From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!uunet!math.fu-berlin.de!Germany.EU.net!mcsun!mcsun!seunet!kullmar!pkmab!ske Thu Apr 16 11:34:03 EDT 1992
Article 5049 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!uunet!math.fu-berlin.de!Germany.EU.net!mcsun!mcsun!seunet!kullmar!pkmab!ske
>From: ske@pkmab.se (Kristoffer Eriksson)
Newsgroups: comp.ai.philosophy
Subject: Re: syntax and semantics
Message-ID: <6757@pkmab.se>
Date: 10 Apr 92 23:44:48 GMT
References: <1992Apr8.215800.18021@mp.cs.niu.edu> <1992Apr9.204735.21732@psych.toronto.edu> <1992Apr10.013123.24515@sophia.smith.edu>
Organization: Peridot Konsult i Mellansverige AB, Oerebro, Sweden
Lines: 102

In article <1992Apr10.013123.24515@sophia.smith.edu> orourke@sophia.smith.edu (Joseph O'Rourke) writes:
>In article <1992Apr9.204735.21732@psych.toronto.edu> michael@psych.toronto.edu (Michael Gemar) writes:
>
>>Heck, if everyone decided to interpret *your*
>>bank's computer as playing chess instead, how would you prove them wrong?
>
> I don't think it is possible to consistently interpret Neil's
>bank computer as playing chess.

Even if it was possible to interpret the bank's computer as playing chess,
I think that if the input to the bank's computer still consists of the
banking transactions the bank's customers are performing, and the computer
still runs the usual banking program, then I still have a case for claiming
that the computer system is performing bank duties rather than playing chess.

If the computer received chess moves as its input, the case would change.

It may be true that if you look at the computer alone, where arbitrary
numbers are going in and out, the computer could not be said to be
performing any specific task except arbitrary number manipulation that
could equally well be interpeted as either bank duties or chess playing
(with a bit of luck).

But I would like to widen the picture, and look at the combined system
formed by the computer and the source of its inputs. In every business
oriented computer system, a certain number of manual routines are also
included. In this case, when for instance a customer enters the bank
office and makes a deposit of cash into his account, the bank clerk
enters that amount of cash into the computer, and not some chess move or
something else. Or there could be some automated device that takes the
money and counts it and reports the amount to the computer. These numbers
are causally linked to the real money, and can be said to represent that
money in the sense of being a recording of that money. This link exist
independently of any interpretation given by us humans external to the
system.

The numbers in the computer still record the deposited amount of money,
no matter how much every one else claims that they are really about a
game of chess, and the customer should be able to make the case that he is
entitled to the amount of money that those numbers record.

Sure, those same numbers might be used to accidentally solve some chess
mystery too with suitable reinterpretation, but they originated as money,
and if they represent anything in the physical world, it is that money and
not some imaginary chess game.

Think of a closed system that takes money, real money, in one end, outputs
balance statements in the middle, and outputs the money in the other end
again once you request it. Somewhere inside, it contains a computer with
a banking program, that controls the operation of the system. Now, why
should I not claim that it is handling money rather than playing chess? After
all, it does take real money as inputs and outputs, and it does not move
chess pieces around. And the bank office in combination with their computer
is just such a system.

In his Scientific American article, Searle writes:
> First, symbols and programs are purely abstract notions: they have no
> essential physical properties to define them and can be implemented in
> any physical medium whatsoever. The 0's and 1's, qua symbols, have no
> essential physical properties and a fortiori have no physical, causal
> properties.

As far as programs as static, formal specifications, and bare symbols on
their own are concerned, he is right.

But when the program is implemented in a specific physical system, and
external inputs are supplied, something more happens. The symbols and the
system are no longer merely abstract notions; they are given physical
existence, physical properties, and causal properties. The inputs are
connected to specific parts of the external world (not just any part you
want to interpret into it) and will record states of that world and offer
the program the possibility of manipulating a representation of that world,
made up of symbols causally connected to that world. The system can produce
something that is in some way connected to that world, although exactly how
of course depends on the program.

Now, to complicate things further, suppose we take the closed banking box
described above, and encode moves of a chess game in the pattern of money
we feed the box, and we do it in such a way that the for instance the
balance statements can be interpreted back into new moves in that game, or
inte some other chess related answer.

Has the box now turned to playing chess game, or is it still handling money?
Well, I don't see how anyone could deny that it does still handle money as
much as before, since it is still rattling of money entering and leaving
the box. However, I would say that the combination of the box and the
inputting process that takes chess movements and encodes them into money
that is entering the box, and decodes the results exiting the box, does
play chess (or solves some other chess problem), although perhaps in a
somewhat expensive way.

The system embodied by the box alone handles money. The system embodied by
chess inputs, an encoding process, and the box, is playing chess. The system
embodied by the computer inside the box, without the other devices, is
manipulating numbers. The transistors in the computer are switching 
electron flows. Behold the full-blown systems reply (which also implies
the robot reply).

-- 
Kristoffer Eriksson, Peridot Konsult AB, Hagagatan 6, S-703 40 Oerebro, Sweden
Phone: +46 19-13 03 60  !  e-mail: ske@pkmab.se
Fax:   +46 19-11 51 03  !  or ...!{uunet,mcsun}!mail.swip.net!kullmar!pkmab!ske


