UPLOAD: advocatus diaboli

John K Clark (johnkc@well.com)
Sat, 11 Jan 1997 20:22:26 -0800 (PST)


On Thu, 09 Jan 1997 Eliezer Yudkowsky <sentience@pobox.com> Wrote:

>Would you stop that? Nobody is attacking the Church-Turing
>Thesis, and if we were it wouldn't necessarily have any
>religious implications.

I assume you don't want to debate this point. You said the same thing before,
I responded at some length explaining exactly why you were wrong, you said

>It is possible that:
>1) Neural assemblies are the "bits" of the brain.
>2) Neurons require complex systems to simulate.
>3) Single synapses are the basic causal tokens of the brain.
>4) [1..3]s are the causal tokens of the brain, but don't map
> easily to Turing bits.

And maybe LTP has nothing to do with memory, and maybe Penrose is right, and
maybe the Pope is right, and maybe pigs will fly. We can play the "maybe"
game all day long, but in all of that only one thing is new, the most popular,
most detailed, most highly developed, theory that explains how the brain
stores memories has been shown to have a much smaller storage capacity than
was thought before January 28 1994. Call me crazy if you want but I think
that is a rather significant development.

>RAM doesn't map well to Turing tape.

If you mean it's not practical then what you say is true, but absolutely
nothing about A Turing Machine is practical, it is after all not a real
machine but a mathematical construction that helps us understand things.

If you mean that RAM should be put in a special category because Turing
Machines have extra trouble with it, then what you say is nonsense. A Turing
Machine with two tapes and two heads is provably equivalent to a solitary one
with one tape and one head, in fact it's equivalent to any Turing Machine with
a finite number of heads and a finite number of tapes, it's also equivalent to
a machine that uses a 2 dimensional plane instead of a tape, it's equivalent
to a random access Turing Machine.

Incredibly even a Non-Deterministic Turing Machine is equivalent to a
regular run of the mill single tape single head Turing Machine, although it
would be enormously faster, that is, it would be faster provided that what
everybody thinks but can not prove really is true, that P is not equal to NP.

John K Clark johnkc@well.com

Version: 2.6.i