Editor's note: The Supreme Court is scheduled to hear oral arguments Monday in a case that could help clarify what constitutespatentable subject matter. Many technology companies are watching the outcome closely.
Raleigh-based software company Red Hat filed an amicus brief on the case, arguing that restrictive patent laws can hurt innovation. This essay is by Rob Tiller, a vice president and assistant general counsel with Red Hat.
In a world of constantly accelerating technological change, economic prosperity depends on innovation. To support such innovation, it is vital that our patent system be well-calibrated, so that overly broad patent monopolies do not choke innovation. In the last several years, patent standards have been relaxed by the courts, which has created a patent system that hinders innovation in the software industry.
The Supreme Court now is considering a case, Bilski v. Kappos, that may address this critical problem.
The problem is well-recognized in certain industries and in scholarly literature but is only now coming into view for a wider public. Under federal law, patents may not be granted for certain kinds of inventions, such as laws of nature, mathematical algorithms and other abstract intellectual concepts. The logic of this is simple. Patents are supposed to encourage, rather than block, innovation. If patents were allowed on, for example, E=mc2, an entire line of scientific research and technological innovation would be blocked from development.
When the software business developed in the 1970s and 1980s, software was generally thought to be unpatentable, and there were few software patent applications. In the 1990s, however, federal court decisions began allowing patents inareas that had previously been considered too abstract for patenting. The results included a flood of patents on computer software. There are now more than 200,000 computer software patents, and the number continues to grow by tens of thousands a year. Modern programs may involve thousands of discrete components that may arguably infringe previously issued patents. This creates a patent minefield for software developers.
As patent lawyers know, proving patent infringement does not require showing that the accused party copied the patent holder's product or had knowledge of the patent. A new invention can be completely original, in the sense that it is the product of independent creativity, and also infringe a patent. In fact, most infringement cases involve no allegation that the defendant wrongfully copied someone else's invention.
There is no practical way for software developers to be certain that their freshly conceived code does not infringe some prior patent. The boundaries of individual software patents are typically vague and uncertain, and there is no reliable, cost-effective way to clear new products by searching existing patents. This means that an act of creating innovative (or non-innovative) software always entails the risk of an infringement lawsuit.
Such lawsuits can be ruinously expensive - including, for an average-size case, millions of dollars in attorneys' fees. Large software companies have developed defenses against some patent threats, including obtaining their own patents that they may use to bring countersuits if attacked. This strategy is only available to well-financed companies. Even large companies face increased litigation risks from businesses with no purpose other than exploiting patents. These businesses - called non-practicing entities, or, less politely, patent trolls - buy patents not with a view of producing products, but rather so that they can demand ransom from operating companies.
The Supreme Court has agreed by taking the Bilski case to address the nature of patentable subject matter. The lower court decision acknowledged that prior case law, including the Supreme Court cases involving software, excluded from patenting abstract inventions such as algorithms. Using language from prior Supreme Court decisions, it determined that a process must be either "tied to a particular machine or apparatus" or must "transform a particular article into a different state or thing."
This "machine-or-transformation" test makes sense. It is important, though, that it not be interpreted too broadly. Some have tried to argue that "tied to a particular machine" means running computer software on computers. Lower court and administrative decisions have generally rejected this view, however. The Supreme Court should do the same, and make clear that patent law does not cover algorithms, including algorithms implements as software running on computers. This decision would be good for the economy and good for innovation.
