W3C home > Mailing lists > Public > www-talk@w3.org > May to June 2002

RE: The Standards Manifesto

From: Bill de hÓra <dehora@eircom.net>
Date: Thu, 23 May 2002 01:47:57 +0100
To: "'Dan Brickley'" <danbri@w3.org>, "'Aaron Swartz'" <me@aaronsw.com>
Cc: <www-rdf-interest@w3.org>, <www-talk@w3.org>, "'Tim Berners-Lee'" <timbl@w3.org>, "'Simon St.Laurent'" <simonstl@simonstl.com>
Message-ID: <002401c201f3$7c72bc50$887ba8c0@mitchum>
 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

> -----Original Message-----
> From: www-rdf-interest-request@w3.org 
> [mailto:www-rdf-interest-request@w3.org] On Behalf Of Dan
> Brickley  
> 
> There is no right way to do this. No process is perfect. 
> W3C's process, however, is both documented and evolving in 
> response to change. These are both fine qualities. If you 
> have specific feedback on your experience as a working group 
> member and/or interest group member, there are non-megaphone 
> ways of bringing them to the attention of W3C and the W3C 
> Team. Sometimes shouting can be counter-productive.
> 
> Your message below reads uncomfortably close to a direct 
> attack on your RDF Core WG colleagues, who I would assert 
> already 'know their stuff', share a 'concern about simplicity 
> and the Right Thing'. Fortunately the proceedings of RDF Core 
> (mailing lists, irc, meeting minutes, issue list) are a 
> matter of public record, so readers of your rabble rousing 
> inspirational manifesto can take a look at how the group has 
> worked over the last year.

I appreciate where Aarons' coming from, but I'm in agreement with
Dan's post re RDF. As for the general state of the W3C...well, it's
a consortium. At least it's not a cartel. The best you can do is be
really smart technically, really persistent, and hope mutual
self-interest averages out.


> Since this is the RDF Interest Group list, I've a suggestion. 
> Could you take the time to recast your manifesto as a 
> proposal for making the RDF Interest Group a more effective, 
> useful forum to complement W3C's existing Working Group 
> machinery? Or do you really believe the whole thing is rotten 
> and needs replacing wholesale?

Speaking as someone who doesn't have their face pressed against the
RDF coal-face these days (and somewhat ex-cathedra, apologies for
jumping in, I have a soft spot for RDF :).

Really short version: no denotation without notation!

Short version: amend the current charter - develop an XML syntax
from the ground up  - clearly admit RDF is an investment - hound
the WS community to write WSDL in RDF (thus unifying DAML-S and
WSDL) - deprecate backward compatibility for the time being - if
RDF is simple make it obvious - the W3C and its members are not the
chokepoint.

Long version:

0) The people involved, selfish interests with fiduciary duties,
W3C dropping the ball, et al, these are red-herrings for RDF.
That's to say taking RDF into a radically new process won't fix
much that can't be fixed now. RDF will sink or swim on its own
merits and the shrewdness of the people that want to see it widely
deployed, the process isn’t the problem.

1) Wrt to process, the scope of the wg charter is imbalanced.
Unless formalization comes into play, it's been pretty tough to get
things pushed through against a "bug fixing" scope. 

2) There's something out of kilter when a wg has to invent a new
syntax to skip around the existing XML one, because the wg has
difficulty using the XML, yet taking a broad-axe to it is deemed
"out of scope". The existence of N3 the traffic on this list, and I
think two (?) simplified XML syntaxes should have set the alarms
bells off long ago. The wg, and the syntax subgroup in particular,
has done /remarkable/ work to get the current syntax wd to the
state it's in under the circumstances. 

3) A charter that untied the wg's hands to address syntax could
quickly see RDF underpinning any incumbent WSDL Recommendation.
Read: everything very much comes up roses if we unify WSDL and
DAML-S via an RDFXML. WSDL is very much a killer app for RDF. And
no, a non-normative wsdl2rdf screenscraper or transliteration
doesn't cut it. WSDL needs to be written /in/ RDFXML. If this isn't
a tactical goal for the semantic web initiative and a strategic one
for the WS side of the house, well, it should be. 

4) Wrt to deployment, RDF's costs are frontloaded. We think it's
going useful, some day, because we have a notion that information
in RDF form is highly repurposable and easy to merge
(serializations notwithstanding). I haven't seen much by way of
acknowledgement that RDF is a pension plan for your information,
and surely it wouldn’t hurt any to get this message across some
more. 

5) This strategy may pay yet great dividends. You would think it
makes sense because information tends to stick around longer than
applications do. On the other hand, information in itself is not
the core function of information technology, building applications
that process and generate information are. 

6) I recognize that formalizing RDF has been valuable (the
logicians told us we needed to do it, mucked in, and we have good
reason to believe that was a good thing). However the MT is not how
the next million RDF users will come into the fold; that will be
best done through a vastly simplified XML syntax clearing the path
for basic applications and manipulations. Sugar coating what's good
for you comes down to syntax. 

7) [An aside: The argument that syntax doesn't matter when you've
got an MT is ostensibly a correct one, albeit crashingly trivial.
It's on a par with saying programming languages are Turing
complete, so programming languages ultimately don't matter. Any
developer, or anyone who's followed the traipse of XML, will be
aware that syntax matters a great deal. As an aside to the aside, a
W3C blessed RDF API (UML not IDL thanks!) wouldn't hurt any.]

8) As for backward compatibility being a tough call. Backward
compatibility IMO seems important when you have a sizable installed
base to protect. Backward incompatible changes to technologies with
large user bases are to be resisted, i.e. you don't muck about with
the Internet (ipv6) or the European monetary system (Euro), unless
you feel justified in thinking things will plain come apart without
the change. However a sizable user base is precisely what RDF does
not have. [Warning: bogus reductio argument follows.] Consider the
number of RDF users and dependent technologies 5 years from now;
the numbers today will likely as not be a fraction of that (if
they're not, then we may as well all go home now ;). If five years
isn't enough, try seven or ten or fifteen years, and ask if
backward compatibility for RDF makes sense today, given its current
state. I suggest it does not, and backward compatibility for RDF is
something that needs to be justified, not assumed. The required
resistance will likely as not arise of its own accord in due time
if and when RDF sees wide deployment. [Warning aside, if relatively
few people are using your technology you have scope to break it to
develop a future market] 

9) Even the "it's too complicated" viewpoint seems bogus; the
syntax is and always has thrown sand into people's eyes. If RDF is
brain-dead simple, shouldn't there reasonably be a brain-dead
simple XML syntax as a consequence? XML Schema is far more
complicated than RDF+MT. My instinct is that WSDL is more
complicated or par with RDF+MT. However their syntaxes can be
grasped, and the needs are clear enough that tools for tools are
being developed for WSDL and Schema without much pause for thought.
 There are claims that it is in essence, simple stuff, and I'm
inclined to believe them. It beggars belief then, that something 
as simple as RDF can justify such obfuscated markup. 

Bill de hÓra

-----BEGIN PGP SIGNATURE-----
Version: PGP 7.0.4

iQA/AwUBPOw76OaWiFwg2CH4EQLLKwCgv5YakRjONFHJl4pHR723V8Z/BV8AoKr3
xmM2mCxFzc36whjTiUI0jKS4
=i7rt
-----END PGP SIGNATURE-----
Received on Wednesday, 22 May 2002 20:50:09 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 27 October 2010 18:14:27 GMT