X-Message-Number: 11686
From: Thomas Donaldson <>
Subject: replies to Bob Ettinger and Mike Perry
Date: Thu, 6 May 1999 23:30:09 +1000 (EST)

To Bob Ettinger:

I feel uneasy about the possibility of some unknown factor requiring that
we must necessarily be biological. The problem is that research on how
consciousness works in our brains and those of animals close to us is
going on right now, and it's actually likely that in a relatively few 
years we'll have a good account of how consciousness works. And then
we'll know whether we must be biological or not.

It is true that no computer neural net works like a brain; the major 
difference is that brains are continually rewiring themselves. They 
qualify as neural nets, yes, but SO FAR no one has devised a means to 
do the same thing in a nonbiological system. And I do think that this
design feature is likely to be very important.

But my major problem with the notion of a virtual person in a virtual
world is that it is so blithely assumed that a virtual world can be made
that would be good enough to come anywhere close to reality. I find that
close to totally out of the question. As I've said before, just how are
we to do that unless we understand reality --- which so far we do not,
not as completely as we'd need to to make such virtual worlds.

To Mike Perry:

We may be getting somewhere. However when I said that parallelism was
necessary I meant just that. If you wish to consider hypothetical but
impossible sequential computers, you're welcome to do so, but if a
conscious being is to work by means of such a computer then it becomes
hypothetical and impossible too.

The reason I say this comes simply from hardware limitations. If you
look at what Intel's latest PC chip can do, it's proudly said to be
able to operate several hundred times faster than previous chips. That's
fine for its purpose, but even several hundred times just isn't good
enough. We may make our computers quite differently, and still keep
them sequential, but the universe, and even a human brain, involves
the SIMULTANEOUS operation of billions of objects (100s of millions in
human brains, but the human brain gets input from many other things).
Ultimately the speed of a single processor is limited by physics.

I will add, as I said to Damien Broderick, that a quantum computer is
one more variety of parallel computer (it works simultaneously on all 
the many possible values of a quantum object). But we can do calculations
as to THEIR speed, too, and you'd need a large quantum computer to 
emulate the universe or even a person. Some people even think that 
we are quantum computers, though I believe the evidence for that is
weak.

Ever since I got into parallel computing professionally, I've felt that
all the theories of computing which look only at sequential models 
have little validity when we try to do any practical computing (I also
believe that parallel computing should be done EARLY in computer 
science courses). Sure, they may be correct as mathematical constructions,
but they ceased to be relevant years ago. It's one sign of this that to
describe a parallel computation as if it were done by one (theoretical) 
sequential computer easily becomes far more complex than the original
parallel program. Correct answers are useless if we must wait 1 million
years to receive them, and particularly useless if what we are trying to 
compute is (say) the weather for tomorrow.

That is why I say that any device which comes close to human beings in
terms of awareness (not to mention abilities) will have to be parallel.
It will need a good deal of other things too, and would only qualify as
a computer in the sense that (say) a transistorized radio or car 
qualifies as a computer.

			Best and long long life to all,

				Thomas Donaldson

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=11686