> -----Original Message-----
> From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Tim Bray
> Sent: Saturday, January 08, 2005 10:14 AM
> To: Bill de h�ra
> Cc: 'Atom WG'
> Subject: Re: Atom extensibility, RDF, and GRDDL

--snip--

> The people who insist that you have to have to buy into a 
> graph-theoretical KR model to be able to use the word "extensible" live 
> in a universe that is self-consistent, but it's not the one where I 
> work. -Tim

Having stepped back from both Atom and RDF a bit during the last few months,
in part because I want to make sure I'm not wearing X-coloured glasses too
often, but also because I find that I get too invested in and stressed out
by some of the more heated discussions :), I've been trying to avoid jumping
into this thread, but think its about time that I chime in and make my
feelings known on this subject.

It has long been apparent that the Atom community is not ready for RDF.
While much of the industry is finally starting to see the benefits of
semantic web technology, hence the emergence of terms like "semantic
integration" (as an evolutionary step beyond current data integration
techniques/tools/etc.), Atom and its community just aren't there yet.

RDF is unfortunately one of those things that is difficult to understand
until you've been bitten by one or more of the issues that it addresses and
find yourself in a position to "get it". Once you need it, you "get it".
Once you've "gotten it", you use it. Once you've "gotten it", it also
becomes frustrating to watch other people suffer while not "getting it".

You have the Atom community on one side, looking at RDF and thinking "the
syntax is dumb, the model is unnecessary, XML is enough." Then you have the
RDF pundits on the other side thinking "How can these guys not get it yet?"
And this isn't meant to slight the Atom community in the least. The Atom
community doesn't get it yet because it hasn't needed it yet, and
unfortunately, as I mentioned before, RDF is one of those things that is
hard to really "get" until you really need it [1].

And, as I've come to feel over the last few months, this can actually be
seen as a _good thing_:

Atom (or its successor) will adapt to (or simply somewhat align with) RDF
when its community "gets it", which will occur if and only if / when and
only when they _need_ it.

The problem is that in the meantime most of the arguments from the RDF
pundits are not only largely misunderstood or just seen as irrelevant, they
act _against_ the intentions of the RDF pundits by coming off as the
rantings of zealots of those want codified things that are obvious and who
live in some strange world where everything is a graph. Not that this is the
case, just that it ends up appearing as such. And I can safely say both of
these because I am one of those zealots. :)

With that in mind, here's what I would suggest:

Henry, keep posting your periodic messages trying to softly help people
understand RDF (and OWL, as many of your posts rely on it). Everyone seems
to have missed how your most recent posts showed that an OWL reasoner and
the rules provided to it would keep an application from even having to
_implement_ logic to decide what entries should be merged or kept separate,
and they all seemed to miss the fact that the included locations could be
deemed equivalent and then merge together any information and/or relations
ascribed to each of the two expressed locations. These are worthwhile things
for people to understand, so keep at it. At least one of us needs to keep
trying to help the community "get it", and in recent months I feel you have
been doing the best job delivering that message in a clear, even-handed
fashion.

In general, the RDF pundits need to step back and recognize that until such
time as the community has been bitten, needs, and subsequently "gets" the
semweb, the best thing you can do is recognize that an RDF version of Atom
will need to be created as a mapping, that mappings will also need to be
created for any Atom extensions, and to spend your time helping Atom (and
its extension authors) avoid doing things that will directly hamper RDF
mapping efforts. Get working on those mappings, keep them in sync, keep the
feedback loop going between the main effort and the mapping efforts, and
work to create apps that will help the community edge closer to "getting
it".

Jeremy

[1] This is worsened by the fact that the specification documents were
largely HORRIBLE for several years, and because the intentions behind and
syntax resulting from the RDF/XML effort is often misunderstood, but those
are subjects for another place and time. :D


Reply via email to