Perspective: The patent threat to the Web

The penultimate step in a yearlong battle over patents on Internet standards came last week, when the World Wide Web Consortium Patent Policy Board voted to recommend a royalty-free policy.

The World Wide Web Consortium (W3C) draft recommendation has not yet been published. After public comments are solicited, the draft will be presented to the organization's director, Tim Berners-Lee, and an advisory board representing all of the consortium members. Large patent-holding companies may present a dissenting minority report, but Berners-Lee and the board are likely to approve the royalty-free proposal.

As one of the three representatives of open source and free software on the patent policy board, I welcome the new recommendation. Developers of the Apache Web server, the Linux kernel and GNU system, and other popular free software will continue to be able to implement W3C standards in competition with proprietary software.

Had the decision gone for so-called "RAND" patents--licensed with "reasonable and non-discriminatory terms," but sometimes requiring royalty payments--the effect would have been to create a tollbooth on the Internet, owned by the largest corporations, collecting a fee for the right to implement open standards.

Open-source developers, who do not collect royalties--and thus cannot afford to pay them--would have been locked out entirely. Smaller companies that develop proprietary software would have been at a disadvantage, compared with the largest corporations, which cross-license their patent portfolios to each other and thus would not be burdened by royalty payments.

A royalty-free policy doesn't assure that standards won't be covered by patents, because a patent may become known after a standard has been adopted--especially a patent applied for in the United States, whose government is almost unique in allowing patent applications to remain secret for an extended period. Under the new policy, the W3C would probably withdraw a standard so affected, and re-engineer it to avoid the patent.

Open-source developers, who do not collect royalties--and thus cannot afford to pay them--would have been locked out entirely.

A royalty-free policy also may scare the largest companies away from participation in standards working groups, for fear that they will lose lucrative licensing rights. Working groups that create new W3C standards will require their members to file a declaration granting royalty-free rights to anyone for "essential claims"--patent claims that would unavoidably be infringed while implementing a standard. This will prevent "patent farming," the practice among patent holders of deliberately embedding their patents in new standards as a revenue-generating strategy.

However, the grant of royalty-free rights does not apply to any use of the patent other than to implement the standard. Thus, patent holders will still have lots of opportunities to sell licenses to developers who wish to practice their patents for any other purpose.

The worst implication of the royalty-free patent policy is that standards working groups will choose an algorithm that everyone can implement over a technically superior algorithm with restricted rights. But that compromise is in keeping with the purpose of enacting a standard: a procedure that everyone is to follow the same way, in order that all parties may interoperate with each other. Besides, it's not the fault of the standards body that a system of software patents that forces technical compromises has been imposed upon them. They didn't make the bad laws, and they have to live with them.

The purpose of patents is to advance technology by providing a reward for the creation and disclosure of new processes, and to provide for that information to enter the public domain once the patent expires. There is little proof that software patents actually promote technology, and much suspicion that they actually hinder it by preventing later developers from being able to leverage upon earlier work by other researchers.

Since software algorithms are mathematical in nature, I question whether they are inventions at all, or simply discoveries of a pre-existing mathematical foundation of the universe and thus not appropriate for coverage by patents.

The W3C's decision will resolve only a single battle in a much broader war.

There are many other problems with software patent law. The 20-year term of patents is so long, compared to the duration of a "generation" of computer technology, that patented algorithms do not enter the public domain while they are still useful to the public. And patent applications should be published immediately, rather than kept secret for years, as the United States allows.

Secret applications lead to "submarine" patents, which surface years after developers commit to commercial use of an algorithm in the belief that it is unrestricted, and put those developers in the unenviable position of having to license the patent, write their way around it, or quit.

Contrast the role of software patents to that of copyright. A copyright holder can legally prevent the unlicensed duplication of his work. That copyright holder's competitor still has the right to create and distribute his own program. But a software patent can block competition entirely by making it illegal for that competitor to produce and sell his own implementation.

The W3C's decision will resolve only a single battle in a much broader war. A similar royalty-free policy must now be enacted by the Internet Engineering Task Force (IETF) and many other organizations. Some standards bodies may decide to buck the trend and act as playgrounds for large patent holders.

Those organizations will argue that by allowing patent royalties, they will always be able to choose the best algorithm for any job. It will then fall to the market to decide which organizations it will follow. This battle must also be taken to the various governments and treaty organizations that produce bad law permitting the patenting of software and business systems, and continue to do so.