OStatic

One of the big threats to open source software - so far more threatened than carried out - is software patents. As the software universe becomes hedged about with thousands upon thousands of patents, it becomes increasingly impossible to create software without violating someone's patent. A case currently in front of the United States Court of Apeals for the Federal Circuit, In re Bilski, offers some hope that the patent madness will be reined in. Here's what's going on and why you should care.

The bare facts of the case are moderately simple: Bernard Bilski attempted to patent a method of managing the weather-related risks of commodity trading. Because ideas themselves aren't patentable, what's being claimed here is a "process" - one of the muddier terms in patent law. The USPTO rejected the patent, Bilski appealed, and the case ended up at the court responsible for hearing patent appeals. They took the unusual step of ordering a hearing en banc (in front of the whole court, instead of a subset of the judges) and inviting additional briefs, which may indicate that the Court is considering using this case to clarify the laws.

Three of the briefs are particularly interesting, offering advice to the court on limiting the scope of "business process" patents (which category includes many software patents). First, the End Software Patents project argues that the Appeals Court has misread the appropriate Supreme Court decisions in the past, and that algorithms are in the main not patentable. They also bring in their core economic argument that software patents threaten to destroy the economy by presenting huge liabilities to every business that uses computers.

Also interesting is the brief filed by the Patent Office itself (available via the Intellectual Property Law Blog). They argue that the Supreme Court has narrowly limited the meaning of the word "process" in the original legislation is meant to apply to processes that transform an article to a "different state or thing" (like grinding wheat to flour, or smelting iron ore to steel) and that transforming information doesn't fall under this heading. They ask the Court to "clarify" some earlier opinions (that is, throw them out) to make this clear.

Finally, our friends the EFF have filed their own amicus brief. They go along with the Patent Office as to which processes should be patentable, but go further to offer a set of factors that the Court can consider in resolving disputes. Were these tests to be applied, most current software patents would fail.

The Court's decision is due next month. While it's impossible to predict what they will rule, it's easy to read the en banc hearing and the request for briefs from interested parties as encouraging signs. Whatever is decided, there will still be thousands of existing software and business process patents to worry about - but this case has the potential to stem the tide, and to make earlier patents easier to challenge.

Do you think the current patent system deters software development? 

• patents