Software Patents as a Barrier to Scientific Transparency: An Unexpected Consequence of Bayh-Dole

Transcription

1 Software Patents as a Barrier to Scientific Transparency: An Unexpected Consequence of Bayh-Dole Victoria Stodden & Isabel Reich Department of Statistics Columbia University Intellectual Property Scholars Conference Cardozo Law School August 9, 2013

2 Computation is Becoming Central to Scientific Research 1. enormous, and increasing, amounts of data collection: CMS project at LHC: 300 events per second, 5.2M seconds of runtime per year,.5mb per event = 780TB/yr => several PB when data processed, Sloan Digital Sky Survey: 8th data release (2010), 49.5TB, quantitative revolution in social science due to abundance of social network data (Lazier et al, Science, 2009) Science survey of peer reviewers: 340 researchers regularly work with datasets >100GB; 119 regularly work with datasets >1TB (N=1700, Feb 11, 2011, p. 692) 2. massive simulations of the complete evolution of a physical system, systematically varying parameters, 3. deep intellectual contributions now encoded in software.

3 Software in Scientific Discovery JASA June Computational Articles Code Publicly Available of 20 0% of 35 9% of 32 16% of 29 21% Reproducible Research Movement: Have data and code are conveniently available at the time of publication Conflict between incentives to patent academic code and the scientific method?

4 IOM Evolution of Translational Omics: Lessons Learned and the Path Forward March , IOM releases report, Recommends new standards for omics-based tests, including a fixed version of the software, expressly for verification purposes.

5 IOM Report: Figure S-1 The fully specified computational procedures are locked down in the discovery phase and should remain unchanged in all subsequent development steps. The fully specified computational procedures are locked down in the discovery phase and should remain unchanged in all subsequent development steps.

6 Bayh-Dole Act (1980) Promote the transfer of academic discoveries for commercial development, via licensing of patents (Tech Transfer Offices), Legislators blind to the coming digital revolution, and the impact of software patents on code release for reproducibility. Implications for science as a disruptor of openness norms: delay in openly revealing code/methods, or closed code, obfuscation of methods submitted for patents (Bilski v Kappos (2011)), alters a scientist s incentives toward commercial ends, instead of the traditional production of science as a public good.

7 Academic Patenting Total Number of Software Patents filed by the top 23 University Patent Filers, Software Patents as a Percent of the Total University Patent Portfolio,

8 Software Patenting by Rank 2002 Rank 2008 Rank Assignees Software Overall Software Overall University of California Massachusetts Institute of Technology Wisconsin Alumni Research Foundation University of Texas Stanford University California Institute of Technology Johns Hopkins University Columbia University Georgia Tech Research Corp University of Illinois University of Pennsylvania

9 All Software Patents by Year (Academic and Non-Academic) PTO Code Definition 341 Coded Data Generation or Conversion 345 Computer Graphics Processing 370 Multiplex Communications 706 Data Processing: Artificial Intelligence Data Processing: Database and File Management or 707 Data Structures Electrical Computers: Arithmetic Processing and 708 Calculating Computer-aided Design and Analysis of Circuits and 716 Semiconductor Masks Data Processing: Software Development, Installation, 717 and Management

10 Modeling Software Patent Growth Yt = βt + Ind.Collabt + Ac.Pub.2to5t + Ac.Pub.1t + t + εt where t = , Yt : number of software patents filed by the 23 universities indexed by year, Ind.Collabt : the number of patents with an industry researcher as co-author, Ac.Pub.2to5t : counts the number of patents that refer to an academic article within 2 to 5 years of the data of filing, Ac.Pub.1t : there was a referenced academic publication within a year of the patent filing date (ie. exempt from being classified as prior art if they are within one year of application date).

11 Model Coefficients for Academic Software Growth Independent Coefficient Standard p-value Variable Estimate Error βt Ind.Collabt Ac.Pub.2to5t Ac.Pub.1t t n=659

12 Model Interpretation Researcher collaboration with industry drives up patent numbers, related academic publication appear to have little effect, time is significant, ie. patenting levels increasing over time.

13 Model Improvements Include other covariates (non-software patents by university, dummies for each university, pervasiveness of computation in research, journal and funding agency code requirements,..), model time series directly, incorporate policy shocks, changes in TTO structure,?

14 Dual Licensing Distinguish between industry and academic research applications for licensing, Standard for academic research, in particular results associated with published results, and publicly funded research, Policy: Code for academic use simple to download, with associated licensing terms (no interaction with TTO, or TTO facilitates).

15 Other Legal Barriers to Code Transparency in Science HIPAA (Health Information Portability and Accountability Act) and privacy regulations, Copyright, Collaboration agreements with industry, Hiring agreements, institutional rules, National security.

16 References RunMyCode.Org: A Novel Dissemination and Collaboration Platform for Executing Published Computational Results, with C. Hurlin and C. Perignon, IEEE escience eson workshop, Sept Reproducible Research, Guest editor for Computing in Science and Engineering, July/August Enabling Reproducible Research: Open Licensing for Scientific Innovation Reproducible Research: Tools and Strategies for Scientific Computing, July 2011 available at