Citation: Matthew Sag, Predicting Fair Use, 73 Ohio State Law Journal 47 (2012)
In a nutshell:
In Predicting Fair Use, Matthew Sag argues that fair use is far more predictable than its critics assume, presenting an empirical study of more than 280 district court cases showing that transformative use and partial copying are significant predictors of fair use outcomes, while commercial use and the nature of the copyrighted work are not.
Summary
Predicting Fair Use confronts the familiar charge that fair use is a lottery, or as Lawrence Lessig put it, merely “the right to hire a lawyer.” Where earlier empirical work by Barton Beebe studied the reasoning of fair use opinions after the fact, this study asks a different question: can the outcome of a fair use case be predicted from objective characteristics of the dispute that are apparent to the litigants before trial? Sag built a dataset of more than 280 fair use decisions from U.S. District Courts between 1978 and 2011, supplemented with information about the parties and attorney ratings from the Martindale-Hubbell directory, and reduced doctrinal claims about fair use to twelve testable hypotheses covering the statutory factors, the “underdog” theory of fair use, and industry effects.
The regression results confirm the centrality of transformative use. Sag operationalizes transformativeness as “Creativity Shift,” coded where the plaintiff’s work is creative and the defendant’s is informational or vice versa. A creativity shift nearly doubles the predicted probability of a fair use win, from 33% to 62%; adding partial copying raises it to 68%. Direct commercial exploitation of the plaintiff’s work cuts the predicted probability to 29%. By contrast, commercial use in the general sense, the creative or unpublished nature of the plaintiff’s work, and industry separation between the parties have no measurable effect on outcomes, evidence that district courts have quietly abandoned the Sony presumption against commercial fair use.
The article also tests the common characterization of fair use as a subsidy favoring politically and economically disadvantaged users. The data point the other way: defendants win more often when the plaintiff is a natural person rather than a corporation, and defendants with less experienced counsel than their opponents do worse, not better. Sag concludes that fair use is an integral part of the copyright system, as available to Fortune 500 companies building search engines as to struggling artists.
Why read this article?
Predicting Fair Use is a useful primer on the four statutory fair use factors and the case law elaborating them, including a clear account of the difference between a transformative use and an infringing derivative work (illustrated with Pride and Prejudice and Zombies) and of how courts stretched the transformative use label from parody in Campbell to recontextualization and nonexpressive uses like visual search engines. The article also shows how contested doctrinal propositions can be translated into testable hypotheses, and it is candid about the limits of the method, with an extended discussion of selection effects and circuit-level variation in win rates. The concluding section connects the predictability question to comparative copyright policy through the U.K.’s Hargreaves Review, which declined to recommend a U.S.-style fair use exception partly on uncertainty grounds. A statistical appendix documents the case selection, variables, and alternative model specifications.
Further Reading
Barton Beebe, An Empirical Study of U.S. Copyright Fair Use Opinions, 1978–2005, 156 University of Pennsylvania Law Review 549 (2008) – The first empirical study of fair use case law, coding how judges apply the four statutory factors in their written opinions; Sag’s study builds on Beebe’s work but shifts the focus from judicial reasoning to pretrial case characteristics.
Pamela Samuelson, Unbundling Fair Uses, 77 Fordham Law Review 2537 (2009) – This article sorts modern fair use jurisprudence into policy-relevant clusters, arguing that the case law is more coherent and predictable than commonly believed, a conclusion Sag’s statistical analysis independently supports.
Wendy J. Gordon, Fair Use as Market Failure: A Structural and Economic Analysis of the Betamax Case and Its Predecessors, 82 Columbia Law Review 1600 (1982) – The classic law and economics account of fair use as a response to market failure and high transaction costs, one source of the underdog intuition that Predicting Fair Use tests and rejects.
Pierre N. Leval, Toward a Fair Use Standard, 103 Harvard Law Review 1105 (1990) – The article that introduced transformative use, later adopted by the Supreme Court in Campbell; Sag’s data show that the concept Leval proposed has become the strongest statutory predictor of fair use outcomes.