78 Brooklyn Law Review (2013)
59 Pages Posted: 17 Jan 2012 Last revised: 25 Oct 2013
Date Written: September 24, 2011
Many contemporary treatments of the patent system begin with Fritz Machlup’s damning with faint praise. If we did not have a patent system, it would be irresponsible, on the basis of our present knowledge of its economic consequences, to recommend instituting one. But since we have had a patent system for a long time, it would be irresponsible, based on our present knowledge, to recommend abolishing it. and conclude that for all its imperfections, the patent system is still worth keeping. Patent may introduce costs and inefficiencies, this analysis goes, but since patents serve a necessary function as incentives to innovate, we must bear and mitigate their costs. In the case of software patents, I challenge the incentive side of the equation: Patents do not provide a useful incentive to innovation in the software industry, I contend, because the patent promise ill-suits the engineering and development practices and business strategies of software production. The problem is not merely inefficiency in implementation of software patent, but a structural mismatch between where the incentive applies and how software innovation happens. Even an ideally implemented software patent . well examined, fully disclosed and enabling, properly scoped in light of the prior art would fail to serve the incentive functions intended by the Constitution, the Patent Act, and standard patent theory. Previous scholarship, whether critical or congratulatory of software patents, has largely failed to examine the structure of software *Fellow, Berkman Center for Internet & Society at Harvard Law School and Princeton Center for Information Technology Policy. Research performed while a Fellow at Silicon Flatirons Center at University of Colorado School of Law was funded by a generous grant from Brad Feld to the Silicon Flatirons Center. All opinions are those of the author.
Previous scholarship, whether critical or congratulatory of software patents, has largely failed to examine the structure of software development and the institutional specifics of patent’s operation in this industry. I therefore look at these mechanics: How is the incentive function of patent believed to operate? How does it operate in the software industry? Does the tool serve its goals? Addressed head-on, even before compounding the issue with side-effects and unintended consequences, I conclude the answer is no, patents do not encourage software innovation.
Part I describes the nature of software development, its sources of innovation, and its business environment. This part draws on sources from engineering, computer science, business, and strategy literature, as well as the experiences of commercial and open source software developers. I discuss several ways in which software development differs from the canonical model of manufacturing widgets, the challenges of going from idea to implementation, including prototyping, revising to meet user needs, and debugging. This part shares with the New Institutional Economics literature a focus on the nuts and bolts of how systems function, to identify common frameworks. While market dynamics differ among segments, we can identify commonalities derived from the underlying nature of software.
Part II reviews existing legal theories of patent incentives and innovation. It formalizes the mismatch between incentive theory and software patent practice. In many of the accounts that attribute value to software patents, there is a circularity. Startups say patents are important because venture capitalists (VCs) demand them, whereas VCs who look to patents as a signal of capacity or uniqueness are in fact seeing a show aimed at the potential funders rather than to the customer market or a demonstration of genuine novelty. Prospect theory (Kitch) does no better to validate software patents. The claim is staked too early to give the proprietor a useful coordinating or notice function. Additional justifications or explanations for patent, such as knowledge modularity, are at best ambivalent.
Part III applies the theories to software more specifically. Where do software developers and venture capital backers seek patents, and how do individuals and firms use them? Looking particularly at the timing of patent’s intervention in the system, I conclude that it encourages invention, not innovation; idea-generation, but not implementation, debugging, and deployment. A player focused on patenting can obtain numerous patents without developing any of the technologies to useful levels of deployment or disclosure, yet leaving a minefield for others who actually deploy software. Hence the troll problem is particularly acute in the software field. Since initial invention is rarely the bottleneck to economically useful software development, software patents more often become entangling thickets than productive incentives. Here I also analyze the alternatives to patent available to protect software development: trade secrecy, copyright, first-mover advantage, market complements.
Part IV, finally, uses this analysis to reflect on the institutional dynamics of patent law. Close analysis of software patents and software development adds to the debates over technology-specificity versus uniformity and the proper role of courts versus congress, and assists in better framing the question how to promote the progress of science and useful arts.
Suggested Citation: Suggested Citation
Seltzer, Wendy, Software Patents and/or Software Development (September 24, 2011). TPRC 2011. Available at SSRN: https://ssrn.com/abstract=1985780
By Mark Lemley