Last semester, I taught Comparative Intellectual Property Law in London, and I enjoyed the opportunity to think about different ways of structuring IP regimes. One of the more interesting differences is the use of jury trials in U.S.intellectual property litigation. Other countries are much less likely to have juries pass on such questions as the obviousness of an invention, the confusion created by different trademarks, or the similarity of two copyrighted works.
Whether juries are capable of making these determinations is ultimately an empirical question, and it is one that Jamie Lund from St. Mary’s University School of Law has sought to answer. Her recently posted paper on the “lay listener” test in music composition copyright cases suggests that our trust in juries may be poorly placed. I like her article, An Empirical Examination of the Lay Listener Test in Music Composition Copyright Infringement, lots.
Music copyright is a strange bird. When you hear a new song on the radio, that song is generally protected by two different copyrights, one in the underlying composition (composition copyright) and one in the particular recording of the composition (recording copyright). The composition copyright protects the author’s use of melody, harmony, rhythm, and lyrics to create a musical work, while the recording copyright protects the performer’s decisions regarding phrasing, style, genre, tempo, key, timbre, and orchestration. These are separate copyrights, often owned by different people, to which different sets of rights attach.
The bifurcated music copyright creates a number of difficulties, perhaps the most challenging of which is ascertaining whether a song by artist B infringes one or both or neither of the copyrights in a song by artist A. If A alleges that B violated the composition copyright of her song, how should the law determine whether B copied from A and whether B took “too much” of A’s song? The solution that copyright law typically adopts is the jury-centered “Lay Listener” test. Because the market for the music is the consuming public, courts believe that jurors are best positioned to determine whether the defendant took “so much of what is pleasing to the ears.” Arnstein v. Porter, 154 F.2d 464 (2d Cir. 1946).
The difficulty is that the lay public, and the jurors that come from it, have only indirect access to the underlying composition. Given the state of musical education, they are typically unable to read sheet music, so they only experience the composition through the recording. Courts have responded to this dilemma by allowing jurors to listen to the recorded versions of the plaintiff’s and defendant’s songs, usually unguided by expert evidence. But this means that some aspects of the musical work the jurors are hearing (those relating strictly to performance) are irrelevant to the task of comparing compositions. Jurors are asked to decide both whether the defendant actually copied from the plaintiff as opposed to independently creating the work or copying it from another source (the Copying in Fact inquiry) and, if so, whether the defendant’s copy of the song was “substantially similar” or took too much of the heart of the plaintiff’s song (the Substantial Similarity inquiry).
Lund suspected that jurors might not be particularly adept at answering these questions, and she set up a delightful study to test whether this was the case. Her main hypothesis was that aspects of the musical recording would unduly influence jurors’ likelihood of finding infringement of the musical composition. The study, in brief, compared the responses of two different sets of subjects to questions about Copying in Fact and Substantial Similarity between two songs that were actually litigated. The first set of subjects heard the two songs performed in a similar manner (tempo, orchestration, key, and style), while the other group heard the songs performed differently. Lund has posted the recordings to her website. Take a listen.
With one of the pairs of songs, Lund strongly confirmed her hypothesis. Subjects who heard the songs performed similarly were much more likely to report a higher degree of similarity between the compositions, higher likelihood of copying, and higher degree of substantial similarity. For example, when subjects heard the songs performed similarly, 86% of them thought the songs were substantially similar, but when they heard them performed differently, 85% thought the songs were not substantially similar. That’s quite a reversal! (Note that Lund’s data on a second pair of songs were not quite as striking, although this could have been due to order effects, the underlying similarity of the songs, or other experimental factors).
These findings should cause serious concern not just for music composition copyrights, but for the role of juries in IP cases more broadly. Lund discusses a number of possible solutions to the problem including using expert evidence, special verdict forms, or multiple recordings of the same piece of music. But as Lund herself notes, these tools may have little effect considering the poor understanding that jurors are likely to have regarding essential issues of a copyright lawsuits, such as the meaning of “originality” in copyright law and the relationship between litigated works and their public domain forbears. Maybe, and I never said this last semester, the U.S. should follow the French example and get rid of juries in IP cases. My other proposal is to have jurors watch this video before hearing the songs.