When Biopharma Meets Software: Bioinformatics at the Patent Office

Scholars have spilled much ink questioning patent quality. Complaints encompass concern about incoming applications, examination by the U.S. Patent and Trademark Office (“USPTO”), and the USPTO’s ultimate output. The literature and some empirical data also suggest, however, that applications, examination, and output may differ considerably based on technology. Most notably, although definitions of patent quality are contested, quality in the biopharmaceutical industry is often considered substantially higher than that in information and communications technology (ICT) industries. This Article presents the first empirical examination of what happens when the two fields are combined. Specifically, it analyzes the creation and early history of a USPTO examination art unit (AU 1631) that reviews interdisciplinary inventions at the intersection of the biological and information sciences. We explore private value and quality metrics in an early cohort of incoming applications assigned to AU 1631, comparing the applications’ performance on these metrics to a group of applications assigned to a related art unit for more traditional software. We then explore the marginal value of the examination process by comparing examination in AU 1631 with that of a matched set of applications assigned to the traditional software art unit. Our results show that, on almost all conventional measures of patent value and quality, incoming bioinformatics applications were substantially different from, and “better” than, traditional software applications. Moreover, when we compared examination of applications in the two art units that had been matched on these dimensions of private value and quality, applications in AU 1631 experienced significantly more rejections, particularly notice-related rejections, than the conventional software applications. The notable exception was in the area of nonobviousness, where the prevailing law at the time made interdisciplinary, or “recombinant,” inventions presumptively nonobvious. Potential causal explanations for the higher rejection rates in areas other than nonobviousness include “biotechnology-specific” guidelines then in place at the USPTO as well as the higher educational attainment of examiners in AU 1631. Our results contribute to the empirical literature on factors that affect patent examination quality, particularly with respect to notice. They suggest that technology-specific examination guidelines and educational level not only have an impact, but that this impact can “spill over” into other technologies. The results also demonstrate, at the level of the art unit (an important but relatively understudied unit of analysis), the empirical theme of substantial variation in what the USPTO receives and how it processes what it receives. We conclude by discussing potential policy implications, including a link to the literature on how examination should be conducted when (as is increasingly the case) the art in question is an interdisciplinary, team-based field.

Saurabh Vishnubhakat & Arti K. Rai

Harvard Journal of Law & Technology

January 2015

External Link