The Legal Landscape: Copyright, Fair Use, and Archival Best Practices

Copyright and fair use issues in recorded sound collections have been written about extensively over the past twenty years, especially in the Tim Brooks–edited “Copyright and Fair Use” column of the ARSC Journal and the spate of publications emerging from the Council on Library and Information Resources since the establishment of the National Recordings Preservation Board in 2000.³ Reports from the Register of Copyrights also indicate that, though we still await changes to the code, the U.S. Copyright Office is fully aware of the shortcomings of Title 17 to address the rapid changes in technology and society that have enabled mass digitization and dramatically increased the (potential) accessibility of public archives.⁴ In the absence of legislative action, the judiciary has taken the lead in supporting accessibility. Courts have ruled repeatedly in favor of fair use in recent years, particularly—but not necessarily—when the use is “transformative” (unfortunately, a still ill-defined term).⁵

The libraries and archives community has recognized for some time that we must act quickly to preserve our aural heritage, most urgently because the fast-decaying magnetic tape format predominates in sound collections.⁶ This cannot be accomplished by locating rights holders and seeking permissions for every item. It is well known that securing permissions from rights holders for the use of archival materials is a difficult and labor-intensive process for textual materials.⁷ Due to the overlapping layers of rights in recorded sound materials, this difficulty increases exponentially. In the Bentley's holdings, for instance, the vast majority of sound recordings are unpublished and noncommercial, and at least one right—that of the engineer/producer of the recording—is orphaned in nearly every case.⁸ Although not the focus of this article, other legal considerations, such as accessibility for users with disabilities, exist in providing access to digital content online.⁹

This inarguable fact forces a direct confrontation between archival ethics and the law. Section 108 exemptions for libraries/archives preservation reproduction do not apply to recordings fixed before 1972 because they are not subject to federal statutes. Without the guarantee of these reproduction rights, duplication for preservation becomes illegal infringement. As the National Recording Preservation Board succinctly puts it, “Were copyright law followed to the letter, little audio preservation would be undertaken. Were the law strictly enforced, it would brand virtually all audio preservation as illegal.”¹⁰ An orthodox adherence to Title 17 ensures sound recordings will wither and die on the vine. This would be nothing less than a fundamental dereliction of duty on the part of archives, charged as they are with the preservation of records for the benefit of current and future generations.¹¹

Fortunately, emerging best practices within U.S. libraries and archives and codes promulgated by legal institutes and professional associations¹² endorse using Section 107 fair use rights to go forward with the proactive digitization of at-risk materials and many orphan works. The Library Copyright Alliance (LCA) (representing the American Library Association, the Association for College and Research Libraries, and the Association of Research Librarians) now argues that legislation is no longer necessary for their members' uses because other legal developments, primarily regarding copyright's fair use doctrine, are sufficient to allow for many uses of orphan works. Consequently, fair use evaluations are being undertaken more frequently and by more institutions, and, as the LCA report notes, “The more they engage in these activities, the more confident libraries become with their fair use analysis concerning the mass digitization of presumptively orphan works.”¹³ Additionally, Eric Harbeson points out that Section 504(c)(2) of the copyright code allows a “secondary line of defense,” after fair use, for nonprofit schools, libraries, and archives: the “reasonable belief” defense.¹⁴ The code makes clear that libraries and archives cannot be held liable for statutory damages for copyright infringement if the infringer “believed and had reasonable grounds for believing that his or her use of the copyrighted work was a fair use.”¹⁵ As of August 2017, no archives have been sued over digitization projects for access, as pointed out by Peter Hirtle.¹⁶ Some may consider the Hathi Trust case an exception to this, but it is important to point out that in this case, an archives was not sued and that search and accessibility were considered transformative enough and fair use.¹⁷

Harbeson also explains that fair use evaluations like these are becoming more and more common has legal implications. Widespread practice in a community reflects “community standards” as much as more formal best practices codes; and community standards, rooted in court rulings, can provide a strong basis for reasonable belief. “Even if a court later rules that practice to be unfair,” Harbeson writes, “the fact that the practice has been widely understood to be fair can point to a reasonable belief that it was fair. This gives use communities a certain power over their destiny: they can, to some extent, shape fair use standards in ways that might or might not be in their interests, by following common practices.”¹⁸ But what are today's common practices among institutions engaged in mass digitization?

Toward Community Standards: Common Practices at Five Institutions

After reviewing the literature on current best practices standards, recent judicial opinions, and other articles encouraging a bolder approach to fair use than we currently employ at the Bentley, we determined that Section 107 limitations on exclusive rights provide far more flexibility in allowing users access to digitized content than Section 108, and we therefore developed policies and tools to leverage our Section 107 rights. We next reviewed the actual practice at four U.S. research libraries/archives and one U.K. institution to get a better understanding of emerging community standards in allowing public access to digitized archival materials.¹⁹ These institutions were the Wellcome Library, Duke University, the University of North Carolina at Chapel Hill, the UCLA Special Collections Library, and Pacifica Radio Archives. They reflect a diversity in size, geography, and collection composition; most primarily digitize and provide access to textual materials, but we found that all provide useful lessons for our project.

Our study first looked at the Wellcome Library (WL) digitization project, a massive endeavor to make 30 million pages of archives and library materials accessible by 2020.²⁰ Although the application of fair use exemptions differs in the United Kingdom (where it is known as “fair dealing”²¹), enough overlap exists to make an examination of the WL's general principles and approach useful.

The project is unusually risk-comfortable. The fundamental rule for archival materials is managing rather than eliminating risk, which means in practice that no attempt will be made to contact all rights holders, but only those who “appear to present a risk.”²² Permission from the named creator of a collection is assumed to already be in place; the analysis concerns third-party copyright holders within specific collections. The WL's process focuses on anticipating litigants and emphasizes identifying the higher-risk copyright holders (for example, the author/creator is a well-known literary figure, broadcaster, or artist; or the author/creator/literary estate/publisher is known to actively defend copyrights; or the relationship between the holding institution and the author/creator/publisher is awkward). Next, a search is made for these higher-risk entities, and permissions are sought. The WL manages access through different layers of restrictions, including agreeing to terms of use, user registration, and limiting some items (especially potentially sensitive materials) to the premises. Finally, a strong takedown policy is in place and easily visible.

The four other institutions studied reinforce the WL's general approach—anticipating and heading off litigants—and use several safety valves to assuage aggrieved parties by responding quickly to any concerns that arise. However, the WL risk-management strategy hinges on aggressive permission-seeking, which is extremely resource intensive. Institutions without such resources, unable to locate and petition hundreds of rights holders, cannot reasonably assume as much risk nor open as much material online as the WL.

Our study of the policies of Duke University focused on a description of the principles behind the AdViews project, a digital collection of historical television commercials, by Kevin L. Smith, former director of Scholarly Communications.²³ Smith argues strenuously against what he deems “self-censorship” in the archives and library worlds and for a much higher tolerance of copyright infringement to keep digital collections complete and to fulfill institutional missions. Like the Wellcome Library, Smith argues for a risk-management strategy centered on anticipating and heading off potential litigants more than observing the letter of the law. His principles are, first, to reduce the number of risky items a collection contains and, second, to reduce the number of people likely to want to sue. The strategies he suggests are simple: ask permission from those most likely to object to digital display (not every rights holder, but large, prominent ones); have a take-down policy prepared in advance; and recognize that a strong fair use argument will support many collections, especially when such an argument is a “back-stop” or “last line of defense” as part of a larger strategy.

Smith's article echoes the general method of the Wellcome Library but is noticeably more parsimonious regarding how many entities to search for and seek permissions from. Both reinforce the idea of a community standard around seeking permissions selectively and implementing a strong take-down policy should objections arise. They differ on how high the risk must be before seeking permissions, with Smith notably more liberal, a difference largely attributable to the greater elasticity of the U.S. fair use principle than the U.K. analog, fair dealing.

Our study of the University of North Carolina (UNC) focused on the methods articulated in its project to digitize manuscripts in the Southern Historical Collection, particularly the “decision matrix” employed.²⁴ The matrix is a scoring tool to create priorities for digitization at the collection level. A nine-page questionnaire, though its questions skew heavily toward manuscript materials, is designed to encompass all media types in the special collections. It takes a holistic approach to evaluating collections, examining not just rights and restrictions but content, research value, and risks for obsolescence/degradation. In particular, the UNC decision matrix contains many more questions about donor agreements and restrictions than our previous examples. After evaluating a collection with the matrix, it is sorted into one of four groups according to its assessed level of risk. As a last step, selected collections are evaluated at the item level for personalities in the materials known to be, or thought likely to be, litigious.²⁵ Unlike the Wellcome Library, UNC's method does not include tiered levels of access for users, but rather is fully open.

The Special Collections Library (SCL) at UCLA includes a risk assessment procedures report as part of a larger body of instructions that details the library's entire end-to-end digitization process, from selection to web architecture to user experience.²⁶ Whereas previous examples primarily emphasized risk reduction, the SCL stresses fair use and its thorough documentation. The layers of documentation provide proof that due diligence is performed and is copiously recorded and vetted. The implication, of course, is that proof of serious, thoughtful fair use judgments will indemnify UCLA Special Collections (as a public, nonprofit, academic library) from infringement suits. Permission requests play a part in UCLA's digital project workflow but are not the linchpin that they are for the Wellcome and Duke libraries.

Pacifica Radio Archives (PRA) “risk factor assessment”²⁷ tool is a straightforward chart that asks the archivist to sort each item, at the item level, into one of three categories of risk (low, medium, and high) for each of seven questions. This document is useful because it is media-specific, pointing to several essential questions pertaining to sound recording as a whole, with several particularly tailored to radio—a major portion of our collection under review. Although PRA's tool approach focuses mainly on risk, its risk matrix model was significant to our review method.

Based on the common practices of these institutions, we developed a review workflow that includes four tools to assess risk, to evaluate the strength of a fair use defense, and to document due diligence and reasonable belief. They are discussed in greater detail in the following sections.

Since the completion of this initial research phase, we have learned of more recent work in mass digitization projects for audiovisual materials and efforts to provide access by addressing rights. The American Archive for Public Broadcasting (AAPB) recently presented its work on exposing digital collections at scale by using fair use, broadcast genres, and tiers or “buckets” of different types of access.²⁸ Likewise, Indiana University's Media Digitization Preservation Initiative 2 (MDPI) is another example of innovative approaches to provide access to digitized audiovisual content. Its efforts have focused on a rights review management system.²⁹ Both of these examples show a growing need to address rights status for digitized audiovisual materials; the need and desire to increase access and scalability; and a need to understand how practices in this field are evolving.

The Genre Method

Before the development of review tools, we set out to analyze the content we would be evaluating for rights. Existing metadata for this content revealed that available genre information could inform and improve review processes and tools. Our research had shown that different levels of inherent risk subsided in different genres of material and presented different requirements when it came to evaluating rights. For example, potentially infringing creative, profit-earning works (symphonies, poetry readings, etc.) is far riskier than infringing extemporaneous conversations. The former are copyrightable, while the latter are not; the former have a profit incentive (and a solid constitutional protection guarantee);³⁰ the latter generally do not.

It became clear that larger, umbrella groupings based on genre could be formed with recordings of the same base risk level. Accordingly, we identified the following five primary genres: spoken word, oral history, music, documentary, and literary. This “genre method” became a way of grouping content to more efficiently assess risk and fair use for thousands of recordings.

The reason the genre method works is that most collections are not nearly as eclectic as they may appear at first. We have found that about nine out of ten of our collections are completely or very nearly homogeneous in primary genre. This allows us to scale our risk evaluations, because we can now conduct them at the collection level rather than at the item level. When a collection consists of more than one primary genre, it must be divided accordingly into homogeneous subgroups for evaluative purposes.

In short, genre is the first factor to consider when determining how risky expanding access to a recorded sound collection is, for two reasons: different general risk levels inhere in different genres, and the specific questions a reviewer must ask when reviewing a collection vary according to genre. This second fact led to the first of four modifications to the Information Gathering Questionnaire that resulted in five separate questionnaires, one tailored to each primary genre.

This study of our data allowed us to gain a good overview of our test batch and prioritize our rights review process. Knowing unpublished spoken word recordings generally are the lowest risk and have strong fair use defenses, we planned to review these first (25% of the total number of collections tested). Published spoken word recordings generally are low risk and have the strongest fair use defense (38% of collections), and so were prioritized second. Next came documentaries, oral histories, music, and finally literary readings. Already, we could see that, considering the spoken word genre alone, nearly two-thirds of our collections were potentially low-risk, strong fair use defense collections, and needlessly being restricted.

The Four Tools

In setting out to develop our own tools and workflow, we determined that a strategy of managing, not avoiding, risk was best in line with our institutional goals. As such, our judgments needed to account for both the likelihood of a use attracting litigants (termed the risk level), and the strength of our fair use defense should a complaint arise (the fair use strength). To achieve this, we developed four tools to review the contents' rights status, risk level, and fair use defense strength. We applied these tools after determining the genre groupings in a collection.

• Information Gathering Questionnaire: documentation of the collection's gift agreement, known rights holders, existing restrictions, etc. (see Appendix 1).

• Fair Use Assessment Checklist: identification of potential issues related to fair-use factors (i.e., the purpose of the use, nature of the copyrighted works, amount/substantiality of the portion used, and the effect of the use on the works' market value) (see Appendix 2).

• Fair Use Statement: documentation of the potential strength of fair use exemptions for the collection (weak, medium, or strong), with an explanation of the findings based upon responses to the checklist (see Appendix 3).

• Risk Assessment Rubric: placement of the collection on a spectrum of risk (high, medium, low, or none) based upon the risks associated with public access to content under U.S. copyright law (see Appendix 4).

First is the Information Gathering Questionnaire, which collects details such as the existence of a gift agreement, whether any restrictions exist, and whether any potential copyright holders are known or identifiable. This step always requires consulting the donor file and the finding aid and sometimes requires consulting the catalog record and the audio preservation database and listening to samples of the item. This tool is an adaptation of the one used by UNC in the Southern Historical Collections project, additionally informed by UCLA practices.

Table 1. The Five Primary Genres

View Fullscreen

Next, the Fair Use Assessment Checklist evaluates the collection for the strength of its fair use defense, balancing each of the four fair use factors outlined in Section 107 of U.S. copyright law: purpose and character of use, nature of the copyrighted work(s), amount and substantiality of the portion to be used, and effect of use on the potential market for the work. It is adapted from a schema for evaluating the four factors employed by UCLA, based on the original by Kenneth D. Crews.³¹ This checklist has achieved widespread community use and is additionally employed by Cornell University, Columbia University, and many other institutions. It has been lightly modified for our purposes.

Third, the Fair Use Statement tool asks the analyst to examine the checklist and judge whether the four factors weigh toward a strong, medium, or weak fair use defense. The analyst then drafts a statement explaining the rationale for the ruling on each of the four factors and the overall judgment. This tool was inspired by UCLA practices and is especially important for documenting good-faith belief as well as due diligence. Of special importance is questioning the “transformativeness” of the use. At the Bentley, we ask: Have modifications been made? How is value being added? And is the proposed use markedly different than the original use?

Finally, the Risk Assessment Rubric is used to assess a specific risk level. Each collection/subgroup is sorted into one of four categories: High, medium, low, or none. The analyst drafts a brief “risk assessment statement” summarizing the rationale for the assigned level of risk.

While a general and preliminary risk level can be assigned to a collection on genre alone to prioritize the digitization and the risk evaluation processes, the Risk Assessment Rubric provides a more thorough and refined judgment based on the specific characteristics of a collection (the genre method is detailed in the following section). It focuses on copyright law itself without regard to fair use defenses, aiming to elucidate the degree of exposure to legal action the wide release of the collection might engender. The rubric is primarily drawn from a similar tool in use at the Pacifica Radio Archive, with modifications informed by the practices at all four of the other examined institutions.

Implicit in these tools is that the Bentley draws a distinction between risk level and fair use strength. We define risk level as the likelihood of attracting complainants or litigation. We define fair use strength as our judgment, informed by case law, that our use would be found fair in a hypothetical court challenge. While these two attributes are related, it is useful to disentangle them and evaluate them separately.

Risk Levels

In assessing potential risks under copyright law, the Curation Division defined four discrete risk levels to help quantify the Bentley's potential exposure to legal action. The criteria were lightly revised throughout the pilot phase, resulting in these final definitions (see also Appendix 4, Risk Assessment Rubric):

• No Risk: Indicates that the deed of gift allows for digitization and/or assigns the creators' copyright to the Regents of the University of Michigan. It is also employed when the collection (or a specific grouping thereof) DOES NOT include items with third-party copyrights. In general, public domain status is not evaluated because virtually no sound recordings are in the public domain as federal copyright law does not apply to pre-1972 sound recordings.

• Low Risk: Generally denotes that the deed of gift allows for digitization and/or assigns copyrights to the Regents of the University of Michigan but also indicates that the collection DOES include some items with third-party copyrights. This level is also applicable to collections (or subgroups thereof) for which the creator is dead and has no literary estate, and content was created with academic and/or personal intent.

• Medium Risk: Used with collections (or subgroups thereof) that include excerpts or small amounts of creative works but do not use “the heart” of a work. This level is also employed with radio programs that lack station identification and collections with many orphan works. Medium risk may also be assigned to collections where the author/creator is alive or known to have a literary estate, but materials were created with academic and/or personal intent.

• High Risk: Those collections in which rights have not been transferred; releases do not exist; the author/creator is a well-known literary figure, broadcaster, or artist; the materials appear to have been published/broadcast and/or prepared for commercial gain; a non-University of Michigan–owned station ID is announced; large amounts of third-party creative works are present; or any other indications of restrictions or special sensitivity are noted.

Fair Use Strength Levels

An evaluation of the strength of fair use defenses for the digitization and dissemination of sound recordings complement the levels of risk associated with copyright law. Our checklist is modeled on a basic template used by libraries at Cornell, Columbia, UCLA, and many more institutions. The process is a balancing test that assesses whether a proposed use “favors” or “disfavors” a fair use exemption in each of the four factors.³² As the Columbia Copyright Advisory Office (CCAO) notes, “This flexible approach to fair use is critical in order for the law to adapt to changing technologies and to meet innovative needs of higher education.”³³ Not all factors need to weigh either for or against fair use, but overall, the factors will usually lean one direction or the other. The CCAO also points out that “the relative importance of factors is not always the same” and individual analysis is necessary to reach a conclusion. There is no hard and firm rule regarding fair use. In our collection, the most salient factors have tended to be nature of copyrighted work and potential market effect, as our purpose for use (educational) and amount to be used (entire) are usually consistent.

The Curation Division has established three separate fair use defense categories (see also Appendix 2, Fair Use Assessment Checklist):

• Strong Defense: Indicates a majority of items “favoring fair use” have been checked. That the proposed use is transformative (factor 1) and that no significant market effect is expected (factor 4) are particularly important. A “strong” assessment also usually pertains only to factual items (factor 2) that do not infringe on third-party rights in large or substantial amounts (factor 3).

• Medium Defense: Indicates a balance between the two columns of the checklist (“favoring” vs. “disfavoring” fair use). The chief distinction between a medium and a strong defense is most often that the proposed use is not transformative (factor 1) or that the work infringed upon is creative or unpublished (factor 4).

• Weak Defense: Applies to proposed uses that cannot reasonably be viewed as transformative (factor 1) and/or that are likely to have a market effect (factor 4). It also applies to unpublished works for which rights have not been transferred and to oral histories with no releases and/or clear provenance (factor 2).

The Review Process Workflow in Brief

Figure 1 summarizes the stages of the rights review process discussed. The full assessment workflow can range from ten minutes to nearly an hour, depending on the size and complexity of the collection. For a complete, detailed description of the rights review and risk assessment workflow, please see Appendix 5.

View Figure

• Download as Powerpoint
• Download Figure

The Review Process Close Up

Appendix 6 provides four sample evaluations. Each represents a collection with a different risk level: none, low, medium, and high. This section will walk through parts of those assessments to illustrate how determinations are made, some typical characteristics of specific genres, and where gray areas reside that can make judgment calls difficult for the analyst.

The Hilary Whittaker Papers (see Appendix 6.1) represents a typical no-risk collection: a small, homogeneous collection (in this case, one item); no third-party rights; creator transferred copyright to UM Regents; and a clear gift agreement detailing this transfer is on file. Small collections allow for sampling a larger percent of the collection to increase certainty that no third-party rights are present. In the case of Whittaker's recording, an on-the-spot description of an earthquake in Morocco, the entire file can be heard in under three minutes. Because no third-party rights subside in the item, no fair use analysis or risk assessment is necessary.

The Karoub Family Papers (see Appendix 6.2) likewise is typical of a low-risk collection, many of which are small (this one is three items). Rights have been transferred to the regents and a gift agreement exists. However, as is the case with many low-risk collections, this collection cannot be assessed at no risk because third-party rights may be contained within. Imam Karoub reads religious poetry and religious songs on the tape, and, while these works were probably created before 1923 (given their devotional nature), it is possible they were not. No greater level of certainty can be attained at this time due to another challenge to performing fair use analysis: foreign languages. It is, of course, not reasonable to expect the analyst to be fluent in every language contained in the archives. Nevertheless, linguistic ignorance has the effect of increasing uncertainty and, therefore, risk. The analyst must make a judgment call about the likelihood of third-party rights being infringed based on the date of the recording, the type of content, and the description. This point also emphasizes the importance of adding good descriptive metadata to each item during the digitization or QC process. This greatly reduces the need for sampling, and the evaluation process moves rapidly. Inadequate metadata can sometimes slow the process to a crawl, as the analyst must sample many different time points in a file to confirm homogeneity of content and genre.

Things get trickier at the medium-risk level. A good example is the Henry Russel Lectureship Committee Papers (see Appendix 6.3). The committee itself is a university organ, and the regents hold copyright to its papers. But in these recordings—keynote addresses by junior faculty detailing their research—the committee probably holds no more than the right of audio engineer; each respective faculty member owns the main content—scripted speeches. The Risk Evaluation Rubric shows the two criteria that place this collection in the “medium” category, the most relevant being “The author/creator is alive or known to have a literary estate, but materials were created with academic intent.” The Fair Use Assessment shows the factors favoring and disfavoring fair use almost equally balanced. That these works were delivered publicly, are used for educational purposes, and will have no significant market effect supports fair use of these lectures. But the creative nature of the works and the use of the “heart” of the works disfavor fair use. On the whole, the fair use defense likely comes down to the “transformative” factor: would contextualizing these lectures together and opening access to them as a historic group be “transformative enough”? The “transformative” factor has become perhaps the most preeminent in recent legal rulings, but it is notoriously difficult to judge. The analyst must assess whether the use seems defensibly transformative. In this case, the analyst concluded it did not, and the balance of factors resulted in a fair use defense of “medium.”

Evaluating high-risk collections can be very simple, or it can be as complex as medium-risk evaluations. For example, all music and literary sound recordings (poetry readings, radio plays, and so on) begin at the high-risk level and rarely escape it. Only one-third of collections in the music genre and no collections in the literary genre have a risk level lower than “high” (see Figure 3). For these two genres, risk assessment tends to be straightforward. This point will be discussed further in the Findings and Recommendations section.

The Ruth Ellis Papers (see Appendix 6.4) highlights the importance of good description and sampling recordings when in doubt about their content. At first glance, the collection appears to be low or medium risk, as it is in the spoken word genre, a broadcast interview, and made for public media (NPR), which is less risky than commercial radio. But an important concern for previously aired materials is whether there is a market for it—or more loosely, whether it can still generate income. The NPR website, where ads are displayed to generate revenue, still hosts this interview, though seventeen years old. The analyst must do this sort of work when the fourth fair use factor, “significant market effect,” is in question. In this case, our use could indeed negatively impact an important NPR income stream. Second, the amount proposed to be used (factor 3) clearly exceeds what is necessary for educational purposes: Ellis only appears in a few short clips within one of four segments in this hour-long program. Researchers interested in Ellis would only need those clips or at most that one segment. For all of these reasons, this collection earns a “high” risk rating and a “weak” fair use rating.

It is important to again emphasize that the strength of a fair use defense and the overall risk level are not synonymous. That is, a low-risk item does not always have a strong fair use defense; likewise, the use of high-risk items is not always weakly supported. Risk level and fair use defense strength must be considered separately.

Phase Three: Making Access a Reality

We were pleased with the results of our pilot project and proceeded to apply our review process to all our digitized recordings.

The assessments prepared through August 2017 cover 5,583 of 6,392 (87%) of the digitized sound recordings in the Bentley Historical Library. The BHL holds over 22,000 sound recordings on physical media; the assessments thus cover nearly a quarter of its audio holdings. Table 2 summarizes these findings.

Table 2. Item Counts per Risk Level–Fair Use Combinations

View Fullscreen

Access Levels

With this data in hand, we could clearly see that almost half of our digitized sound recordings are low or medium risk. We next needed to decide how much material to open and in what way. As the Bentley's mission is, first, to serve the educational and research needs of the university community, and in considering the capabilities of our current access platform, the Bentley Digital Media Library, the library decided to implement a tiered system of access that would allow three different degrees of streaming access to Bentley digital media content (see Table 3).

Table 3. The Three Access Categories

View Fullscreen

Next, we needed to decide which risk–fair use combinations to associate with each access category. Given that there are ten distinct combinations of risk level and fair use strength, we were able to make a more nuanced division of the access categories, applied as shown in Table 4.

Table 4. Relationship between Risk Level, Fair Use Defense, and Access Profile

View Fullscreen

Combining the data in Table 2 with the access profiles in Table 4, we see that 28% of the digital sound recordings we have reviewed could be opened to public access, and an additional 15% could be opened to the university community. The 13% of sound recordings that have yet to be reviewed appear to be composed of a comparable mix of genres as the portion that has already been reviewed. Overall, the Bentley has the potential of making 43% of our digitized sound recordings more available to our users.

Take-Down Policy

Under Section 108 of U.S. copyright law, the Bentley Historical Library has a right to provide on-site (i.e., “reading-room only”) access to digital surrogates of analog sound recordings. For all recordings, clear and actionable take-down procedures will allow the Bentley to mitigate risks associated with providing access and respect the prerogatives of rights holders. As Kevin L. Smith writes, “a responsive take-down policy will inevitably have the effect of preventing most complaints from ever becoming lawsuits.”³⁴ A good take-down policy is a near-universal practice among hosts of digital collections, and each of the five institutions studied for this project all had clear and easy-to-find notices.

Outcomes

The Bentley launched this project during the summer of 2016 with the aim of developing tools and a process to review sound recordings at scale and increase access to them. After a year, the Bentley had developed tools and a review process, and had completed a pilot review project for 5,583 recordings. This resulted in the following outcomes:

• Provided an opportunity to reconceptualize rights review, digitization, and curation work in an integral way, as well as associated workflows.

• Completed rights reviews for nearly a quarter of the Bentley's sound recordings.

• Increased access to 43% of the Bentley's sound recordings.

• Produced rights review documentation demonstrating due diligence and ability to facilitate reference requests and other inquiries regarding our sound recordings.

• Developed robust project documentation including rights review tools, reports, collection assessments, workflow guidelines.

Given the legal nature of this project, the Bentley is submitting this review process, and access determinations made, to University General Counsel for approval and additional institutional support.

Findings

Through this project we have also identified four key findings.

Music and Literature—Lower Priority

Our project has shown that certain genres have inherent traits that usually prevent expanding access and can be considered lower priorities for review. Musical performance and literary works can be very homogeneous compared to most genres, as both tend toward high risk. These show inherent risk and tend toward weaker fair use defenses. Review of musical performances showed that about two-thirds of music collections would remain limited to the reading room. Figure 2 also demonstrates higher risk levels for these two genres. For this reason, we recommend these two genres be considered a lower priority for review and reading room access be applied as the default access level. In this way, review efforts can be made more efficient by focusing attention on other genres that might result in wider access.

View Figure

• Download as Powerpoint
• Download Figure