W3C home > Mailing lists > Public > uri@w3.org > September 2002

Re: Possible new issue: Things with and without identity?

From: Al Gilman <asgilman@iamdigex.net>
Date: Mon, 09 Sep 2002 20:10:54 -0400
Message-Id: <>
To: Miles Sabin <miles@milessabin.com>, uri@w3.org, rest-discuss@yahoogroups.com

At 07:17 PM 2002-09-09, Miles Sabin wrote:
>There's also a brief discussion of the relation between stuffs and
>things, but that's irrelevant here: RFC 2396 talks about things from
>the outset, so individuation is already presupposed.

Yes, but erroneously so.  This language in RFC 2396 was heuristic, and the
idea that individuation and a persistent referenceable identity are
intrinsic and universal in the Web, while central to the thinking of some
people, did not at that time nor at this one reflect a globally valid
interpretation of all the uses of URIs.

And the distinction between things and stuffs is not at all irrelevant to
building a reasonable engineering foundation for semantics for the Web.

On the web we need to be able to refer to both stuffs and things.

We just discovered we needed to factor ACSS into perturbations of the
text-to-speech transform, which is analogous to font characteristic
manipulation, and sonicons that are used as punctuation for phrasal elements.
Analogous to quotation marks and other punctuation which appear in the final
stream as characters but this is because these controls are of such
antiquity that there are standard glyphic characters for them.

The moral here is that this points straight at a distinction between text and
fills, which are styled as stuffs, and list elements and paragraphs and
such, which are styled as things.

If we comprehend this distinction at the outset, it would make the whole job
of characterizing styling a lot easier, and the result more like natural
language semantics.

In particular, for my dump on how regarding a search URL as a reference to
an identity is spurious, see


It is just much more direct to view a search URL as subclassing a stuff and
the server as returning references to things that exemplify the stuff that
the user's URL described.

The data of the 'dark Web' is more practical to approach as stuff typed by
tuples of properties than as entified things.  Or at least that view matches
application semantics better.


<et_cetera ad_nauseam/>

In any case, Miles, there are requirements in semanticizing the web both for
entities without identity, to refer to found instances matching a known
pattern, and for stuffs, in which you find the instances, as in the
examples above.  So web processing revolves around both things and stuffs,
and our logical vocabulary is not going to do the job if it doesn't deal
directly with both.

So please don't blow off what you know about logic because of what we didn't
know then.  We now have enough evidence that we should have known better.


>Stephen Cranefield wrote,
> > Miles Sabin <miles@milessabin.com> wrote:
> > > At issue is the first sentence of the informal definition of
> > > resource in RFC 2396 1.1,
> > >
> > >   A resource can be anything that has identity.
> > >
> > > "that has identity" is redundant because *everything* has identity
> > > in the only reasonably straightforward understanding of identity,
> > > ie. the logical truth in all but the most obscure formal systems
> > > that,
> > >
> > >   (Vx) x = x
> >
> > A discussion of the philosophical notion of identity can be found in:
> >
> > Guarino, Nicola and Chris Welty. 2000. Identity, Unity, and
> > Individuality: Towards a formal toolkit for ontological analysis. In,
> > Horn, W. ed., Proceedings of ECAI-2000: The European Conference on
> > Artificial Intelligence. pp. 219-223. Berlin: IOS Press. August, 2000
> > http://www.ladseb.pd.cnr.it/infor/Ontology/Papers/LADSEB02-2000.pdf
>It's an interesting enough paper, but it's about unity, integrity and
>persistence of physical objects rather than identity. And that's a good
>thing, because their theory appears to be expressed in terms of
>first-order predicate calculus with identity initially, then later a
>simple first-order mereology and set theory, both of which presuppose
>identity: so if it really was supposed to be a theory of identity it'd
>be hopelessly circular. There are formal systems around which don't
>take identity as a primitive, but, like I said, they're somewhat
>That said, if the qualifier in RFC 2396 said something like,
>   A resource can be anything that is integrated (for some value of
>   "integrated")
>it would at least be adding something non-trivial. But I'd strongly
>advise against going down that route. Integrity is a slippery enough
>concept for physical objects (eg. it's extremely difficult to
>distinguish is-attached-to from is-part-of without running into all
>kinds of awkward edge cases) never mind the kind of electronic
>artifacts which are RFC 2396's primary (tho' maybe not exclusive)
>There's also a brief discussion of the relation between stuffs and
>things, but that's irrelevant here: RFC 2396 talks about things from
>the outset, so individuation is already presupposed.
Received on Monday, 9 September 2002 20:11:05 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:04 UTC