FEATURES & COMMENTARY

San Diego, CA — As Gary H. Anthes reported for IDG, internet pioneer Vinton Cerf and his wife, while dining on roast veal one evening, were discussing cattle drives of the mid-19th century. They wondered what had happened to legendary routes like the Chisholm Trail.

Cerf’s house is equipped with a radio LAN, so he was able to connect to the Internet via a wireless laptop computer at the dinner table. “Within a few queries, we got a wonderful dissertation with pictures and maps and histories of the trails,” he said. “I thought, ‘My God, the world’s knowledge is just sitting there.’ What’s our culture going to be like when you can find out literally anything in a few seconds, when the brilliance of every human being is suddenly available to you?”

Cerf’s vision isn’t so far-fetched. And if that vision is to be realized, there’s little doubt that the World Wide Web Consortium (W3C) will be a major driving force. Technical specifications developed by the W3C, most notably Extensible Markup Language (XML), are morphing the Web into a second-generation architecture. One likely to eclipse even the phenomenal success of the Web in the 1990s.

But despite nearly universal praise for its work, the W3C draws some criticism for its methods and concentration of power at the top, namely Tim Berners-Lee, creator of the Web and founder and director of the W3C.

Nevertheless, even those who don’t fancy the W3C’s operating philosophy acknowledge that its agility is rare in a standards group and that its specifications bear a “moral majesty.”

Semantic Web

The W3C is mapping out technology to support a “semantic Web,” in which all the world’s knowledge becomes computer-accessible. “Querying a database is not exciting,” Berners-Lee says. “But querying a database that gets linked so as to query the whole planet is very exciting.”

The W3C, based at MIT and research centers in France and Japan, last year took a giant step toward that goal by publishing XML. It can describe Web pages with far more power than Hypertext Markup Language (HTML), the Internet programming language developed by Berners-Lee in 1990 at CERN, the European Laboratory for Particle Physics in Switzerland. Unlike HTML, which describes the structure of a page, XML allows developers to make up their own tags, or metadata, to describe information content on the page.

The W3C is also working on the Resource Description Framework (RDF), a language that uses XML to enable application and content developers in different domains to share vocabularies, their own metadata, in ways that allow them to link diverse databases.

XML and RDF promise to make the Web much more powerful by enabling search engines to “understand” the meaning of information. No more getting a million hits on topics you don’t care about, while missing information you want because it doesn’t happen to include the keywords you specified.

Closing the Gap

The Web has propelled the Internet to 60 million nodes in the past five years, according to Scott Bradner, a senior technical consultant at Harvard University and an area director at the Internet Engineering Task Force (IETF). “The Web filled a hole we didn’t know we had; we geeks were doing just fine,” he said. “The W3C deserves an awful lot of the credit for that.”

The W3C also fills a hole left by the older IETF standards body. Bradner says the groups jointly decided several years ago that the IETF would handle low-level topics such as the Web protocol HTTP, while the W3C takes on issues closer to the application. “They are upper-middleware, and we are lower-middleware and below or, as someone put it, underware,” Bradner said.

But the groups differ in other ways as well. In fact, the W3C doesn’t consider itself a standards body at all, preferring to think of itself as a research and development organization, a kind of techno-think tank. It develops open-source software for demonstration purposes or when it feels the marketplace isn’t meeting a critical need.

For example, the W3C developed the first Web browser/editor combination called Amaya, and a flexible and extensible Java-based Web server called Jigsaw. Anyone, even nonmembers, wishing to use, improve or build a product around Amaya or Jigsaw can download the source code at the group’s Web site.

Out of the Loop

The IETF, which doesn’t develop software, is a loosely structured, grassroots-like group from which standards bubble up after being shaped and critiqued by anyone who cares to participate. The W3C is a more structured, less open coalition of 339 software vendors, large user companies and others who pay $5,000 or $50,000 (depending on size) in annual dues and sign an agreement vesting final decision-making authority in Berners-Lee.

But some people aren’t entirely happy with that arrangement. MCI WorldCom Inc. withdrew from the W3C after two years when it concluded that membership wasn’t worth $50,000 per year. “The structure of the W3C didn’t lend itself to quite the degree of freedom to contribute that the IETF does,” says Cerf, MCI’s senior vice president for Internet architecture. “We found it difficult to get points across and to influence what was happening.”

“We were never completely comfortable with W3C acting as a standards body, with its decision model based ultimately on the personal preferences of the director,” says John C. Klensin, distinguished engineering fellow at MCI. “We’ve tended to prefer Internet standards work to be done in bodies that more clearly use an open consensus process rather than in limited-membership consortia of any sort, including W3C.”

But Klensin praises the W3C’s ability to move quickly. “The W3C approach is probably optimal for the design and development of sample advanced technology, especially when it addresses problems two or three years ahead of current products, while the IETF approach is far better for actual standardization,” he says.

Power of Standards

Indeed, Berners-Lee says one of his goals in setting up the W3C in 1994 was to make it more nimble than the IETF. “Always, standards processes have been too slow,” he said. “The IETF has a particular set of processes, and in some circumstances they work very well and under other circumstances they don’t.”

As for his power to affect millions of users, he’s unapologetic. “Members give a mandate to the consortium to do things, and the director has an executive responsibility to get them done,” he said. “But there are a whole set of checks and balances, and there is even a process to review the process.”

The Boeing Co., a W3C member, has an enormously complex and geographically dispersed computing environment, but it’s unified by an intranet with 175,000 users, an extranet with 26,000 users and more than 2,000 Web servers. “The Web is of immense importance to us,” said Ann Bassetti, Web products manager at Boeing.

Boeing joined the W3C to get early information on Web developments and to influence them as a user, Bassetti said. “For us, interoperability is crucial. Without the W3C, we would not have standardization of protocols, and individual vendors would dominate with their proprietary formats.”

Whether one calls them standards or, as the W3C prefers, recommendations, HTML, XML and other technical specifications from the consortium have virtually the force of law. “They have a moral majesty behind them, a moral hegemony,” said Carl Cargill, director of standards at Sun Microsystems Inc. and a member of the W3C Advisory Board. “And they have a great deal of acceptance among users.”

Cargill said the W3C’s members, including Sun, are, in effect, a giant “advisory committee” to Berners-Lee, whom he calls an “honest and reasonably open individual” and an “impartial adjudicator.”

But Cargill acknowledges that the director has tremendous power in an arena increasingly vital to the IT community. “Tim’s in a position where, if he shakes his head yes, somebody can lose $1 billion, but if he shakes his head no, somebody else loses $1 billion,” Cargill says.

“If you have a strong opinion on X, and Tim doesn’t share it, well, it may be a little tricky to get a standard out,” Bradner said. “(But) Tim is the key to ensuring a consistent architecture, to keeping things from fragmenting.”

Asked if anything worries him about the future of the Web, Berners-Lee said, “Fragmentation. If TVs end up with one version of HTML, and normal Web browsers end up with a different version, that would be a mess. That could come about commercially if a monopoly player decides it’s going to try to tweak the standards so everyone has to follow a little behind. That’s a constant threat, because there’s a huge commercial incentive to try to carve out a piece.”

The answer, he said, involves us doing our job and buyers doing their job as informed by the press.

World Wide Works-in-Progress

The W3C has 57 full-time employees and more than 600 people from member organizations around the world assigned to 50 groups working in four domains: architecture, user interface, technology and society and the Web accessibility initiative. Projects the organization is working on include the following:

— HTTP-Next Generation, to offer greater flexibility and performance for distributed applications.

— Ubiquitous, device-independent access to the Web via television, mobile phones, pagers and the like.

— Metadata and a semantic Web that uses better ways to describe and catalog information to enable smarter search and retrieval than what is currently available.

— Platform for Privacy Preferences Project, which helps Web users learn and influence the privacy policies of Web sites. Web content accessibility guidelines to help Web developers make content accessible to people with disabilities.

— XML (a joint project with the IETF), a way to cryptographically sign XML documents.

============================================================