Weblogs: Atom

Atom and the W3C

Friday, June 04, 2004

The first Atom Community Meeting took place on Wednesday 2 June, organised and hosted by Tim Bray. On the agenda was a discussion of whether Atom should go the IETF or W3C route, or both. Tim's notes on the meeting, and the IRC log of the event document much of the discussion.

Compatability with the W3C was the main focal point of the first half of the meeting. Tim Bray suggesting that inertia is a key ingredient of the standards organisation decision. In Tim's words:

We've invested a lot of work in the IETF option, so the W3C would have to win by a more than 5% margin to make it attractive to me.

Matt May, one of the W3C representatives attending the meeting has made an excellent effort to prove there is more than a 5% benefit to doing Atom within the W3C.

Passion and commitment

What is evident in Matt's posting is that there is a passion about Atom inside the W3C. This passion is centered around the technical benefits offered by the Atom package. It is interesting to note that the W3C people who've come together to make this approach to the Atom community are from separate Working Groups. Each of them have independantly seen a benefit of working with Atom.

Their commitment to Atom if we decide not to go to the W3C is the same level of commitment most of the current Atom community offers - we all have full-time jobs mostly unrelated to Atom, and we do Atom stuff in our spare time. Its that ad-hoc committment that keeps the specification vendor-neutral (alongside Sam and Tim's excellent company-independant approach to guiding Atom along its path).

Working on Atom as a day-job

The extra special bit going to the W3C offers is bringing onboard some seriously professional experience in a day-job capacity. That, in my opinion, is how serious they are taking Atom - its worth allocating people in a full-time role to bring it to fruition.

Looking at the experience of the people representing the W3C in this discussion, they are amongst the cream of the technical staff. Matt May has seriously impressed me technically within his remit on the Web Accessibility Initiative. Dan Brickley strikes me as very tecchie, so to does Eric Miller. Karl Dubost I've known over the last year as a die-hard web standards evangelist, and he's another person who has seriously impressed me with the quality and assuredness of his work.

Of course, it would be absolutely ideal to have people like Mark Pilgrim, Joe Gregorio, Mark Nottingham, Ken MacLeod working on this as part of their day jobs, I doubt this is going to be remotely feasible.

Having some technical people in the W3C interested in Atom - interested enough to convince their management, even Tim Berners Lee - offering to help out as part of their day jobs - that's a bonus I've not seen from the IETF.

The Atom and W3C community

The one big downer for me is what I see as the fundamental incompatibility of the organisation of Atom and the W3C.

The Atom community is an open developer and user community. Its an ad-hoc community, with people dropping by to add their value, perhaps disappearing for a few months, then reappearing. These people still add a lot of value to the Atom process. What matters in this community is what people do. Working code, drafting a specification, demonstrating workable alternatives, taking the initiative on certain issues - that's the attributes that make the Atom community work.

The W3C is organised around member companies, invited experts and interest groups. It seems to be a strict identification of roles people play. Trying to map people from the Atom community to a proposed W3C structure creates a situation where the number of Invited Experts will probably about four times the number of the other Working Group members.

But with a defined role of Invited Experts comes the obligations outlined by the W3C - their member of good standing describes these obligations. Its a heavy price to pay compared to the free-flowing nature of the current Atom community.

The ideal is for the W3C to allow the Atom structure to remain as is - a tiny group of "full-time" members, a massive group of contributors who discuss, specify, code, test, analyse, correct, improve the specifications. This massive group evolves, grows, shrinks based on time availability and interest of its members. It is a hands-off approach, rather than a top-heavy, but it is clearly working.

The community of companies flaw

The incompatibility of communities reveals itself in the weird situation that within the W3C, Sam Ruby and Mark Pilgrim consistute one voice instead of two (in a voting situation). That's a heavy sacrifice to make, its not a price I would willingly pay.

Ideally, if I could have my own way, I'd be an Invited Expert with the express proviso that any time a voting situation arises, I would delegate that vote (and other privileges) to Mark Pilgrim. But just having to do that is fighting against a structure that doesn't work for us.

On a real world note, I can't meet the minimum requirements of an Invited Expert, nor would I qualify as one. But I hope the hypothetical scenario demonstrates the necessity of getting around the member company problem.

The W3C representatives have mentioned a few times that the structure of Working Groups can be flexed and changed. I would like to see the W3C guys propose a structure that is more amenable to how the current Atom community operates. That is a discussion I feel needs to be done before we make a decision about the W3C.

The RDF and WSDL groups

Mark Nottingham raised the issue of the RDF and WSDL groups wanting to bring their standards into Atom. The W3C representatives have clearly committed to Atom regardless of whether RDF is chosen or not.

However, I do expect the discussion to take place again. Even though we've previously decided not to go the RDF route, and most of the RDF experts are still on board accepting that decision. We will need to have that discussion, this time with the RDF experts. It would be naive of us to expect otherwise.

Conclusion

I do feel like working with the W3C does offer us benefits. I stated on IRC I prefer the working code and rough consensus approach of IETF - I prefer to talk with working code since specification discussions are too abstract for me to follow comfortably. But the visibility and resources of the W3C are very tempting for an emerging standard.

There is a heavy price to be paid for working within the W3C. This is particularly in mapping the current Atom community to the W3C roles and groups. There's not a one-to-one match. And the second problem is treating Sam Ruby and Mark Pilgrim as one entity instead of two committed and productive developers. Ironing those two issues out will certainly guarantee a Yes to the W3C from me.