Hesitation over software patents in Canada

Last Wednesday (July 26), the Federal Court of Appeal (FCA) handed down its highly awaited decision on the patentability of computer-implemented inventions (software patents). In Canada (Attorney General) v. Benjamin Moore & Co. (2023 CAF 168), the FCA rejected the test proposed by the Intellectual Property Institute of Canada (IPIC) in the first instance decision, Benjamin Moore & Co. c. Canada.

On the one hand, according to the Court of Appeal, Justice Gagné of the Federal Court should not have included IPIC’s proposed framework in the operative provisions of her ruling because it was not requested in the parties’ application. In addition, in the FCA’s opinion, establishing an analytical framework for software patents would be premature, as many issues have yet to be properly considered by any court in Canada.

So, back to square one? Not quite…

Review of the facts of the Benjamin Moore case (2022)

This FCA decision—the most recent in the Canadian software patent saga—establishes a modicum of order against a highly unusual backdrop. For those of you who have just joined the debate, here is a short summary of the facts.

As part of a national initiative, Benjamin Moore applied for two software patents in Canada in 2010.¹ The software in question was designed to help users navigate a colour bank based on criteria such as “emotion” and “harmony”. In October 2014, both applications were rejected on the grounds that they dealt with non-patentable subject matter. By applying an analytical framework that the Canadian Intellectual Property Office (CIPO) referred to as a “problem-solution” approach, the CIPO examiner concluded that the inventions in question were akin to applying mathematical formulas to a database. Mathematical formulas, however, are not patentable. In May 2020, the Patent Appeal Board confirmed CIPO’s decision. In November 2020, Benjamin Moore appealed the decision before the Federal Court. In an unusual move, during a hearing on March 30, 2022, the Attorney General immediately acknowledged, in light of the Choueifaty²decision handed down in the interim, that the Commissioner of Patents did not apply the right test and agreed to send the patent applications back for review. In another unusual move, IPIC intervened as a third party on the grounds that the case transcended the parties’ interests because CIPO examiners continued to apply the incorrect software patentability criteria despite repeated criticism by the courts. In so doing, IPIC sought to take the matter further with a view to making the process of granting software patents more predictable, i.e., by setting out the analytical framework that CIPO should apply when examining such cases. In a surprising judgment, Justice Gagné agreed with IPIC’s request: the patent applications were sent back for review based on IPIC’s proposed analytical framework.

This declaratory judgment recognized the need for IPIC’s proposed analytical framework. Overnight, this framework became the applicable test for software patents.

IPIC’s intervention in Benjamin Moore (2022) or the need to clarify the analytical framework for software patents

This was not the first time that IPIC intervened as a third party in a patent case. This type of participation is not always looked on favourably by the courts, particularly with regard to administrative reviews. Moreover, last Friday (July 28), in the case of Taillefer v. Canada,³ Madam Justice McDonald rejected IPIC’s intervention after applying the FCA’s three-step test.⁴ In her decision, she stated that “to the extent IPIC intends to make submissions on the issue of the proper approach to statutory interpretation, as noted by the Federal Court of Appeal above, such an intervention is inappropriate”.

However, a number of patent agents and jurists were delighted in 2022 to see that a more predictable analytical framework had emerged. Clearly, IPIC’s intervention was unusual, but we must revisit the long-running Canadian software patent saga to fully grasp the problem that IPIC was attempting to resolve.

It should be noted that the Patent Act remains silent on the topic of software. In it, an invention is defined as “any new and useful art, process, machine, manufacture or composition of matter, or any new and useful improvement in any art, process, machine, manufacture or composition of matter”.⁵ When software is being patented, we must fall back on a creative description of exactly what is involved, e.g., by describing it as a process or a storage device containing instructions that, once carried out by a computer, produce specific effects. The lack of a proper legislative framework for software leaves a void that the courts must fill by interpreting a law that was last overhauled when Windows 95 was released, along with its 13 floppy disks!

In the early 1970s, the Patent and Copyright Office, tasked with examining patent applications, issued a notice for patent professionals explaining that, in its view, software was non-patentable subject matter. However, in the wake of United States v. Waldbaum,⁶ the Patent Office showed more openness, though it backtracked in 1978 in a bid to harmonize its practices with American and British jurisprudence at the time. In 1981, the Schlumberger case⁷ became the landmark decision in the area of software patents, with the court ruling that software that relied on the simple execution of a mathematical formula by a computer could not be patented. The Patent Office’s narrow interpretation led to the rejection of large numbers of patent applications. When these cases were appealed, however, most of them were deemed to contain patentable subject matter and were sent back to the examiners for further review. In 1984, CIPO’s directive became more permissive after amendments were made.

In 2009, however, the Patent Office rejected an application that had been filed by Amazon, on the grounds that it dealt with non-patentable subject matter. In 2011, the FCA took the Patent Office to task for using a simplified test for evaluating software patents.⁸ The Patent Office appeared to be applying certain jurisprudence principles too generally, without taking the relevant facts and context into account.

Amazon’s patent dealt with the use of cookies to facilitate online purchases without authentication (the famous “one-click” patent). CIPO had difficulty understanding that an invention could be entirely intangible.

The patent was finally granted. However, instead of following the FCA’s recommendations, CIPO issued new practice notices in 2012 and 2013 without taking recent jurisprudence into account. The contradictions between the practice notices and the jurisprudence led to unpredictability in the patent review process, during which examiners either adhered to the practice notices or to the jurisprudence.

In summer 2020, the Federal Court criticized CIPO in the Choueifaty case,⁹ explaining that the problem-solution approach described in CIPO’s Manual of Patent Office Practice was not in line with Canadian jurisprudence. In the aftermath of this decision, CIPO modified its practice notices by re-establishing certain principles that had actually been prohibited in the Amazon decision, such as applying principles of section 27(8) of the Patent Act to the notion of an “actual invention”, rather than to “the subject-matter defined by the claim”.¹⁰

In short, while the complexity of software grows exponentially, patent law is drawn one decision at a time, while inventors bide their time while grappling with various uncertainties. When IPIC intervened as a third party in the Benjamin Moore case and proposed an analytical framework, the hope was to put an end to 50 years of confusion and unpredictability.

Unease in the aftermath of Benjamin Moore (2022)

Understandably, this declaratory judgment produced a certain sense of unease among CIPO and the Attorney General, who appealed the matter in fall 2022. Following the Federal Court judgment, CIPO argued that it would be difficult for an examiner to adapt the analytical framework to new situations without risking a contempt of court charge. In addition, since the initial Federal Court application filed by Benjamin Moore was not aimed at obtaining a declaratory judgment, the Attorney General argued that the Court should have adhered to the conclusions sought by Benjamin Moore and the Attorney General. The appeal of the judgment thus focused on whether it was appropriate to issue a declaratory judgment that imposed an analytical framework at the request of a third party.

Judicial minimalism — Benjamin Moore (2023)

In its 41-page decision handed down on July 26, the FCA reiterated the importance of judicial minimalism while granting the Attorney General’s request to do away with CIPO’s analytical framework. The FCA was unequivocal when it came to including the analytical framework in the mechanism provided for in the lower court ruling:

Given the lack of a detailed analysis in the FC Decision, the Federal Court’s statement that the current version of MOPOP (as amended by PN2020-04) was irrelevant, and the lack of consideration of all relevant case law, it appears to me that it did little more than “leapfrog” the test to our Court by including it in its judgment instead of its reasons. This exercise, which is quite different from dealing with an issue that was not necessary to determine the appeal before it in obiter, is inappropriate. It does not involve “judicial courage”, as IPIC argued before the Federal Court.¹¹

The FCA also noted that “interveners cannot seek a remedy that was not sought by the parties themselves”.¹² Moreover, it stated that “subject to limited exceptions, unless a request to include a particular framework for all computer-implemented inventions in the judgment is a remedy specifically sought in the notice of appeal, it should normally not be considered”.¹³ Finally, recognizing that the Federal Court does have the power to render declaratory judgments, the FCA noted that even if an application is duly submitted, the court’s “discretion to grant such relief can only be exercised after considering the four-part test set out by the Supreme Court in Ewert v. Canada, 2018 SCC 30 at paragraph 81”.¹⁴

Rejection of IPIC’s analytical framework

After concluding that she could have rectified the lower court judgment by applying the four-part test, Justice Gauthier held that IPIC’s proposed framework was not adequate and that too many open questions remained that should be analyzed by the courts before any such analytical framework could be imposed.

The decision, however, offered some obiter advice by emphasizing the need for patents to add to the body of human knowledge. In this regard, Justice Gauthier wrote “[The] difficulty often lies in determining where the discovery lies, i.e., what new knowledge has been added to human wisdom” and further on, “If the only new knowledge lies in the method itself, it is the method that must be patentable subject matter.”

Now that IPIC’s proposed framework was rejected, practitioners will have to fall back on previous decisions to determine whether software is patentable or not.

The challenge posed by software patents

To understand why CIPO is reluctant to grant software patents at times, we must bear in mind that patents are an exception to the principle of competition. Section 79(5) of Competition Act¹⁵ stipulates that exercising a monopoly obtained via a patent or other intellectual property rights is not an anti-competitive act. This privilege, which is granted to the patent-holder for a 20-year period, is designed to provide compensation for the public disclosure of the invention, with potential benefits for society as a whole.

The exchange principle is not a recent development: most patent systems in current use were inspired by England’s Statute of Monopolies adopted in 1623. The odds are, however, that the English in the 17^th century had no inkling that software applications would ever come into existence.

Historically speaking, patented inventions include a tangible component: a machine, a device, the transformation of raw materials, etc. In contrast, unpatented inventions often have an intangible quality, e.g., abstract ideas, mathematical formulas, esthetic considerations, music, etc. When software applications emerged, sophistry became inevitable!

Moreover, the notion of “software” itself is hard to define. Software is more than the code used to put together an application (code is copyright-protected, rather than patent-protected). For example, two programmers following the same requirements and specifications will no doubt end up creating different implementations, compiled to different machine instructions. Once these instructions are executed, however, what could be regarded as the same software will emerge. In some ways, patenting a software is akin to patenting certain aspects of its specifications.

When we are interested in a mechanical invention and we describe its fundamental components, we realize that it is made up of parts with specific shapes and sizes attached to each other. In contrast, when we break a software application down into its fundamental components, we find various states, thresholds and conditions, which are often described using abstract concepts and mathematical formulas. Unfortunately, these components are dangerously akin to a type of subject matter formally excluded by the Patent Act. Section 27(8) thereof states that “No patent shall be granted for any mere scientific principle or abstract theorem”. This exclusion is designed to ensure that scientific principles remain universally accessible with a view to fostering research and innovation. Therefore, examiners are duty-bound to reject applications that appropriate scientific principles.

Depending on whether an examiner focuses on the overall objective of the software or on its individual components, he or she will conclude that it is patentable or not. The difficulty lies in recognizing that, as a general rule, software is more than a set of mathematical formulas. In other words, a much broader view is needed, i.e., one that looks at the forest as a whole, rather than at the individual trees. That is what the Canadian courts have referred to as “purposive construction”.

In the Benjamin Moore case, the FCA described purposive construction as a “difficult exercise even for judges”¹⁶ and pointed out that if errors do occur, it is because “CIPO and the Commissioner simply did not properly understand all of the subtleties of this difficult exercise”.¹⁷ Striking a more positive note in the same paragraph, the judge offered reassurances that the need to apply this method should no longer be problematic because “[as] noted by the AG, the Commissioner did not appeal the decision in Choueifaty, as it did help clear up some misunderstanding in respect of purposive construction […]”. Whether or not we find that reassuring, we have to acknowledge that IPIC’s intervention might not have been the ideal solution to this problem, particularly when we realize that Benjamin Moore was taken hostage by a debate that transcends the private sphere. No doubt, IPIC’s efforts do deserve recognition. But perhaps we should also be asking ourselves whether allowing the courts to draw up an analytical framework one decision at a time is an effective way of establishing software patent law, especially when such a piecemeal manner of progression undoubtedly comes at the expense of the inventors.