2. Software innovations are clearly patentable

Martin Campbell-Kelly and Patrick Valduriez went a long way toward debunking the inappropriateness and undesirability of software patents in their 2005 research paper, A Technical Critique of Fifty Software Patents.[1] Rather than focus on notoriously bad software patents, such as Amazon's infamous one-click patent, the researchers examined 50 "good" patents, which they define as those "sufficiently useful or prominent to attract a large number of citations." Among the characteristics they examined were obviousness, disclosure, inventiveness (as opposed to "strategic" patents), breadth, duration, and whether the patents also required copyright protection. They concluded that the patents were generally not obvious nor overbroad; they represented true innovations; but they last too long and require more disclosure; and many also needed copyright protection in addition to patent protection.[2]

The overall quality of software patents was the focus of research conducted in 2007 by John R. Allison of the University of Texas McCombs School and College of Engineering, and Ronald J. Mann of the Columbia Law School. In The Disputed Quality of Software Patents,[3] the researchers studied the inventiveness of individual software patents and concluded that "software patents as a group have remarkably low numbers of prior art references and forward citations." The authors claim that their studies refute the claim by many in the software industry that "large firms are plagued by a mass of low-quality patents obtained by smaller firms." They assert that "there is no substantial difference between the patents obtained by the largest and most experienced patentees and those obtained by their less well-capitalized competitors."[4]

An obvious criticism of the research by Allison and Mann is that it was conducted primarily in the 1990s, although the period studied extended to 2002. Also, the research looked only at the inventiveness of the software patents and then extrapolated a "quality" and "value" score based on the number of claims and prior art references in the patents, and the number of forward citations (references to the patent in subsequent applications) each received. This may in fact substitute a "quantity" indicator for a "quality" indicator.[5]

Another is that the researchers seem to gloss over the implications of one of their findings: that "software patents often include separate sets of claims that characterize the invention as a method (or process), a machine or apparatus (or 'device'), and a 'system'."[6] They offer three possible explanations for this duplication, including the need of software patent applicants to demonstrate some "physical transformation" in their invention.[7] The authors fail to suggest a fourth possibility for the overlap: that software is simply unlike other technologies, so attempts to apply existing patent models are like trying to put square pegs in round holes.

b. 35 U.S.C.S. §§ 101

[1] 9 Marq. Intell. Prop. L. Rev. 249, Summer 2005.
[2] Ibid.
[3] 85 Wash. U. L. Rev. 297, 2007.
[4] Ibid.
[5] Id. at 303.
[6] Id. at 335.
[7] Ibid.