| |||||||||||||
KS2009: Legal issues
Lawyers tend to be a rare sight at kernel summits. At the 2009 meeting,
though, the group heard from two legal representatives on the subject of
software patents and how to deal with them. It was a bit of a
slow-starting session, with some members of the group clearly skeptical of
the entire concept. In the end, though, it ran well over the allotted time; attendees
are clearly interested in learning more about the challenges presented by
software patents and what can be done to protect Linux from the worst
threats.
The leaders of this discussion were Karen Copenhaver, counsel for the Linux Foundation, and Keith Bergelt, head of the Open Invention Network. The first topic was defensive publication. Software patents are a relatively recent creation in the United States, and patent examiners lack a database of prior art to compare applications against. The result is the issuance of vast numbers of patents for techniques which have long been known to developers. Patent examiners are not required (or able) to perform a full literature search - much less a search of existing source code - so techniques which are not in the patent database are essentially invisible to them. Defensive publication is the practice of writing up an invention in the bizarre form of English used for patents and entering it into the USPTO database. A publication in this form establishes prior art which can easily be found by patent examiners. That, in turn, should help to prevent the issuance of patents for the covered invention and to protect the publishing company from patent suits over that invention. The OIN would like to help the community in the creation of these publications; ideally, it could be done in a way which has little impact on developers' time. Peter Zijlstra asked: why does the patent office grant applications for techniques which have been covered in text books since the 1970's? The answer, beyond the lack of a comprehensive prior art database, had to do with high workloads and high rates of turnover in the patent office. Will the upcoming Bilski decision in the U.S. Supreme Court make this problem go away? Karen answered that she didn't know of any patent lawyers who thought that this decision was really going to change things. It might make it harder to get bad patents in the future, but it won't, on its own, make the vast pile of existing patents go away. In the long run, we might get more help from the changes currently underway in the patent office; changes in how examiners are evaluated and compensated will hopefully improve patent quality over time. Worry was expressed about the potential for defensive publications to function as a sort of troll bait. If somebody holds a patent which covers techniques described in a defensive publication, that somebody knows that the publisher is likely infringing on that patent. There didn't seem to be a good answer to that question. Developers were also curious about how they could know when they have done something which merits a defensive publication. The answer was that if the project was hard for the developer, then it may well be interesting enough to be worth publishing. There was a brief description of what the Open Invention Network actually does. It seems that there is a thriving market for patents. Companies which have patents are trying to realize cash from them, while those which are currently under patent attack are looking for weapons for their counterattacks. The OIN keeps an eye on this market and attempts to scoop up patents which might be threatening to Linux - or to those who might attack Linux. These patents remain in reserve, to be deployed against companies which file suits targeting Linux. OIN's patent portfolio, at this point, is a potent weapon against companies which make and sell products of their own. It's less useful against patent trolls, though, who are immune to injunctions and have little to countersue for. It's the trolls who are the worst threat against Linux. When patents exist, one can try to get them invalidated by challenging them at the patent office, but this is an expensive and risky strategy. Any failed attempt to invalidate a patent can end up strengthening it instead; it can also nullify any prior art used in the attempt. There are a huge number of patents, most of which will never be asserted; there's no way to try to invalidate them all. Finally, an invalidation attempt can lead to willful infringement damages. All of this adds up to a simple conclusion: frontal attacks against patents are often not the best way to go. What can be done instead is to work around the patents. Workarounds are a complicated topic in their own right; unfortunately, the session ran out of time before it could really be discussed. So there is still no real policy on when patent workarounds should be merged into the mainline, though the unofficial policy is that they are acceptable only if they do not impair functionality. One final question had to do with the threat against individual developers: have any developers been targeted in a patent suit? Thus far, it seems that this has not happened; patent trolls normally go for targets with deeper pockets. But we were reminded that the Linux Foundation maintains a defense fund for that eventuality; should a free software developer end up in that situation, they may have more resources available to them than they might think. (Log in to post comments)
|
|||||||||||||