In the recent decision of Bancorp Services v Sun Life Assurance Company of Canada (U.S.), the United States Court of Appeals for the Federal Circuit has attempted to reconcile the eligibility of software and business process patents with previous decisions from the United States Supreme Court. The holding also makes qualifying patent eligibility in business process claims tougher due to the requirement of being more than mathematical formulae.
In Bancorp, the Court has ruled that patents outlining a software system that administers and tracks the value of life insurance policies (5,926,792, 7,249,037) are invalid. The Court of Appeals ruled that the patents do not include a computer as an integral part of the patent as claimed and that the process described could be conducted, albeit less efficiently, manually. Without a foundation that includes a computer, the process claimed in the patent becomes an abstract process, which fails the criteria established by the United States Supreme Court in Diamond v Chakrabarty. With the Bancorp ruling the Court has singled out business process patent claims that misidentify a computer as being integral to the patent, otherwise known as Beauregard claims, as probably unpatentable, potentially culling the number of software patents that will pass through the U.S. Patent and Trademark Office.
The Bancorp case highlights two interesting trends in software patents. First, the Court has maintained the distinction between software that is strictly algorithm-based, processes that are made more efficient by running on a computer, which would therefore be abstract and not patentable, and function-based, processes that build on top of the functionality provided by a computer to provide a new innovative product, which would be patentable. The Court was careful to note “that ‘inventions with specific applications or improvements to technologies in the marketplace are not likely to be so abstract’ as to be ineligible for patent protection”. This opinion is consistent with judgements over the past years, as seen in Research Corporation Technologies v Microsoft Corporation and CyberSource Corporation v Retail Decisions. Although consistent, these decisions do not draw a clear dividing line between algorithm and function-based software. As a result, there is no principled way to determine when software transitions from a mathematical base into a functional base. Some rightly argue that it never does. This confusion could benefit from a business process patent version of “The man on the Clapham omnibus” test; can an average software developer recreate the patented process, or would he/she need a computer to complete the task? This test could potentially provide a practical definition to the two distinct views of software; however one would cringe at the thought of a developer trying to manually imitate the process of creating a halftone image based on a visually pleasing guideline created by an algorithm (the patent claim in question in the Research Corporation Technologies v Microsoft Corporation case). Realistically, the Courts will probably keep the distinction undefined and allow policy to further define it.
The second trend of interest concerns the increasing limitations being placed on process patent eligibility by both the United States Court of Appeals for the Federal Circuit and the United States Supreme Court. This was not always the case. At one point the Court of Appeals had been a “consistent supporter of broader and stronger patent rights”, including business patents, leading to a large number of software patent applications. This view of eligibility for business patents existed from 1994, with the In re Alappat decision, until 2008 and the Federal Court’s decision in In re Bilski. The Bilski decision signalled a change for the Court of Appeals and both courts have since supported more stringent qualifications for business process patents. The recent United State Supreme Court’s remand of the Ultramercial v Hulu ruling (previously covered by IP Osgoode) continues this trend.
This Bancorp decision is unlikely to modify the view of Canadian courts concerning software patents. The Canadian Patent Act explicitly lists abstract theorems as ineligible for patent protection. S. 27(8) of the Patent Act states that a patent will not be granted “for any mere scientific principle or abstract theorem”. The decoupled relationship between Canadian and U.S. views of business process patents might mitigate any effect this holding might have on Canadian law. While the U.S has a long history of case law and approved business process patents, until the recent Amazon 1-Click ruling late last year there was doubt to whether software patents were legal in Canada. The lack of support from the courts south of the border should be noted however, and could indicate stricter Canadian eligibility requirements in the future.
The holding in this case could go a long way in damming the “flood” of software patents seen in the U.S. over the past 15 years. This will no doubt make the many organizations who argue against granting software patents very happy, but there is still a case that the idea of allowing patented innovation to occur in the medium of software will “promote the progress of science and useful arts”.
Mark Bowman is a JD candidate at Osgoode Hall Law School.