Math You Can't Use: Patents, Copyright, and Software

Chapter 4. Patenting Math

Imagine a continuous line of inventions, with physical machines built from transistors and diodes at one end and pure mathematics at the other end. Any given piece of software falls somewhere along this spectrum. The line between the patentable and unpatentable items along this continuum should meet three basic criteria: physical machines should be patentable subject matter, pure mathematics should not, and whatever distinction is made between the two categories should be clear and unambiguous. Software may not fit the U.S. Code’s definition of patentable subject matter for two main reasons. The first is that software is math, and it is universally agreed that pure mathematics cannot be patented. The second is that software has no physical manifestation beyond symbols on paper or bits on a hard drive, whereas it is generally assumed that patents apply to the manipulation of physical objects. In its first rulings on the subject —Gottschalk v. Benson (1972) and Parker v. Flook (1978)—the U.S. Supreme Court endorsed both arguments, thus ruling that the patentability line should be drawn at physical machines only. The turning point for the physical manifestation question came in Diamond v. Diehr (1981), a case about a rubber-curing machine with a significant amount of software. The Supreme Court ruled that this invention was indeed patentable because of its physical manifestation—the patentability line was moved to inventions with a physical component of Patenting Math CHAPTER FOUR 44 04-4942-2 CH4 10/27/05 3:12 PM Page 44 any kind. In the wake of this decision, the number of software patent applications to the U.S. Patent and Trademark Office (USPTO) using some sort of physical terminology increased: instead of claiming “a method to calculate,” applicants claimed “a general-purpose computer on which is programmed a method to calculate.” Some of these technically physical inventions were granted patents, and some were not. The Court of Appeals for the Federal Circuit (CAFC) convened to clarify the issue and in In re Alappat (1994) ruled that these rewordings made the invention a physical device. In fact, if the author of the patent was a little careless and forgot to use the right wording, the patent examiner was obliged to insert the correct terms. Then in State Street v. Signature (1998), the CAFC drew the current line regarding subject matter: a pure mathematical algorithm may not be patentable, but when it has any useful application, it becomes patentable. This means that if applicants assign real-world names to the variables in their equations, they meet the requirement. Even this line has not held, and many patents do not even bother to disguise their mathematical algorithms with a real-world application. Loopholes Many patent advocates believe that mere technicalities about form should not prevent an applicant from getting a patent. The CAFC, some say, is stocked with pro-patent judges who wrote these rulings to simply close the technical loopholes that kept software from the patents it deserves.1 The change of wording may look like a silly trick, but it is intended to shut down what the judges seemed to feel was a silly objection to begin with. However, the objection is not simply a technicality. Because it is difficult or impossible to distinguish between applied and pure math, patent rules that allow applied math send the law down a slippery slope with the patenting of abstract mathematical procedures at its end. The physical manifestation rule, defined appropriately, could be an excellent way to draw the line. The courts were unable or unwilling to distinguish between what most would consider a clear physical manifestation (like a rubber-curing machine) and a trivial physical manifestation (like writing to a hard drive); the ambiguity of the line again led to PATENTING MATH 45 1. On the CAFC being stocked with pro-patent judges, see Jaffe and Lerner (2004, p. 105). 04-4942-2 CH4 10/27/05 3:12 PM Page 45 patentable math. However, the state machine and the states into which that machine can be placed are easy to distinguish, and drawing the patentability line between the two makes machines patentable, leaves math unpatentable, and depends on objective standards rather than judgment calls. Math All the arguments about fostering innovation still apply to mathematics. The best theorems are those that a mathematician spends months working on, tirelessly trying possibilities and dead ends until finally reaching a conclusion...