Much of the modern economy is driven by software development. Companies are creating and refining new apps that run on mobile devices, and using machine learning to provide users with personalized user interfaces and content. Consumers expect intuitive and polished interfaces. After devoting significant energy and resources to developing software, many companies seek to protect their intellectual property. Unfortunately, a bewildering legal gauntlet confronts them. At the threshold of this gauntlet lies a major obstacle: is their software even “eligible” for patent protection?
The 1952 patent act (and later court decisions) established as a default that patents could be filed for “everything under the sun made by man.” Courts created exceptions for natural laws (these are discovered, not created), as well as mathematical algorithms (these were too abstract to be eligible for patenting). While computer hardware is clearly patent eligible, many questioned whether software could be eligible because algorithmic steps in computer code can be performed or followed in the human mind, but performed much more efficiently with a computer. Computer code is often expressed in writing (which can be protected with copyrights), but it has functional results when performed on a computer. As standardized personal computers spread, software innovation accelerated using relatively generic computing hardware.
These issues came to a head in the 1990s with the question of whether business methods, performed using computers, could be patented. In 1998, the federal court that hears patent appeals held in the State Street Bank case that useful, applied algorithms could be patented and refused to categorically exclude software-implemented business methods. The following decade saw an explosion of software-implemented business method patent applications and the U.S. Patent Office had to hire thousands of new patent examiners to address this demand.
In the mid-2000s, some academics and interest groups (such as the Electronic Frontier Foundation) used examples of broad business method software patents to complain this system was not fair or economically helpful. They argued that rather than providing reasonable incentives to inventors to the benefit of all, software patents actually stifled the economy. They also helped popularize the idea of a patent “troll.”
Of course, patents have always established a “negative” right—a right to block others from infringing on a legal “claim” describing the innovation in the context of a patent description. Patents have never required that their holder manufacture or sell products under the patent. Many inventors throughout history have chosen to sell the patent rights to others who may be better at manufacturing or commercializing. For many entrepreneurs, this is the only viable option because to start their own company invokes the difficulties of raising capital, establishing manufacturer and supplier relationships, fending off threats from large companies, and complying with government regulations (to name a few). However, once an inventor sells a patent or the rights to further patents, some patent acquirers are “non-practicing entities” or “NPEs”—they use the patents to sue other companies without producing related products or software themselves.
Large corporations claimed they were frequent targets of patent lawsuits by these NPEs. Unlike start-ups, their business model no longer depended on protecting a few key innovations. Thus, large corporations tended to favor the arguments against software patents, even though they often held their own large portfolios of such patents for at least defensive purposes. Persuaded by some academics, interest groups, and many large companies, a significant portion of the public came to view software patents negatively.
This background led to two major changes in patent law. First, the “America Invents Act,” billed as patent “reform,” was signed into law on September 16, 2011. It made many changes, but among the most important was to make it easier and cheaper to attack issued patents using administrative judges at the patent office. Second, the Supreme Court held in the Alice case of 2014 that some software patents were too abstract and therefore were not eligible for patenting. Shortly after these changes, one of Google’s lawyers was appointed to lead the U.S. Patent Office. For the next several years, it became much more difficult to obtain and enforce patents on software innovations, especially those involving business methods. One former patent judge referred to administrative judges at the patent office as “patent death squads.” Not only were federal and PTO judges more skeptical of software patent eligibility, but software patent examiners themselves were much more likely to reject new patent applications on this basis.
Recently, there have been some signs that the pendulum may be swinging back toward a more neutral view of software patent eligibility. In the last two years, several appeals court decisions have upheld eligibility of several software inventions despite arguments they were too abstract, under a handful of theories. The new director of the U.S. Patent Office highlighted some logical problems with the way some courts and examiners were applying the law of patent eligibility. He also spearheaded new Patent Office guidance and training materials for its employees that streamlines issuance of software patents, focusing on whether the patent claims are useful. Two major patent specialty associations have suggested draft legislation with similar goals.
It is too early to tell how far the pendulum will swing, given the forces at work on this important question of public policy. However, as software innovation continues apace, those who obtain patents on their work will continue to have a business advantage that may provide the edge when seeking funding. Courts, academics, and ultimately the American public will all have a say in how much patent protection can be used to reward this type of innovation.