Software patent developments: a programmer's perspective.

Author:Gibby, John A.

    Patent law has evolved over the past twenty-five years such that computer program inventions are now much more likely to fit into one of the four types of subject matter(1) eligible to receive patent protection. Under the Examination Guidelines for Computer-Related Inventions(2) (the Guidelines), as promulgated by the United States Patent and Trademark Office (the "PTO"):

    * A useful, programmed computer is a statutory product (machine or manufacture).(3)

    * A claim for a data structure or computer program, recorded on a computer-readable memory, that can be used to direct a computer to function in a particular manner will usually be a statutory machine or article of manufacture.(4) This means that not only a computer's internal RAM or ROM, but also floppy disks and CD-ROMs containing new and useful programs, are patentable subject matter.

    * A claim that requires the performance of one or more acts is a statutory process, if that process results in a fully-disclosed or generally-knowable practical application involving a physical transformation outside the computer, or if it is "limited by the language in the claim to be a practical application within the technological arts."(5)

    Under these Guidelines, it is now possible to draft machine, manufacture, or process claims for most computer program inventions. However, this state of affairs has not come about without a struggle.

    The United States Supreme Court has not provided much guidance in the development of computer program patent law. The Court has interpreted section 101 in the context of processes or methods in only three cases,(6) all of which concerned only simple programs implementing mathematical algorithms.(7) The job of developing the patent law for other, more complex types of programs, such as word processing programs, has fallen to the PTO and the United States Court of Appeals for the Federal Circuit.(8) Unprepared for problems such as examining prior art software and afraid of being buried under an avalanche of software inventions, the PTO responded with attempts to classify software as unpatentable subject matter.(9) The Court of Appeals for the Federal Circuit has forced the PTO to rethink that position, resulting in the current Guidelines.

    Although much progress has been made in adapting the patent law to deal with software inventions, that law has become overly complex. This article reviews current law and suggests improvements including a simplified approach to the subject matter issue and a higher standard for section 103 nonobviousness(10) to be applied to software inventions.

    The issues examined in this article are similar to those addressed in Maximilian R. Peterson's excellent article Now You See It, Now You Don't: Was It a Patentable Machine or an Unpatentable "Algorithm"? On Principle and Expediency in Current Patent Law Doctrines Relating to Computer-Implemented Inventions.(11) This article, however, offers several new ideas and perspectives, including: (a) a rationale for shifting emphasis from patentable subject matter to the section 103 "nonobviousness" requirement; (b) exposition of a "continuum of claim limitation/abstractness" approach to better understand which computer program inventions should and should not be patentable;(12) (c) additional exposition of the analogy between computer programs and mechanically-implemented machines; and, (d) the general perspective of an experienced computer programmer.

    This article discusses the current state of patent law and why courts have had difficulty formulating it. This article also discusses new ways to understand software patent issues and to improve patent law as it relates to software. The courts have been floundering due to a lack of understanding of the nature of software inventions and the process of creating software programs. Courts have thus formulated tests that seemed appropriate at the time but were later discarded or reinterpreted. The law in this area is complex as Illustrated by the current Guidelines, and successful claims drafting for software invention patents has become tricky. A reexamination of the case law and the underlying purposes of the Patent Act can serve to simplify the law and to shift the emphasis from patentable subject matter to the nonobviousness of the software invention.

    The Peterson article recognizes the dichotomy between the PTO's treatment of mechanically-implemented versus software-implemented inventions and the injustice that results when a software invention is denied protection that would be awarded to an analogous mechanically implemented invention.(13) He sees a partial solution to this problem in the "enactment of appropriate legislation as a vehicle for achieving desirable public policies."(14) Peterson, however, points to a dichotomy between machine and program, whereas the appropriate dichotomy is between abstract idea and concrete implementation. In addition, he fails to recognize that for software the whole issue of patentable subject matter should take second place to the nonobviousness issue. Peterson's vision of a comprehensive study determining legislative improvements to the current patent law might not be necessary if the courts and the PTO improve their understanding and application of basic, already established patent law principles. When it comes to software inventions, mechanically-implemented inventions should not be dichotomized; the principles and policies that govern both types of inventions should be the same.


    1. Gottschalk v. Benson and Mathematical Algorithms

      The United States Supreme Court first addressed the issue of patentability of certain computer program inventions in a 1972 case, Gottschalk v. Benson.(15) It is important to understand exactly what Benson claimed.(16) Benson had claimed an efficient method, implemented on a general-purpose digital computer of any type, for converting a number expressed in binary-code decimal ("BCD") format into the same number expressed in pure binary numerals.(17) Thus his claims were limited by the requirement that they be implemented on a computer. Benson asserted that the claims were patentable under the "process" category; the issue was thus whether Benson's invention was a section 101 "process."(18)

      The Court first cited the rule that abstract concepts, principles, mental processes, or scientific truths cannot be patented, but that the practical application of one of these might be patentable.(19) In other words, a patent cannot cover the idea of doing a particular process, or the discovery of a new concept; it can only cover a particular way of doing that process, or an application of the discovered concept to "a new and useful end."(20)

      The Court then set out to determine whether the claimed invention was more than an abstract idea or scientific truth.(21) It is a scientific truth that numbers in BCD format can be converted to binary form, and this truth clearly cannot be patented. In the Court's view, however, even the method of effecting this conversion (e.g., by hand on paper, by mechanical process, etc.) also is a scientific truth waiting to be discovered. From this perspective, a way of doing long division would be a scientific truth waiting to be discovered. At first glance, Benson's claims do not look like a scientific truth; the claims were for a very specific and detailed method, with steps Re "masking out said binary `1' in said second position of said register," and "shifting the signals to the left by two positions."(22) The steps one would go through to write out a long division problem, however, would also be specific and detailed. Thus, these "methods" (including algorithms) are considered to be scientific truths waiting to be discovered by anyone who does not yet know them.

      The Court, however, reached a different conclusion about method or process in Diamond v. Chakrabarty,(23) which involved a human-made bacteria that is used to break down oil slicks.(24) One could argue that this bacteria was also a scientific truth waiting to be discovered: if molecules are put together this way, it is true that Chakrabarty's result will follow. The Court, however, found the subject matter of a live, human-made microorganism patentable.(25) The different conclusion might be explained in several ways. First, the bacteria is physical and is a "composition of matter," one of the statutory categories.(26) In addition, the bacteria has a practical use, all by itself it eats oil. Benson's computer algorithm can only be used within another application. Moreover, the bacteria cannot be described by a math equation. Thus, the Court saw Chakrabarty's claim as "not to a hitherto unknown natural phenomenon, but to a nonnaturally occurring manufacture or composition of matter--a product of human ingenuity `having a distinctive name, character [and] use.'"(27)

      There are practical distinctions as well. People cannot chum out useful bacteria as fast as they can chum out working computer algorithms; so, there are fewer problems dealing with patents for new bacteria. Additionally, it is not as necessary to provide incentives for people to write computer algorithms for solving math problems.(28) Devising solutions to math problems is a necessary step to making other, useful computer applications; so, no appreciable incentive is added by the patent law.

      Thus, a claim for the BCD-to-binary algorithm by itself (i.e., not implemented on a computer), like a claim for the method of long division, would unquestionably be considered a scientific truth.(29) However, Benson further limited his claim by specifying that it was for use on a general-purpose digital computer, and he argued that this limitation was sufficient to make the subject matter patentable.(30)

      In considering whether the claimed invention was an abstract idea, the Court noted that the "process" claimed was abstract in that it covered a computer method of converting BCD to...

To continue reading