bestkungfu weblog

Atom/W3C redux

Filed in: tech, Web, Fri, Jun 4 2004 09:19 PT

Atom logo

The Atom community meeting happened at Sun on Wednesday. Eric Miller and I came along for the ride to represent W3C as a swell place to create a spec, but we generally failed to, as they say, win hearts and minds. That’s too bad, and I feel that’s a failure on my part, so here’s a little bit on what the hell we were thinking.

We at W3C like Atom. In fact, we like it so much that we want to work on it. We like it so much that we’re offering a staff member to play the role of process sherpa. We’re opening our doors, not unlike what we’ve done in the past for specs we thought were important. We’re doing it because we get it, and we care about it, and we want to help it succeed when it’s made a standard.

I know what we offered, and it’s still on the table. A W3C partisan like myself lays it out thusly: a W3C Working Group, and probably also an Interest Group; a W3C team contact who works 30-40% of the time facilitating; a royalty-free patent policy; open membership to anyone who wants to participate; conference bridges and meeting space; systems for meeting management, logging, straw polls and questionnaires; input from accessibility, quality assurance and internationalization experts; and publicity for the spec as it progresses.

The IETF side (and I’ll be happy to fill in what I’m missing) offers: an IETF Working Group which takes effect immediately upon approval (W3C would likely take six weeks to approve and set up the group); a patent policy which they argue is either superior to or not substantially different from W3C’s; limited process constraints; input from security experts; and open participation to anyone.

Both groups offer the opportunity for public comment, and require that those comments be addressed. (Yes, really. You can issue a formal comment on any spec W3C is advancing.)

Back at the meeting, I think everyone danced around the elephant in the room, which is time to market. My sense is that the Atom community feels they want as little outside pressure on their process as possible, and the W3C’s more formal consensus process looks more onerous than IETF’s rough-consensus-and-running-code model. Of course that’s appealing to people writing running code. But is it enough for a large group with a wide range of viewpoints, and a history of forking when views diverge? It’s natural that a formal process takes cycles, but it’s not true that those cycles are wasted. We require issues to be resolved and documented (publicly) in order to show our work. We require a charter listing the scope and timing of a working group’s deliverables to keep specs like Atom from growing a kitchen sink, and limping out years late. It takes some discipline to make sure decisions are made fairly, compromises are fair and successful, and everybody is working together toward a common goal. Can you do it faster without lots of rules? Maybe, but the larger the spec, and the more hands in it, the more risky the venture gets.

Tim Bray said, essentially, that the argument is one of inertia. At this point, he said, W3C would need to be more than 10% better than IETF in order to change course. And that makes sense: Atom went to IETF, put effort behind getting it approved, and, quite honestly, we didn’t exactly put our best foot forward by introducing ourselves with a message to IETF saying we think we’re a better choice. Sorry about that, guys. It was either that or buy Google ads on ongoing.

We know that Atom wants to get on with it, and I can’t blame them. But the early parts of the process aren’t what’s important: it’s the end product. A W3C Recommendation comes with a test suite, which is hugely beneficial on specs that require a great deal of interop. It comes with proof of interoperable implementations, developer testimonials, and publicity. Not bad for free.

That’s why I don’t think it was a bad idea to jump in. Atom 0.4 isn’t days away from publication, as far as I can tell, and the group seems to be functioning just fine in the interim. Whichever way Atom goes, the worst hit they’ll take is a reformatting of the document. Having a running working group in a standards body only matters when a group is ready to publish.

A straw poll was taken by a group hum. It sounded like five or six to one preferred IETF over W3C, though surprisingly enough, that one was Bob Wyman, someone I pegged as the least likely convert I could think of. And what he said to wrap things up was that, sooner or later, he felt it needed to go to the W3C, so Atom may as well start there. Mind you, he said it in a kind of exasperated fashion, but I think he’s right, and that’s the reason I’ve put a lot of work into making the case both inside the W3C and out. I don’t stand to gain anything personally or professionally by “winning” Atom for W3C. I just think it’s the right thing to do.

Those of us at W3C who have been working on this each saw a hole that Atom fills, whether it’s a simplified publication model, a more accessible means of getting to the core content of a site, or a great way to ease users into the upside of the semantic Web. And we think Atom should be in an environment where it can cross-pollinate with Web specs like XHTML and RDF. (Though that doesn’t mean going back to RSS 1.0 at all.) I think that what we offer over IETF, all told, does meet Tim’s better-than-10% criterion.

If Atom goes to IETF, and that’s how it’s looking, Atom remains a great idea, and we wish it well. But I still believe that the W3C is the best place for Atom to take root.

Comments are closed.

Powered by WordPress (RSS 2.0, Atom)