UPLOAD: advocatus diaboli

John K Clark (johnkc@well.com)
Sun, 12 Jan 1997 21:03:05 -0800 (PST)


On Sun, 12 Jan 1997 Eliezer Yudkowsky <sentience@pobox.com> Wrote:

>Challenging the Church-Turing thesis does not make me a
>religious lunatic

I don't thing you're a lunatic Eliezer, religious or otherwise, I just think
you're wrong.

>Being a religious lunatic does not automatically imply that
>I am wrong

As I said before, I don't think we have a soul but I don't find the idea
crazy, what is crazy is that information processing doesn't cause
consciousness and the soul doesn't either.

>It is possible that consciousness is computable, but that
>the brain projects an energy field which survives the death
>of the neurons.

Well we know the brain can do just that, it generates a very weak
electromagnetic field, and if it is not absorbed by the Earth's ionosphere it
will survive until The Big Crunch, but I don't think that's what you mente.
You're talking about some mysterious immortal energy field that is responsible
for consciousness, fine, but in that case the energy field is identical with
the traditional concept of the soul.

>It is possible that consciousness is noncomputable, but that
>it still goes out like a light with the removal of the brain,
>and is nothing more than the interaction of the noncomputable
>physical particles making up the brain.

Even if I give you the very large and completely unsubstantiated point that
atoms are black boxes of noncomputable stuff, it wouldn't change my theory
that information is the essence of what we are. One point I do refuse to give
you, because it would mean that almost all Physics done in the last 90 years
is garbage, is that noncomputable or not, one Hydrogen atom is identical to
another Hydrogen atom.

If I find the information on the position and velocity of every atom, every
noncomputable black box, in your brain by using a destructive scan, and if I
then use that information to put new atoms in the same positions, then that
new brain is you. Unless of course you have a soul.

>There are known mathematical concepts, such as "oracle
>machines", which violate the Church-Turing thesis in a
>technical sense by being able to solve the halting problem
>for lesser oracular ordinals, such as Turing machines.
>These mathematical concepts have no religious implications.

Near as I can tell these concepts have no implications of any sort. Not long
ago two French scientists, Bournez and Cosnard, published a paper about a
super Turing Machine, an analog automation that could do things no Turing
Machine could. The trouble is, Analog Computers do not exist, never have
never will. I can think of 4 reasons:

1) Matter, energy, momentum, spin, and electrical charge are all
unquestionably digital, they come is discrete package. It could well turn
out that space and time do too. When you build your analog computer you
must make use of all these things.

2) An analog computer MUST have an infinite number of internal states, so it
must be built with infinite precision, but quantum mechanics tells us that
is impossible.

3) If your analog computer is operating at any temperature above absolute
zero (and it will be) it will be subjected to thermal vibrations further
reducing it's precision.

4) Unlike digital machines, any errors in a analog machine are cumulative.
This is very serious in all but the simplest calculations.

The best we can do is build pseudo analog devices with limited precision and
a FINITE number of internal states, that is a pale reflection of the analog
concept. A digital computer can simulate a pseudo analog computer perfectly,
but the reverse is not true. I admit that there are some Real numbers that a
Digital Computer can not deal with, Turing proved that, but it can handle a
lot more than a pseudo analog computer can. A digital computer can be made to
work with arbitrary precision, but the precision of an pseudo analog computer
is strictly limited by the laws of physics. At times it may be more convenient
to use a pseudo analog device than a conventional digital computer, but that
is an engineering consideration not a philosophical principal.

>Quantum randomness, if truly random, is not computable in
>the strict sense. This has no religious implications and
>does not even *impinge* on AI

I agree, it has nothing to do with intelligence or Free Will or religion.
So what's your point?

>As for your explanation, I saved it and have it flagged in
>my list of favorite messages. It's very interesting. The
>only problem is that it applied as well to ones and zeroes
>as any hypothesized soul

Not so. Axioms should always be as simple and as self evidently true as
possible. Both a bit and the soul are equally simple objects, neither is made
of internal parts, however there is a difference. What the bit does is just
as simple as what a bit is, what the soul does on the other hand is VERY far
from self evident, this equally simple object, does things that are
astoundingly complex. I can not imagine anything more complex. If this is
really true then it's magic and I'm going to church.

John K Clark johnkc@well.com

Version: 2.6.i