RE: My action item on Moby Dec, issue 14, etc

Hi Tim,

> -----Original Message-----
> From: Tim Bray [mailto:tbray@textuality.com]
> Sent: 19 September 2002 17:15
> To: Norman Walsh
> Cc: www-tag@w3.org
> Subject: Re: My action item on Moby Dec, issue 14, etc
> 
> 
> 
> Norman Walsh wrote:
> 
> > The error, I think, is that you've brought representations into the
> > picture. The important part about "Absolute URI references are
> > unambiguous: Each absolute URI reference unambiguously identifies one
> > resource." isn't the consistency of the representations you can
> > dereference, it's the fact that they're fully qualified and globally
> > unique. That's why we can paint them on billboards, write them on
> > busses, and flash them in commercials.
> 
> OK, I think that what you're saying is that a resource is simply that 
> which is identified by a URI.  I agree with this and think it's 
> consistent with the 2396 definition too.  Given this, our "principle" is 
> a tautology and not in the slightest worth saying.

Hmmm... I think your shifting the ground. That "a resource is simply that
which is identified by a URI." may be a tautology, but the prinicple Norm
quoted was "Absolute URI references are unambiguous: Each absolute URI
reference unambiguously identifies one resource." which I think is worth
articulating as a principle of Web Architecture (if it is a principle that
'sticks').

> Spin it another way: the URI, and the representations you can (maybe) 
> get with it, are all there is.

Sure... a resource is experienced through the representations exchanged with
it and what other resources ['meta-data resources'] have to 'say' about it.
If you are not interested in the what these other resources have to say,
then the URI and representations exchanged with a given resource is indeed
all you have.

> There is no point in arguing about the fundamental nature of what the URI
identifies and what the 
> representations represent,

Agreed. I've got to this point by concluding that TimBL's expression of the
issue, that HTTP URI (no fragment) must not/should not be used it identify
abstract concepts or non networked real-world artifacts (people, cars,
mountains, whales...) is secondary to a concern ambiguity in the use of URI
(HTTP URI).

> because (a) you can never know, and 

Yes, from the point of view of experiencing a resource through exchanges of
representations with that resource, and from the point of view that the
resource itself and others can lie or disagree about the nature of a given
resource. The URI assignment may authoratively assert the nature of a
resource, but I guess a resource is in the eye of the beholder... and what
they choose to believe it to be may be influenced by others, but
fundementally is under their own control.

> (b) it 
> doesn't matter.

What matters (I think) is being able to tell that assertions, made at
different times and by different entities, are being made about the same
thing - even if those assertions are contractictory. It does matter that at
least within some context of use (spatial, temporal... possibly universal) a
given identifier or reference identifies or refers to the same thing.

> Furthermore, in the context of using URIs to build KR systems a la RDF, 
> the notion that you can banish ambiguity by architectural fiat is simply 
> wrong and dangerous to the future of the semantic web. -Tim

Hmmm... can you un-pick this a bit. Can you expand on "dangerous"? It seems
to me that the 'intended' universal absense of ambiguity in the relationship
between an identifier and a resource was part of the original design of the
Web.

Cheers,

Stuart

Received on Friday, 20 September 2002 10:22:21 UTC