On 10 Aug 2009, at 17:20, David Nyman wrote:
> 2009/8/10 Bruno Marchal <marc...@ulb.ac.be>:
> Bruno, I'm broadly in agreement with your comments, and merely
> re-emphasise a few points below on which I'm being a stickler.
> I have some further comments and questions on step 8.
>>> In this light
>>> it becomes self-evident that any and all explanatory entities -
>>> physical, computational, or whatever - are severely restricted to
>>> domain of epistemology.
>> I don't see why. I would not put arithmetic in epistemology, or only
>> in a very large sense of epistemology, the epistemology of the 0-
>> person views!. It seems clearer to accept them ontologically like in
>> the usual practice of math. Could be only a vocabulary problem here.
> Yes, I normally wouldn't dispute this point, but but I'm being a
> stickler here. In the strictest sense the ontological equivalence of
> anything whatsoever to the indexical OFP can only be an assumption -
> albeit one that might be justifiable in the case of the best theory.
OFP, OFP, ... ? Ah Ontological First Person.
Actually in "indexical OFP" I could argue that "indexical and
ontological" are just reminders on point where we already agree (and
agree with the universal machine discourse, after some definition and
But with comp, the first person does not necessarily identifies with
any representation, like we don't have to identify ourselves with a
body, but we can bet on "staying the same person in the sense that I
stay the same person after drinking a cup of coffe" (SSPSSSPDCC, if
you want :) for some digital body transformation.
The 1-I, the 1-person, does not need to identify itself to any body,
or representation (actually: the 1-person cannot do that, it can
identify itself only to itself as a person).
So the 1-person bet on some 3-person (the body, the code relatively to
its most probable universal computation), and on its invariance for
some (digital) body transformation.
Assuming comp, the reasoning leads to make consciousness a larger
invariant (and then you have more choice and latitude in the
transformation, for example with amnesia, or false souvenirs, etc.).
But strictly speaking (I am also a stickler), the first person can
never identify herself to *any* representation, she share this with
the 0-person ONE, or the non differentiate (arithmetical) truth. The
knower does not know who he is. Relatively to probable histories, he
makes bets "all the time".
> Beyond this, that the unique qualitative nature of the OFP *is* as it
> appears, is in principle outside the scope of explanation itself.
No! This is the "miracle" of comp. Machine cannot not discover the
incommunicable part of their experience, but they can, assume of bet
on comp, and justify that why it has to be so. That is AUDA. The gap
is justified from inside. It is a consequence from the fact that
machine can prove their own incompleteness theorems, and even study
the geometry of their ignorance. The ultimate gap remains unavoidable,
so you are right saying that the unique qualitative nature of the OFP
is outside the scope of the explanation, but that fact, is an "easy"
theorem on and by the machine which introspect herself.
To sum up:
The unique qualitative nature of the OFP *is* beyond the scope of the
explanation-comp theory. But that very fact *is* in the scope of the
>>> IOW - as Bruno says above - they are
>>> theoretical constructions.
>> Yes, but this does not mean those construction does not refer to
>> something real independently of us, and this is what I assume for
> I agree, as above that it is the whole point of our endeavours to say
> that the construction *refers* to something real. But I think perhaps
> that the something thus referenced is not best characterised as being
> real *independently* of us, but rather *constitutive* of us and our
> (most general) environment.
I agree with you, but this can be said among enlightened people who
understand the whole stuff.
Before the reasoning, you could be suspected to put the horse behind
With comp, numbers, or finite things like combinators etc. have
clearly a relation with us, but a priori it is simpler to state their
laws without referring to us.
A number is even if and only its square is even. This is a law about
numbers. Those are the type of truth which we have to state as not
depending on us <here and now>, even if it depends on us, or are us,
>> Even if the whole existence get annihilated, 17 would still be
> I understand that it is justifiable to take this as your point of
> departure and don't really wish to make an argumentative point out of
> it. Nonetheless, in passing, perhaps I have a more radical intuition
> of annihilation than you. One can waste a lot of breath speculating
> on 'nothing' because, strictly I guess, there can be nothing at all it
> can refer to.
This I do not understand. There are many nothing everywhere, and other
absence, and I am open that absolute nothingness could be conceivable,
a bit like theories having no models. It seems you just point here on
a difficult open question.
> I could demonstrate this, given infinite time, simply
> by flatly rejecting *any* survivor of such annihilation that you or
> anyone cared to propose, to the crack of doom. On this basis, even
> '17 is prime' is a goner.
I still don't see why or how you could do that, except by convincing
me that Peano Arithmetic is inconsistent.
>>> So far so obvious. But - as has again been recognised
>>> immemorially -
>>> solipsism is a dead-end and hence we seek a theory to capture the
>>> relation between the OFP and its environment. But immediately we
>>> faced with the notorious 'explanatory gap', and it seems to me that
>>> its most precise expression is in the gap between ontology and
>>> epistemology. Indeed, what conceivable strategy could raise these
>>> theoretical constructions - to which the OFP uniquely lends
>>> - to the ontological certainty of their host? Is there a coherent
>>> to conceive what it could mean to *be* a theoretical entity (as
>>> opposed to postulating or observing one)?
>> It is the point of saying "yes" to the doctor. You don't say yes
>> because the new brain is a good modelisation of your brain, but
>> because you bet it will enact yourself completely, relatively to the
> Yes indeed, this is my point. There is no way to *conceive* in
> advance what it would mean to *be* such an enactment (i.e. to be
> *sure*) so you can only bet that saying yes will not affect the state
> of the indexical OFP.
Yes, thanks to the "(i.e. to be *sure*). Yes. Of course we can
conceive in advance what it could mean to be such an enactement. It
would be like me or you but with souvenirs of going to the digital
hospital and coming back, or like using teleportation, and
communicating themselves at the speed of light in the neighborhood,
like those very words.
>> You may confuse the reality of number, and the reality of machine/
>> theories talking about those numbers. Numbers are not viewed as
>> theoretical construction. The theoretical construction are our
>> theories on the numbers. It simplifies things.
> I agree that this assumption simplifies things, and as you say it is
> one shared by all mathematicians. But again, in the final analysis,
> numbers can only be 'viewed' as ontologically real, not *known* to be.
> But this is true of any assumption whatever, and I freely concede
> your points about the simplicity of the assumptions in the case of
>> All theories demands faith, but the faith needed for understanding
>> that 17 is prime is not comparable to the act of faith needed to say
>> yes to the doctor.
>>> It's no use appealing to
>>> notions of 'what it's like to be a brain' - nor what it's like to
>>> be a
>>> COMP-quale - because we can never say that it is 'like anything to
>>> the stuff of epistemology.
>> Assuming comp we can still say that it is like you feel right now.
>> Only zombie cannot understand, but if they are good zombie, they will
>> have no problem to fake that they understand.
> Yes, *assuming* comp. We cannot *know* what it is like to be a
> comp-quale, but we may have sufficient faith to bet that it's like
> 'what you feel right now'.
Yes. Actually this is even a problem for real practitioners of comp,
they may come to believe they know that comp is true, and then they
will be inconsistent. Comp entails the existence of sort of possible
future "delusion", it may already be a way to keep alive the past
delusion. There is a sort of paradox here a machine clever enough to
understand or bet on an artificial brain can understand the artificial
brain is not needed. The problem is that you don't need to understand
the implication of saying "yes doctor" for saying "yes" to the doctor.
A common problem with technologies, already related to their
>>> Hence we must see our theorising and
>>> observing - in physical, computational, or whatever terms - *in
>>> relation* to ontological certainty, not as constitutive of it.
>> That's right.
>> I thought this was obvious.
> You may have heard the following story. A professor of mathematics
> enters the lecture room with a sheaf of papers and writes a complex
> theorem on the blackboard. He turns to the students and says "ladies
> and gentlemen, this of course is obvious". He then shuffles his
> papers, looks at the board again and continues more doubtfully "at
> least, I think it is obvious". Then he stares fixedly at the board
> for ten minutes without speaking, looking increasingly uncomfortable.
> Finally, he rubs out the theorem and leaves the room. The students
> are nonplussed, but remain in their seats. Thirty minutes later, the
> professor re-enters the room, looking disheveled but happy. He writes
> exactly the same theorem on the blackboard with a flourish, turns to
> his audience and announces triumphantly "ladies and gentlemen, I was
> right - it IS obvious!"
I guess it was obvious, but not obviously obvious :)
Yes in math there is the obvious (that nobody understand)
Then there is what is obviously obvious (for those who still does not
And so forth.
Normally "obvious" means, assuming a non sadistic teacher:
- it follows in one step from the definition or from the last theorem.
Making it a very relative notion.
>>> Rather, they stand in some theoretical
>>> relation to RITSIAR, but strictly on the epistemological side of the
>>> explanatory gap. They are 'real as far as theory takes us', or if
>>> further jargon is unavoidable: RAFATTU.
>> Well frankly this will depend of the first "T" of RAFATTU. It depends
>> of the theory.
>> With the comp theory, quarks, electrons, planet and galaxies are not
>> ontologically real.
>> With string theory, they may be real.
> Yes, real in the relative sense we have been discussing.
OK. And number are real in that sense, too.
Usually, even physicists, have less doubt about the existence of even
numbers than on the standard modal of elementary particles.
>>> My argument against the *physical* instantiation of a computational
>>> mind (i.e. in any non-eliminative sense) rests on the claim that the
>>> very arbitrariness of possible physical instantiations of a given
>>> computation (cf Hofstadter) violates the criterion of direct
>>> supervention on *specific* physical entities and relations from
>>> a class of emergent phenomena inherits physical - as opposed to
>>> mental (and hence egregiously question-begging) - stability.
>> I think that this what the movie graph argument makes necessary.
> Good, I'm glad we agree.
>>> Naturally, all this is per physics as ordinarily understood.
>>> Tolerating such a violation is tantamount to accepting (and this is
>>> notoriously claimed by Hofstadter et al) that *any* arbitrarily
>>> assembled set of physical entities deemed to be in the required
>>> 'functional' relation (e.g. - famously - in an anthill) necessarily
>>> stabilises exactly the same 'mental state'. AFAICS this is by
>>> quite sufficient to reveal such a 'mind' as intrinsically
>>> unphysical -
>>> and a fortiori un-RITSIAR.
>> Hmm... things are more subtle, but in a first approximation this can
>> be useful.
> Remember, I mean 'unphysical' in terms of standard physical theory.
> By un-RITSIAR, in this instance, I mean that in these strictly
> physical terms (i.e. without comp) such a mind doesn't even exist -
> hence 'a fortiori'.
>>> My argument assumes, however, that - per physicalism - a running
>>> computation (as opposed to its specification)
>> It is important to distinguish a program (immaterial static object)
>> and a computation (an immaterial "dynamic" object, which can be
>> or infinite, but is best handled when accepting it is infinite,
>> because in that case a dynamic can be defined by a function from N to
>> anything: it is a sequence of things). If not, people get troubled by
>> the existence of running computation in platonia/arithmetical truth.
> Ah, but my argument attempts to distinguish a computation (immaterial
> dynamic object) and an implementation of a computation (material
> dynamic process) - again, per standard physical theory - as a
> refutation of standard comp *in these strictly physical terms*. My
> point is, that per physicalism, a computation must be implemented in
> some physical mechanism in order to have any real - i.e. physical -
> effects (at least this was true the last time I did any programming).
> Hence the existence of 'immaterial objects' in this case is simply
> irrelevant to any effects that would be strictly justifiable as
> ontologically real, per physicalism.
Actually I do disagree with this, and Peter Jones made good point
here. If you were true, UDA could be simplified a lot. Physicalism
does not prevent dualism form/matter at all, like immaterial software
and physical hardware. Physicalism can still explained the existence
of the many immaterial being like nations, countries, persons (being
fuzzy on the 1-3 distinction), games, music, etc.
But comp, pushed on its logical conclusion prevent physicalism to
explain the 1-person stable observations. We have to explain the
appearances of all observables protagonists from the relation between
numbers. There is a problem of vocabulary deciding if matter disappear
or not. It is easier to say that physicalism is shown inconsistent or
epistemologically empty. But this is not an entirely trivial
proposition to demonstrate; it is full of traps, and without QM, nor
Post or Turing, I would never have believed this, nor find the
argument. UDA needs the universal machine concept.
> Consequently, I agree that the
> reversal of ontological primacy you stipulate is necessary to save
Comp doesn't let much choice: like biology got an evolutionary aspect,
physics got an evolutionary aspect in logic and arithmetic. Comp just
show that things are far more complex than physicalist, or even some
mathematicalist, tend to think.
Comp provides a relatively easy theory of mind (computer science, self-
reference logics and their intensional nuances) but present a big
problem : extracting the qubits from a relative dreaming property of
> Your argument however seems to be based not on the physical
> implementation but the 'immaterial' computation to which it is
> postulated - per standard comp - to be equivalent. Is this right?
> (More on this below.)
This is unclear. UDA1-7 can be based on the physical implementation of
the devices presented in the protocol of the thought experiment. UDA-8
eliminates that assumption.
>>> necessarily requires
>>> *some* physical activity to transform inputs to outputs (e.g. in
>>> of logic gates). Step 8, however, seems to take a step beyond
>>> this by
>>> proposing that a running computation can take the form of (as
>>> to merely being described by) a machine *state*: i.e. without the
>>> requirement of activity.
>> No, activity is required. But activity is just a function from N to
>> set of states. The movie graph show that a machine cannot distinguish
>> physical activity from such an arithmetical activity. A computation
>> an infinite set of numbers such that there is a universal number
>> generating that sequence.
> When you say that "activity is just a function from N to set of
> states", you again seem to refer to 'immaterial activity'. It seems
> to me that what you are saying amounts to this:
> If it is the case that, per comp, it is the 'immaterial' activity of
> the running program, regardless of specific implementation,
Careful: the mind needs specific implementations, but below its
substitution level, all implementations acts simultaneously, from its
perspective (in case it decides to take a look below its level, he
should see this).
Once the reversal is done the word "physical" is reduced to something
very specific, which includes a (perhaps to high) first person
> implements the function and hence the mind, then this is
> indistinguishable by the machine from it simply *being* the function
> and hence the mind.
The mind (the 1-mind, not the numbers) is distributed in the set of
all computations, you cannot really attach it in any single
computation, unless you have reasons to think it is correlated to your
own histories. Remember that at UDA-8 we abandon the physical
supervenience for a comp supervenience, which is hard to describe
without the minimum amount of math I was talking about.
I may insist that the 1-person, nor its consciousness, is never
attached to any thing which can be represented. Only its third person
vehicles are, and the person can make relative and local bets on those
vehices and their relative stabilities.
This is recover and made consistent by the nuance between the prover
Bp and the knower Bp & p (and the feeler, Bp & p & Dt); which by
construction are the same, yet they cannot know that.
> Standard comp is then seen to refute - or at
> least make irrelevant - its own basis in materiality. Is this right?
comp + physicalism is either contradictory or epistemologically
empty (person eliminativist).
The proof is constructive. It gives the place where you can studied
where the appearance of physics come from.
This will provide jobs for mathematicians of the next millenia.
>>> And in any
>>> case, in what way is step 8 intended to extend intuition beyond my
>>> argument, which - as I have tried to show - also elicits the insight
>>> that the direct supervention of 'functional' relations on functions
>>> themselves - not on their arbitrarily-defined physical tokens - is
>>> central to the recovery of 'mind' from computation.
>> See above.
> Have I succeeded in answering my own question?
I let you judge.
>> The discovery of the universal machine is the creative bomb which
>> makes comp possible and plausible. Universal machine, like computers
>> and brains, are not trivial mathematical object at all. To study comp
>> without computer science, is like doing cosmology without QM and GR.
>> Of course we can have the deepest intuition right, by experience, but
>> to make a sharble verifiable theory, I am afraid we cannot dismiss
>> some math ...
> Vous avez raison cher maitre, j'en suis sur.
You mean "Vous avez raison cher maître, j'en suis sûr."
Stickler in French is "pointilleux", which means literaly asking for
the correct accents!
I cannot wait for your questions on step 8 :)
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To post to this group, send email to firstname.lastname@example.org
To unsubscribe from this group, send email to
For more options, visit this group at