The Thicke case has me thinking about one my favorite/least favorite parts of copyright law: music copyright. There’s been much made of the fact that the Gaye copyright was limited to the sheet music. Now, I think that holding was probably right, resting as it does on the 1909 Act. But the way that music compositions and sound recordings are treated in copyright law has troubled me for some time. All this kerfuffle over the Thicke trial led me to revisit a paper I started some time ago but never managed to finish, in which I argue against the dual-copyright approach to music.
The rest of this post is just an extended excerpt, so read on only if you’re really into the more tortured details of music copyright.
The mere coexistence of two works on one object is not intrinsically troubling. The argument can be made (though I hope to debunk it shortly) that although the works are both copied in the information contained in a recording, those works are not coextensive: a sound recording consists in information that goes beyond mere composition. This line of reasoning, apparently embraced by the few courts to consider the question as well as by the Nimmer treatise, holds that there are certain aspects of a sound recording that can be considered pure acts of performance or production. These additions—timbre, sound engineering, etc.—differentiate a recording from a composition because they are not annotated in a traditional composition (that is, sheet music) and moreover, the suggestion is, such features are fundamentally incapable of annotation. Put more harshly, if it cannot be expressed by staff notation, it is not compositional.
The reality is far less cut and dry, and courts struggle when confronted with the need to distinguish “composition” from “recording.” The district court’s decision in Newton v. Diamond is illustrative of this confusion. In that case, the Beastie Boys had licensed a six-second sound recording of flutist James Newton’s song “choir,” but had neglected to secure the composition rights. Newton, who owned only the composition rights, sued for infringement. The court, in attempting to distinguish between the sound recording right and the composition right “filtered out” what it thought were elements intrinsic to the captured performance, as to leave behind only those elements intrinsic to the composition. After filtering, the court found that what remained was too elementary to reach the threshold of originality and therefore not protectable by copyright.
Filtering is not at all new to copyright infringement analysis. While not formally the law in all circuits, the act of filtering out unprotectable elements has been widely adopted, especially in the case of computer software and other highly functional or fact-based works. The problem with the court’s analysis in Newton is not that it chose to use filtration as a method (that much seems well-reasoned and defensible, if not legally required), but rather that it took the highly unusual step of attempting to filter a recording from a composition. To take that step requires certain assumptions, namely (a) that the protectable aspects of sound recordings and musical compositions are necessarily different and (b) that the two can be adequately distinguished. The court proceeded with its analysis without discussing the reasonableness of these assumptions, or even acknowledging their existence.
Copyright exists only for original works of authorship. That protected works must be original is a constitutional requirement and the “sine qua non” of copyright. Originality consists in independent creation and “some minimal degree of creativity.” It follows that any sound recording right must consist only in original works of authorship. While the originality requirement is modest and easily met, it remains essential to copyrightability. It is not clear that sound recordings have much basis for “originality” when the compositional elements have been entirely filtered out. This problem is exposed by conducting a filtering analysis that no court, to my knowledge, has ever had occasion to perform. That is, by doing the opposite kind of filtering than what was done in Newton, that is, filtering compositional elements out of “performance” elements, we should, in theory, be left with the pure, copyrightable heart of a sound recording. What remains must contain sufficient creativity to meet the constitutional standard for copyrightability, but it is not clear that it will do so in all cases. There is, of course, a great deal of creativity inherent in musical performance, but much, if not most or all of that creativity, inheres in elements that could only in fairness be considered compositional. A performance may alter rhythm, key, arrangement, and much more from the underlying composition, and those distinctly compositional alterations will create much, if not all, of what we find creative about the new recording. What is left? Nimmer points us to “distinctive voices” and “timbre” and nothing else. The House Report on the matter could not even get that far—it assumed the problem away, concluding:
As a class of subject matter, sound recordings are clearly within the scope of the ‘writings of an author‘ capable of protection under the Constitution, and the extension of limited statutory protection to them was too long delayed. Aside from cases in which sounds are fixed by some purely mechanical means without originality of any kind, the copyright protection that would prevent the reproduction and distribution of unauthorized phonorecords of sound recordings is clearly justified..
In the case of instrumental electronic music and MIDI files where timbre may be programmed and fundamentally duplicable, there might be no originality whatsoever outside of compositional elements. Despite the originality deficit that filtering exposes, no court appears to have ever found a sound recording unoriginal.
The Newton court never considered this problem in part because it found “performance techniques” to be sound recording elements rather than compositional elements. This finding is certainly mistaken. “Performance techniques” are well capable of being annotated as part of a composition, which are often littered with such annotations. For example, scores for stringed instruments can be replete with pizzicato, sul ponticello, sul tasto, tremolo, staccato, vibrato, etc., etc., ad infinitum. In fact, the central technique in Newton—singing into a flute as to produce multiple pitches simultaneously—was noted in the score. Other techniques the court found essential to the originality of the sound recording, but absent from the composition, were similarly capable of annotation and thus more properly considered compositional elements. Specifically, the court refused to consider portamento as a compositional element. It found that, “[a]s Plaintiff’s specific techniques . . . do not appear in the musical composition, they are protected only by the copyright of the sound recording.”
The court treats the line between composition and recording as a movable one. If a composer annotates a portamento, it is compositional. If he does not, it is part of the sound recording. It cannot be argued that this method of analysis embodies a principled distinction between composition and recording—it allows that one element of a song (performance technique) can be either recording or composition, but not both. How can this be?
Perhaps a more generous reading of the Newton decision could provide an answer. In that case, the performer and the composer were one and the same person. But, if the court thought them of as distinct persons (for the purpose of creating sound precedent), a motivating factor might emerge. Had the composer omitted the portamento and given the sheet music to the performer, who proceeded to add the portamento in creating a sound recording, it would seem only fair to credit the performer with the originality exhibited in the addition. The Newton court may have been afraid of carving away the protection of elements of creativity in performance, thereby giving a composer rights over that which he has not sown. This is not, however, how the law works.
Instead, the performer’s recorded rendition consists of two works derivative of the original composition: the performer’s derivative composition, and the performer’s derivative sound recording. That is to say that the performer’s compositional additions do not become part of the sound recording right merely because they were born of performance. There is already in the law a solution to this problem: performers’ compositional additions may be protected as compositional works where authorized by the rights holder in the original work. The difficulty lies in securing that authorization, which is unlikely to be given in many, or perhaps even most, instances.
A less troubling, but still much-criticized, decision was reached in Bridgeport Music, Inc. v. UMG Recordings, Inc. There, the Court of Appeals for the Sixth Circuit found that various elements of George Clinton’s “Atomic Dog,” which was composed and recorded more or less simultaneously, qualified for protection as part of the composition right even though they were not written in the sheet music. As in Newton, the plaintiff in Bridgeport only had composition rights in the song at issue. And, as in Newton, the court applied a filtration analysis. However, the Sixth Circuit’s filtration only acted to remove unprotectable elements from consideration and did not even attempt to filter the recording out of the composition. Instead, the court found quite simply that “the song was composed and recorded in the studio simultaneously and, therefore, . . . the composition was embedded in the sound recording.”
This holding troubles the authors of Nimmer on Copyright. The treatise opines, “It stretches matters too far to conclude that everything on the recording forms part of the musical composition.” This view is not unfounded. If composition and recording rights overlapped, then what purpose would the recording right serve? But, at the same time, a musical composition is simply annotation which, when “read,” produces sound. The 1976 Act makes it clear that machine-readable annotation—grooves in a record, binary encoding in an MP3—is still a copy of a musical composition even if it is not human-readable. Why should not all the information contained in those grooves and digits be considered compositional? And if it cannot or should not be, how can we draw a principled distinction between the two?
 Nimmer on Copyright, supra note 37§ 2.05[A].
 See Complaint, Newton v. Diamond, 2001 WL 34396090, ¶ 34 (C.D. Cal.).
 Id. at 1256. It is worth noting that this author is not troubled by the result in Newton which reflects very real de minimis concerns, but rather by some of the more troubling technicalities of the court’s reasoning.
 See Nimmer on Copyright, supra note 37§ 2.05[A]. The leading case on filtration explicitly borrows the method from the Nimmer treatise. Computer Associates Int’l, Inc. v. Altai, Inc., 982 F.2d 693, 707 (2d Cir. 1992).
 I use the word “unusual” advisedly. My research identifies only three courts that have taken this approach: the district court in Newton, the Ninth Circuit in Newton’s appeal, and the southern district of New York in Poindexter v. EMI Record Group Inc., 11 CIV. 559 LTS JLC, 2012 WL 1027639 (S.D.N.Y. Mar. 27, 2012) (slip opinion). The Poindexter opinion does not explicitly cite any authority for its filtration analysis, but relies heavily on the Newton district court opinion for its understanding of the composition/recording dichotomy. See Poindexter, at n. 3.
 17 U.S.C. § 102(a).
 U.S. Const. art. 1, § 8, cl. 8; see also Feist Publications, Inc. v. Rural Tel. Serv. Co., Inc., 499 U.S. 340, 346 (1991).
 Feist, 499 U.S. at 345.
 See L. Batlin & Son, Inc. v. Snyder, 536 F.2d 486, 490 (2d Cir.1976).
 See Nimmer on Copyright, supra note 37§ 2.05[A].
 House Report No. 94-1476, H.R. REP. 94-1476, 56, 1976 U.S.C.C.A.N. 5659, 5669.
 For just some of the notation that may be used in writing for stringed instruments, see Eric Starr, General Bowing Techniques, netplaces, http://www.netplaces.com/music-composition/writing-for-strings/general-bowing-techniques.htm (last visited Apr. 30, 2012).
 Newton, 204 F. Supp. 2d at 1250 n. 3.
 However, many derivative arrangements are made under the terms of the compulsory licensing provisions of § 115 and its private sector counterparts, which do not provide for copyrights in arrangements made under a compulsory license. See 17 U.S.C. § 115(a)(2).
 585 F.3d 267, 275 (6th Cir. 2009)
 The vocal tracks were recorded spontaneously by George Clinton who had “been partying pretty heavily” and required the support of two co-songwriters to keep him in front of the microphone. Id. at 272–3.
 Id. at 274.
 Id. at 276,