W3C Invitation

[Michael Champion <mc@xxxxxxxxxxx>]

Just some thoughts as an Atom outsider but someone who spent more than 
6 years on several W3C WGs, including 18 months as a co-chair:

- Does it seem a bit odd that W3C made this initiative by asking the 
IESG not to approve the Atom work because they wanted it, rather than 
approaching you folks to see if you wouldn't rather standardize Atom 
under the W3C umbrella?  Or were there some back-channel activities?

[Restating some stuff I posted to Danny Ayers' weblog comments ...]

- The traditional  W3C process is  far too heavyweight a process for 
Atom. It took about 3 years to move SOAP 1.1 to 1.2, and in that time 
SOAP 1.1 became so entrenched (and its real problems sorted out by 
WS-I) that 1.2 has little mindshare. Likewise, 5 years after the XML 
query workshop, XQuery 1.0 is not all that close to Recommendation 
status.   There are activities such as XLink that take years of time to 
become Recommendations and are more or less ignored.  I know that W3C 
realizes that things must change, and there was discussion of a more 
lightweight process (e.g. to attract some of the WS-* work that is now 
done in an ad-hoc process by a vendor coalition)  when I was involved.  
I don't think there is a track record for that yet; basically Atom 
would be the guinea pig.

 - Just getting a WG off the ground in W3C is a multi-month process 
even without opposition, and this proposal is sure to generate plenty 
of that.  The real value of the W3C is that it front-loads a lot of the 
pain -- it takes awhile to get a WG going, but once it is going there 
is a W3C professional staff to help (I can't praise the W3C employees 
I've worked with highly enough!), there is a "benevolent dictator" to 
make final decisions (although of course you may not agree with him), 
and there is a large community to help with the truly nasty details 
such as internationalization.  Likewise, patent policy is a non-issue 
-- you don't have any freedom, but that takes an infinitely complex and 
controversial topic off the technical agenda.  The trouble is, you lose 
a lot of people and enthusiasm while going through that front-loaded 
pain.

-  Finding anyone insane enough to chair this hornets nest who has the 
skills necessary to do a good job and the time/energy needed to make 
this a success will be difficult at best.  The only way you'll make 
progress, IMHO, is to have both an agreed upon decision making process 
and a chair that will enforce it, even when prominent people yell,  go 
to the media, stalk out, get their boss to call your boss and twist 
arms, try to get the TAG or Tim Berners-Lee to overturn a majority 
opinion, etc. (That stuff happens, don't kid yourself!).

- The successes of W3C during the browser wars were possible because 
the *individuals* got along just fine (at least in my experience) even 
if their employers were waging war by press release. Likewise, there 
was a strong sense of urgency that there was only a narrow window of 
opportunity, and everyone's vision of a world with conflicting 
"standards" from MS, Netscape, Sun, etc. was far more horrific than the 
thought of making compromises.  I think even the most casual observer 
of the RSS/Atom world would be inclined to doubt that these conditions 
are duplicated <duck>

- It is likely that the WG would contain people who would prefer no 
Atom/RSS Recommendation to one that didn't accept their pet feature or 
worldview. The W3C consensus guidelines make it relatively easy for 
such people to "lay down in the road" in front of a majority. 
Historically, the only way forward that W3C WGs come up with in this 
situation is to bloat the spec: I want option A, you want option B, so 
let's compromise and do both. You'll have to figure out how to prevent 
that scenario up front.  I don't have experience with the IETF process, 
but legend has it that running code tends to run over people who lay 
down in the road :-)  W3C (at least during my DOM days during the 
browser wars) explicitly said that implementations of draft specs had 
no special status.  (Needless to say, that rule didn't really and truly 
apply to certain dominant vendor(s), but  let's not go there ...)

- Likewise, I'll guess that there are people in the W3C who would 
prefer no Atom Recommendation to one that does not pay due deference to 
some existing W3C Recommendation or other -- the Semantic Web stuff 
perhaps, SOAP perhaps, REST, who knows what all?  You may be faced with 
unpleasant choices between bloat, delay, and writing off the time 
invested in W3C at some point down the road.  Ultimately, if Tim 
Berners-Lee doesn't like what you come up with, it's his prerogative to 
keep it from being a Recommendation (although that has never yet 
happened AFAIK).

- Finally, as others have pointed out, the W3C mainly  is a consortium 
of big vendors, and the processes are geared to that reality.  Any 
member company can send representatives to a WG; conversely  people who 
are employees of a member company and want to participate in a 
hypothetical Atom WG would have to be appointed by their Advisory 
Committee representative.  That doesn't reflect the reality of how 
RSS/Atom got to where it is now.  Eric Miller indicated that all these 
things can be fixed up in a WG charter, but again that just front-loads 
more pain to the charter-writing process.