An empirical study of U.S. copyright fair use opinions, 1978-2005.

• Author: Beebe, Barton

Section 107 of the Copyright Act of 1976 establishes the affirmative defense to copyright infringement of "fair use," by far the most enigmatic doctrine in U.S. copyright law and by far the most important. Without it, much of our economic and communicative action would constitute copyright infringement. Yet despite the importance of the fair use defense, and despite the enormous amount of scholarly attention that it has received, we continue to lack any systematic, comprehensive account of our fair use case law and the actual state of our fair use doctrine. Instead, our conventional wisdom derives from a small set of conventionally agreed-upon leading cases. This Article presents the results of the first empirical study of our fair use case law to show that much of our conventional wisdom about that case law is mistaken. Working from a data set consisting of all reported federal opinions that made substantial use of the section 107 four-factor test for fair use through 2005, the Article shows which factors and subfactors actually drive the outcome of the fair use test in practice, how the fair use factors interact, how courts inflect certain individual factors, and the extent to which judges stampede the factor outcomes to conform to the overall test outcome. It also presents empirical evidence of the extent to which lower courts either deliberately ignored or were ignorant of fair use doctrine set forth in the leading cases, particularly those from the Supreme Court. Based on these descriptive findings, the Article prescribes a set of doctrinal practices that will improve courts' adjudication of the fair use defense.

INTRODUCTION I. BACKGROUND A. Section 107 of the Copyright Act of 1976 B. Courts' Mechanical Application of the Four-Factor Test II. SUMMARY STATISTICS A. Distributions of the Opinions 1. Distribution of the Opinions by Year and Venue 2. Distribution of District Court Opinions by Posture 3. Distribution of the Opinions by Subject Matter B. Reversal, Dissent, and Appeal Rates C. Fair Use Win Rates III. INTERFACTOR ANALYSIS A. Correlation and Regression Analysis B. Word Count Analysis C. Stampeding IV. INTRAFACTOR ANALYSIS A. Factor One: Purpose and Character of the Use 1. The Commerciality Inquiry 2. The Transformativeness Inquiry 3. The Bad Faith Inquiry 4. The Preambular Purposes Inquiry B. Factor Two: Nature of the Copyrighted Work 1. The Creative/Factual Work Inquiry 2. The Published/Unpublished Work Inquiry C. Factor Three: Amount and Substantiality of the Use D. Factor Four: Effect on the Market V. CONCLUSION APPENDIX: THE COLLECTION AND CODING OF THE OPINIONS INTRODUCTION

Section 107 of the Copyright Act of 1976 establishes the affirmative defense to copyright infringement of "fair use." (1) This affirmative defense represents the most important--and amorphous--limitation on the otherwise extraordinarily broad rights granted to copyright owners under section 106 of the Act. (2) Much of our economic and communicative action would constitute infringement of those rights but for the defense of fair use. It is the ill-defined and unpredictable point of flexibility, the "dangerous supplement," (3) that makes the rest of our copyright law possible--and bearable. (4) Section 107 offers no definition of "fair use," and none is offered elsewhere in the Act. The section does, however, enumerate four broadly worded factors that courts "shall" consider in determining whether a use is "fair" and thus noninfringing:

(1) the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes; (2) the nature of the copyrighted work; (3) the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and (4) the effect of the use upon the potential market for or value of the copyrighted work. (5) For better or worse, these factors form the core of our fair use doctrine and functionally define what fair use is. For all of the ambiguity of their statutory language, they are what stand between us and what some have called, with only a touch of hyperbole, the "tyranny of copyright." (6)

Section 107, and the concept of fair use more generally, have attracted an enormous amount of scholarly attention, ranging from broad social-theoretical treatments (7) to precise analyses of the leading cases and their impact. (8) Nearly all of this commentary has been highly critical of section 107's four-factor test and how courts have applied it. (9) Yet, remarkably, we continue to lack any systematic, comprehensive account of our fair use case law. (10) Instead, like the "great men" (11) approach to history, we pursue a "leading cases" (or "usual suspects") approach to fair use. This anecdotal method, one essentially of connoisseurship, derives conventional wisdom about our case law from a limited aristocracy of hand-picked opinions appearing primarily in the U.S. Reports--or in the student casebooks. (12) Whether these opinions have any influence on or are representative of the true state of our fair use doctrine as it is practiced in the courts remains an open, and strangely unasked, question.

To answer this and other questions, this Article presents the results of the first empirical study of our fair use case law. (13) Working from a data set consisting of all reported federal opinions that made substantial use of the section 107 four-factor test from the January 1, 1978, effective date of the Copyright Act through 2005, the Article shows that much of our conventional wisdom about our fair use case law, deduced as it has been from the leading cases, is wrong. The dual purpose of the Article is systematically to induce from the population of section 107 opinions what our fair use doctrine actually is in practice, and critically to induce from that population of opinions what it ought to be.

The Article proceeds as follows. Part I provides background on section 107 and identifies in the data the curious origins of the highly routinized fashion in which judges typically apply its four factors. Part II sets forth summary statistics on the 306 opinions studied, including their distribution along the dimensions of time, venue, posture, and the kind of copyrightable work at issue. The Part further reveals the full extent to which the Second and Ninth Circuits dominate our fair use case law. It also identifies and seeks to explain the remarkable increase in the prevalence of fair use summary judgment opinions that began in the mid-1990s and has continued to the present, and the just as remarkable absence of fair use bench trial opinions during the same period. It then reports the reversal, dissent, and appeal rates in the fair use case law and finds that notwithstanding strong conventional wisdom to the contrary, none of these are especially high as compared to other areas of law. Finally, and perhaps most interestingly, the Part sets out the fair use win rates (i.e., the proportion of opinions that found fair use) in the case law. For certain postures, these rates were exceptionally low. The Part seeks to explain why.

Parts III and IV then focus on how judges used the four-factor test. Part III engages in an interfactor analysis of the test as applied. It focuses on the factor outcomes and shows how these outcomes interacted with each other and with the overall test outcome. Correlation and regression analyses reveal the surprising extent to which the outcomes of the first and especially the fourth factors appear to drive the outcome of the test. The Part also reports the results of a word count analysis of the opinions across time. It concludes by considering the extent to which judges "stampeded" (14) the factor outcomes to conform to the overall test outcome. The data show that, again, notwithstanding strong conventional wisdom to the contrary, and notwithstanding the examples of certain leading cases, judges do not generally stampede the factor outcomes. Rather than make a fair use determination first and then "align the four factors to fit that result as best they can," (15) courts appeared quite willing to call the factor outcomes as they saw them, even when those outcomes did not support the overall test outcome. This is a highly encouraging finding.

Part IV then turns to an intrafactor analysis of the fair use test as applied. It looks inside each of the four factors to determine which subfactor considerations animated courts' determinations of the factor outcomes and, through those factor outcomes, the outcome of the overall fair use test. Regression analysis, as well as simple descriptive statistics, shows that certain subfactor findings were far more important to the outcome of the fair use test than is generally thought, and that others thought to be important actually had no significant influence on the outcome of the test. For example, a finding that the plaintiff's work was factual in nature or published appeared strongly to influence a court to find fair use, (16) as did a finding that the defendant's use was for a noncommercial purpose. (17) Meanwhile, the data suggest that a finding that the defendant's use was for a commercial (rather than noncommercial) purpose had no significant influence on the outcome of the test, and a finding that the defendant used the entirety of the plaintiff's work was far from dispositive. More generally, and more disturbingly, the Part shows the extent to which lower courts repeatedly either deliberately ignored or were ignorant of basic, albeit dictic, Supreme Court precedent. Indeed, the data show that the lower courts repeatedly and systematically inverted Supreme Court dicta to favor the defendant, so that if the Court stated, for example, only that "not x" favors the plaintiff, the primary lesson the lower courts would draw from this is that "x" favors the defendant. This may come as a pleasant surprise to fair use maximalists, who are generally pessimistic about the...