Maths you can patent: introducing an asymptotic computational complexity standard in software patent eligibility analysis under European, UK and US patent law

The patent system aims to promote the progress in useful arts by rewarding inventive technical contributions. Practically every field of technology today is driven by software and most Fourth Industrial Revolution (4IR) inventions are indeed computer-implemented. Since the dawn of the software industry, however, the patent system has been perceived as an imperfect fit for software inventions for at least two reasons.

The first is based on a misunderstanding of the Church-Turing thesis and the Curry-Howard correspondence between computer programs and mathematical proofs. This isomorphism is said to mean that software is abstract, non-technical, and, in fact, nothing more than non-patentable pure mathematics. The second is that software innovation is incremental because software is built with reusable components and a software invention’s net contribution to the art is therefore minimal. Largely owing to these views, algorithmic inventions have become a perplexing patentable subject matter which lies on a spectrum between abstract ideas and concrete technical implementations.

The thesis locates the problem in the alienation of software patent eligibility doctrine from the nature of computation, as understood in computer science. The patent eligibility doctrine can be seen as a gatekeeping tool which should limit the scope of protection to the real-world causal effects of an invention. To measure these effects, this project suggests the introduction of a non-trivial asymptotic computational complexity standard into the patent eligibility analysis.

The patent system should encourage the development of not just more software, but more efficient software. By requiring software patent claims to delimit their computational complexity standards, the eligibility doctrine would become more predictable in performing its gatekeeping function.