Opinion-The Importance of Non-Obviousness for Software Patent Applications

by Keith Kevelson, Summer Law Clerk 2020

The Supreme Court’s 2014 decision in Alice Corp v. CLS Bank International (523 U.S. 208) laid out the standard for the patent eligibility of software. These standards had and still have significant consequences for software inventions. The rules and standards for software eligibility announced in Alice have framed patent eligibility in terms of the novelty requirements found in 35 U.S.C. 101. Patent eligibility of software inventions in the United States now largely hinges on the extent to which the USPTO considers the software to be an abstract idea. I argue below that obviousness under 35 USC 103 is a better legal standard for evaluating software, rather than its perceived abstractness.

The Alice framework is rooted in a two-step test first announced in Mayo v. Prometheus (566 U.S. 66). That two-step test analyzes:

1. If the claims of the patent are directed to a patent-ineligible concept, and
2. If so, do the specific claims' elements, considered individually and collectively, "transform the nature of the claims into a patent-eligible concept."

As a clarifying standard for applying Mayo to software, the Alice Court wrote “The method claims, which merely require generic computer implementation, fail to transform that abstract idea into a patent-eligible invention.” Alice (2350). This statement implies that software may be nothing more than "generic computer implementations" and, therefore, ineligible as an abstract idea.

There is a valid argument that software is never purely abstract software and there is no such thing as a "generic use" of software. The early historical examples of software, Hero of Alexandria's automaton manuals, Banu Musa's "Book of Ingenious Devices", and Ada Lovelace's manual to calculate Bernoulli Numbers with the Analytical Engine, were all specific uses of hardware or mechanical computers. In each of these early examples of software, someone describe instructions for a machine that already existed. There has never been useful software without some associated hardware in existence. The rise of the electronic computer following the work of Claude Shannon, Alan Turing, and others did not alter this relationship. The increasing complexity of computers required scientists and engineers to create instructions with increasing complexity. With the increasing complexity of computers and software, came the ability to create further complexity in both computers and the associated instructions. This resulted in creative loop when scientists and engineers continued the advancements in hardware and software. As Uri Vool and Michel Devoret wrote in Introduction to Quantum Electromagnetic Circuits, Int. J. Circ. Theor. Appl. 2017; 45; 897-934, "programs are groups of instructions to electric circuits."[1]

The patent system is designed to drive innovation, however, the administration in the field is not equally innovative. For example, published application US10713582B2 addresses methods for using quantum computers and US Patent 10,713,582 is a software patent for a quantum computer that the general public is not used to seeing. The CPC classification code for both, G06N 10/00, makes no distinction between quantum computing hardware inventions and quantum computing software inventions. Applying the Alice standards of "generic implementation" broadly means these quantum computing method patents, which ultimately are software based, risk invalidation on eligibility grounds. While the United States Patent and Trademark Office (USPTO) has interpreted the Alice standards in a manner that has allowed patents for quantum computer software without much controversy, shouldn’t software that results in so many new and beneficial uses for conventional computers be patent-eligible?

Software for conventional computers has the potential to detect if people are having heart attacks or strokes, guide driverless cars, improve COVID-19 vaccines, and put men on Mars. If inventions of quantum-computing software are not excluded as ineligible under the novelty requirement, why treat conventional software inventions differently? In the current patent landscape, many software inventions implemented on quantum computers are treated as patent eligible while identical software inventions are treated as ineligible when implemented on conventional electronic computers. The novelty of the software has nothing to do with the novelty of the machine it's implemented on.

In part, the problem has been caused by the fact that the Alice framework is exceptionally vague. The Court did not clarify exactly what constitutes a "generic implementation of a computer" or what kinds of software are "transformative." The USPTO examiner’s manual, MPEP 2106.05(a), cites to Enfish v. Microsoft, in its effort to clarify that "In computer-related technologies, the examiner should determine whether the claim purports to improve computer capabilities or, instead, invokes computers merely as a tool. " However, this standard is incidentally vague and circular because any use of a machine improves it in some sense. The fact that a given software does not increase bus speed or increase disk space does not mean it doesn't improve the machine. All software manipulates the hardware’s electric circuits to make a computer perform tasks that it would otherwise be unable to perform in the absence of the software and this, by definition, an improvement.

As a policy matter, Alice’s vague standards have harmed software innovations. While one might expect some small-time, independent inventors to be less likely to pursue patents, a far more damaging consequence has been deterring inventors from disclosing innovations for fear of losing their IP without and recourse by way of patent infringement. For example, Jarek Duda, a Polish computer scientist was shocked when he learned that Google was trying to patent work based on a compression algorithm that he allegedly developed, see US20170164007A1. Duda apparently assumed that the compression algorithm was not patent-eligible after Alice. Google responded by alleging that it was attempting to patent an invention, that it derived from Duda's algorithm, In the context of the America Invents Act’s first to file system, such a scenario can have frightening consequences for developers because Alice the standards for claim construction and infringement are as vague as those for eligibility. Developers without access to aggressive legal teams will fear that should they disclose their algorithms in an open source environment, others will patent uses and applications of the algorithms in ways that will hinder the developer’s freedom to operate. Given the very limited protection that copyright affords software for any practical purpose other than preventing outright plagiarism, this will lead to software developers preferring trade secrets over patents and denying the public the disclosure or teaching function of patents.

A better standard for considering the patentability of software pertains to obviousness, not novelty and the vague transformative step. The reason why software patents have drawn the ire of so many developers is because many pre-Alice software patents were obvious. Amazon's infamous one-click shopping patent (US5960411A) is the well-known example of a granted software patent that most developers considered obvious at the time. The patent system works best when inventors can rely on common sense for at least a general idea of the kinds of inventions that are patentable. Developers and inventors cannot determine if the use of a new compression algorithm to increase the information storage capacity of computers is sufficiently transformative under Alice, but they can conceptionally test whether it is obvious or not. The worst possible intellectual property environment is one where inventors fear disclosing their ideas without any form of protection.

Alice has sewn doubts in the minds of developers that not only cause hesitation, but also make it hard for developers to see the benefits of software patents in many situations. For example, had CD Projekt pursued a patent strategy when attempting to implement many of its novel technologies and techniques, Cyberpunk 2077 would have likely had far fewer glitches due to the feedback it would have received once the applications had published. Software patents incentivized Allen Rozman, an independent inventor with a small team, to develop and disclose his cyber-security system (USRE43103E1) because he knew he could either license it to large companies or sue for infringement; his heirs successfully sued Google for infringement (https://www.reuters.com/article/ip-patent-google/google-ordered-to-pay-co-inventors-20-million-in-patent-case-idUSL1N1FY0FY). Vague patent eligibility guidelines from Alice have created a hostile innovation ecosystem where talented developers are not working on important problems with software because they don't see a way to protect it. Ryan Terrill Holmes and his team at a small Montreal company, Descript Inc., have invented software that uses a computer's sound card to artificially synthesize extremely realistic speech with an infinite number of voices (US10564817). Without a patent system capable of protecting Holmes' artificial speech innovation when disclosed, the innovation would likely remain a secret and the public would be at an infinitely greater risk of being misled by audio deepfakes. Most developers would need to consult lawyers to see if respective patents for Holmes' speech synthesizer software and Rozman's security system would survive under Alice. Worse, the lawyers could struggle to conclusively say if the speech synthesizer and cyber-security software are sufficiently transformative under Alice. As a result of the ambiguity, the developers would have to tailor their developments to avoid potential infringement suits. However, every developer would acknowledge that Rozman and Holmes' respective software-based inventions are non-obvious. Furthermore, most developers, except for a few followers of Richard Stallman, would readily acknowledge that the respective inventions of Holmes and Rozman deserve patent protection to encourage innovation. Without strong patent protection for software-based inventions, our most talented software developers and engineers like Allen Rozman and Ryan Holmes will be keeping their inventions secret instead of disclosing them. As SCOTUS merely interpreted 35 USC 101 in the Alice decision and did not articulate any constitutionality standard, Congress may amend the Patent Act. Congress should act and amend the Patent Act to establish an obviousness-based standard for software patent eligibility that provides legal clarity and protects innovation in the software sector.