Hal wrote
>¿ (At) 12:16 +0100 28/01/99, Marchal Ècrivait (wrote) :
>>Bruno : Put in another way : with QM you have many "worlds", but, a
>> priori, not
>>as many than with COMP. With COMP I think we must justify, in a purely
>>non empirical way, why the vast majority of computationnal state we are
>>living seems to be quantum computationnal state.
>
> Hal : I agree that this is one of the tasks ahead for this model, but it
>is going to be long term.
May be. My feeling is that computer science can provide theoretical shortcuts.
> Hal : One direction is to rule out alternative
>physical models by showing that they don't seem to be able to support
>life. Max Tegmark has a diagram at http://www.sns.ias.edu/~max/toe.html
>showing that small variations on our current physical laws would seem
>to make life much more difficult.
B: I agree, although there is a weakness in this approach due to the restricted meaning of the term "life" we can give. This restriction comes from our carbon-being prejudice.
But it is certainly an interesting procedure. And the anthropic argument here provides indeed useful information.
Hal :
>Another approach would be to try to create simpler universes which would
>still be able to support life. We could even do computer simulations to
>try to show that life seems to be evolving. Failure to successfully
>demonstrate realistic evolution would be evidence that you needed more
>like the complexity of our own quantum laws.
That would take a long time too. Why not generate all (simple and complex) universes ?
I will try be more precise because it will perhaps put some light on a question you ask in another mail (where you quote Wei Dai in the series "RE:consciousness based on information").
>Maybe you could have a continuous infinity of Turing machines, but I'm not
>sure you could have a continuous infinity of Turing machine *programs*.
>
>It would depend on whether it made sense to have an infinitely long
>program. If we required legal programs to have some kind of syntax, say
>a balanced number of parentheses, or some kind of end symbol (which is the
>approach adopted by Greg Chaitin in his considerations of similar issues)
>then any syntactically legal program would be finite. I don't believe
>this impairs the universality of the TM by the standard definitions.
>
>If all TM programs must be finite, then there can be only a countable
>infinity of programs. If we allow the notion of an infinitely long (and
>infinitely complex) computer program, then there would be an uncountable
>infinity of them.
>
>Which of these views seems correct?
It is a consequence of Church thesis that there is a little program able to generate and execute all programs, with any number of inputs, including all the finite approximation of the reals, complex, etc...
I call it the universal dovetailer (UD). Basically, it generates the first program (in let us say Fortran) and execute it a little, then it generates the second program, go the first to execute it a little, then he execute a little the second, then he generates the third, go back to the first, etc.
That program will emulate all program in all possible (with Church thesis) programming language. In particular it will emulate all quantum computer, it will dovetail on all the solutions of DeWitt-Wheeler equation, etc..., With all real and complex value for any constant, etc... (because it dovetail on the "fan" (the set of all the finite approximation) of the real and the complex, etc.
UD exist. (I have written one in LISP, for exemple).
Of course, it is not really useful, you must wait billions of years before anything interesting happens.
I do not even believe that our local "branche of multiverse" is sufficiently robust to support the infinite execution of the UD. For exemple, we need at least an ever self expanding cosmos.
But for the sake of the argument, let us suppose that a UD is executed, and let us suppose our cosmos never evolves neither toward a big crunch, nor a vast cold quasi-emptyness, so that the UD never stops.
*** take some rest here ***
Now, suppose COMP true, i.e. I can survive with a digital brain (or body, or universe, but the thought experience are simpler with a brain).
In that case you are, like an amoeba, duplicable. That is, your instantaneous state can be saved, your original body can be destroyed, and you still survive to a reconstitution.
If you are reconstituted in two place A and B simultaneously, then, before the experience, you cannot predict where you will feel to be. This is a first person objective indeterminism. I call it mechanist interpretation.
Because I don't want to be too long, I refer to
http://iridia.ulb.ac.be/marchal/ for simple thought experiments showing
that :
1) the quantification of mechanist indeterminisme does not change if a delay of reconstitution is introduced at B (for exemple). That is pretty obvious, in fact, at least in this discussion-list.
2) the quantification of mechanist indeterminisme does not change, if real environment are replaced by virtual one, i.e simulated by a computer.(Pretty obvious too, remembering we postulate COMP of course).
3) if I am reconstituted in ten similar instance in A, and one instance in B, then the quantification must take this fact into account, in particular it is more likely that I will find myself in A. (This is less obvious, although Mallah seems to express that view somewhere in the discussion list).
*** ok so far ? ****
Now remember we are in a cosmos in which history a UD never stops.
In such a cosmos, to make a simple prediction, because of 1), 2) and 3) you must take into account all your (infinite) reconstitutions. In fact the probability you stay in "the real" cosmos is zero, for each instant you can predict you will be virtually reconstituted by the UD.
And a priori mechanist indeterminism explodes ... (but see my last mail where I suggest it would be a too easy way to refute mechanism).
Note also, that because of 2) you must really take into account (for computing the relative probability) the whole execution of the DU. There will be a continuous set of semblable turing machines (and of course also of less semblable) Turing machine.
And this (I hope not too long nor too short mail) should answer your question in "consciousness based on information or computation).
But concerning : (I quote Hal again)
>We could even do computer simulations to
>try to show that life seems to be evolving. Failure to successfully
>demonstrate realistic evolution would be evidence that you needed more
>like the complexity of our own quantum laws.
If you are rigth here, you should be able to extract a kind of quantum UD, perhaps "the equation of the universe" from the UD, with an explanation why this special natural dovetailer provides much more instantiation of the (relative) you than the rest of the UD (or of course (as I say to Gilles) you can try refuting COMP).
It is plausible that indeed the solution is here, in a "natural UD". In that case the theory of everything could be string theory, for exemple, but with COMP, it would mean that string theory could be deduced from computer science or (number theory).
[All the counterfactual stuff in my work is used for showing you don't need a "real" execution of the UD. You can also, in the manner of James, use some strong form of Ockam razor. After all, with Church thesis the UD exist like PI, and the probabilities are language or machine independant].
A+ Bruno.
Received on Fri Jan 29 1999 - 14:05:53 PST