This article examines the issues inherent in patenting software-related inventions, particularly where mathematical algorithms are concerned. Software-related inventions are difficult to patent because they often contain mathematical algorithms, and mathematical algorithms are not patentable subject matter. The PTO recognizes that this approach -- simply identifying the algorithm-containing software as non-patentable -- precludes the need to address the more difficult test of whether software-related inventions are new or not obvious over prior art. The author proposes an improved examination procedure to replace the PTO's current three-step test. The existing PTO examination procedure entails determining whether the claim merely recites a mathematical algorithm, and, if not, whether the mathematical algorithm is within the prior art and is not obvious. The very nature of mathematical algorithms, however, makes it very difficult to determine obviousness, and critics of the current approach feel that the PTO does not have the means to perform the obviousness analysis correctly. The first step of the improved test is to determine whether the claim as a whole is for a mathematical algorithm, or whether a mathematical algorithm is only one of the claim components. If the whole claim is for a mathematical algorithm, the invention is not patentable. However, if only one of the components is a mathematical algorithm, the second step is applied by assuming that the mathematical algorithm is within the prior art, even if it is not. The last step is one of determining whether the claim is new and obvious over the prior art -- a step which is greatly simplified through having assumed that the algorithm is within prior art. As a result, the PTO avoids having to determine whether the algorithm is within prior art. By using abandonment as prior art under 35 U.S.C. § 102(c), the mathematical algorithm is eliminated from consideration. Subject matter that is not claimed, but is disclosed, is presumed abandoned and dedicated to the public. Under this rule, the "old combination" doctrine is used to support the premise of the 102(c)/103 bar to patentability. The U.S. Supreme Court's use of the concept is discussed and related to the proffered approach to examining patent applications covering software-related inventions.
Alan P. Klein, Reinventing the Examination Process for Patent Applications Covering Software-Related Inventions, 13 J. Marshall J. Computer & Info. L. 231 (1995)