The software-related and business-method patent markets are depressed, and there is little hope for a bailout. A federal appeals court decision, In re Bilski, rendered a multitude of business-method patents illegitimate, and the Supreme Court's decision in the case, which could come as soon as next week, could cause even greater damage. The United State Patent and Trademark Office (USPTO), at least, seems to be anticipating a more restrictive decision.
Subject-matter patentability, however, is not the only threat to these patents. For years, patent professionals writing and prosecuting software and business-method patents blithely used a provision in the Patent Act that allows inventors to claim new software by reciting how it functions. Now claims with such elements routinely fail a new requirement for specific algorithms corresponding to each function.
The third storm to hit these patents is the more-stringent requirements for nonobviousness from the 2007 Supreme Court case of KSR Int'l Co. v. Teleflex Inc. That case did not single out software and business-method patents, but these patents suffer mightily under the Supreme Court's standard.
In Bilski, the Federal Circuit declared a method eligible for patent protection only if it (1) is tied to a particular machine or apparatus, or (2) transforms a particular article into a different state or thing. The court did not explain what a "particular machine" was, and said nothing about claims to machines or articles of manufacture (e.g., programs on a disk). The district courts and the USPTO, however, have sought to fill that vacuum.
One district court, in FuzzySharp Technologies, Inc. v. 3D Labs Inc., Ltd., held that a computer was not a particular machine. A different district court, in H&R Block Tax Services, Inc. v. Jackson Hewitt Tax Service, Inc., agreed, and went further to invalidate apparatus claims as well.
Not to be outdone, the USPTO in ex parte Gutta found both apparatus and article of manufacture claims involving mathematical algorithms ineligible for patent protection unless the claims met a new test it devised. Under this new test, a claim must be limited to a tangible practical application that results in a real-world use ("not a mere field-of-use label having no significance"). In addition, a claim may not encompass substantially all practical applications of the mathematical algorithm either "in all fields" or even in "only one field."
The Federal Circuit has not considered these cases, and the Supreme Court decision on Bilski may render them moot or trailblazing. These cases, however do demonstrate the feeding frenzy Bilski created by providing a club for these patents.
Rejecting the black box: The rebirth of the algorithm
Not so long ago, patent professionals believed they had an easy way to protect computers running inventive software. A provision of the Patent Act allows one to claim a component according to its function. Although that provision limits the scope of such claim elements to the structure disclosed in the patent corresponding to the function (and equivalents of such structure), the courts initially placed little restriction on that nature of the structure.
In one early case where the invention related to a mechanism for controlling a modem, In re Hayes Microcomputer Products, Inc. Patent Litigation, the structure was simply a microprocessor. The claims in the patent included a "timing means," and the court said that an instructional flowchart provided sufficient structural disclosure for the microprocessor. In another case where the patented device reconstructed spatial current distributions in biological objects like patients' brains, In re Dossel, the structure was an inherent computer executing unspecified but "well-known" algorithms.
In 1999, however, the rules changed. The Federal Circuit decision in WMS Gaming Inc. v. International Game Technology held the structure corresponding to means-plus-function elements for software was the algorithms in the specification that corresponded to the recited functions. In WMS Gaming, the corresponding algorithm (program) was not equivalent to the accused program, so there was no infringement.
Patent practitioners thought the lesson from WMS Gaming was to avoid specific algorithms that might restrict claims. They could not have been more wrong. A series of recent cases has found that the lack of specific, detailed algorithms rendered claims with means-plus-function elements indefinite.
In one 1998 case, Aristocrat Techs. Austl. PTY Ltd. v. Int'l Game Tech., the court considered the following slot-machine claim term: "game control means being arranged to pay a prize when a predetermined combination of symbols is displayed in a predetermined arrangement of symbol positions selected by a player." The patent merely referred to "appropriate programming" to perform the claimed function. The court held that the patent failed to provide a specific algorithm to correspond to the means-plus-function element, and so invalidated the claim. Aristocrat, though, was just the beginning.
The noose tightened later that year. In Finisar Corp. v. DirecTV Group, Inc., a claimed "database editing means" performed the very detailed functions of "generating a hierarchically arranged set of indices for referencing data in said information database, including distinct indices for referencing distinct portions thereof," and "embedding said indices in said information database." The "algorithm" in the specification was identical to the function. Since the claim called these operations functions, they were not "algorithms," and the claims were indefinite.
Later that year, the patent owner in Net MoneyIN, Inc. v. Verisign, Inc. argued that the disclosure of a bank computer was sufficient structure for the functions of comparing account and amount data to certain data structures, and generating an authorization indicia if credit was available since all bank computers performed these functions. The court invalidated the claims, however, because there was no specific algorithm.
Most recently, in July 2009, the Federal Circuit invalidated claims in Blackboard, Inc. v. Desire2Learn, Inc. containing an "access means." The lack of an algorithm to perform the atomic function of providing access doomed those claims.
Also recently, the court considered a claim with 10 means-plus-function elements in Encyclopedia Britannica, Inc. v. Alpine Electronics, Inc. Despite the narrowness of the claims, the court found no specific algorithms for the two most-detailed elements, "means for providing access to...related information" and "means for retrieving...textual information and interrelated graphical information." Even though the two means were so specific that the functions were coextensive with the algorithms, the court found the claims invalid.
After KSR, combining old elements is not enough
KSR raised the bar across the board for patentability by making it easier to prove claims obvious. The Federal Circuit has affirmed the USPTO's efforts to use the KSR tool to reject software and business-method patents using prior art that several years ago would have fallen short.
Following KSR, the court sanctioned the combination of disparate and incompatible references in the case of In re Gleizer, where the court held "non-analogous art directed to an entirely different problem" close enough since they vaguely explained that common sense would have led a person of ordinary skill to arrive at the invention. The invention eventually printed out a shipping label after an "automated transaction method" occurred between a buyer and seller, whereas the reference discussed freight calculations.
Similarly, in a case where the patent was directed to an online communication terminal, In re Mettke, the patent was found to be invalid. The court concluded that an article, which indirectly discussed the combination of ideas similar to the patented invention, was sufficient to invalidate the patent. The patentee argued that the article was not precise enough to warrant the decision, but the court held that the teaching in the article was close enough to support the ruling.
Thus, two major trends have emerged: (1) arguments against combining references because prior art is nonanalogous have little chance of success, and (2) an inventor's best hope for an invention is to find a feature that does not exist in prior art.
Business-method and software patents currently have few friends, as several companies (e.g. Microsoft) and industries (e.g., banks and insurance companies) would like to see their demise. Perhaps the climate will only change when those companies and industries face increasing competition and find the absence of protection to be a handicap, not a boon.