but MGA is complete, as an argument. MGA is simpler somehow, and Maudlin got the idea, independently of the MGA, in one of his sum up of Olympia-Klara argument.

However, on page 25 he states "Olympia has shown us at least that some other level beside the computational must be sought."

Maudlin's starts from MAT. I start from MEC. This explains the difference of wording. But both Maudlin and me show, essentially, the incompatibility between MAT and MEC. To keep MAT, you have to abandon MEC, either by adding new axioms to it (but Maudlin is already aware that this can hardly work), or by just weakening MAT, but this leads to the result I describe in this list, but which I published in the eighties (in french)).

"Our Olympia demonstrates that running a particular program cannot be a sufficient condition for having any form of mentality"

Assuming MAT this is correct. You might be obliged to introduce a substantial soul, or to change MEC for another hypothesis, making us non Turing emulable. More easily, just abandon MAT. In that case, the (Gödel, Löb, Solovay) self-referential constraints illustrate that you cannot not recover physics from arithmetic indeed, without changing anything in MEC. This makes the MEC theory 100% testable empirically: extract physics from MEC, and compare it to the inferred physics from nature. I have extracted the logic of the measure one for the observable laws, and it is indeed arguably close with quantum (ortho) logic. To be sure, this tests not really (pure) MEC, but MEC + the classical theory of knowledge and matter, of the platonicians (Theaetetus (Plato), and Plotinus).

The main point of his complex examples seems to be that the same output supervenes on two very different mechanisms, but this does not force a reversal.

I'm afraid you miss Maudlin's point. His point is that you can reduce the physical activity as much as you want to, for any particular (conscious) computation, by putting the needed counterfactuals in physically, as far as computationally, inactive pieces of device, (Maudlin's Klara), so that it makes the (physical) supervenience thesis incompatible with MEC. This has nothing to do with multiple realizabity, or functionnalism, at that step.

MGA is more easy, because it substituted a physical computation into a physical process mirroring exactly the physical activity of the physical computation for a particular computation, but doing no more any computation at all. It means that consciousness supervenes on the abstract computation, not on his physical implementation. It is not a so easy point because it is important to distinguish a computation and a description of a computation, even if in Arithmetic this can be considered as equivalent (but machine cannot really see that equivalence for herself). This is made utterly clear by using the logic of self-reference. For p (sigma_1, that is computable) we have the truth that p is equivalent with provable(p), but no sound machine can prove that. Löbian machine can prove p -> provable(p) for p Sigma_1, and this helps a lot for interviewing the machine on the UD.

MGA is *a* subtle point. People take more time for this than for UDA1-7. They forget it also more quickly. I am afraid. And the conclusion might be hard to swallow: with comp, mec, CTM, consciousness does not supervene on the physical activity, but on some (abstract, relatively) implemented person (the turing universal person). There is a Galois connection: the less information is treated by that person, the more possibilities she has access to.

Could you tell me the central piece of the logic as you see it in simple terms.

I can tell you them in so simple terms that any Löbian machine can understand it.

With UDA and MGA I ask *you* the human, really. It is already a decomposition of the arguments in simplest terms. I have no problem to explain UDA1-7. UDA_8, which is MGA, is quite intriguing but makes its curious point when people look close enough. In a way MGA transforms the physical supervenience into a Searle type error, a confusion of level.

Read cautiously and tell me at which step of the UDA reasoning you have any trouble with. UDA needs only a passive understanding of Church thesis (to get the universal dovetailer, the key role of the Sigma_1 arithmetical sentences).

With AUDA I ask the ideal self-referentially sound machine. For this you need to read Gödel 1930, 1931, Löb 1955, and Solovay 1980, and some others. Then there are some good textbooks on the formal self- reference (Boolos 1993, 1979).

The whole thing is made possible thanks to the closure of the partial computable function for the diagonalization procedure.

Comp is perhaps a more precise version of CTM in the sense that it assumes and exploits Church thesis and thus the theory of partial computable functions, computable sets, and relation with mathematical logic, arithmetic, etc.

The arithmetical reality emulate all possible "dreams", and physical reality is a first person plural (sharable: it really means that we are collectively here) dreams.

In simple term MEC => "matrix", like in the movie, except that thye first person plural sharable matrix are defined (from inside, and in the limit) by the additive and multiplicative structure of positive integers.

Any first order logical specification of a universal system (in the Post, Turing, Church, Kleene sense) will do. Even just a polynomial diophantine equation of degree 4 will do (by Jones, Matiyazievitch results).

Then you can extract the many invariant by the invariance on the self- referential change. The laws of physics does not depend from the initial basic ontological theory. To start from the quantum is treachery. It is like the choice of a special universal number (the arithmetical code of a universal quantum dovetailer (or other modular functor)).

But the UDA+MGA reasoning invite your to derive that special u from a matter theory where matter appears as a modality of self-reference by the universal machine. The advantage, (forgetting the "have to"), is that, well Penrose was wrong on Gödel, Löbian machine can, on the contrary, already prove their own incompleteness: if I am consistent I cannot prove it, and actually, it is mathematical trivial, starting from a theorem by Solovay to show that machine can access to the gap between truth and their own provability.

A bit like I can explain to you that IF I am conscious, then I have no means to prove that fact to you.

That gap, I call it the Solovay G/G* splitting: it gives a couple of modal logics: G axiomatizes what a lobian machine can prove about herself, and G* axiomatizes what is true about the machine. At some level(s) (0 in modal logic, 1 in arithmetic) it is axiomatizable and decidable. When you define the points of view in modal term from the Gödel arithmetical provability predicate, your logics inherit the G/G* splitting which allows you to distinguish the communicable part with the private part. As expected the material variant splits into quanta and qualia. How close those quanta isolates from a rush application of the Theaetus in arithmetic (made possible by Gödel's theorem) remains to be seen.

At the CIE, I published an (accepted) paper which explains all those intensional variants of Gödel's provability predicates in term of Plotinus hypostasis, making explicit that a strong Löbian machine can study the complete propositional theology (i.e. prrof and truth) of a simpler machine, and then lift here to herself as far as self confidence obtains. It makes also explicit that Mechanism is closer to a (neo) Platonist theology than an Aristotelian one.

Andrew, I think the singularity has already occurred. It is the discovery of the universal machine. Babbage, or Turing, or Post, or Kleene, etc. The quantum computer and the Löbian person are or should be (assuming MEC) consequences of it.

The bit -> qubit is a two way path, but in the bit -> qubit direction you can split the logic, (by just taking into account the reflexive incompleteness: it gives (arguably) qualia and quanta. If the quanta works well, it will indirectly justified the qualia. Consciousness got also a role, of self-acceleration, again by a speed up theorem of Gödel, mainly, which describes transfinite autonomous progression "from G to G*", in term of their arithmetical contents.

I am just taking the mind body problem seriously and the hypothesis that there is a level where I am Turing emulable, by using what Turing, Kleene, Myhill, Feferman, and Gödel proved on the subject. This is comp! CTM is fuzzy about the level. Comp is a weaker assumption than CTM, I think now.

You get Löbianity from universality by adding the induction axioms. It is very cheap, but non trivial. Peano Arithmetic, Zermelo-Fraenkel set theory, ... are example of Löbian machines.

UDA+MGA is for the human babies,
AUDA is for the Löbian babies,

There is no simple terms. The problem is highly complex. My work is before all, an attempt to formulate the mind/body problem mathematically, by using an hypothesis (comp) which eventually (by MGA!) makes computer science fundamental. Matter is generated somehow by the gap between computer science and computer's computer science, if you look close enough using mathematical logic. I am not a genious: from Gödel to Solovay, they made the dirty and hard work. I describe this in the long version of the thesis where I sketched very closely Solovay's demonstration, and the same for all the main use of the second recursion theorem of Kleene.

Enjoy, and ask *any* question, on any part. You might try to read sane04 entirely. You understand already the first person indeterminacy, which shows a pure indeterminacy occur from machine self-duplicability. Do you agree that delays of reconstitution does not change the first person experience? Etc. It is a reasoning, a deductive reasoning. A transformation of the mind body problem into a computer science theoretical body problem (and thus a conceptual explanation where the laws of physics could come from).

For AUDA, think about me as the guy saying: oh look we can already listen to the universal machine, or: guess who come to dinner tonight.

Sorry for being too long, or to short.



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 
For more options, visit this group at 

Reply via email to