Perhaps the only doctrine of copyright law that its critics love is that of fair use.

Professor Barton Beebe from Cardozo law school recently presented his statistical research on 271 opinions from the federal courts between 1978-2005 which made “substantial use of the Section 107 four-factor test for copyright fair use.” The goal of Professor Beebe’s work is to find how the four factors are weighed, how they interact and drive the outcome of cases. The cases included seven from the Supreme Court, 77 from appellate courts and 187 district courts. 21% of the opinions addresses motion picture or television mediums, 12% of opinions between 1990-2005 involved computer software and 12% involved the Internet.

Professor Beebe states: “unclear whether fair use win rates” should be “considered disappointingly low or reasonably high.” 29% of 41 preliminary injunctions found fair use, while 31% of bench trials did. 83% found no fair use in the 23 where plaintiffs sought summary judgment with no cross-motion by the defense. Of 34 cases where defendants sought summary judgment, 77% found fair use.

Here is a summary of the four-factor fair use test in fair use and general conclusions of the presentation for each:

1. The Purpose and Character of the Use: has a strong correlation with the overall outcome of the fair use test.
2. The Nature of the Copyrighted Work: “a relatively large percentage of opinions that failed to consider factor two or found it to be not relevant.”
3. The Amount and Substantiality of the Portion Used: nearly all opinions that gave this factor to the defendant eventually ruled for the defendant on fair use.
4. Effect on the Market: this factor is “nearly decisive” whether it tilts for the plaintiff or defendant, with a slightly higher correlation with the plaintiff.

One of the main observations in the presentation is that courts treatment of the four-factor test makes it “especially amenable” to statistical analysis: “Courts often rehearse the platitude that the test is not to be applied mechanically, but the data suggest that over time, the test has indeed become more and more mechanical in application.” Of the opinions analyzed, 65% explicitly stated, “this factor favors/disfavors a finding of fair use.” 34% engaged in the practice of concluding Section 107 analysis with an explicit summary of how each factor was applied.

With 271 opinions, and four factors per opinion coded in for statistical analysis, Professor Beebe had 1084 assessments of how each factors (dis)favored fair use; only 5% were unclear. Further, while courts state the non-exhaustiveness of the Section 107 factors, and that “other factors may be considered,” only 15% of 271 cases considered additional factors; with half of them rejected the additional factor(s) as irrelevant or unpersuasive.

Correlations between factor one and factor four are “fairly strong,” suggesting that they are “closely-related considerations that drive the outcome of the test.” Outcomes under factor two correlate “very weakly with outcomes under the other factors, which lends further support to the common assumption that, in practice, factor two plays a largely peripheral role in the fair use analysis.”

According to Professor Beebe, courts tend to find that the first, third and fourth factors point in the same direction. He reasserts an observation by Professor Nimmer: “judges who uphold fair use almost always find that three, if not four, of the factors incline in its favor; judges who deny the fair use defense almost always find that three, if not four, of the factors incline against it.” This raises the question of stampeding.”

What is stampeding? Well, 37% of 187 district court cases found that all four factors favored the outcome of the test. Of the 113 cases where no fair use was found, 47% found all four factors favored the outcome. Stampeding was “less pronounced in opinions which found fair use or otherwise denied a plaintiff’s motion for summary judgment.” 65% of 187 district cases found that the first, third and fourth factors “favored the outcome of the test.” Professor Beebe summarizes his correlation of stampeding:

While I know of no statistical way to show that courts are indeed putting the cart before the horse when they engage in a Section 107 analysis, the strong evidence of stampeding is at least consistent with Nimmer’s description.

posted by Noel Le @ 4:03 PM | DMCA, Fair Use

Link to this Entry | Printer-Friendly | Email a Comment

TrackBacks (0)

Listed below are links to weblogs that reference Predicting Fair Use:

TrackBack URL for this entry: http://blog.pff.org/movabletype/mt-tb.cgi/3440