W3C home > Mailing lists > Public > public-sw-meaning@w3.org > October 2003

Re: what matters is what's said, not what's meant

From: pat hayes <phayes@ihmc.us>
Date: Thu, 9 Oct 2003 20:18:34 -0500
Message-Id: <p06001f20bbabb5d7f074@[10.0.100.25]>
To: Sandro Hawke <sandro@w3.org>
Cc: public-sw-meaning@w3.org

>  > 1. If you derive a conclusion from a contradiction, then nobody is
>>  responsible for it but you: a derivation from a contradiction adds no
>>  weight to the conclusion at all, so if you publish it then that is
>>  just like publishing a brand-new assertion; you, the publisher, can't
>>  appeal to any other authority.
>...
>>  Im sure there are others, but the first one is enough to handle your
>>  case, since if I publish a contradiction, then its not going to be
>>  any SW-style use to anyone else: they can't use it to do anything
>>  that they couldn't do on their own.
>
>While I agree with that, it doesn't help with my original concern,
>which is how the use of a particular URI in an RDF graph affects how a
>receiver of that graph might behave.

I bet that it has an immediate effect on it. If being useful to 
others is any kind of reason to put info on the SW (and if it isnt, 
why would anyone do it?) then suddenly your stuff is useless: my 
guess is that people will let you know about that fairly fast, in 
practice.

>If imagine you say to me (in the
>document "pat"):
>
>      pat:Donny a bio:Dingo.
>      pat:Donny a bio:Cat.
>
>I might well follow the bio link, and while I'll be nice and not hold
>you to everything bio says

Oh, but if we follow that elegant device Tim has implemented, then 
you should, because I will have imported stuff from bio. If I didnt 
import it, then Im not publishing a contradiction: Im just 
disagreeing with people. (You could still come to a poor opinion 
about my reliability, of course, just on that basis.)

>, ... what should I make of the fact that
>you've said Donny is an instance of two disjoint classes?

That I am not saying anything useful? That what I say is useless, and 
cannot be relied upon? Seems fair enough.

>   Hrm.
>Neither "pat" nor "bio" is inconsistent, but the combined set is
>(assuming OWL semantics and that bio really used OWL).
>
>It seems to me that as a reader I'm not going to be too happy with
>this graph you sent me.

I think you might conclude that this source isn't very reliable, and 
decide not to trust it. That's what I would do. I know that if I rely 
on it, I might as well be generating random fictions, because the 
rules require that I can't cite it as a source for any conclusions I 
might draw from it.

>  What have I really learned from you about
>pat:Donny?  What can I do with what you said?  Not a lot, I think.
>Ewwww.

Right, nothing at all, in fact. But then, if so, why....

>So working back, I end up saying you SHOULD NOT say things like that,
>and MUST NOT do so knowingly.

If saying a contradiction amounts to saying nothing, then why should 
it be prohibited? There are contradictions in the RDFS and OWL test 
cases documents, and they are quite useful.

It might be hard to satisfy that SHOULD NOT stuff. Suppose I publish 
A which imports B which then gets changed and now A is inconsistent. 
My bad?? Has to be, right? B might still be consistent, and Im the 
one importing it: B's owner need not even know I exist. Yet my 
inconsistency is inadvertent: am I supposed to monitor all my 
imports? All my import *closures*?

Why not just say in this case: what I said has been reduced to 
nothing, rendered free of content, by B's change, not that I am now 
in violation in some way. If I care about what I publish, this gives 
me plenty of motivation to fix it, but fueled by the normal social 
conventions of information publication on the Web, rather than by a 
TAG 'rule'

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@ihmc.us       http://www.ihmc.us/users/phayes
Received on Thursday, 9 October 2003 21:18:36 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:42:15 GMT