X-Message-Number: 8528
From:  (Thomas Donaldson)
Subject: Re: CryoNet #8515 - #8520
Date: Wed, 3 Sep 1997 03:16:16 -0700 (PDT)

Hi guys!

I have a problem with Turing machines also. You see, I too have been involved
with computers for some time, particularly parallel computers. These hardly
classify as Turing machines because they have more than one processor (though
we can criticise the idea in other ways too). But parallel computers or
circuitry (like the MMX that Intel is so proud of) become so pertinent not
for theoretical reasons but for very practical ones: if we tried to make a 
true Turing machine to do the calculations which a large Intel Paragon might
do, we'd not live long enough for the Turing machine to complete those
calculations. 

Any real attempt to come to grips with the world needs to use several levels
of abstraction or it misses out on some important things. Turing machines
are fine for some kinds of theoretical study of computing; but there also 
comes a stage when we must put restrictions on them: how much memory, how
much data. So far, in understanding how our brains work, computers haven't
been a very good analogy. And yes, besides working in parallel (neural nets
are a special form of highly parallel computer --- so very parallel some 
might wonder whether they are even computers) our neurons are also analog
machines. Not digital. 

Incidentally, the latest (July-Sept 1997) issue of IEEE Concurrency, to which
I subscribe as a member, has a nice article about the computer system which
beat Kasparov, and introduces us to some of the programmers and engineers 
who built it. It used several IBM RS7 machines, all linked together, and 
in addition each such machine had some special hardware designed specifically
to chase down the possibilities of chess moves. My respect for the engineers
who built it (they also had a chess grandmaster to help) increases: the
system was hardly simple.

At the same time, a comparison of Kasparov's brain with the computer tells
me that we have a long way to go before we could really claim to make something
equal to a human brain. (Yes, I think we could eventually build a DEVICE equal
to a human brain --- whether it is a computer raises other issues). Not only 
wasn't Kasparov's brain specialized to play chess, but there's lots more
miniaturization needed just to make the machine of more or less similar size.
And the lack of specialization is critical too: we could hardly hope to 
make a device capable of everything a human brain does simply by adding 
specialized processors for each one.

Nonetheless it was a nice article, and gave an overview of the system and 
the program it ran. For those who want to find out the nitty-gritty here,
rather than just spin dreamy theory.

			Long long life to all,

				Thomas Donaldson

PS: Yes, I know it's late, but I had to get up for another reason and I
decided to check out my email while I was up.

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=8528