Fair Use and Sound Recordings: Lessons from Community Practice

COPYRIGHT AND FAIR USE GENERALLY

The purpose of copyright is specified in the constitution: “to promote the Progress of Science,” a term which at the time meant learning and culture generally. As Justice O’Connor wrote for the Supreme Court majority in the landmark Feist decision, “The primary objective of copyright is not to reward the labor of authors, but ‘[t]o promote the Progress of Science and useful Arts.’”¹⁰ As we have already seen, sometimes the copyright bargain breaks down and the statutory monopoly undermines the very progress it is meant to promote. Fair use trumps copyright protection in these circumstances. Based on years of judge-made law, Congress codified the fair use doctrine using four non-exclusive factors that judges (and others) are directed to consider:

1. The purpose and character of the use, including whether the use is commercial or for nonprofit, educational purposes
2. The nature of the work used
3. The amount and substantiality of the portion of the work used
4. The effect of the use on the market value of the work

Courts are to weigh these factors together in light of the purposes of copyright. Courts today routinely look to a powerful concept known as transformative use as a key part of the fair use determination. The idea of transformative use was first articulated in 1990 by a federal judge named Pierre N. Leval,¹¹ and it was strongly endorsed shortly afterward by the Supreme Court, which called transformative uses “the heart of fair use.”¹² A use is transformative if it takes existing copyrighted material and uses it for new, socially beneficial purposes. While some transformative uses involve altering or editing the underlying works, courts have also endorsed uses where entire works are used without alteration for dramatically new purposes. Search engines, for example, often require copying millions of works to create a useful search index.¹³ Books about historical subjects sometimes reproduce full images to illuminate their subject.¹⁴ When a use is for a transformative purpose, courts ask whether the amount of the original work used is appropriate to that purpose. Courts overwhelmingly find uses fair where they find the purpose is transformative and the amount taken is appropriate to the purpose.¹⁵

Legal scholars have shown that courts tend to care about what is “normal” in a given practice community. When communities routinely acquiesce to license demands, courts may take this practice as evidence that such uses are generally not fair and that payment is therefore appropriate (Gibson 2007). On the other hand, where courts see an established norm or practice of making unlicensed uses, they may infer that such uses are fair and legitimate and pose no special threat to copyright holders.¹⁶ Facts about custom and practice are not decisive, of course. Courts know that sometimes licenses are sought for reasons of expedience or risk management rather than genuine legal obligation,¹⁷ and that sometimes licensing is avoided for reasons other than the user’s justified entitlement. Nevertheless, evidence that a practice is broadly approved and justified in light of community mission will favor fair use in the right circumstances. This report is an effort to bring together the considered opinions of related communities about activities that recorded sound collections would like to pursue.

¹⁰ Feist Pubs., Inc. v. Rural Tel. Svc. Co., Inc., 499 U.S. 340, 349 (1991).

¹¹ See Leval 1990.

¹² Campbell v. Acuff-Rose, 510 U.S. 569, 579 (1994).

¹³ See, e.g., Authors Guild v. HathiTrust, 755 F.3d 87 (2d Cir. 2014).

¹⁴ See, e.g., Bill Graham Archives v. Dorling Kindersley Ltd., 448 F.3d 605 (2d Cir. 2006) (reproducing full color concert posters and tickets as part of Grateful Dead timeline was fair use); Warren Pub. Co. v. Spurlock, 645 F.Supp.2d 402 (E.D. Pa. 2009) (reproducing complete magazine covers in biography of artist was fair use).

¹⁵ The ascendance and present dominance of the transformative use framework is explained in detail in Netanel 2011.

COMMON MYTHS AND MISUNDERSTANDINGS ABOUT FAIR USE

Before pursuing this approach, we should ask what direct guidance, if any, may be available from decided fair use cases. Certainly, court decisions help put to rest some common myths and misunderstandings that recorded sound collections, like memory institutions in general, may have about relying on fair use.

One common misunderstanding of fair use is that the key criterion for fairness is the amount of the work that is used, and that the amount must be slight to qualify as fair use. In reality, whenever courts consider the extent of unlicensed use as a component of fair use analysis, as they are directed to do by the statute, the inquiry is contextual. And, contrary to the “folk wisdom” of copyright, there is no rule that automatically bars the use of works in their entirety. Instead, where uses are transformative, courts will look to the purpose of the use to determine whether the amount is appropriate. Thus, for example, a teacher who insists that students in a course on the history of popular music listen to a selection of recordings, in their entirety, in preparation for class may have a strong fair use claim, especially if this is recognized as an appropriate instructional practice among teachers.

Another source of concern for would-be fair users is the widely held belief that unpublished works (a category that includes some musical and a larger number of non-musical recordings) have a fundamentally different and more protected status under the law than published ones. While there was some basis for this distinction years ago, when publication was a more significant legal event in the copyright system, there is very little reason to treat unpublished materials substantially differently for fair use purposes today.¹⁸ The key, as always, is the relationship between the nature of the work and the user’s purpose. A work’s unpublished status may weigh against fair use if the purpose is purely exploitative or frivolous (so that the author’s interest in keeping the work obscure is not outweighed by the user’s or the public’s interest in publication). It may favor fair use if the purpose is scholarly or critical and the public interest in learning from new information trumps the author’s decision to keep information hidden.

Likewise, there is no basis in fact for the proposition that because many recordings were originally produced for entertainment purposes, they are somehow shielded from fair use. While courts do consider whether a work is primarily factual or creative, the key is context. If the user is “covering” a copyrighted song discovered on an old record, the fact that both the original and the ultimate purpose of the use were for entertainment may be highly relevant. By contrast, a scholar using a historical popular song to make points about cultural history has little to be concerned about on this score.¹⁹ And it is worth noting that when considerations such as whether the work is unpublished or creative in nature do come up, it typically is in connection with the second statutory factor (“The nature of the work used”), which is typically marginal in its effect on the final outcomes of fair use cases.²⁰

Finally, memory institutions often worry needlessly that they could be liable for others’ downstream uses when they provide access to works in their collections. For example, if someone obtains a copy of a recording from a sound recording collection under false pretenses then sells CDs of it on eBay, would the collection be liable somehow if the sales constitute copyright infringement? Put simply, it is almost impossible that a collector, an institution, or an employee of an institution could be held responsible for the bad acts of others who abuse their access to copyrighted materials. Legal doctrines of secondary liability require control over the infringing activity, knowing encouragement of the activity, or direct profit from the activity, none of which is likely to apply to downstream uses of materials made available for free or at cost under a defensible fair use rationale.

¹⁸ Following a series of court decisions suggesting fair use may never apply to unpublished works, Congress amended the fair use provision of the Copyright Act to add the savings clause: “The fact that a work is unpublished shall not itself bar a finding of fair use if such finding is made upon consideration of all the above factors.”

¹⁹ See, e.g., Sag 2012 (observing that cases involving “creativity shift”—“where the plaintiff’s work is creative and the defendant’s is informational, or vice versa”—are very likely to be ruled fair).

²⁰ For more on the oft-overlooked second fair use factor, see Kasunic 2008. Kasunic argues persuasively that the second factor could play a more compelling role in fair use decisions if courts explored the nature of the work with more nuance, but so far the courts have not taken his advice.

WHAT THE CASELAW TELLS US

Unfortunately, the courts have little direct guidance to offer on the specific question of the scope of fair use for recorded sound. Certainly, federal judges have heard a variety of cases regarding use and reuse of sound recordings, but the vast majority of these cases have to do with commercial infringement by competing musicians, commercial “pirates,” or online platforms. In most of these cases, fair use has not been a serious issue. The most that a recorded sound collection can learn from them is that it is not fair use to copy and distribute musical recordings in ways that serve general consumer interest in commercial recordings in direct competition with copyright holders.

On the other hand, it is clear that contemporary trends in fair use jurisprudence do apply, with full force, where sound recordings are concerned. Thus, a recent decision from the Second Circuit Court of Appeals, which is widely respected as an “expert” circuit on copyright matters, found that a financial news agency had wide fair use rights to distribute the full recorded audio of a newsworthy earnings call, even though the copyright owner had expressly forbidden redistribution.²¹

Such cases “on point,” however, are few and far between. The guidance the courts have to offer is substantial, but largely indirect. Another way for recorded sound collections to understand how fair use will apply to their activities is to look at analogous uses by similar groups of similar materials. Thus, in what follows, we will discuss analogies derived from judicial caselaw in the context of best practices statements formulated by curatorial communities with which recorded sound collections share interests and concerns.

One question remains: Does fair use, a doctrine developed in federal court, apply to claims relating to the use of pre-1972 sound recordings in potential violation of state law? Only one court has considered this question explicitly, but the affirmative answer it gave was definite enough to command respect.²² Moreover, the logic of the decision seems inevitable: the Supreme Court has said that fair use is a limit on copyright that is required by the First Amendment of the U.S. Constitution.²³ Since no state law can trump the Constitution, a court hearing a state copyright dispute should apply some version of the fair use exception as a limitation to state law just as it would in a case involving federal copyright law. Indeed, courts applied fair use as a judge-made limitation to copyright for more than a century before the doctrine was expressly codified in federal law in 1976. State courts currently apply a similar “transformative use” test as a First Amendment defense against state law right of publicity claims.²⁴ They should apply the fair use doctrine to pre-1972 sound recordings in the same way, and with an eye to the way federal courts have done so.

21 Swatch Grp. Mgmt. Servs. Ltd. v. Bloomberg L.P., 756 F.3d 73 (2d Cir. 2014).

22 EMI Records Ltd. v. Premise Media Corp., 2008 N.Y. Misc. Lexis 7485 (N.Y. Sup. Ct. Aug. 8, 2008) (applying fair use to state common law copyright).

23 See Eldred, 537 U.S. at 190.

24 See Bartholomew and Tehranian 2013.

SOME GENERAL CONSIDERATIONS—INDICIA OF GOOD FAITH

Before exploring fair use considerations that are particular to specific recurring situations, we would like to highlight a suite of core fair use best practices that recur across a range of situations. These practices can be characterized broadly as practices indicative of good faith. Some courts have made much of this notion where would-be fair users are seen to have “unclean hands” because their activity involved some other wrongdoing,³⁴ but more recent cases have been less concerned with this consideration.³⁵ Nevertheless, communities of practice that develop their own standards of good practice in fair use have repeatedly emphasized the importance of acting in good faith as an aspect of satisfying professional norms. Accordingly, many of the best practices documents include requirements or strong recommendations that fair users engage in a series of good faith practices when fair use is invoked to make copyrighted materials available to all or some members of the public.

The first of these practices is that, wherever possible, materials used fairly should be properly attributed, according to conventions of the field. We see this articulated in virtually every context and by virtually every community, from libraries to online video makers to documentary filmmakers to poets. The practice of attribution shows good faith by giving credit, directing interested parties to source material, fostering future scholarship and reuse, and helping to put creators themselves on notice when their works are incorporated in new uses. Research also shows that “creators are willing to sacrifice significant economic payments in favor of receiving attribution for their work” (Sprigman et al. 2013). Attribution thus helps ensure that the fair use bargain is not entirely one-sided by giving creators a genuine and valuable benefit.³⁶

Another very common practice across communities is to show due consideration to concerns about privacy and other potential harm to vulnerable third parties who may be affected by a planned use. Previously unpublished or obscure materials may contain information that could be damaging if revealed, and many fair uses can go forward without making unnecessary disclosure of this kind of information. Archivists, librarians, scholars, and teachers have all agreed that, where possible, materials that are made public on the basis of fair use should be screened to avoid unnecessary harm of this kind.³⁷

A third common practice is to provide a mechanism to get feedback from interested parties (copyright holders, performers, subjects, family members, and so on). This need not be a “notice-and-takedown” mechanism,³⁸ per se, and in fact many communities have found that this engagement is just as likely to be positive (yielding expressions of appreciation for the project, offers to provide related materials) as it is to be negative (takedown requests, privacy concerns, and the like). The important thing is to make it easy for anyone with questions or concerns about a given project to communicate with someone on the project team.

A fourth general practice is to provide users with some basic information about copyright and fair use as they apply to the kinds of uses you intend to foster (for example, teaching, scholarship, or study) by making materials available. This typically includes notice that users are responsible for their own downstream uses of materials. These notices and educational efforts help foster a healthy respect for copyright as well as awareness of the purpose and value of fair use.

³⁴ See, e.g., Harper & Row v. The Nation Enters., Inc., 471 U.S. 539 (1985).

³⁵ See Campbell, 510 U.S. at 585 n. 18 (leaving the relevance of good faith an open issue). Academics have argued that good faith is irrelevant to fair use. See, e.g., Leval 1990, 1126 (“The [fair use] inquiry should focus not on the morality of the secondary user, but on whether her creation claiming the benefits of the doctrine is of the type that should receive those benefits.”); Frankel and Kellog 2013.

³⁶ Of course, for some orphaned works full attribution will be difficult or impossible; all that is required is that the user make reasonable effort.

³⁷ An example that came up more than once in our conversations with recorded sound collections professionals was ethnographic recordings of spiritual rites and similar culturally sensitive materials.

³⁸ A part of the Digital Millennium Copyright Act requires certain online intermediaries to follow very strict “notice-and-takedown” procedures in order to avoid liability for the acts of their users. To comply with these procedures, a service provider must immediately disable access to user-posted material when they receive a compliant notice. These provisions, codified at 17 U.S.C. § 512, do not apply to recorded sound collections and others who use materials themselves, but many have adapted analogous procedures as a safety valve to give third parties a way to register concern without resorting to more formal, high-stakes measures.

ELECTRONIC ACCESS TO RARE/UNIQUE MATERIALS FOR OFF-SITE RESEARCHERS/USERS

Interviewees from recorded sound collections told us repeatedly that one of the most important things they could do to better advance their various institutional or personal missions would be to make the rare or unique items in their collections more readily available to remote users and researchers. Practical barriers to this activity are lower than ever thanks to the growth and power of the Internet, together with the explosion of service and technology providers catering to storage and streaming of even large, high-quality files. The connection to mission is clear, as recorded sound collections exist to provide access to culture.

The legal barriers, however, can seem daunting at first glance. More than any other activity, provision of access to remote users may seem at first to resemble high-profile cases involving online file sharing. On closer inspection, however, the superficial resemblance dissolves.

First, the courts have made it clear that large-scale file sharing is a disfavored activity, and that fair use cannot be invoked to justify it. By contrast, however, the Copyright Act actively favors the provision of research copies to students and scholars. Sections 108(d) and (e) of the Copyright Act allow libraries to provide patrons with copies of works where the library reasonably believes unused copies of the work are no longer commercially available at a “fair price.” While these provisions are helpful where they apply, their application is limited: they exclude musical works, for example, and they only apply to libraries whose collections are open to researchers in general. Still, the existence of this exception shows that Congress looks favorably upon provision of copies for research in appropriate circumstances. This, in turn, supports claims of fair use.³⁹ Courts have blessed similar claims in the context of unpublished manuscripts,⁴⁰ and should be equally open to fair use claims for recorded sound materials.

Second, and crucially, the sound recordings that our interviewees seek to make available are almost by definition materials that have no commercial value, and may never have been commercial objects. These recordings are sufficiently old, rare, ephemeral, personal, technical, arcane, and so on, that no appreciable segment of consumers is interested in paying for access to them, and consequently no market actor has an interest in making them available for a reasonable price. Academic studies show that even for music that was at one time popular, the vast majority of sound recordings quickly become unavailable from commercial outlets.⁴¹ As several interviewees told us, it is in these circumstances, where commercial markets have truly failed, that the imperative to provide access to researchers is strongest.

The provision of copies to remote individual users for research and study, considered as a possible fair use, has been a recurrent topic of best practices codes over the last decade. Principle Four of the Code of Best Practices in Fair Use for Academic and Research Libraries (“Libraries’ Code”) provides that, “It is fair use to create digital versions of a library’s special collections and archives and to make these versions electronically accessible in appropriate contexts” (20). The Statement Of Best Practices In Fair Use Of Collections Containing Orphan Works For Libraries, Archives, And Other Memory Institutions (“Orphan Works Statement”) states that “providing access to their collections [is] the ultimate goal of all [memory institution] activities” and that “fair use [is] available to them for this purpose” (27). It goes on to describe a series of detailed best practices for providing access to collections materials, addressing acquisition, seeking clearances, and more. The Statement of Best Practices in Fair Use of Dance-Related Materials (“Dance Heritage Statement”) provides that “Furthermore, where Dance Collections can assist academic efforts from afar by making scarce, oneof-a-kind, or out-of-print resources available through secure electronic, streamed, or other digital technology, or conventional models of information delivery, these efforts fall within fair use” (14–15).

The strongly positive community norms around this activity share several key characteristics. One is that the fair use case will be much stronger where the works are not commercially available or were never intended for commercial exploitation. For example, the Libraries’ Code cautions that, “Providing access to published works that are available in unused copies on the commercial market at reasonable prices should be undertaken only with careful consideration, if at all” (20). Wise use of technology to shape access, while not always required, was recommended as a way to ensure that access is commensurate with a legitimate purpose. Streaming might be preferred to downloading, for example, or lower quality downloads over higher, depending on context.

39 See Band 2012.

40 Sundeman v. Seajay Soc’y, Inc., 142 F.3d 194 (4th Cir. 1998) (finding fair use where collector provided a copy of the full text of an unpublished manuscript to a scholar in order to facilitate scholarship without endangering the fragile original).

41 See Brooks 2005 (finding that “On average, rights owners have made available 14 percent of the historic recordings that they control” for a sample of recordings released between 1890 and 1964).

ELECTRONIC ACCESS TO COLLECTED MATERIALS FOR AFFILIATED STUDENTS AND INSTRUCTORS IN SUPPORT OF TEACHING

In addition to supporting research and promoting interest in collections using themed exhibits, interviewees expressed interest in finding ways to support teaching by making collected recordings available to students in relevant classes at educational institutions. For example, university libraries frequently support courses by placing professor-selected collections of recordings on electronic reserve sites available only to authenticated enrolled students. This functionality was seen as a natural part of a library’s traditional mission to support teaching and learning by collecting materials that would be useful to students as a complement to textbooks and other materials that students traditionally purchase for class. While interviewees showed particular enthusiasm for making rare or unique holdings available for this purpose, there was also interest in making items from the broader collection available in support of teaching (as, for example, documentation of trends in popular music).

Section 110(1) of the Copyright Act substantially immunizes the use of copyrighted materials in connection with live classroom teaching from copyright liability. Many other teaching practices that recorded sound collections can support, especially those enabled by new technology, are not covered, however. Fortunately, “Teaching” is one of the examples of a core fair use that is mentioned in the preamble to Section 107, and courts have cited the importance of educational use as a compelling “purpose” in fair use analysis. Indeed, its centrality is so widely assumed that—in practice—rights holders seldom challenge educational uses.

Recently, however, courts have had a chance to weigh in on fair use in this context in two cases, AIME v. UCLA⁴² and Cambridge Univ. Press v. Patton (the Georgia State e-reserves case),⁴³ both with relatively positive outcomes for educators that were consistent with the best practices. In the UCLA case, an association of video vendors sued the university for copyright infringement in connection with its practice of making films available securely online to authenticated students, via a streaming media server, when their professors had assigned those films for class. The case was not decided on fair use grounds because the films at issue had been licensed to UCLA in a way that the court found allowed streaming, but along the way the court did acknowledge that there was at least a reasonable argument to be made that streaming the films was fair use.

The decision in Cambridge Univ. Press is much more detailed, and focuses almost exclusively on fair use. The very specificity of its focus, however, limits its value as a source of general guidance. On one hand, the question of whether the particular uses involved (excerpts from scholarly monographs posted for students at instructors’ direction) are fair is discussed in extensive detail—hundreds of pages across the district court and appellate court decisions. However, both decisions depend to a large extent on the finding that while these materials may have been written and published with an “academic market” in mind, as a practical matter, no mechanism existed by which most of the actual uses involved in the case could be licensed. Although the district court opinion found the majority of Georgia State University’s uses were fair, the appellate decision was much more ambivalent, announcing an intentionally vague set of standards that gives little indication of how lower courts should evaluate similar cases in the future. Be that as it may, much of the reasoning was entirely consistent with that of the best practices documents insofar as the court urged caution where educational materials are being used for educational purposes. As of this writing, the case has been sent back to the district court for a second round of fair use evaluation.

Many of the communities that have devised fair use best practices have a strong direct or indirect interest in enabling teaching from primary sources. Thus, because research libraries support the teaching missions of their institutions, the Libraries’ Code clearly favors this practice, subject to important qualifications and limitations.⁴⁴ The Code of Best Practices in Fair Use for Media Literacy Education (“Media Literacy Code”), which deals primarily with the K-12 classroom, also expresses strong support for exercising fair use to make all kinds of media “available to learners, in class, in workshops, in informal mentoring and teaching settings, and on schoolrelated Web sites” (10) and affirms that, “Under fair use, educators…can integrate copyrighted material into curriculum materials, including books, workbooks, podcasts, DVD compilations, videos, Web sites, and other materials designed for learning” (11). The OCW [Open CourseWare] Code describes several contexts in which teachers can use copyrighted works online in support of teaching. The Dance Heritage Statement also affirms that providing electronic access in support of teaching is a fair use (14– 15). The Society for Cinema and Media Studies’ Statement of Best Practices for Fair Use in Teaching for Film and Media Educators (SCMS Teaching Statement) includes a general recognition that “Educators engaged in distance education teaching may rely on the general protection afforded under the fair use doctrine…to create an educational experience for online students that is comparable to that of their face-to-face classroom counterparts” (161). The Poetry Code provides that, “Under fair use, instructors at all levels who devote class time to teaching examples of published poetry may reproduce those poems fully or partially in their teaching materials and make them available to students using the conventional educational technologies most appropriate for their instructional purposes” (11). Last, but certainly not least, the Music Library Association has issued a Statement on the Digital Transmission of Audio Reserves that is largely consistent with these best practices documents (Music Library Association 2010).

These communities of librarians, educators, and collectors have identified several best practices for sharing materials in support of education. First, musical materials made or marketed expressly for use in support of teaching, such as CDs designed to complement specific textbooks or anthologies compiled for teaching, should be used in this context only with great caution, if at all. The extent of the work used should of course be justified in relation to the teacher’s pedagogical purpose; where excerpts will serve just as well as entire works, excerpts are preferred. Relatedly, access should be carefully managed so that only enrolled students can obtain relevant materials, and only for the duration of the course.

The good news for recorded sound collections is that the materials they hold are seldom intended solely for educational use, nor are they being actively licensed for that purpose. It also should be noted that the appellate court in Cambridge Univ. Press specifically acknowledged the applicability of transformative use arguments in the educational context,⁴⁵ so the logic of the best practices concerning the use of non-educational materials to support teaching is still very much available to educators and their allies. Indeed, the Cambridge Univ. Press courts endorsed some of the same considerations identified in the best practices, such as the importance of choosing an amount that is justified by educational purpose, limiting access to students currently enrolled in the relevant class, and providing information about fair use to teachers and students so that they can make responsible decisions about using the material.

⁴² Assoc. Info. & Media Equip. v. Regents of the Univ. of Cal., 2012 WL 7683452 (C.D. Cal. 2012).

⁴³ Cambridge Univ. Press v. Patton, 769 F.3d 1232 (11th Cir. 2014).

⁴⁴ Libraries’ Code, 14 (“It is fair use to make appropriately tailored course-related content available to enrolled students via digital networks.”).

⁴⁵ Cambridge at 1263 n.21

PRESERVATION/FORMAT-SHIFTING

Most interviewees saw digitizing and other reformatting of sound recordings in support of preservation as a high-priority and low-risk fair use activity. Several studies have shown the pressing need for mass-scale reformatting of sound recordings trapped in fragile and outdated formats, but the scope of Section 108’s express provision for preservation is widely seen as narrow and burdensome.⁴⁶ In particular, the idea that a particular copy must already be damaged or deteriorating before it can be preserved has been criticized as tantamount to requiring preservation “malfeasance.”⁴⁷ Another source of frustration with Section 108 is the narrow definition of “obsolete format,” which rules out any format where compatible equipment is available on the market, regardless of whether the equipment is scarce, expensive, or adequate to professional needs. Fair use is clearly available as a supplement to permit preservation where 108 stops short. The text⁴⁸ and legislative history⁴⁹ of the Copyright Act are both explicit on this point, and community best practices agree.

Courts have considered preservation in at least two cases, finding fair use explicitly in one case and implicitly in the other. In Sundeman v. Seajay Society, an author’s estate brought suit against the owner of a copy of an unpublished manuscript for making and distributing copies of the manuscript to a scholar and to the University of Florida. While there were several uses at issue in the case, the Fourth Circuit ruled that it was fair use for the Seajay Society to make and distribute copies of the manuscript for the scholar and the university as a form of preservation, in order to minimize the risk of harm to the rare and fragile original manuscript. In the HathiTrust case, the Second Circuit did not expressly find that preservation, in itself, justified digitizing millions of books—ruling instead that facilitating search was a sufficient independent basis. However, in the district court, Judge Baer had ruled that preservation was an eligible fair use,⁵⁰ a ruling that was not reversed on appeal. Judge Chin’s district court opinion in the related Authors Guild v. Google case also signaled approval for preservation as a legitimate fair use purpose.⁵¹

Some of the community statements that have endorsed fair use for preservation include the Libraries’ Code (17ff.),the Orphan Works Statement (26), the Dance Heritage Statement (8ff.), the Visual Resources Association Statement on the Fair Use of Images for Teaching, Research and Study (“VRA Statement”),⁵² and the Online Video Code (7). These communities agreed that preservation is highly socially beneficial and (when separated from questions of access) poses no threat to the market prerogatives of copyright holders. Libraries and memory institutions agreed that preservation was a core mission that should not be unduly hindered by copyright concerns. At the same time, these institutions agreed that copying for preservation was unnecessary when a particular copy was not unique and a suitable replacement was readily available at a reasonable price. Statements also suggested that originals and preservation surrogates should not circulate simultaneously, lest the preservation function become a pretext for multiplying usable copies when no other fair use justification is available. When preservation surrogates are created to facilitate access (and fragile originals are taken out of circulation), the Libraries’ Code suggests limiting general circulation of preservation copies to “authenticated members of the library’s patron community.”⁵³

⁴⁶ See, e.g., Council on Library and Information Resources and Library of Congress 2010, 120-21 (“Section 108…has failed to keep pace with best practices currently followed by the audio engineering and the federally and privately funded restoration communities.”); Association for Recorded Sound Collections 2005 (“Regarding preservation, the Association believes that current copyright laws and regulation should be modified to eliminate many of the restrictions present in the law.”)

⁴⁷ See Loughney 2011 (“To deliberately delay preserving a culturally, historically or aesthetically important sound recording until it is in a deteriorated condition is a foolhardy practice that could constitute malfeasance on the part of a professional librarian or archivist.”)

⁴⁸ See 17 U.S.C. 108(f)(4).

⁴⁹ See H.R. Rep. No. 94-1476 (“[T]he making of duplicate copies for purposes of archival preservation certainly falls within the scope of ‘fair use.’”).

⁵⁰ The appellate court did not comment on this ruling, so Judge Baer’s ruling appears to be the last word on the issue.

⁵¹ See 954 F.Supp.2d 282, 293 (2013) (lauding the Google Books scanning project because, “It preserves books, in particular out-of-print and old books that have been forgotten in the bowels of libraries….”).

⁵² VRA Statement,10. Note that the VRA Statement was not developed using the usual process of interviews and small group discussions with community members, nor was it facilitated by any of the usual team at American University. Rather, it was authored by a committee of the Visual Resources Association with input from a panel of distinguished IP experts. Nevertheless, the principles described in the VRA Statement are consistent with the norms identified by the various American University-facilitated statements.

⁵³ Section 108(c) and (d) bar off-premises circulation of digital preservation copies “to the public.” Some commentators have suggested that allowing off-premises circulation to a subset of users, such as faculty, students, or affiliate researchers would not constitute circulation “to the public.

COLLECTING ONLINE MATERIALS

While the situations discussed so far have involved materials already acquired, interviewees also described encountering copyright uncertainty in connection with building their collections. Most of this uncertainty had to do with acquiring materials that are (or may be) available only online, including various kinds of ephemera. For some of the works in question, such as musical recordings sold (or, perhaps, “licensed”) exclusively through online stores such as iTunes or Amazon, the purchase is made subject to “Terms of Use” or an “End User License Agreement” (EULA) that often include terms that preclude lending, preservation, and other activities typical of an institutional collector. As discussed in the introductory material, courts have generally allowed agreed-upon contractual terms to trump fair use rights, even where the contracts are seemingly one-sided, take-it-or-leave-it affairs, as EULAs and other electronic licenses generally are.⁵⁴ Accordingly, the existing best practices literature can do little to assist a user who is bound by hostile contract terms, at least as the law currently seems to stand.

Not surprisingly, neither the statute nor the caselaw addressed the acquisition of electronic content by memory institutions. But fair use certainly is available for materials posted or distributed online without a contractual limitation on use, and communities have described some useful ways to think about collecting and using materials from the web and other born-digital works. First, the Libraries’ Code has said that it is fair use to “create topically based collections of websites and other material from the Internet and to make them available for scholarly use” (27). The ICA [International Communications Association] Code found that it is fair use for individual scholars to assemble personal research collections that include material from the Internet, reasoning that, “The materials in question, generally topical or even ephemeral in character, are transformed by collection or organization into a research corpus, which exists for a new and fundamentally different purpose” (12).

Norms for collecting materials in this way are fairly straightforward. The Libraries’ Code counsels that collectors represent any captured materials accurately in the way they would have appeared online, and include information about when and how the material was captured from the web. Attribution should be provided not only for the creator of captured material (as discussed above) but also for the proprietor of the website— or, by extension, any other online source. While not required, the Libraries’ Code suggests that having a clear and consistent policy about bot exclusion headers⁵⁵ will bolster a fair use case, as will an effort to collect as comprehensively as possible within a given topic, theme, or other collecting rationale. While the ICA Code was written by and for individual researchers, the limitations on its personal archiving principle may also be instructive. Collections should be reasonably related to a specific research or collecting interest of the collecting institution or individual; for example, a state university might reasonably try to collect sound recordings posted online by local artists in the state. As with other collections, institutions should make clear to users that they are responsible for using materials to which they are granted access legitimately—including use for scholarship and research.

⁵⁴ A leading case is Bowers v. Baystate Techs., Inc., 320 F.3d 1317 (Fed. Cir. 2003), in which the court found that a shrink-wrap license barring reverse engineering trumped the established rule that reverse engineering software is fair use. Judge Dyk’s dissent in the case is instructive, and, we think, the better view.

⁵⁵ Typically stored in a file called “robots.txt,” a bot exclusion header allows website proprietors to signal their preference that their site not be automatically copied or indexed by companies like Google. Requests for exclusion can be motivated by a variety of concerns, often having to do with audience, bandwidth, and server capacity rather than copyright. Research institutions have found that uncritically following the instructions of every robots. txt file is not a reliable way to discern the wishes of site proprietors regarding archiving. Quite often robots.txt files will exclude only parts of websites, which can result in substantial unjustified omissions from collections. See Legal Issues, International Internet Preservation Consortium (n.d.). In at least one case, the Libraries’ Code has helped libraries move to more nuanced policies. See Gray and Martin 2013.

DATA-MINING/NON-CONSUMPTIVE RESEARCH

Exponential growth in computing power together with the shrinking cost of storage has made it increasingly feasible to create and process comprehensive databases of all kinds of media. To date, the leading use of such databases is the creation of a search engine to locate particular terms in a corpus, the way a search engine helps users find relevant websites on the Internet. As computational power grows, however, the kinds of information that can be gleaned from analysis of large corpuses grows with it. The power of “big data” seems nearly impossible to overestimate.

For recorded sound collections, the potential of digital analysis manifests itself in several ways. First, it makes possible the creation of new search tools that allow researchers anywhere to identify whether and where particular recordings or kinds of recordings can be found, based not on textual information but on the characteristics of those recordings themselves. This is the promise of digitized search. Second, technology enables the computer analysis of recordings to facilitate an increasingly common mode of research sometimes called “distant reading.”⁵⁶ Interviewees described existing projects that analyze music, metadata,⁵⁷ and lyrics,⁵⁸ and expressed keen interest in exploring ways to leverage computer analysis, including providing databases for this purpose, so that scholars and students could learn more from recorded sound collections. Scholars have called uses like these “non-consumptive” or “non-expressive.”⁵⁹ The use is “non-consumptive” because the works are searched or analyzed—usually in large quantities—by a computer, rather than being “consumed” (i.e., read or heard) by a human being in a way that might implicate the market prerogatives of a copyright holder. The outputs of a computer analysis are “non-expressive,” i.e., they are facts about a work or a corpus—the frequency and proximity of particular words, phrases, notes, and so on.

The courts have been quite clear in their endorsement of fair use for search engines and similar uses. Early cases were about mass market Internet search engines run by companies like Google and Amazon, but later cases have considered uses in more specialized contexts such as plagiarism detection tools⁶⁰ and, most recently, in the Authors Guild v. HathiTrust case, search and text mining for scholarly use. While the most familiar search engines deal primarily with text, courts have also applied the fair use doctrine to search engines that scan images and display “thumbnail” images as results,⁶¹ as well as search services that crawl television programming.⁶² Extending the logic of these cases to cover sound recordings poses no particular conceptual challenge.

The Libraries’ Code addresses these uses directly and finds that it is fair use “to develop and facilitate the development of digital databases of collection items to enable non-consumptive analysis across the collection for both scholarly and reference purposes” (25). The principle is grounded in a strong community consensus that such uses are well within the realm of what libraries and memory institutions have always done to add value to their collections—extracting factual information about items in the collection that helps researchers and users understand collections and locate material that will best serve their research needs.

Best practices that impose limitations on such uses are mainly directed to ensuring that materials digitized and collected for non-consumptive purposes are not repurposed without independent justification. For example, while non-consumptive research may be sufficient to justify mass digitization and computer analysis of a collection of sound recordings, a new and separate fair use argument (or a statutory justification, or a license) would be required before making the same works individually available for listening. Also, when creating a public-facing search or research tool, it is important that any portions of text displayed as search results (often called “snippets”) are carefully chosen to ensure they serve the research purpose and do not unduly intrude on the ordinary market for access to the works.

56 See, e.g., Moretti 2013.

57 Scholars interested in this issue have created The International Society for Music Information Retrieval, a group that hosts an email list and annual conferences for presentation of research. See http://ismir.net. Work in this area is spread across many leading research institutions and funded by groups including the National Science Foundation in the United States and JISC in the United Kingdom. See Byrd and Fingerhut 2002.

58 One example that came up several times was Tahir Hemphill’s “Hip Hop Word Count” project, a Kickstarter-funded database of rap lyrics that has since become part of an educational project called the Rap Research Lab. For more information, see Hemphill’s website, http://staplecrops.com. Information about his Kickstarter campaign is at https://www.kickstarter.com/projects/1801076626/ the-hip-hop-word-count-a-searchable-rap-almanac.

59 See, e.g., Sag 2009.

60 A.V. ex rel Vanderheye v. iParadigms, LLC, 562 F.3d 630 (4th Cir. 2009).

61 Perfect 10 v. Amazon.com, 508 F.3d 1146 (9th Cir. 2007)

62 Fox News Network, LLC v. TVEyes, Inc., No. 13 CIV. 5315 AKH, 2014 WL 4444043 (S.D.N.Y. Sept. 9, 2014).

DIGITAL EXHIBITS AND EXHIBITS FOR THE PUBLIC

A distinct but related recurring use scenario, as significant for other cultural institutions as it is for recorded sound collections, involves the assembly of a group of recordings into a curated collection or exhibit available on the open web and designed to facilitate exploration of a particular theme, such as a genre, an era, a performer, or a geographic region. Interviewees suggested that recorded sound collections could facilitate teaching in formal settings such as public school classrooms or informally by reaching individuals via the Internet. Ideally, such collections will include rich metadata and contextual information as well as the collected recordings, giving instructors or individual browsers some of the information they need to make sense of the recordings and to derive value from the collection.

Although the courts have not addressed the fair use status of virtual (or, for that matter, physical) exhibits, it is widely assumed that such uses should be considered a form of privileged public education, to which fair use applies with full force. Not only is the inclusion of copyrighted material in the context of an exhibit an obvious example of transformative use, but also such exhibits pose little or no threat to the copyright owners’ legitimate interest in commercializing their intellectual property.⁶³

Almost every relevant community that has undertaken to articulate best practices in fair use has described some version of this practice as a legitimate exercise of fair use rights. Both the Libraries’ Code and the Orphan Works Statement address this type of use. Principle Four of the Libraries’ Code, which addresses digitizing special collections, is also relevant here, but Principle Two deals more specifically with the creation of exhibits, both online and in physical space. Principle Two provides that, “It is fair use for a library to use appropriate selections from collection materials to increase public awareness and engagement with these collections and to promote new scholarship drawing on them” (16). Similar principles occur in the Code Of Best Practices In Fair Use For Poetry (Poetry Code),⁶⁴ the Set of Principles in Fair Use for Journalism (Journalism Principles),⁶⁵ the Dance Heritage Statement,⁶⁶ The Code of Best Practices in Fair Use for OpenCourseWare (“OCW Code”),⁶⁷ and the Code of Best Practices in Fair Use for Online Video (Online Video Code).⁶⁸

These statements share several key characteristics. They emphasize the value added by informed curation as well as additional information and commentary posted along with collected materials. Most communities require some minimal information (attribution, clear indication of the theme or purpose of the exhibit) as a threshold matter, and agree that providing more information (historical context, critique, etc.) will strengthen any fair use claim.⁶⁹ Second, the statements indicate that the amount of any given work posted should be appropriate to the educational purpose of the exhibit or collection. Where excerpts of works will suffice as illustrations of a given theme, posting of entire works will tend to weaken a fair use argument. Indeed, the Libraries’ Code suggests, “[U] se of a work (other than a single image) in its entirety is likely to require a special level of justification.” Third, several statements suggest that provision of additional tools to allow for engagement with the collection will strengthen a fair use case.⁷⁰ Fourth, the absence of commercial availability is again a factor that some communities have found relevant in deciding what materials to include.

63 In this sense, the exhibitor is in the same position as the publisher of a book surveying the history of the Grateful Dead, which the court found had no (negative) effect on the commercial efforts of concert promoters whose posters were reproduced in the book. See Bill Graham Archives v. Dorling Kindersley Ltd., 448 F.3d 605, 614 (2d Cir. 2006).

64 Poetry Code, 13 (Principle Six: “Under fair use, an online resource (such as a blog or web site) may make examples of selected published poetry electronically available to the public, provided that the site also includes substantial additional cultural resources, including but not limited to critique or commentary, that contextualize or otherwise add value to the selections.”).

65 Journalism Principles, 14 (Situation Six: “The use of copyrighted material to promote public discussion and analysis can qualify as fair use.”).

66 Dance Heritage Statement, 10 (Situation 2, Exhibits: “Given the significant cultural and educational function provided by Dance Collection exhibitions and displays, fair use should apply in instances of this kind, both for the materials on display and for the items used for ancillary materials.”); 15 (Situation 5, Digital Information Exchange: “Creating web-based resources and engaging in digital delivery of selected materials can be an appropriate way to extend the Dance Collections’ mandate to deliver their culturally valuable materials.”)

67 OCW Code, 11–14 (describing fair use of copyrighted materials in online course materials known as “OpenCourseWare” for purposes of criticism, commentary, and illustration.).

68 Online Video Code, 7 (“Four: Reproducing, Reposting, Or Quoting In Order To Memorialize, Preserve, Or Rescue An Experience, An Event, Or A Cultural Phenomenon.”)

69 See, e.g., Dance Heritage Statement, 17 (“the less extensive the indications of significant added value are, the weaker the fair use claim will be.”).

70 See, e.g., Journalism Principles, 14 (“The journalist (or outlet) should make available tools and forums designed to encourage participation by news consumers.”); Orphan Works Statement, 32 (“provide users with specialized search tools”);

TRANSFER OF COPIES TO THIRD PARTIES IN SUPPORT OF DOWNSTREAM FAIR USES

Many interviewees described receiving requests from a variety of users— e.g., musicians, documentarians, journalists, artists, writers—who sought copies of sound recordings for use in their own creative cultural activities that go beyond research, study, or teaching. While supporting academic and educational uses is certainly a core part of what many libraries and memory institutions do, they also have a mission to provide points of cultural reference for the current generation of creators. Indeed, one of the principal reasons for the fair use doctrine is the necessity that successive generations of creators incorporate material from existing works.⁷¹ For these users, a stream or other technologically limited mode of access would not suffice, nor (in some cases) would a partial copy. Professionals charged with managing recorded sound collections generally have a mission to support fair and legitimate use of collections materials, which can go beyond mere provision of access in many contexts. The legitimacy and importance of these uses have been recognized not only in the statute but also in the variety of best practices statements that have been created to date.

Notably, criticism and commentary are mentioned expressly in the Copyright Act as the kinds of uses that often qualify as fair,⁷² and the courts have made it clear that intermediate uses of copyrighted material that enable or facilitate ultimate legitimate uses can be considered fair in appropriate circumstances. One notable example is the copying of large amounts of computer code in order to engage in lawful “reverse engineering;”⁷³ another is the unauthorized reproduction of millions of student papers to fuel a digital engine for detecting plagiarism.⁷⁴ The same principle can and should be applied to uses in the cultural sector.

Indeed, there are compelling arguments in favor of providing copies in good faith to would-be fair users. This situation is susceptible to some of the same arguments that apply where remote researchers seek access to rare or unique items simply for study. Section 108 provides some limited allowance for making copies that become the property of the requesting user, and there is good reason to believe that fair use could apply where Section 108 leaves off (to musical works, for example). Section 108 also shields libraries from liability for patrons’ use of on-site copying equipment, a measure that helps to facilitate unsupervised patron copying. Also, in many cases the materials that users request are rare or unique, meaning they are subject to compelling arguments about lack of market harm: no market exists and commercial exploitation may not have played a role in their creation.

The variety of fair uses that might be made of sound recordings is virtually limitless, but practice communities have identified several recurring varieties of fair use in their domains that also exist where recorded sound collections are concerned, and these creator communities recognize such uses as fair in all of their best practices statements.⁷⁵ Reproducing portions of copyrighted works as evidence, illustration, or documentation, as well as objects of commentary, is a recurring fair use activity across multiple best practices statements.⁷⁶

Educators have described scenarios when fair use allows incorporation of copyrighted material in their lectures and curriculum materials,⁷⁷ as well as when students should employ fair use as part of their coursework.⁷⁸ Users present these arguments to sound recording collectors, and request access to collections materials to facilitate fair uses. Community practices associated with providing copies to users for research as well as creating online exhibits could easily be adapted to apply here. For example, recorded sound collections should follow the general practice of including information about copyright and fair use with each copy, and signifying clearly that the user is solely responsible for their own uses of copyrighted materials. A click-through acknowledgment of the user’s responsibility to comply with copyright is a related practice recommended in the Orphan Works Statement. If possible, the nature and quality of the access provided could be limited in ways commensurate to the needs of the user, though a user who needs a downloaded copy for fair use purposes may need higher quality than a researcher.

⁷¹ See, e.g., Leval 1990, 1109 (“First, all intellectual creative activity is in part derivative. There is no such thing as a wholly original thought or invention. Each advance stands on building blocks fashioned by prior thinkers.”).

⁷² 17 USC 107 (“fair use…for purposes such as criticism, comment… is not an infringement of copyright.”).

⁷³ See Sega Enters. Ltd. v. Accolade, 977 F.2d 1510 (9th Cir. 1993).

⁷⁴ See A.V. ex rel Vanderheye v. iParadigms, LLC, 562 F.3d 630 (4th Cir. 2009).

⁷⁵ See, e.g., the Documentary Filmmakers’ Statement of Best Practices in Fair Use (“Documentary Statement”), 4 (“Employing Copyrighted Material As The Object Of Social, Political, Or Cultural Critique”); Poetry Code, 11 (“Under fair use, a critic discussing a published poem or body of poetry may quote freely as justified by the critical purpose; likewise, a commentator may quote to exemplify or illuminate a cultural/historical phenomenon, and a visual artist may incorporate relevant quotations into his or her work.”); Journalism Principles, 11 (“When Copyrighted Material Is Used In Cultural Reporting And Criticism”); Online Video Code, 5 (“Commenting On Or Critiquing Of Copyrighted Material”).

⁷⁶ See, e.g., Documentary Statement, 4 ff. (“Quoting Copyrighted Works Of Popular Culture To Illustrate An Argument Or Point”); Journalism Principles, 11 (“Use Of Copyrighted Material As Proof Or Substantiation In News Reporting Or Analysis”), 12 (“When Copyrighted Material Is Used As Illustration In News Reporting Or Analysis”); Online Video Code, 7 (“Reproducing, Reposting, Or Quoting In Order To Memorialize, Preserve, Or Rescue An Experience, An Event, Or A Cultural Phenomenon”); Code Of Best Practices In Fair Use For Scholarly Research In Communication (“ICA Code”), 10 (“Scholars may invoke fair use to reproduce copyrighted material where it serves to explain or illustrate their scholarly insights or conclusions about communications in relation to social, cultural, political, or economic phenomena.”).

⁷⁷ See, e.g., Poetry Code, 10–11; Dance Heritage Statement, 13-15; Media Literacy Code, 10-12; OpenCourseWare Code, 11–14; SCMS Teaching Statement, 157–61; VRA Statement, 11–12. 78 See, e.g., Media Literacy Code, 12–13; VRA Statement, 13–14.