On 23 Dec 2011, at 00:19, Russell Standish wrote:
On Thu, Dec 22, 2011 at 12:06:54PM +0100, Bruno Marchal wrote:
On 22 Dec 2011, at 00:30, Russell Standish wrote:
Maybe you haven't explained the 323 principle properly. My
understanding was that if a program did not need register 323 in
to be conscious, then it would be possible to remove that register,
and still run that program.
However, if that register is required by the program in order for it
to handle counterfactuals, then it is quite possible that it is
for consciousness, and that removing it will render to program
unconscious. IIUC, the latter case is not covered by the 323
The 323 principle *is* that later case. It says explicity that if a
program does not need some register FOR A SPECIFIC COMPUTATION
(supposed to be conscious, with sup-phys) then that specific
computation will remain conscious when the register is removed.
I was correct that you were arguing against 323.
You have to explain how an inactive machinery has a physical
activity in a computation.
Because in the latter case, the "inactive" machinery is not really
inactive. It is only inactive in one branch.
If the activity of the piece in another branch plays an active role in
the computation "here and now", then it is part of the computation,
and I have to ask the doctor that the artificial brain takes the other
branches into account.
Either that other mechanism is Turing emulable, or not. If it is,
MGA will apply, unless you are using a supervenience violating 323.
This needs magic, it seems to me. The kind of magic whose existence
would make me doubt that I can survive "qua computatio".
By the sounds of things from what you've described above, I am using
supervenience violating 323. And no, I wouldn't describe it as
magical, no more magical than the fact that the irreverible 2nd of
thermodynamics supervenes on reversible Newtonian dynamic.
It is different. We have an explanation of why 2nd emerges, from the
laws of big numbers.
But you are asking for a physically inactive object (with respect to a
computation) to have a physical role.
(Although you seem to abandon that idea for saying that the film
(projection in time and place) is conscious).
But I don't
need to provide a mechanism to find a flaw in the argument - the
argument needs to prove that no such mechanism can possibly exist, and
I don't think it does that.
If such a mechanism exist, it can be simulated. if not, comp is false.
In any case, Maudlin's argument does not assume the 323 principle,
Yes, that's the role of the Klara, but it concludes that we have to
provide a physical activity to something which, for a particular
computation is physically inactive, and functionally equivalent to the
absence of those piece. He shows that the consciousness is not related
to physical activity, but to the abstract computation.
Quantum computing is a specific process of exploiting entangled
states. I do not mean that, and think it unlikely that nature works
that way (contra Penrose). Supervenience over multiple MW branches
does not entail that sort of quantum computing.
But if would make comp false if that kind of supervenience is not
Why? Turing emulability of the physics that supervened on is surely
Unless it plays a role at some level, like you suggest by attributing
a physical activity to something which is not used in a branch, but
might be used in another branch, for physical reason. In that case,
*such* physical acitvity is relevant, and has to be taken into account
in the artificial brain, as it will be in the UD.
in the 3rd person dovetailed POV. And, I find it hard to think
dovetailer as conscious.
It does not need to be conscious. It needs only to instantiate
consciousness. As it does, even in "time and space" when it is a
concrete UD running in a primary physical universe, and that is all
we need to get the epistemological contradiction: primitive matter
has no interaction with consciousness. It becomes like invisible
horses pulling cars.
I have no problem with the statement "primitive matter is like
invisible horses pulling cars". But this is not the MGA.
? It is, or should be.
Because the MGA aims to show something stronger - that primitive
matter is actually incompatible with COMP.
It shows it to be useless both for consciousness and the appearance of
Why does supervenience over multiple parts require that the parts
interact? Are you invoking emergence here?
If they don't interact, they don't play any role in the computation.
How does that work?
By definition of a computation. Like with QM, independence is
isolation. Computation relies on the causal, or arithmetical,
relations between parts. Non interacting parts can play a role in the
measure on computations, not on any singular computations.
Consider a dovetailer program running in some well defined
physical machine. In
particular, the states of the machine, its transition table and
machine code are all defined in terms of whatever physical process
used to instantiate it.
At each step, the dovetailer program runs one step of a program,
switches to the next program (or next input) as appropriate.
Now consider the sequence of states the machine passes through. All
state transitions are legal (exist in the transition table),
it wouldn't be a program.
So we could consider the dovetailer as running two steps of a
before switching to the next. Or running three steps of a
before switching to the next. Indeed, there must be 2^\aleph_0
valid ways of
partitioning the running program into dovetailed programs, the vast
majority of which presumably do not dovetail on all possible
One way of partitioning the UD so that not all programs are executed
is to only switch tracks if the machine's next instruction is
instruction 0. This means that all programs beginning with
1 are never executed.
Does this not mean that the fact that the UD instantiates conscious
programs is purely in the eye of the beholder? So to say that
dovetailing preserves supervenience would be an incorrect
What matters is that you dovetail on conscious programs, and the UD
dovetails on all programs. Why would a dovetailer missing a program
have to generate the consciousness of that program?
That a dovetailer does not instanciate the consciousness of the
conscious program would mean that delays of reconstitution, or
virtuality, prevents the existence of consciousness. Are you
backtracking on UDA step 2, and step 6?
We are always focussed on step 8. What I am saying is that there is
no unique interpretation of the UD is actually doing.
It is, from the first person points of view, and that is what counts.
the UD still executes all programs anyway.
Necessarily. (We accept Church's thesis 'course).
One way of avoiding the
above example is to note that the UD will also execute all programs
that differ from another program only by ignoring the first step (or
ignoring the first input).
Could it be proved that a UD always executes all programs (albeit in
different order and possibly different measure), regardless of the
partitioning of UD*?
Yes. Consciousness does the partitioning, like consciousness chooses
between "I am in W" and "I am in M" after a self-duplication. By the
compiler_interpreter theorem the UD is quite robust with that respect.
This would differ from a non-universal dovetailer, of course, but
could follow from the "miracle" of Turing completeness being closed
This still does not eliminate the fact that supervenience in the case
of a consciousness generated by a UD is on the emulated physics
(Brent's point), not on whatever the UD is constructed from.
So this doesn't contradict the possibility that the Multiverse is
"primitive matter". The "sire, je n'ai besoin de cet hypthese"
argument is something completely different, it is not what the MGA is
trying to address.
Indeed, it is somehow worst. It is not " Sire, I don't need that
hypothesis", it is "Sire, I cannot use that hypothesis".
It shows that primitive matter has nothing to do with any perception.
It makes primitive matter a God-gap type of explanation.
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