Re: Thinking ahead to next Tuesday's telecon...

On Mon, Jun 29, 2009 at 5:52 PM, Harry Halpin<hhalpin@ibiblio.org> wrote:

> The reason I think the ontology is so big is that in general, the
> issue with lots of these specs (URI specs, especially those mentioning
> "conceptual mapping", even bits of Fielding's thesis, and even the
> HTTP specs) is that modelling the entire spec is a bit difficult, as
> the specs were written relatively informally.

No this isn't the reason. The reason is that I want to put all
definitions together in a single picture, to make it easier to choose
between them. So ultimately it has to do not with a large number of
things or parts, but with a large number of distinct definitions from
many specs that must be consciously either discounted or reconciled. I
don't want to make yet another new theory of everything; that would be
too easy and would not help with integration and conflict resolution.
I want to pick and choose from what's already present. As I have said
before the final ontology will be much crisper - we agree on this -
but no crisper than is correct.

Another purpose in making all those boxes at the top is to put in
everyone's face what an embarrassment this is. So I will fight any
suggestion that this ontology is a "proposal".

My comment about you thinking representations are ephemeral comes from
your IRW ontology:

"The realization of a message encoding that 'goes on the wire'
according to an interaction protocol (e.g. http) in order to resolve a
Web accessible resource."

which I misinterpreted as saying that they only go on the wire and not
elsewhere. Sorry about that. Maybe you could clarify what's meant by
"realization" (I thought have that would be: "made real" or "made
physical") so I don't get misled next time I come across this.

What's the difference between a message encoding and a realization of it?

Note that in HTTP, representations (which are entities) are subject to
a transfer-encoding, so what goes on the wire is actually an encoding
of an entity, not the entity itself. The entity body itself has a
content-transfer-encoding applied to it, and is in turn sometimes
subject to a character encoding. Then the characters may encode yet
another thing, via application of the content-type, and so on... so
"encoding" is a slippery and prolific relationship, and piling
"realization", which sounds like another kind of encoding, on top of
the stack doesn't seem to shed much light.

Regarding equivalence of distinct entities -- well, the generic
resource theory already *is* a theory of equivalence - at each time a
generic resource has a set of equivalent representations. If it has to
be built on another theory that needs its own notion of equivalence
that would be a failure. Better to just build genont on a theoriy of
things (representations) that are adequate to capture, in their own
identities, the foundation needed to explain "fixed", so that you can
effectively leverage intuition, ordinary language, model theory and
inference and not repeat have to repeat equality at an intermediate
level of the ontological tower, so to speak. (Assuming we want to
pursue genont and equivalence theories at all, that is.) Requiring an
equivalence theory for almost *any* class seems to me a failure to
find the right thing (ontologically speaking) to which each
equivalence class would correspond.

Best
Jonathan

Received on Monday, 29 June 2009 22:37:59 UTC