- From: John Gilmore <gnu@toad.com>
- Date: Sun, 30 Sep 2001 20:10:25 -0700
- To: www-patentpolicy-comment@w3.org
- To: djweitzner@w3.org, timbl@w3.org
- Cc: sob@harvard.edu, gnu@toad.com
To summarize my comments: The comment deadline should be extended. The Working Draft policy would undermine the creation of widely available open standards, by providing patent-owning companies the opportunity to derail the W3C process into channels that benefit their private interest at the expense of the public interest. Non-RF standards would largely preclude open source implementations, such as the original Web server software written at CERN. Patents on software are a dubious concept legally; many are invalid; and the concept does not exist in most worldwide jurisdictions. These issues and dangers are barely mentioned in the draft policy. The draft policy should be revised to require RF licensing of all W3C-recommended technologies. I see from the sudden influx of comments at http://lists.w3.org/Archives/Public/www-patentpolicy-comment/2001Sep/ that the free software world has woken up to the challenge posed by this committee of proprietary software companies writing proposed changes to the W3C standards process. However, most of the comments are probably hurriedly made, since they were being written on the last possible day, with almost no advance notice (that was actually *noticed* by the affected community). I hope that the W3C will at least extend the deadline for comments, so it can hear more closely reasoned comments from all sides on the issue. RAND terms basically outlaw open source implementations, because they permit per-user or per-copy fees to be required. Such licensing terms are incompatible with open source implementations of the W3C recommendations, because open source licenses permit each end-user to make unlimited numbers of copies of the implementation, without further permission or payment. It might be legally possible for an open-source implementation to buy a one-time "buyout" license to the required patents, for all maintainers and users of their software, if the patent holder was willing to offer such a license. But the proposed W3C RAND terms would require that the exact same one-time buy-out price and terms be offered to every other implementer without discrimination. Since per-copy royalties from licensees of varying success would vary widely, a "one price fits all" buy-out license would be unlikely to be offered. Either it would eliminate the potential gain from the very high end of the market (receiving a per-copy price from Microsoft, for example), or if it did not, would be too expensive for the rest of the market to ever pay. I understand that RAND terms are common among the national and international standards bodies. I've served on standards committees. I also understand why these bodies are full of fights among organizations to get THEIR technology adopted as a standard. It's so THEY can get the windfall profit from forcing everyone else to license it. The companies have rigged the government standards game so that big companies with big patent portfolios get extra advantages from standardization. (Even if a big company has to pay someone else for one standard, it will be the winner on some other standard. The little guys and the non-patent-owning public always pay and never receive payment. Thus big companies generally favor forcing the market to use patents, even though the public interest may not be thereby served.) It would be much preferable if, like the AES and DES encryption standard competitions, and IETF standards today, any participant in the standards process with a patent is required to license it for free and unlimited use if it is accepted as part of a W3C standard. This gives such donors the first-mover advantage of having set the standard, understanding it better than others, already having compatible products in the market, etc. But it does not make everyone else in the world pay them for the privilege of using the standard. This is the RF model. I believe that "RF" is the W3C policy today, and I believe it should remain so. I'd be interested in hearing good reasons to change it. I don't see any in the Working Draft. In fact, the Working Draft points out "Consensus Points" in section 2.2, but is eerily silent about the points of non-consensus, and about the motivations behind those points of non-consensus. Clearly the draft policy goes beyond the Consensus Points, by advocating that patent holders be granted the option to impose per-copy fees on users of a W3C standard (if they can manipulate a standards committee to take such a vote). But somehow the discussion of what motivated the PPWG to do so is not part of the public record. Perhaps there is a hint of commercial interest in this motivation, perhaps not. I can conceive of principled positions such as, "W3C should be able to standardize on good technologies that no company will give up its patent rights in." I suspect from the lack of representation in the PPWG by end-users of standardized products, or by creators of open-source software that uses W3C standards, that it was not this principled position that provided the motivation. But the whiff of vendor greed isn't the real problem. Even such a principled position would need substantial factual backing in order to prevail in an honest argument. No such backing is apparent in the Working Draft. So far the W3C has had no apparent problem in coming up with technologies that it can standardize and that do not require payment of patent royalties. In areas where this is not true, the best course for the public may be to wait to standardize, or to encourage freely available research and development to fill the gap. As a past example, the Web community invented the PNG file format, after it became clear that the patented GIF format was not being made available on RF terms. As an even earlier example, there were many proprietary information publishing and navigation systems such as Compuserve, Dialog, Lexis, and Nexis, but the Internet community invented a much better open standard (HTTP and HTML), which has come to serve the public interest with many orders of magnitude more benefit than the proprietary ones ever did or will. The document states, "our goal is to affirm the Web community's longstanding preference for Recommendations that can be implemented on a royalty-free (RF) basis. Where that is not possible, the new policy will provide a framework to assure maximum possible openness...". The document does not address the concern that the new policy will actually undermine the community's longstanding preference for RF standards, by making an alternative for-pay standards process available to patent holders. Under the RF policy, companies still had strong incentives to contribute technology, even if it would become royalty-free if selected (the first-mover advantage, etc). Under the new policy, companies have less incentive to do so, since if they merely bluster and threaten long enough, they can force the standards process down a path that favors their private interest, to the detriment of the public interest. The Working Draft also does not explicitly state that all existing Working Groups have a policy of "RF". Instead it is silent on what the policy of existing working groups is, though it requires each new Working Group to define its policy. The Working Draft also provides in section 5.3 a policy for changing the policy of a working group, but there is a huge flaw in it. If the working group changes its licensing mode, then "licensing committments made by Working Group members under the older charter are void". This means that even when W3C has received written assurances in existing working groups that any controlling patents will be available royalty-free, companies can squirm out of these committments by merely arranging that a majority of (commercial companies in) a working group vote to change its policy. Then all those existing assurances blow away, and each such company is free to charge whatever "reasonable" fees it likes, despite its earlier assurances. The dubious nature of many existing and upcoming software patents also provides a problem that the Working Draft does not address. Many software patents are of dubious validity due to pre-filing disclosure of their method, or due to obviousness to others skilled in the art. However, at least in the US, the process of contesting such claims is painful and expensive, since the public is not notified of pending patents, yet once a patent is issued, it is legally presumed to be valid. Suppose that the W3C was interested in standardizing a method of looking up information about musical recordings, and a company such as Gracenote, for example, proffered its dubious patents in an attempt to derail or control such a standard. It's not clear that the W3C process contemplated by the Working Draft would be able to essentially thumb its nose at the patent-holder and say, "So sue us if you're serious, and we'll get your patents invalidated." Instead, it might be forced down some less useful course of action by the mere threat of patent action by an unscrupulous (or merely misguided or greedy) patent holder. A further complication is that many software patents are only valid in a small number of jurisdictions (principally the IP-lawyer-dominated United States). It would be inappropriate to jeopardize a worldwide standard because of patent issues in a few countries that have broadly overreaching intellectual property laws. Instead, a better course would be for patent holders from those countries to waive their patent rights in W3C-recommended technology -- which is what the RF policy does. To summarize, the Working Draft policy would undermine the creation of widely available open standards and widely available open implementations, by providing patent-owning companies the opportunity to derail the W3C process into channels that benefit their private interest at the expense of the public interest. RAND standards would largely preclude open source implementations, such as the original Web server software written at CERN. These dangers are barely mentioned in the draft policy. The draft policy should be revised to require RF licensing of all W3C-recommended technologies. John Gilmore Entrepreneur and Civil Libertarian (Author of GNU Tar, previous maintainer of GNU GDB, co-founder of open source company Cygnus Support, co-founder of civil rights group the Electronic Frontier Foundation, co-founder of the Usenet "alt" newsgroups) http://www.toad.com/gnu/
Received on Sunday, 30 September 2001 23:10:29 UTC