Re: "destabilizing core technologies: was Re: An RDF wishlist

Dan and all, hello.

On 2010 Jul 1, at 11:30, Dan Brickley wrote:

> Yes, you are right. It is fair and interesting to bring up this
> analogy and associated history. SGML even got a namecheck in the
> original announcement of the Web,


> So, I think I'm holding an awkward position here:
> * massive feature change (ie. not using triples, URIs etc); or rather
> focus change: become a "data sharing in the Web" community not a
> "doing stuff with triples" community
> * cautious feature change (tweaking the triple model doesn't have many
> big wins; it's simple already)

This is a very thought-provoking argument.

There's a third position: no feature change, but change our goals.

Myself, I finally 'got' SGML (and by implication XML), when I struggled through the HyTime spec (anyone remember that?), and finally really properly understood the extent to which it was the _abstraction_ of structured information that was the point of it, and nothing to do with what the tags looked like, or which parser you were using, or any of that mess.  It may be HyTime that the RDF data model is analogous to, and not XML after all.

Because of the many-fold annoyingnesses of RDF, it's far too easy for us to be distracted by issues which are to some greater or lesser extent syntactical (which arguably includes the 'literals as subjects' argument).

One big point about RDF, for me, is that if you can prove (perhaps only to yourself) that you can round-trip information from its source, into triples, and back, then you have proven something useful and interesting about the RDF model in question, _even if_ you never actually publish those triples.  If you can serialise those (abstract) triples into an EXIF header, a FITS header (my game), JSON, XML, HTML, or whatever, then you completely know the semantics of that serialisation, and that there isn't anything missing.  The analogy with HyTime is therefore that it helps the understanding or discipline of a designer or standards author, not that it's anything anyone else would want to see.  (HyTime was addressed to DTD authors, not SGML authors).

Thus I can imagine setting up a service which stores its 'knowledge' as RDF-style triples (because that's a good fit to the heterogeneous nature of the information in question), but makes it available, and documented, only as JSON or microformats/RDFa.  It's fundamentally an RDF application, but there isn't a single triple visible from the outside.

Thus we might not have to change the spec, just drop our expectation that anyone who isn't an 'information architect'[1] will ever read it.  Changing who we regard as the 'users' might also release us from some obsessing about creating user-accessible tools.

Best wishes,


[1] Ooooo!  Hats off!

Norman Gray  :

Received on Thursday, 1 July 2010 14:35:43 UTC