From newshub.ccs.yorku.ca!ists!torn.onet.on.ca!utgpu!news-server.csri.toronto.edu!rpi!uwm.edu!rutgers!mcnc!aurs01!throop Mon Jun 15 16:04:50 EDT 1992
Article 6222 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!torn.onet.on.ca!utgpu!news-server.csri.toronto.edu!rpi!uwm.edu!rutgers!mcnc!aurs01!throop
>From: throop@aurs01.UUCP (Wayne Throop)
Newsgroups: comp.ai.philosophy
Subject: The Turing Test is not a Trick
Summary: still unclear as to what is the TTT, and why
Message-ID: <60807@aurs01.UUCP>
Date: 11 Jun 92 21:19:56 GMT
References: <1992Jun11.154029.29686@Princeton.EDU>
Sender: news@aurs01.UUCP
Lines: 85

> harnad@phoenix.Princeton.EDU (Stevan Harnad)
> [.. pen pal scenario ..]
> Under these conditions, the problem of how the pen-pal would
> be able to provide even the kind of WRITTEN performance (TT) that in
> the rest of us draws on our sensorimotor capacities and experience (as
> in saying what is really in our pockets at any time) shows a real and
> serious (I think fatal) limitation of the TT, relative to the TTT; this
> problem has nothing at all to do with tricks.

I don't understand.  The TTT has all the same problems, just at
somewhat different points.  Inside the TTT-testee could be tricks of
any sort, to simulate kinesthesia, proprioception, digestion, or
whatever feature the TTT judge is trying to distinguish from human
norms.  Just as "inside" the pen-pal there could be tricks to simulate
broader sensorimotor experience.  I see no significant difference
here.

The TTT and the TT differ ONLY in what aspects of the testee
it is possible to directly check.  They are virtually identical
in what aspects of the testee it is possible to indirectly check.

Thus, in prefering the TTT over the TT, it seems to me that one
is arguing that the extra things directly checkable are relevant.
The only relevance I have seen advanced for them, however, is that
they make the test harder to pass.  But this seems very strange.
Making the test harder to pass is no particularly worthwhile goal
in and of itself.

> (Nor does it have anything to do with handicaps:  Think of how far we'd
> get in designing a system that had the TTT-capacity of a car if we
> allowed ourselves to aim just for TTT-indistinguishability from a car
> that's out of gas, or has a seized crankshaft, or no wheels. 

Yet if it were an electric car, it wouldn't matter if it were out of
gas, and it wouldn't have a crankshaft.  If it were a ground-effects
machine, it wouldn't matter if it had no wheels.  Perhaps a motorcycle
with a trailer would do for the purpose.  If what matters about a car
is that it's a vehicle to get from place to place, insisting that it
have "gas" or "crankshaft" or "wheels" is simply irrelevant.

Similarly, if what's important about a human is their communicative
ability, it doesn't matter if they are speaking english or signing
ASL.  The fact that someone is deaf and signing ASL has NOTHING to do
with "handicaps" in this context.  It's just that insisting that a
communicating human have working ears is irrelevant. (Again, Sacks'
"Seeing Voices" has much of interest to say about the "prelingualy
deaf", and their linguistic neurological development.  And again, this
example is NOT about "handicaps".)

So the question is, what's important about the TTT?

I still have no answer to just what is supposed to constitute the TTT.
Let me make a small list:

- Can the testee be disqualified for lack of hair?  
  If so, how fast should it grow, and what colors are acceptable?
  Upon what areas of the body should it grow, and upon what areas
  should it NOT grow?
- Can the testee be disqualified for excessive servomotor
  noise while walking?
- Can the testee be disqualified for lack of sweat?
  If so, how about for out-of-range salinity of sweat?
- Can the testee be disqualified for out-of-range postural
  abnormality (eg: the ability to sit completely still for
  long periods of time, or the inability to wiggle its ears?)

None of these have anything at all to do with "handicaps".  None of
them really have to do with a picky definition of what is an average
human: I'm talking about things well outside the human average.  I'm
not trying to be cute here, I really and genuinely have very little
idea just what the TTT entails.  Just what advantage are the judges
supposed to take of the fact that the testee is in the same room with
them?  Most advantages I can see that they have, don't seem worth having.

The point is, "everything a car can do", or "everything a human can do"
are both hopelessly vague.  It is absolutely clear to me that certain
features of what current cars can do are irrelevant to their car-ness.
Similarly, it is clear to me that certain features of what humans
can do are irrelevant to their intelligence.  Unless one makes at least
some attempt to say what is relevant and irrelevant about car-ness,
"everything a car can do" is meaningless.  Similarly, unless one makes
at least some attempt to say what is relevant and irrelevant about
intelligence, "everything a human can do" is likewise meaningless.

Wayne Throop       ...!mcnc!aurgate!throop


