RE: exploring ambiguity via the "something-which-has" URI scheme

>  > -----Original Message-----
>>  From: ext pat hayes [mailto:phayes@ai.uwf.edu]
>>  Sent: 08 May, 2003 18:53
>>  To: Sandro Hawke
>>  Cc: Stickler Patrick (NMP/Tampere); GK@ninebynine.org; uri@w3.org
>>  Subject: Re: exploring ambiguity via the "something-which-has" URI
>>  scheme
>>
>>
>>  >  > Even if, according to the MT, the denotation/interpretation of a
>>  >>  given URI may differ, what counts for the SW is that those
>>  >>  denotations/interpretations be compatible.
>>  >>
>>  >>  Incompatible denotations/interpretations are what harm the SW.
>>
>>  Just for the record, this is nonsense as stated.  One wouldn't expect
>>  that two different interpretations would be compatible.  What I think
>>  Patrick meant is that incompatible ASSERTIONS are what harm the SW.
>>  That is debateable, but at least it makes sense.
>
>No. That's not what I meant (at least I don't think so ;-)
>
>If there could not be incompatible assertions, then there could be
>no disagreement, and I think the SW should allow folks to disagree
>about the things that they are talking about in common.

You seem to be saying that people can disagree about facts but they 
have to agree about the objects that the facts are 'about'. That's 
not a viable distinction to make, though. If you allow them to 
disagree, then one of the ways they can disagree is about what they 
are talking about. Suppose for example one of them asserts that A=B 
and the other one disagrees; he thinks that A=/=B. They can't 
possibly be agreeing about what they are talking about, since one of 
them sees two things where the other only sees one thing. Whatever 
the second guy is talking about, it can't possibly be what the first 
one was talking about. (Equality is the meat and potatoes of 
OWL-style reasoning, so this isn't an arcane distinction, it arises 
all the time. )

If you start getting persnickety about criteria of identity then it 
gets harder and harder to establish agreement about what we are 
talking about, eg. for some purposes it makes sense to think of 
Pat-yesterday as a distinct thing from Pat-today. So is there one of 
me, or two of me, or approximately  21,000 of me? A good answer  to 
questions like this is 'yes'. Fighting over which view is 'right', or 
which one has to be declared globally correct, is just going to lead 
to endless pointless quarrels. As long as we can understand one 
another, as long as what we say to one another can be understood in 
either way, we can get along.  Say some RDF triples store somewhere 
is keeping track of my movements and it has things in it like

ex:thing3246  ex:agent http://www.coginst.uwf.edu/~phayes  .
ex:thing3246  daml:timeOf "1100SCT05082003-1100SCT05092003"^^ex:UStimeInterval

then there really is no need for you and I and the rest of the world 
to come to some global agreement that these ex:thing URIs must be 
understood to denote time-slices of people, or must denote people who 
endure for a time but retain their identity (believe it or not, this 
issue is one that blood has been spilt over, books written to refute 
other books, people yelling in the corridors, etc. etc.. There are 
deployed ontology standards in industrial use which take divergent 
views on it.) You can think of them your way and I can think of them 
my way: and you know what, as long as we avoid trying to decide who 
is right, it doesn't make a damn bit of difference to anything. We 
can communicate perfectly well, because everything that you say 
translates smoothly into my way of thinking and vice versa. What we 
actually say to one another can be interpreted in two different ways, 
as talking about different kinds of thing, and that's fine.

>Rather, if there is to be disagreement, if assertions are to be
>incompatible, then it must be the case that they are about the
>same thing.

In a sense that is true, but in that sense there is no need to do 
anything about it. Just the use of a common language (in our case, 
using the same URIs) is enough to establish that the different uses 
are about the same thing in this sense, ie whatever they are about, 
they are both about it, whatever it is, simply virtue of being tokens 
of the same symbol, and the fact that the semantics is attached to 
the symbol rather than the token.  Of course, what you THINK it is 
about might not line up exactly with what I THINK it is about, but as 
long as what we SAY doesn't actually violate any of these extra 
hidden assumptions, we will be OK.

>What is harmful to the SW is when folks make assertions which
>appear to be about the same thing, but are not.

There is no way to do that on the SW or the Web. The assertions are 
about whatever they turn out to be about when we have got enough of 
them all together to try to pin down the referent, if we ever feel a 
need to. Someone might have one thing in mind when they use a URI, 
but what they actually *say* is whatever is said by using that URI. 
What they have in mind doesn't get transmitted by HTTP, so it is kind 
of irrelevant, seems to me. It has the same relationship to what 
actually gets said as what you hoped the code would do has to what 
the code actually does.

Pat
-- 
---------------------------------------------------------------------
IHMC					(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola              			(850)202 4440   fax
FL 32501           				(850)291 0667    cell
phayes@ai.uwf.edu	          http://www.coginst.uwf.edu/~phayes
s.pam@ai.uwf.edu   for spam

Received on Friday, 9 May 2003 22:10:03 UTC