CNET también está disponible en español.

Ir a español

Don't show this again

Christmas Gift Guide
Software

W3C risks patent tussle in standard push

A pending industry standard for voice-activated computer commands is moving ahead, despite potential entanglements from Rutgers University over its VoiceXML patent.

A pending industry standard for voice-activated computer commands is moving ahead, despite potential patent entanglements, dealing a black eye to investigators handling the case's intellectual-property claims.

Software developers said they are concerned that a recommendation for VoiceXML 2.0, which the World Wide Web Consortium advanced (W3C) this month, could fall afoul of a Rutgers University patent that was disclosed to the group more than two years ago. The W3C is responsible for approving most Web software standards, such as how browsers render Web pages, and recently adopted a controversial policy that aims to exclude most patented technology from its specifications.

News.context

What's new:
The World Wide Web Consortium is advancing an industry standard for voice-activated computer commands. Its technology involves a Rutgers University patent.

Bottom line:
Software developers are worried that a recommendation for the VoiceXML 2.0 standard may lead to patent battles like those of the SCO Group, involving expensive and unexpected licensing fees.

More stories on this topic

In a worst-case scenario, a patent impasse with Rutgers over VoiceXML could lead to expensive and unexpected licensing fees for emerging technology that promises to provide a powerful alternative to traditional Web surfing. VoiceXML enables data stored on Web servers to be retrieved, using voice commands and touch tones. It also enables text-to-speech translation, giving Web surfers access to information such as stock quotes or e-mail over the phone.

One industry observer likened the potential fallout to the SCO Group's use of its Unix copyrights to seek licensing fees from Linux vendors.

"When it comes to VoiceXML, developers are not entering this risk-free," said Dan Miller, an analyst at the Zelos Group in San Francisco. "There's always the chance that the originators are going to pull an SCO on them."

The VoiceXML concerns come amid a broad debate over the role of patents in standards, sparked in part by a succession of royalty claims against widely used technologies that emerged only after they became popular. Some commonly cited examples include belated licensing enforcement for the patented GIF and MP3 file formats.

Some angry software developers now argue that standards bodies should reject patented technologies from specifications whenever possible and carefully investigate all submissions to ensure that there are no licensing surprises down the road. Others, however, accept that it may be practically impossible to avoid all patent claims, particularly if standards groups hope to endorse the best available technologies.

No royalties, please
Capping more than a year of heated discussions on the issue, the W3C

In order to enforce the policy, the W3C from time to time assigns patent advisory groups (PAGs) to investigate claims and issue recommendations regarding specific technology.

The W3C so far has appointed two PAGs to resolve disputes over patents relevant to VoiceXML 2.0, winning some significant concessions from at least two patent holders.

The group received assurances from Avaya Communications that its patents were nonessential to the specification. That meant that while Avaya reserved the right to collect licensing fees for the technology, someone could still implement a fully interoperable VoiceXML application while steering clear of the Avaya technology.

Separately, Dutch electronics giant Royal Philips Electronics transferred its relevant VoiceXML patents to ScanSoft, which offered them to the W3C for royalty-free use.

But one potential problem, disclosed to the W3C's VoiceXML working group in October 2001, hasn't gone away: Rutgers' U.S. Patent No. 6,240,448, "Method and system for audio access to information in a wide-area computer network."

The patent raised a red flag, because Rutgers offered it under "reasonable and nondiscriminatory licensing" (RAND) terms rather than royalty-free, putting the university at odds with the W3C's policy for technology deemed essential to the recommendation.

With the W3C failing to win a royalty-free license, its investigators next sought to determine whether the technology in question was in fact essential to the VoiceXML specification.

In May of last year, after VoiceXML had entered its candidate recommendation phase, the PAG queried Rutgers' licensing agent, Fairfield Resources International in Stamford, Conn., asking for its opinion on the matter.

Fairfield acknowledged the query but didn't reply within the two-week deadline the PAG set. The PAG then reported back to the W3C that Rutgers had not answered its query and therefore had made no claim that the patent was essential. In a report to the VoiceXML working group, the PAG then said its work was complete.

The PAG's investigation incensed some patent critics, who called it inadequate.

Wussing out?
"The W3C wussed out on this one," said Carl Cargill, director of standards for Sun Microsystems. "If the W3C claims to have a patent policy, they have the duty and the responsibility to push that policy. It cannot be a trivial effort."

One patent attorney suggested that determining the essentiality of Rutgers' patent would not be prohibitively difficult but that an independent assessment would likely add to the costs of the investigation.

"'Will practicing the method set forth in standard X infringe patent Y?' is a question that can be readily answered by lots of people," said Carl Oppedahl, an attorney at Oppedahl & Larson. "Both documents are a matter of public record and can be scrutinized to whatever extent is necessary, by anybody who chooses to scrutinize them or who chooses to hire competent patent counsel to scrutinize them. If any competent patent lawyer can answer this question regarding documents X and Y--and would expect to be paid for the work--then why should Rutgers do it for free?"

The W3C said that neither Rutgers, its attorneys nor its licensing agent had yet responded to the PAG's request. A W3C representative said the group does not plan to convene another PAG to examine VoiceXML claims at this time and that development of the proposed recommendation will go forward.

"Until a patent owner asserts claims, we don't see an issue," W3C representative Janet Daly said. "The patent holder is usually in the best position to know their own patent. And whether they assert correctly or not is another story."

Rutgers did not respond to interview requests for this story. But an intellectual-property consultant to the university subsequently wrote in a prepared statement that the patent would be offered under RAND terms.

"Rutgers may have relevant and essential claims," Millburn, N.J., attorney Seymour Hollander wrote in an e-mail. Hollander did not respond to e-mailed and phoned interview requests.

One person close to Rutgers scoffed at the W3C's patent policy, saying it amounted to a request for the university to give away its intellectual property and that that would be an abrogation of the public university's duties to protect assets developed with taxpayers' money.

"The people who promulgate standards are members of what could be called the free-love society," this person said.

The patent's co-inventor, Tomasz Imielinski, declined to weigh in directly on the question of essentiality. But he stressed that Rutgers' patent claims are very broad and appear to predate other voice-activated computer technology.

"Regarding the essentiality, I leave it to lawyers," Imielinski said. "But the patent is very broad and covers pretty much any wide-area system of pages, which can be rendered through TTS (text-to-speech) back to the caller. Current work on VoiceXML is based on the same principle which we worked on, almost 10 years ago now, in (the) Audioweb project at Rutgers. And this is one of our top patent claims."

The stakes of the VoiceXML patent fight are uncertain but potentially significant to those who implement the recommendation.

Tellme Networks, a Mountain View, Calif., company that has technology for automating telephone customer service for enterprises such as airlines and phone directories, boasted more than 1 billion VoiceXML requests in December 2003 alone.

Tellme declined to comment for this story.

Like University of California v. Microsoft/> While some compare the conflict to the Linux struggle between SCO and IBM, perhaps the better analogy to the impasse between the W3C and Rutgers is with the lawsuit against Microsoft by the University of California and its Eolas spin-off.

That suit, over a patent the university says covers crucial plug-in technology in Microsoft's Internet Explorer browser, resulted in a recently upheld $521 million jury verdict, which took more than Microsoft by surprise. Since the ruling, the W3C and others have been battling to have the patent invalidated.

Like the university and Eolas, Rutgers has not contributed directly to the standard in question. Imielinski chaired the VoiceXML working group as an invited expert when it launched in February 1999 but stepped down before the year was through. Rutgers itself was not a W3C member, according to the consortium.

Daly downplayed the risk that Rutgers might become VoiceXML's University of California, saying it would be unusual for the school to stay mum on the patent's essentiality while it already enjoyed significant implementation, and then try to assert claims.

But she also stressed that the W3C's patent policy wasn't as black-and-white as it has been portrayed.

"The W3C's patent policy, while making it clear that there's a preference for (royalty-free) Web standards, also makes it clear that we're encouraging those in a climate where patents--especially in certain vertical industries, like telephony--are very prevalent," Daly said.


Special report
Patent politics
Rivalries are set aside in
defense of Internet Explorer


"So, we have to come up with methods where a spec gives the most information to end users about possible encumbrances. It's a clear policy, but the world provides complex problems sometimes."

W3C supporters point out that the patent application was filed during Imielinski's brief tenure as chair of the VoiceXML working group, following demonstrations by Motorola and other companies with similar technology.

But Imielinski counters that a provisional filing came as early as 1995 and that the technology--a system for extracting content from Web pages and playing it back as audio files--dates back to his pre-Mosaic work on voice browsing.

"We were definitely first," he wrote in an e-mail interview.

Imielinski wrote that the content extraction system led to the founding of a start-up, Connotate Technologies, which offers a number of different data-mining and Web services applications and services.

But the original idea behind Imielinski's Audioweb project was closer to the VoiceXML applications of today, including a 1-800 number that would access the Web, "long before there was Tellme," Imielinski said.

Despite the uncertainty over the Rutgers patent, one analyst encouraged VoiceXML developers to go forward in developing applications--with their fingers crossed.

"As a pragmatist, if I see an opportunity to employ this standard, I have to behave as if it's available to me," Zelos Group's Miller said. "I don't think royalty-free was ever a possibility, and you just have to assume that 'reasonable' will indeed turn out to be reasonable."