Fred Chen wrote:

>I appreciate how something like the Universal Dovetailer or equivalent
>programs can generate an infinite set of programs that could include the one
>that describes our universe (including our consciousness).

You are confusing Schmidhuber-like theories with me-like theories (if I 
can say).

In Schmidhuber-like theories there are indeed a program which
generates an infinite set of programs that could include the one
that describes our universe (including our consciousness).

I have explained in length why such approach fails to explain
what is matter, what is consciousness, what is time, what is a universe.
And also that such approach buries the mind body problem.

What I just show is that if we are machines then the physical
appearances *must* emerges from *all* computations  *at once*, 
and that the physical discourses, both first person (including 
uncommunicable qualia) and third person (communicable quanta)
must be defined with a sort of sum on all computations.

>However, Godel's
>theorem applied to this top-down approach would prevent us 
>from being able to recognize that program, or even knowing how 
>to recognize that program.

You are right! That the first law of machine psychology. It is
related to what I call Post-Benaceraff principle. See
http://www.escribe.com/science/theory/m2487.html.
It can be dangerous for Schmidhuber-like approach. It does
not threat my "theorem" :-)
Quite the contrary: Godel's theorem is generalised by the
modal logic G and G*, which axiomatises completely propositional
psychology machine, from which portion of physics can be derived.

>The best we can do is continually narrow down the options, 
>from an infinite subset to a "smaller" infinite subset, as we 
>add more parameters for
>description.

No. Read carefully the recent UDA thread to understand.

>To reconcile with anthropic fine-tuning without white rabbits, I had bought
>into the postulate that we were in the simplest possible universe, in the
>absence of knowing the exact criteria for developing self-aware
>consciousness, but just assuming that some absolute criteria exist.  But
>this begs the questions, what are those criteria and why those criteria?
>Without knowing these criteria, we cannot tell what is the simplest possible
>universe containing consciousness.

With comp we don't need such criteria. We sum on all program executions.
It really looks like a generalisation of Feynman integral. In particular
it should be an integral ...

Bruno






Reply via email to