Re: W3C response to proposed Atom Publishing Format and Protocol (atompub) working group

[Eric Miller <em@xxxxxx>]

Hi Mark,

Thanks for your thoughts and I appreciate articulating some of these 
concerns. Comments follow...

On May 13, 2004, at 3:09 PM, Mark Nottingham wrote:

> > > - Open participation and consensus
> > >
> > >   An Atom Working Group in the W3C could charter itself to operate 
> > > in the public, and invite participants who are not from W3C Member 
> > > organizations at the chair's prerogative. We recognize that much of 
> > > the work done on Atom has been by non-Members, and want to assure 
> > > those participants that they may take part fully in a W3C Atom 
> > > Working Group.
> > >
> > >  Generally, each participant identified as an individual entity or 
> > > company has a single vote. Decisions are made by consensus and will 
> > > give the possibility to respect the process that the Atom group has 
> > > been able to achieve. Participation in this regard is open to both 
> > > W3C Members and Invited Experts from the community.
>
> This gives a large amount of power to the chosen chair; there may be 
> people who feel excluded.

If the charter was written in the way stated above you are correct in 
that it gives additional power to the chair.  The chair (and possible 
co-chair) work with the team contact. Good chairs and good team 
contacts however try and make sure people aren't excluded. :)

Working Group membership is not an open door; you must present 
credentials and have those reviewed by both the chair and the Team 
contact. Then you can come on in. These processes are in place to help 
assure accountability, productivity and ensure Royalty Free licensing. 
There are hurdles. But these hurdles are relatively small for the 
accrued benefits.

> Even if that can be avoided, the make-up of the Atom community leads 
> me to believe that there will be more Invited Experts than 
> representatives of W3C Members on the WG. Would this be acceptable, 
> from an AC standpoint? In particular, it would be damaging to all 
> concerned if we stalled this effort, only to find out that the AC was 
> not amenable to having what amounts to a non-Member WG formed. Even 
> worse, serious problems could result if the WG came to consensus, but 
> the AC was not happy with that result. Would the Director be willing 
> to prioritise the WG over the AC's advice in such a situation?

There are several W3C Groups where the make up includes many Invited 
Experts (several groups in WAI immediately come to mind). So, this 
group would not be alone in this regard.  Further, a quick scan of the 
Atom participants and it strikes me there are many individuals who 
already work for W3C organizations so this actually helps :).  As for 
the what the director might do in such a situation as you suggest, I 
don't speak for the director. If this is important question to have 
answered I'll talk with him and get back with you.

> Regarding consensus: although I believe they are very helpful in most 
> cases, the W3C's more strict requirements for consensus may turn out 
> to be a disadvantage in the case of Atom. Because syndication has been 
> a contentious topic for some time, and because there are several 
> potential avenues for its development, rough consensus and running 
> code is -- on the face of it -- a more reasonable approach. Can you 
> explain more fully how the consensus process at the W3C would result 
> in a better result, in this particular instance?

I'm a firm believer in the IETF's credo 'rough consensus and working 
code', its had to argue with this to get things up and running.  I 
think Atom is at that place and frankly ready to move on to start 
nailing down the specifics and tighten up the working code. For this I 
find a process for consensus quite helpful. The social change however 
will take work.

I'd expect, however, that there would be an additional responsibility 
on the chairs and staff contact to define some additional guidelines 
for good standings in this case of a group with several IE's. I'll look 
into this in some of the WAI groups to see if they have already done 
this.

> Finally, one of the reasons that the IETF seemed like the appropriate 
> choice was the fact that it's widely regarded as neutral. Some, both 
> inside and outside of the Atom community, perceive the W3C has having 
> a vested interest in the introduction of RDF and/or a relationship to 
> RSS 1.0 (of which you were an author). Whether or not this is true, 
> this perception may make place a cloud over the formation and product 
> of the WG. What steps would you take to dispel this?

I'm a firm believer of technologies succeeding or failing on their 
merit. We all have different background and experiences in building 
applications and working on various specs to help make our jobs easier. 
 The scope of the charter is a handy tool for defining some bounding 
boxes on the technologies that are on the table for consideration.

Yes, I do have a vested interested in RDF and in applications (RSS1.0) 
no doubt about it. But I'm only one person and have one voice (and if I 
don't join the working group, I don't even have that :). For me these 
technologies are valuable because they helped solve various 
interoperability problems I worked on for a long time and now see on a 
daily basis. If people don't have data integration requirements they 
don't need to be using data models to facilitate data integration.

The group decides. This is pro / con aspects of consensus.

> > > - W3C Patent Policy: Royalty Free licensing
> > >
> > >   The process around working groups participation helps ensure 
> > > accountability and encourage Royalty Free licensing of the 
> > > technology. RF licensing in particular is an issue we've heard is 
> > > important to the syndication community. More information about this 
> > > policy is available at:
>
> As you know, the W3C Patent Policy only provides for RF licensing by 
> Working Group Members. Such an end can be achieved in the IETF as 
> well, if the WG adopts a policy of only accepting unencumbered 
> contributions. It is true that doing this work in the W3C will compel 
> non-participating Members to disclose any claims they may have, but 
> I'm not convinced that this would be a sizeable benefit. Am I missing 
> something? This is IP, so I'm sure I am, but I don't believe there's 
> any inherent weakness in the IETF process WRT IPR.

I don't know enough about the IETF's IPR process to comment. I do know 
that having been burned in the past, I really appreciate the process 
W3C has put in place.

> > > - I18n, Accessibility, Device Independence Support and Coordination
> > >
> > >   The Atom group will benefit from the expertise of these groups 
> > > which in turn will help ensure wide deployment regardless of device, 
> > > language, or ability. Further we expect the resolution of any 
> > > coordination issues arising from these groups will be resolved more 
> > > quickly at the W3C.
> > >
> > > - XHTML Cooperation
> > >
> > >   It is important to understand the relationship between the Atom 
> > > API and the Atom document format. Specifically, we are concerned to 
> > > understand how these might evolve (independently or in-step) and 
> > > whether the API can be used with RDF formats such as RSS1, or W3C 
> > > hypertext formats (e.g. XHTML, or mixed-namespace content). Recent 
> > > W3C work (pre-draft at http://www.w3.org/MarkUp/2004/02/xhtml-rdf) 
> > > on XHTML2 and RDF is combining RDF's metadata approach with XHTML 
> > > document syntax, allowing paragraphs of text to be tagged with 
> > > information about their topic, source, author etc. We suspect 
> > > further discussions (perhaps a Workshop) are needed to fully 
> > > understand how the evolution of XHTML-based approaches relate to 
> > > Atom and blog-oriented systems. Since XHTML with embedded metadata 
> > > can be used to directly encode syndicated content, we believe the 
> > > interdependencies between the Atom API and Atom document format 
> > > deserve some attention, both in chartering the work (whether at W3C 
> > > or IETF) and as the work proceeds.
> > >
> > > We believe the cooperation benefits to be great, and the 
> > > coordination cost will be addressed more quickly if this work 
> > > happens at W3C.
> > >
> > > - XML and RDF experience
> > >
> > >   W3C has years of expertise in definition of content format 
> > > language and model description. XML serialization and RDF model are 
> > > areas of expertise at the W3C. Though when it comes to a content 
> > > model, the organization is pretty agnostic (CSS, XML, RDF for 
> > > example are different models). The Atom group is the one defining 
> > > its content model and syntax.
> > >
> > >   Conversely, Atom would benefit from direct liaisons with these and 
> > > other W3C Working Groups to resolve potential conflicts and 
> > > influence future specifications.
>
> I can't speak for the Atom community, but I've heard a number of 
> reservations about the complexity inherent in the XHTML specifications 
> as well as Semantic Web.  While I personally believe that these 
> technologies have a lot to offer Atom, the community hasn't achieved 
> consensus about adopting them, and I don't believe that the choice 
> should be made for them by fiat. Would a W3C charter include mandated 
> relationships with these groups, or particular requirements for their 
> deliverables?

If the Atom community decided it wanted to go to W3C, I expect the 
charter would to help identify scope, related technologies and 
relationships to other working groups that would be advantageous to 
coordinate with. In a group like this, I'd suggest there would be 
strong coordination points with QA, WAI and I18N.  But if your worried 
about 'you must only use RDF', this is not the case. (If the group 
decided it like the RDF Model, but not the serialization for example, 
I'd imagine something like GRDDL 
	http://www.w3.org/TR/2004/NOTE-grddl-20040413/ would be discussed.)

There is a lot a good work that is currently underway in XHTML and 
Semantic Web. I too believe these technologies have a lot to offer Atom 
and its one of the reasons why I think better coordination with these 
groups makes sense and benefits all.

> Also, I note that the IETF and W3C have had several successful 
> interactions in the past. What, in particular, is the disadvantage of 
> pursuing these liaison relationships from the IETF, rather than within 
> the W3C? Superficially, it seems that the only reason they'd need to 
> be deep enough to require close co-ordination by the Team is if there 
> were requirements placed upon the WG by them.

This is certainly an option and one that if the Atom community thinks 
is a good one we could collectively pursue.  My experience with working 
on join organizational efforts in the past, however,  (note: this was 
not IETF and W3C) seemed to be the case where we took on the overhead 
of both, but didn't get the benefits of either.  Feedback from 
individuals that have participated in both in the past would be welcome 
here.

> > > -  QA support
> > >
> > > We have noticed that the Atom community already has a lot of early 
> > > implementation and that you have started a testing effort which is 
> > > already impressive. In the context of W3C, this will accelerate the 
> > > advancement of the Atom spec along the W3C Recommendation track. 
> > > It's great to see that a technology is already so mature with a 
> > > well-built quality process.
>
> Would this take the form of W3C Team members contributing to test 
> cases and/or running code? How much of a Team Member's time would be 
> devoted to such a WG? Having professional staff would a substantial 
> advantage.

I could imagine it might involve W3C Team members, and/or it might 
involve getting additional QA individuals from other W3C members who 
have shown interest in this work to participate. It seems to me the 
later being the best option.

> > > While we recognize the IETF submission, we'd like to hear from 
> > > others within the Atom community regarding the above points, and 
> > > more specifically, their requirements and objectives regarding 
> > > taking this to a standards organization. We'd support proposing a 
> > > W3C Working Group in this area, provided there is support from the 
> > > larger Atom community regarding the points mentioned above and that 
> > > agreement about work areas can be reached with the IETF.
>
> Another concern that I have is the W3C policy of doing substantial 
> technical work in face-to-face meetings. While I agree that this is 
> very productive, it may be too much of a burden for so many Invited 
> Experts, especially considering the W3C's requirement that such 
> meetings be distributed across the globe. I, for example, would likely 
> be participating as an Invited Expert, and would not be able to 
> travel, except when my work plans happen to coincide. The IETF's 
> policy of doing substantial technical work on the mailing list, and 
> not requiring participation in plenary sessions, seems like a better 
> fit.

I don't believe there is a *policy* for doing substantial technical 
work in face-to-face meetings but rather a by-product of being 
face-to-face.  How the group works is stipulated in the charter. Weekly 
teleconferences and active mailing lists, plus IRC, wiki, etc. are all 
tools. The key is to balance the communication tools with the community 
to be the most effective.   Some do best f2f, some prefer more 
asynchronous communication.

> Because of the issues mentioned above, my perception at this time is 
> that it's more natural to have an IETF WG with liaisons to the 
> appropriate W3C WGs. Would this be workable for the W3C?

Absolutely. If the Atom community believes thats the best fit in the 
end.  I think its important to understand the benefits and costs 
associated with this however.

--
eric miller                              http://www.w3.org/people/em/
semantic web activity lead               http://www.w3.org/2001/sw/
w3c world wide web consortium            http://www.w3.org/