The following are my prepared remarks from Milan, offering a partial rebuttal of some of the arguments against software patents. Note that showing that software patents are actually helpful is a task to leave for another day... My actual talk tracked this substance closely but not exactly:

The debate over whether to allow software patents in Europe has been much influenced by perceptions of how software patents work in the United States. Software patents are widely perceived to be a sort of Godzilla monster that will rise from the sea to devour small enterprises in Europe and destroy interoperability and standards. But there is a fundamental puzzle—software innovation in the US is alive and well, as are standards, although we have software patents (and plenty of them). This should be a red flag to Europeans that there has been some oversimplification and exaggeration of problems with software patents.

Now, oversimplification is something of a necessary evil when explaining patents to legislators… they are very busy, but the oversimplifications that one elects to support should be chosen with great care. Otherwise they will stand in the way of both harmonization and fixing the real problems with the patent system.

Today I will examine three perceptions about software patents: 1) That they are harmful to small enterprises; 2)—that they are inherently or uniquely harmful to software development because of unique economic characteristics of software markets and 3) that they are inimical to open standards.

First, then, consider the claim that software patents are inherently harmful to small enterprises. On the one hand, this tends to be true of any complicated legal regime, but the point does not necessarily apply more strongly to patents than to data protection laws, copyright, or taxes.

However, there is no evidence that patents have played much if any of role in the consolidation of software markets in the U.S. Network effects and other factors have played a larger role. And on the US software markets continue to be highly innovative and competitive. In particular, open source is alive and well. The open source community is presently rewriting the linux kernel to remove patent violations, and there seems to be no shortage of high-priced legal help available to the community in the event of lawsuits.

Furthermore, empirical studies of the patent system show that large companies do not enforce their patents as often as individuals or small companies (though note as a caveat that this study was not of software patents only). There are anecdotes about small firms enmeshed in patent difficulties, but also anecdotes about large ones. In any case there is no inherent reason that patents more than any other type of complicated legal regime causes particular problems with market power. This does not mean that there is no room for improvement in terms of the impact of the patent system on small firms; but it does focus the debate on thinking about the patent reform or regulatory streamlining generally, as opposed to thinking about the problem as one that is special to software patents.

There is a rhetorical point here. Patents are susceptible
to being thought of as benefitting big players over small ones because patents, like copyright, are often referred to as conveying a grant of monopoly power. It is true that patents are grants of exclusive rights. But philosophers have also described physical property this way. A restaurant located on a busy street corner may have a particular and unique advantage over its competitors, and may be said to have a monopoly of sorts. But while referring to property rights including patents as monopolies may be a nice rhetorical tactic, it does not mean that patents in fact convey monopoly
power of the sort that would be interesting to an economist or harmful to consumers. Insofar as patents help to encourage innovation they will actually enhance competition.

A second argument that is made about software patents is that patents are inherently unsuited to software and are thus unnecessary or harmful. Software does have some different economic haracteristics from other patentable products. R&D for software, for example, is much cheaper than for pharmaceuticals. And the shelf life of software is very short. This raises the possibility that by the time a software patent is granted, innovation would have passed it by, and its main use would be to harass second-generation innovators. This is a potential problem in the United States because the courts
have not required much in terms of detailed disclosure before a software patent is granted, and have also tended to interpret granted patents broadly, making it hard for the next generation to design around. Furthermore as to patents more generally we have a first to invent rule rather than a first to file rule, shrinking the realm of prior art. Also, various institutional rules and failures make it difficult to defend against patent suits.

The question is, does this mean there ought to be a per se rule against software patentability? That is too much of a leap. While software has some special economics characterists that mean that careless or overbroad interpetation of patent rules can be harmful, this is also true of biotech, semiconductors, and other patentables. Each of these industries has its own peculiar economic characteristics, as do subsectors within each industry. And furthermore the economics within and across industry sectors including software are constantly changing.

General technology neutral patent law is still preferable than getting legislators started down the road of crafting industry-specific patent law. Once that starts, there would be no end to the lobbying for special exemptions and treatments. It would be, to paraphrase Bruno Leoni, a legal war of all against all, carried on through representation and legislation. The better route is for courts to become more sophisticated in understanding how the interpretation of patent law affects innovation in different economics contexts, and to apply that understanding on a case by case basis.

Finally I would like to consider the relationship between software patents and open standards. A couple best-known cases:

JPEG. In 2002 a small company called Forgent pulled a patent on a component of the JPEG standard out of its hat, and has been grazing off the patent settlements ever since. There is some doubt as to the validity of the patent, but no one has litigated the suit. This is clearly a problem; adopting the "loser pays" rule might help. But note that the problem has not been sufficient to bring down the standard. Forgent does not want to push its suits so far, as that will kill the goose that lays the golden eggs.

ebXML: Also in 2002 it emerged that IBM held a patent on an element of the ebXML standard, to the dismay of the standards community, as IBM had been a participant in the standards proceeding. IBM caved to public pressure and licensed the standard royalty free. Another happy ending. In IBM's judgment, they needed the good will of developers and business associates more than the patent revenues.

Again, shows up that there is a need for general patent reform, somehow lowering the cost of defending against questionable patents, continuing to work on disclosure during standard setting. But note that on the whole these horror stories are not so horrible. Standards bodies have worked hard on rules and license terms to prevent a recurrance of the IBM case. The WC3 patent policy is a case in point. And note that standards’making bodies actually tend to favor including patented elements in standards, this will help ensure that the standard incorporates the cutting edge of innovation. To this end they make provisions for companies participating in standards-making to disclose relevant patents and provide for royalty-free or other types of licensing regimes.

To conclude, there are problems with the patent system in the US, as in Europe. Many patents are granted that ought not to be granted, others are interpreted too broadly. There is often insufficient disclosure, particularly with software patents. It is costly and expensive to manipulate the patent system on both side of the ocean, and difficult to defend against possibly exploitive charges of infringement. There is a need for expanded opposition to patents during or shortly after the grant of a patent (oppositions are more common in Europe). All of this, however, is as much a problem for biotech or pharma as for software. There is a need for general patent reform and for greater understanding of how different patent rules affect innovation in different contexts. On the whole, though, it is unlikely that adopting software patents in Europe would harm software innovation in Europe, and there are some important reasons to prefer a technology neutral patent law.

posted by Solveig Singleton @ 5:58 AM | Digital Europe, International, Patents, Software

Link to this Entry | Printer-Friendly | Email a Comment | Post a Comment(0)