X-Message-Number: 11638
From: Thomas Donaldson <>
Subject: more explanation for Mike Perry
Date: Thu, 29 Apr 1999 23:41:45 +1000 (EST)

To Mike Perry:

You have been seduced by the emphasis in computer science on COMPUTING.
The problem with a sequential machine is not that it cannot add 2 + 2
or even (briefly, to a naive subject) reply on the terminal as if it
were a real person. The problem is that a Turing machine, necessarily
sequential, simply cannot emulate a person in the real world because
it fails to be fast enough to do so. Sure, we might use fewer but faster
processors, but our design must inevitably be parallel, and could not
even WORK unless it is parallel. 

And if we produce a machine in a computer-synthesized world responding
only to that computer-synthesized world, we have no more than one more
computer program. It is the confrontation with the real world which makes
a device into one which can have awareness and all the other features;
if it cannot deal at all with the real world, you have at best a fictional
character living in a fictional world. And that fictional character has
no more reality than Cinderella or Puff the Magic Dragon, and no more
claim to awareness than Cinderella or Puff.

A simple example: a character in a computer game is not a real person or
even a life form. But a computer virus, I'd say, qualifies as a life
form (hardly an advanced one, but a life form nonetheless). 

Why this requirement that the device deal with the real world? Because
the real world is not only not predictable, but extremely fuzzy and mostly
unknown --- at least in any practical sense. No amount of pure computing
will answer our desire to find a theory containing quantum mechanics
and general relativity --- or for that matter, go out to the all-night
store and bring us back a hamburger. The real world is effectively
infinite, while any computer character living in a computer-simulated 
world remains finite by definition. A device working in the world can
become aware of all kinds of things which we cannot predict; a device in 
a computer-simulated world has a far more limited range. 

You may if you wish define "awareness" broadly enough that it includes the
"awareness" of an artificial computer construction in an artificial
computer world. Although the computer program running this construction
and the world in lives in is certainly more complex than a work of
fiction, you still have no more than an animated work of fiction. And 
from that point, the question of whether Tom Sawyer has awareness or
Cinderella has awareness becomes pertinent. JUST WHAT DIFFERENCE IS THERE? 

Again, I've written a variety of computer programs, and they will respond
to various kinds of input as I specify (modulo bugs --- bugs always). The
input must naturally have standard forms. Is a word processor or a 
database aware? 

The problem is that if we consider programmed computer constructions in a
programmed world to be aware, then we start running into exactly those
questions. If Tom Sawyer in a computer game is aware, then just how
does awareness arise? Does the mere fact that the program is running on
a computer imply that its characters are aware? Sure, you may say that it
does not, but then WHY NOT? 

I will add that this argument does not apply to an autonomous robot, or
even a semiautonomous robot. Sure, such devices do not have the awareness
that we have, but just like insects they show the beginnings of it. I am
not arguing that we cannot build a device that is aware; I am arguing that
a TURING MACHINE cannot be aware, nor any character in a program running
on a computer (in the current sense of computer). And if we want to create
a device which is aware, a single computer won't do it --- you'll need
lots of them (the exact number depends on the kinds of computers you
use, among other dependencies).

So over to you, now.

			Best and long long life to all,

				Thomas Donaldson

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=11638