Patents, Royalties, and the Future of the Web
October 10, 2001
Social theorists and historians of science and technology tirelessly remind us that technology never exists in a vacuum, that it's always already fully a social production. As historian David Noble says, technology isn't a "disembodied historical force impinging upon" human affairs.
That's never seemed more obvious than it now seems for members of the XML development community, who have been talking lately, not of technical features or challenges, but of legalities and public policy, copyright, intellectual property, and especially patents.
The imbroglio arises from the last-minute "discovery", due in large part to Adam Warner's 28 September document (W3C and the Promotion of Fee-based Standards for the Web), of the W3C's proposed patent policy, of 16 August, followed by the W3C's extension of the public comment on the proposal period until 11 October. The extension came in response to equally last-minute public opposition, sparked by the story showing up at Slashdot, LinuxToday, Linux Weekly News, and other popular tech-centric Web publications and mailing lists.
In what follows I examine the ways the XML and general Web development communities are reacting to the W3C's proposal.
Patents and Royalties
Before looking at community reaction, it's helpful to review briefly some of the relevant legal concepts. First, at least under US law, a patent confers an "exclusive right to make, use, or sell an invention for a specified period...granted by the federal government to the inventor" (Black's Law Dictionary). The US Congress derives its authority to grant patents from the Constitution's patent and copyright clause (Article 1, Section 8, Clause 8): "To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries".
A patent, then, creates a legal, limited monopoly, which is meant to promote scientific and technical progress for the common good. Notice that the point of a patent regime is not to reward the patentee per se, who may or may not benefit financially from the patent, but rather to encourage the disclosure of generally useful, beneficial discoveries and inventions. One way to encourage such disclosure is, in fact, to grant a limited monopoly, from which presumably some financial benefit may result. But neither the patentee's presumed financial benefit nor the grant of limited monopoly are the public policy end of a patent system; they are merely a means to the end of promoting the public good, means that are solely within the powers of the US Congress to determine and impose.
Second, a royalty is a payment made to a patentee for the right to use the patented invention or discovery. A reasonable royalty is one that a person would be able to pay to the patentee and still derive a reasonable financial profit from use of the patented invention.
The W3C's Proposed Patent Policy
The 16 August draft, which had received very little attention from the XML development community before 28 September, despite Elliotte Rusty Harold's note of 17 August, establishes a new licensing model, RAND ("reasonable and non-discriminatory"), which would allow the W3C to create standards the implementation of which could require royalty payments to patent holders.
Also in XML-Deviant
The part of the new policy framework that's been objected to most often and most strenuously is the provision allowing RAND licensing. The W3C -- in "Backgrounder for W3C Patent Policy Framework" -- says that "RAND means that someone may or may not need to pay a fee [in order to implement a RAND-licensed W3C recommendation], and that it is at the discretion of the license holder". Of course the problems with RAND licensing of the core infrastructure of the Web itself are obvious. The W3C tries to ameliorate these concerns by drawing a distinction between fundamental and high-end layers of the Web "stack."
Preservation of global interoperability of the core Web infrastructure is of critical importance. So it is especially important that the Recommendations covering lower-layer infrastructure be implementable on a royalty free (RF) basis. Recommendations addressing higher-level services may be appropriate for licensing on reasonable and non-discriminatory (RAND) terms (W3C's Backgrounder).
But that distinction, between the various essential and non-essential layers of the stack, is invidious, as Simon St. Laurent pointed out.
While some have suggested that this policy only applies to higher-level standards and isn't a threat to the core, I'd like to suggest that HTTP looks like an awfully high-level standard from the perspective of IP (Internet Protocol...) Today's core is composed of yesterday's high-level experiments, and I see no reason that arbitrarily-drawn lines will last as development continues. I also see no integration between this proposal and the Technical Architecture Group (TAG) - it simply isn't clear who decides what is core and what is not.
James Clark also disputed the W3C's distinction.
I think this distinction is irrelevant. If a Recommendation is part of the core infrastructure, then it needs to be RF whether it's low-level or high-level. If a Recommendation is not part of the core infrastructure, then I would question whether the W3C should be devoting part of its limited resources to it. In summary, if it's important for the Web infrastructure enough to be a W3C Recommendation, then it needs to be RF.
The Threat to Open Source and the Web
One theme that's emerged from the XML community is that RAND licensing is a threat to open source, to the non-commercial implementation of non-proprietary standards and, hence, to the Web itself.
Recognition of the threat to small, independent developers was among the earliest community reactions to RAND licensing:
It doesn't seem likely to smell good to developers distributing their code for free, certainly (Simon St. Laurent).
I predict that if following such recommendations _might_ expose people to future legal action, then people will simply go elsewhere, just as people went elsewhere when a mere perception of risk became attached to gopher (Jonathan Borden).
Non-RF standards would have a devastating impact on open-source software, which has historically been very important in the development of the Web (James Clark).
Adam Warner put the nature of the RAND threat in this way:
The prospect of future fees could also have a chilling effect upon free/open source software development. Standards that require licensing fees to implement are, for obvious reasons, totally incompatible with the use of free software. If the free/open source software communities will not be able to rely upon the W3C to pursue royalty-free standards the question has to be raised whether the support of a new institution is appropriate.
In essence [RAND] requires that any company that imposes licensing restriction must impose those restrictions uniformly (the non-discriminatory part of the definition). It appears to follow that non-commercial organizations cannot be given any preferential treatment over commercial organizations since that would be discriminatory licensing.
Eben Moglen, speaking on behalf of the FSF, said,
The World Wide Web cannot exist as a global and uniformly-available facility of human society without free software. Apache, Perl PHP -- and literally hundreds of other immediately recognizable aspects of web technology -- have been outgrowths of the free software production model. Without free software, the web would be a commercialized outgrowth of a few proprietary software producers, and it would be incapable of serving, as it now does, as a force for global egalitarianism.
Because the Web employs no technology not based around completely open standards, software implementing every single facility of Web life can be produced in the free software model, and is therefore available for free modification and improvement all over the world, supplied at the marginal cost of distribution to any programmer -- no matter how financially constrained -- who wishes to produce new facilities and opportunities for users.
"Reasonable and non-discriminatory" licensing of patented technology embodied in W3C standards will eliminate free software production from any area of Web facilities subject to those standards. Such standards will therefore provide a basis to "embrace and extend" the Web under proprietary control, excluding competition from free software, limiting technical innovation and risking the social utility of the Web.
Bruce Perens made clear the illusory nature of the "non-discriminatory" part of RAND licensing.
By requiring "reasonable, non-discriminatory" patent licensing in W3C standards, the W3C would actually discriminate against Open Source, simply because the royalty-free nature of Open Source software is incompatible with patent royalties. This would have a chilling effect on implementations of W3C standards, restricting them only to large companies and over-expensive software.
Richard Stallman also pointed out the internal inconsistency of RAND licensing.
Aside from these substantive changes in policy, the W3C should also stop using the term "reasonable and non-discriminatory", because that term white-washes a class of licenses that are normally neither reasonable nor non-discriminatory. It is true that they do not discriminate against any specific person, but they do discriminate against the free software community, and that makes them unreasonable.
In its claim to be "non-discriminatory," RAND necessarily hurts non-commercial, cooperative, and small independent software developers. And that, in turn, will have a negative effect on the Web itself. Recall that the end of a patent regime is, at least on the US model, technical progress for the public good. While some may not agree with Eben Moglen's larger moral and political goals or claims -- that the Web is a "force for global egalitarianism" or that one must avoid "risking the social utility of the Web" -- given the nature of the Web's historical development, it's clear that RAND licensing would subvert the sole legitimate end of a patent regime, insofar as it discouraged prevented non-commercial participation in the Web's future and ongoing development. You don't have to be a free software radical to understand that whatever hurts free and open development hurts the Web itself.
Tim Bray summarized things succinctly by pointing out that "the *unacceptable* outcome is that there is eventually a piece of infrastructure with the W3C imprimatur that you have to pay toll to...use".
The W3C, It Is A'Changin'
The discussion quickly moved from a rough and ready consensus about the threat of RAND licensing to talk about the W3C itself. Most people agreed that adoption of the proposed patent policy signals an important change in the W3C.
"The W3C is not about little guys", Len Bullard said, "I suspect that is why OASIS is becoming more popular for experimental efforts while the W3C is becoming WebISO".
Respected and important independent Web developer, James Clark claimed that adoption of RAND licensing would negatively affect public perceptions of the W3C.
I also believe that such a policy would further weaken respect for the W3C as an institution and would tend to promote the perception that the W3C is a tool of its corporate paymasters rather than an organization that is trying to do the right thing, whether morally or technically.
Simon St. Laurent took RAND licensing, and the patent policy framework generally, as a change in the W3C's institutional self-understanding.
Unfortunately, the proposed patent policy seems to a serious departure from that mission and the public interest. The introduction of "reasonable and non-discriminatory" licensing of patents -- with real dollar costs imposed by part 5 of the RAND license -- seems like a maneuver by some parts of the membership to make a few dollars off the public and to shut out open source development, at least the kind which lacks corporate sponsorship.
The continuing use of the word "standards" in this document also suggests that the W3C has forgotten its origins as a research group issuing "recommendations". "Standards" deserve much tighter examination and consideration than mere recommendations.
Len Bullard sounded a a similar note:
The presence of the word "standards" all over this document reveals both intent and the outcome. By claiming to be a standards organization, the W3C has overstepped its charter and made its capture a primary business goal. Effectively at this point, it quits being a means to innovate and is now a means to define market.
Perhaps sensing the wider implications for other areas of software development, H. Peter Anvin, noted Linux kernel hacker, also opposed RAND licensing, focusing on the role of the W3C as an arbiter of interoperability.
The proposal of loosening the patent policy of the W3C is an assault on the W3C itself. So far, the W3C has probably become the main reason why the web has not yet devolved into a smattering of incompatible corporate standards...
If RAND is adopted, and subsequently fee-based standards are adopted, W3C will simply be a clearing house for these proprietary standards at no end-user benefit. There will no longer be a drive to develop true standards than anyone can implement, including those that have no ability to pay royalty payments. In the end, we will all be much poorer as a result.
Also in XML-Deviant
There has long been a wide consensus among software developers that the US patent regime is fundamentally broken when it comes to software technology: the wrong things are taken to be patentable, prior art is routinely ignored, patent pooling by large corporations tends to restrict innovation and development by small and independent developers, and so on. Given these, at present, real-world constraints, and given the mission of the W3C, since RAND licensing is objectionable, what alternative patent policies for W3C recommendations would the XML development community urge?
Many XML developers agreed that if, and only if, there are areas of software or web technology that must be patented, the W3C should not be involved in those areas, and it certainly should not offer the auspices of W3C recommendation to patent-encumbered technology.
Don Park said that the "W3C should limit its activities to areas where it can operate with minimal patent hassle". And that idea was widely expressed by other XML developers. Tim Bray expressed this point simply and forcefully.
I think that those standards which have to live in the space where you might have to pay a toll to use them are precisely those which W3C should stay away from. Yes they exist - e.g. the IEEE standardizes lots of things which you have to pay patents to use - but the reason the Web is interesting is that anyone can play without having to pay for permission. I'm not interested in playing RAND games. I'm not interested in a Web where Open-Source efforts are second-class citizens.
Alan Cox, another widely respected (formerly independent) Linux kernel hacker, suggested the W3C simply avoid development areas where patent-encumbrance and RAND licensing were unavoidable. Ideally the W3C would, Cox said, "not 'approve' or 'recommend' or allow its logo to be used on any patent-encumbered item".
As in all such public policy debates, it's important that proposed alternatives actually be implementable. Software developers know this all too well, having sometimes seen their peers chase unimplementable projects and ideas for years on end. The W3C can implement the suggested policy of only developing royalty-free recommendations, Tim Bray suggested, by
- requiring diligent search and disclosure from all members, not just those who participate in particular WGs, for IP that may stand in the way of some task or another
- where such IP exists and the holder isn't willing to grant RF, changing the standard to work around the IP
- use of the bully pulpit and any other leverage the W3C can bring to bear to make it very painful for anyone who tries to set up a tollbooth on W3C output
- declining to enter standardization activities where it appears that RF status can't be achieved
Many members of the XML development community made similar suggestions. It seems clear that the W3C would have the full support of the independent development community if it abandoned a RAND licensing framework and adopted something similar to Bray's proposals.
What are Patents for Again?
It was a welcome and encouraging sign to read so many XML developers, as well as many others, putting the matter of patents, intellectual property, and the W3C in its proper context, namely, technical progress for the public good. Several XML developers in particular made the connection between the Web, XML, and why developers do what they do very explicit.
Steve Newcomb said that the overarching goal of standards efforts should be the public good.
It would be far better to make the public interest the real focus of our standardization activities. Decisions about standards processes, and about standards themselves, should be explained to the public in terms of public benefit, showing why all the possible alternatives would be less beneficial *to the public*.
Robin Cover connected the end of technical progress for the public good to the fundamental brokenness of the current patent regime, as applied to software.
Patent law [with regard to] software is hopelessly otiose in our current century under current law. Period. It no longer serves whatever public Good it might theoretically serve in a previous century under different assumptions. The number of disaffected persons and whole companies (should SW patent law be obliterated by fiat) is minuscule, and in sum expendable. The greater Good by far lies is moving the pre-competitive boundary further forward. One is not required to be a (sic!) 'socialist' to maintain this posture; one is only required to look at the world...to understand that the foundations of technology indeed will NOT crumble if we assign to the dustbin these outdated concept and laws...
Let innovation and competition begin anew in a universe where patent lawyers are as esteemed as Mafia.
Simon St. Laurent concurred with these basic sentiments, suggesting that a recommitment to the public good would put the W3C well on its way to self-reform and renewal.
Overall, I'd suggest rejecting RAND entirely as a first step toward reforming the W3C and converting it from a vendor consortium to an organization which focuses primarily, even exclusively, on the public interest. Rather than relive the 'tragedy of the commons' again, we should focus on building a stronger commons.
Since the extended comment period lasts until 11 October, it's as yet unclear how this mess will be resolved. It's also unclear whether W3C member corporations are willing to compromise on RAND licensing. Some of the corporations involved have staggeringly large revenue streams arising from patent portfolios. IBM, the most obvious example, has some 36,000 patents, from which it generates about $1,000,000,000 in royalty revenue per year. IBM is an interesting case since it's made a very visible investment, perhaps the largest so far, in open source software. So the issue cuts across not only industry lines and alliances, but also across various parts and divisions inside large corporations.
XML developers and others interested in this important issue and in the Web's future should let the W3C hear their voice regarding RAND licensing by 11 October.