Re: Trying again to respond

phayes@cs.uiuc.edu
Message-id: <199305022307.AA28134@dante.cs.uiuc.edu>
Date: Sun, 2 May 1993 18:10:31 +0000
To: sowa <sowa@turing.pacss.binghamton.edu>
From: phayes@cs.uiuc.edu
X-Sender: phayes@dante.cs.uiuc.edu
Subject: Re: Trying again to respond
Cc: dietrich@turing.pacss.binghamton.edu, interlingua@ISI.EDU, pcg@cs.umn.edu,
        phayes@cs.uiuc.edu
John, 

Our arguments are splitting into three: philosophy of mathematics, 
semantics of Krep formalisms, and issues about the Krep 
standardisation effort. As you pointed out in a lucid message 
recently, the semantic stuff does have importance for the 
pragmatics, so they are not entirely distinct, but I will try 
to separate them out so that readers can pick and choose the 
parts which interest them. The mathematical argument is something 
of an aside.

The rest of this message is about the standardisation issue.

Let me be awfully honest: I think this idea of a Krep 'standard' is 
insane. But if we have to have one, it should be as minimal as 
possible, and should contain as few semantic or metaphysical restraints 
as possible. And that seems to rule out CG.

First, let me say that I wish you wouldn't keep using such phrases 
as 'drawing the line' and 'I won't let them say'. We can argue 
(no doubt will) for years about which ideas are correct, but I 
get the uneasy feeling that you want to make sure that everyone 
else agrees with you essentially by appeals to authority: either 
some philosphers, or just your own. You constantly tell us what 
you are willing to allow, or must forbid:

"
>But where I draw the line is with attempts to "identify" w0 with
>the real world.  I am willing to let people say that there is a
>mapping between w0 and some aspects of the world, but I won't let
>them say that any of the elements of w0 are identical to things 
>in the world."

Sorry, but I AM going to say that, and you will just have 
to let me. Take an asprin if you get a headache.

I only began this debate, clogging up my friend's emailboxes, 
when I read in a message from you:

"If Jim and Chris and others ......., then I would be
happy to let them continue using that word.

But as Chris pointed out, there are some people including
David Lewis who are trying to...... In order to prevent them from
doing that, I am insisting that they use the term "model" or
any other word they please other than "world". "

The content of the deleted passages is irrelevant. Many other 
examples could be cited. 

Maybe I am just reacting to a stylistic quirk, and if we were 
just having academic debates, I would make that reasonable 
assumption. But you back it up with electronic behavior. Indeed, 
it seems that you ARE trying to prevent people saying things you 
don't like, or interpreting formalisms in ways you have metaphysical 
objections to. So I feel that we must take it seriously when we 
are talking about a potential national standard.

---------

The aspect of CG which bothers me is its declared relationship to NL:
....
>.... As I pointed out above, I make a very sharp distinction
>between natural language and any system of logic (or kn. rep.).
>I believe that natural languages set the standard for expressive
>power, and logic sets the standard for precision.  What I am trying
>to do with CGs is to design a system of logic that is capable of
>expressing the propositional content of any sentence in NL.

You make the distinction, on metaphysical grounds. But you have said 
that every distinct English sentence has a distinct propositional 
content: therefore, your equi-expressive formalism will amount to a 
transcription of English. But you also reject Montague very sharply, 
so your equiexpressive formalism will not have a model theory defined 
on it, and therefore will not have a clear notion of valid consequence.
As far as I can see, the formalism might as well simply be NL. 
In any case, this is a vision of Krep formalisms which I have argued 
against for many years, and am anxious that it not be somehow made into 
the standard.

> Mike G. and others are welcome to design the syntax of KIF
>in any way they please.  All I want is to make sure that KIF provides
>enough expressive power for me to map the CG constructs into it.

I agree that seems reasonable enough, provided that the meanings of 
these constructs can be made clear. My concern is more the way in 
which any standard might be forced away from having a clear notion 
of interpretation and consequence by having to be 'expressive' 
(my turn to use scarequotes :-)) enough to distinguish, say, active 
and passive constructions in English.

-----------

However, I agree that the first two changes to KIF you list seem
sensible (although I don't speak for anyone at Stanford), 
and agree that the way that metaphysics gets enmeshed with 
grungy practicality is regrettable. I don't think it is wise 
to identify an expression with its denotation, however.

Thats about all about Krep standards. Other (longer) messages will 
pick up the other points. Readers who are not interested can figure 
out which to look at from the Subject lines.

Pat 



----------------------------------------------------------------------------
Beckman Institute                                    (217)244 1616 office
405 North Mathews Avenue        	   (217)328 3947 or (415)855 9043 home
Urbana, IL. 61801                                    (217)244 8371 fax  	  

hayes@cs.stanford.edu  or Phayes@cs.uiuc.edu