Building a patent portfolio is hardly a realistic option for the self-defense of the free/opensource software community. However this community has a certain chance of leveraging prior art to its advantage, especially if it does this in a cost-efficient manner and realises the potential of legal insecurity inherent in the software patent system. So called Defensive Publishing, as has been proposed by initiatives like the Foresight Institute, however appears to be an incredibly silly approach. The most urgent task at this moment (spring 2001) is to time-stamp every CVS archive and every mailing list on which software ideas are developped.
Obtaining and maintaining patents implies high costs:
These costs have to be earned back by exercising monopoly rights. This usually usually means a combination of upfront payment (lumpsum) and per-copy license fees. Free software and per-copy licensing are incompatible. Thus the patenting costs must be earned back by the upfront payment, e.g. 1 million USD in the case of MP3. There is usually no chance of receiving 1 million from anyone in return for a permission to write free software. There may however be a chance of gaining some money from vendors of proprietary software. However this income will at best be slim if you grant a free license to the competion from free software. Under these circumstances, the chances for using patents for defending free software are very poor if not non-existant.
We saw that the proprietary nature of the patent game makes it all but impossible for free software developpers to organise any self-defensive effort within this game. However this same proprietary nature of the patent system generates weaknesses that can be exploited. Free software development produces the largest available body of published prior art. Such ideas are then in the public domain and any patents obtained thereon later are invalidated, once the prior art is found.
Strictly speaking this does not prevent anybody from asserting a trivial and broad patent for which no prior art exists, thus changing nothing about the most important source of danger. But given the difficulty of finding prior art and the uncertainty about whether any patent is really valid, a vast arsenal of hard-to-search prior art can be leveraged as an indirect protection shield for the whole free software community:
Under these circumstances the most urgent thing to do at the moment is to automate time-stamping and provide it as a basic service in all major archives (CVS, Mailing Lists, HTTP etc) where free software and related discussions evolve. Wherever this kind of service is not available, developpers should shift relevant discussions and publications to Usenet, since this is fairly reliably time-stamped by DejaNews.
The silliest thing for free software developpers to do at the moment would help volunteer in helping patent offices to build prior art databases "in order to prevent bogus patents from issuing", as has been proposed by the "Foresight Institutes" with the endorsement by some "top opensource leaders". This so called "defensive publishing" is a boomerang both for the developper who attempts it and to the community. Let us explain this by comparing the "defensive publication" strategy to the normal decentral publication of prior art.
Developpers of proprietary software should also make their work reconstructable day by day, using the same methods as developpers of free software.
Such a proprietary development history archive does not constitute "prior art" and can therefore not be used to directly invalidate the patent. But it can establish "prior use right", meaning that the developper can continue to "sell products based on his idea". He may however not transfer this right to third parties.
So far it is not completely clear what happens to prior use right when the owner of such a right decides one day to publish his formerly proprietary software under an opensource license. Does that constitute a transfer of the right to "sell products"? Or shouldn't any prior use right owner be free to "sell" his "product" to his "customers" in any way he likes, based on the principle that his customers do not have to acquire separate usage licenses? It can be seen here that the assumptions underlying the patent system do not match the reality of information products.
However as long as there are no court verdicts to the opposite effect, we have good reasons to hope that the prior user is free to distribute his work as free software. In that case everyone else is free to circumvent the patent as long as he again lets his development work flow back into this free software.
Thus the prior user of a patented software idea has a particularly strong position. He can either make the idea free for everybody or negotiate a reward from the patentee for not doing so. For the public, prior use right has about half the value of real prior art. This half-value makes it worthwhile for developpers of proprietary software to make their development history archives accessible to the public after a certain period of time. This game, if played effectively, can significantly reduce the incentive to patent trivial innovations for which a lot of private prior use can be expected to exist even if no prior art is found.