Adolescent struggles
W3C walks a political tightrope. Too much secrecy in deliberations, and companies feel excluded. So the working groups, each of which is chartered to produce specifications or prototype software, hold closed meetings, primarily via e-mail, then publish progress reports to the W3C Web site every three months.
On the other hand, too many participants means the sheer number of voices makes delays inevitable. Working groups are expected to be small, typically fewer than 15 members, though groups dealing with hot issues have more. For example, the group chartered to develop specifications for Scalable Vector Graphics (SVG), one of the hottest areas of Web development, has 26 members. Likewise, Cranor says of the P3P working group, "Fifty or more were involved over the life of the project."
It's not just what goes on within the groups that makes consensus hard to achieve. Like other members, Cranor cites a lack of coordination among working groups as a festering problem. "Many other W3C standards were being developed at the same time" as the privacy standards, she says, "and we had to go back and redo work to keep up with other specs from the W3C."
When a working group feels it has hammered out a consensus, its three-month progress reports culminate in a "last call," a term borrowed from bar slang to denote the last chance for dissenters to modify the working draft before it becomes an official W3C Recommendation. These, too, can drag on and on.
"Last calls usually last three or four weeks," Cranor says. But in the case of the privacy policies, "We said up-front this would be six months, because we have to deal with a lot of policy communities not used to responding to W3C reviews."
Coordinating the functions of various working groups requires that each group be solidly coordinated internally. But in fact, pressures and politics occasionally turn deliberations into battles. E-mail lists for soliciting public comment on the progress of the groups often elicit valuable input, but they also bare frustrations and can deteriorate into flame wars.
Mention the word "namespaces" to anyone in the XML community, for example, and prepare to duck and cover. A namespace is a sort of digital phone book in which each name is unique and identifies a single number. In Internet terms these are known as unique resource identifiers. The ubiquitous Web URL, for instance, is a category of URI, since each URL can identify only one page on the World Wide Web. Any breakdown in the URL system would result in chaos.
There is widespread disagreement over how to invent names and define their meanings in the XML language. One way to do this - the method adopted in the W3C's January 1999 Namespace Recommendation - is for Web site developers to arbitrarily create names and give them unique definitions. The alternative - and the option that many contend would best serve the rapidly expanding Web - is to adopt a design in which the meaning of a name might depend on its context.
This may seem like a trivial issue, but it has spawned such heated debates that even Berners-Lee, universally regarded as polite, well-spoken and considerate, became fed up with the level of animosity. On June 22, he began his e-mail to the list with a parody of an XML label: "(flame condescension= "on" spellchecker="0" frustration="98%" )" He went on to write: "When a subcommunity within the Web denigrates, misuses and generally abuses through lack of understanding another part of the Web architecture, it unfortunately falls on staff at W3C to try to hold the Web together. This is no fun."
Berners-Lee continued: "Current notions of 'best practice' for thinking up globally unique names such as 'foo' may be felt to be best practice by a set of people who use and see a very small set of names, but a decade of experience with scalable identifier systems suggests that using arbitrary strings sucks dead puppy dogs' tails."
In addition to widespread surprise at Berners-Lee's having lost his cool, there was another problem with his view: As the W3C's director, he had approved the 1998 Recommendation. In fact, the first sentence of every W3C Recommendation includes the phrase, "has been endorsed by the Director."
Dan Connolly, a respected W3C staff member, noted this contradiction when he responded: "Goodness, Tim! . . . Count to 10 or something next time, please. The fact is, you and I are party to the agreement in the XML Namespaces Recommendation. We neglected to review it closely, and we decided to make it a Recommendation, despite a questionable level of consensus, and despite advice from the [working group's chairman] that we slow down."
Added one member, who asked not to be identified: "Keep in mind that Tim could have addressed this over 18 months ago, but never got around to keeping an eye on it." In fact, Berners-Lee's relationship to the W3C has itself become a point of contention.
"There's angst in some quarters about Tim's role, and his dictatorship of the W3C," says Ann Navarro, chief operations officer at the HTML Writers Guild, which represents some 120,000 Web authors from more than 150 nations. "If he does anything stupid or rash or contrary to the membership, the process will break down. I'm amazed there haven't been more significant problems."
Observes another board member: "Part of the success at the W3C has been someone able to make or push decisions. Tim is having a bad year, but he's had a great career."
Whether the problem is perceived as one man's personality or the institution's structure, people want improvements.
"The business world would like to see these standards done much more quickly," says Laura Walker, executive director at the Organization for the Advancement of Structured Information Standards, which represents corporations that want to apply XML immediately by developing special vocabularies specific to industries.
Even so, Walker says, "We understand how difficult it is to assimilate so many diverse points of view." She adds: "The W3C is genuinely interested in becoming more efficient." To be sure, despite all the squabbling, everyone seems to agree that the W3C has an extremely challenging job and probably does it better than anyone else could. And in general, people respect Berners-Lee personally.
After 10 years in the policy arena with the Electronic Frontier Foundation and the Center for Democracy and Technology, Weitzner says that what attracted him to the W3C was "an environment with enough trust from industry and various advocacy communities around the world to really sort out how to advance the Web in respect to social goals. These are not easy problems."
And besides, animosity is a relative concept. Steve Zilles, an executive at Adobe Systems who has been involved in working groups at both the W3C and the Internet Engineering Task Force, says: "Relative to other groups, the W3C is friction free."
What's more, the group's combination of closed meetings to encourage candid discussion, regular publishing of results and openness to public comment is often cited as a balance that always - eventually - transforms contention into consensus. "Is the W3C working? In this case, yes," says Jon Ferraiolo, another Adobe executive and a member of the SVG working group. He recalls: "We added animation to the SVG language. Within the group, we felt JavaScript was sufficient, but the public came back and said to add declarative animation [applications called by the Web page developer to pull in animation programs]".
"Tim deserves more credit than he gets," Antarcti.ca's Bray says in reference to Berners-Lee. "But the specs are getting bigger, which worries me."
Berners-Lee accepts his role in the increasingly complex evolution of the Web, and he shrugs off the concerns. "Life is such that when you solve a complex problem, you find another one," he muses. "There will always be complex problems."
