Authors Alliance, joined by the Library Copyright Alliance and the American Association of University Professors, filed a petition with the Copyright Office for a new three-year exemption to the DMCA as part of the Copyright Office’s eighth triennial rulemaking process. Our proposed exemption would allow researchers to bypass DRM measures in order to conduct text and data mining research on both literary works that are published electronically and motion pictures. Further details can be found in the full text of the petition, available here.

Text and data mining allows researchers and others to gain new insights into language and culture, scientific inquiry, and civic participation. For example, text and data mining can be used to examine the evolution of language over time or to identify important but overlooked findings in scientific papers.

While the possibilities of text and data mining are great, researchers face limitations in their ability to use the technique. They must either rely on collections of works created by others, which may be missing important, relevant works, or they must build their own collections, a prohibitively time-consuming process in which researchers manually scan printed books or capture video playback in real time.

Authors Alliance believes that researchers building their own collections of works should be able to bypass DRM measures to extract data directly from sources such as eBooks and DVDs. The DMCA prohibits such circumvention, even if the intended use would not infringe anyone’s intellectual property rights. If this exemption is granted, researchers will be empowered to build relevant collections and conduct this important research.

How you can help

Historically, the Copyright Office has scrutinized the factual record when deciding what exemptions ought to be granted. Assuming the Copyright Office considers our exemption request, we will need to make a more detailed submission to the Copyright Office later this year that fully sets out the legal and factual basis for the exemption. If your research has been impacted by the inability to access electronically published literary works or motion pictures, we want to hear from you. Please reach out to us at info@authorsalliance.org and share your story.

Authors Alliance is grateful to Clinic students Jason Francis and Alistair McIntyre, supervised by Clinic Director Catherine Crump and Teaching Fellow Gabrielle Daley, at the Samuelson Law, Technology & Public Policy Clinic at Berkeley Law for their assistance filing this petition and for this blog post.

The Internet Archive has responded to a copyright lawsuit filed by a group of commercial publishers which takes aim at the Controlled Digital Lending (“CDL”) model and the Internet Archive’s (now closed) National Emergency Library. The Internet Archive’s answer to the publishers’ complaint highlights the fair use arguments underpinning the digitize-and-lend model, which has been in operation since 2011 with the support and participation of hundreds of other libraries.

Under the CDL digitize-and-lend model, libraries make digital copies of scanned books from their collections available to patrons (the hard copy is not available for lending while the digital copy is checked out, and vice versa). A library can only circulate the same number of copies that it owned before digitization. Like physical books, the scanned copies are loaned to one person at a time and are subject to limited check-out periods. The Internet Archive launched National Emergency Library in March in response to the COVID-19 outbreak which left the physical collections in libraries inaccessible to patrons; books available through the National Emergency Library were not subject to the “owned-to-loaned” ratio. The National Emergency Library closed on June 16.

The Internet Archive’s answer to the publishers’ complaint explains that the digitize-and-lend model serves the public interest in preservation, access, and research—all classic fair use purposes. Every book in the collection has already been bought and paid for by the libraries that own them, and most of the volumes are out of print.

In a recent statement, Internet Archive Founder & Digital Librarian Brewster Kahle emphasized that “libraries have the right to buy books, preserve them, and lend them even in the digital world” and that learners need digital access to books that libraries own physically. “Controlled Digital Lending is a respectful and balanced way to bring our print collections to digital learners. A physical book, once digital, is available to only one reader at a time. Going on for nine years and now practiced by hundreds of libraries, Controlled Digital Lending is a longstanding, widespread library practice.”

Copyright expert and Authors Alliance Board President Pamela Samuelson agrees, stating “it’s really tragic that at this time of pandemic that the publishers would try to basically cut off access to a digital public library like the Internet Archive is running.”

Authors Alliance has long been a supporter of the CDL model, which helps authors share their creations with readers, promotes the ongoing progress of knowledge, and advances the public good—objectives that are consistent with the mission of Authors Alliance. Several of our members have spoken out in favor of the model. CDL is particularly beneficial for authors whose works are out of print or otherwise commercially unavailable: In the absence of digitizing and lending these books, many would simply be inaccessible to readers. The CDL model is a boon to the authors of these and other books, allowing them to find new audiences online.

Today, Authors Alliance joined with allies to petition to renew the existing exemption to Section 1201 of the Digital Millennium Copyright Act (DMCA) that allows authors to bypass encryption to make fair use of film clips in nonfiction multimedia e-books.

While copyright law generally provides for exceptions like fair use that allow authors to use copyrighted works without permission in certain circumstances, some courts have held that these exceptions do not apply to the anti-circumvention provisions of Section 1201. Evading digital rights management (DRM), even when done for otherwise lawful purposes, may be prohibited by law. However, every three years, the Librarian of Congress is empowered to approve discrete, temporary exemptions from the law in order to carve out space for non-infringing uses caught up in the DMCA’s broad sweep.

In 2015 as a part of the U.S. Copyright Office’s sixth triennial rulemaking session, Authors Alliance joined with author Bobette Buster, the American Association of University Professors, and counsel from legal clinics at the UC Irvine Intellectual Property, Arts, and Technology Clinic and the Samuelson-Glushko Technology Law & Policy Clinic at University of Colorado, Boulder to successfully petition for an exemption that allows authors to circumvent digital locks on Blu-rays, DVDs, and digitally transmitted videos to make fair uses in multimedia e-books offering film analysis. In the seventh triennial rulemaking, the Librarian of Congress expanded the exemption to all nonfiction multimedia ebooks.

Now, as the eighth triennial rulemaking cycle begins, our team has submitted a petition to renew this exemption. Authors Alliance believes that multimedia e-books are an important form of authorship and wants to see authors empowered to fully realize their promise. The freedom to author e-books that incorporate multimedia content, consistent with the core tenets of the First Amendment and academic freedom, remains significant and important. We will continue to track the progress of the rulemaking and provide updates as they become available.

We are grateful to the student attorneys and their supervisors at the Intellectual Property, Arts, and Technology Clinic at UC Irvine School of Law and the Samuelson-Glushko Technology Law & Policy Clinic at University of Colorado School of Law for their work supporting this exemption.

Authors Alliance is grateful to David Hansen, Associate University Librarian for Research, Collections, and Scholarly Communications at Duke University, for sharing this blog post based on his statement to the Senate Committee on the Judiciary, Subcommittee on Intellectual Property on whether the notice and takedown procedures under Section 512 of the Digital Millennium Copyright Act (DMCA) are working. This statement was originally presented on June 2, 2020.

Hansen’s testimony addressed how universities and research libraries interact with Section 512 as both service providers serving a large number of users and as rightsholders who produce creative copyrighted content. The excerpted portion below addresses the first perspective: that of universities and libraries as rightsholders.

* * *

Promoting the creation and dissemination of knowledge is an important part of what our university and our libraries do. At Duke alone, our faculty and other researchers author more than 10,000 articles every year, along with hundreds of books, reports, video content, software, visual works, learning resources, educational programs, and many other types of materials. For virtually all of this content, our primary aim is to get as many people as possible to read and engage with the ideas we are sharing to help increase our collective understanding of the world around us, and of each other. These works of authorship, more than almost any other, fall at the heart of what our Constitution states is the objective of Copyright Law: “to promote the progress of science and the useful arts.”

In most cases and for most of the published research Duke produces, we aim to disseminate these works with no direct financial return; no royalty. If possible, our authors generally want no financial barrier to stand in the way of engagement with their research, operating under the idea that more and faster progress will be made without those barriers. In many cases, we find ourselves licensing around the controls that copyright law automatically provides. For example, more than ten years ago, Duke Faculty voted to adopt an institutional open access policy that provides for free, widespread distribution of research articles that Duke faculty have authored.

Duke leaves the ultimate decision on how to disseminate scholarship to the individual authors. Many authors post their materials to Duke systems (for example, DukeSpace, our institutional repository). But, many also share through nonprofit repositories such as ArXiv or bioArxiv, as well as commercial sites such as ResearchGate and Academia.edu. Because of the variety of content, and the desire to engage our research with the public, we also share content through more popular sites such as YouTube.

Given our interest in widespread dissemination of ideas, for research and academic work our strong preference is a system that is biased toward keeping content up online unless there is strong evidence that an infringement has occurred. The current notice and takedown system does not always accomplish this goal.

First, for some academic works, the ownership of rights is far from clear. Although authors are the holders of those rights initially, they are often asked to license them away at least in part through publishing contracts that are confusing and vary significantly from journal to journal and which can change with some frequency. As a result, some academic authors are unsure of whether they are legally permitted to share their own work online under the terms of their publishing agreement. Many research articles are also subject to pre-existing licenses that attach automatically upon creation—for example, at Duke under our Open Access policy—which provide that authors and their institutions retain certain rights to share and reuse their work. My experience with takedown requests we receive at Duke is that publishers do not take into account pre-existing open access licenses even though their existence is widely known. In the case of a takedown request for an article an author has posted to an online platform, authors can feel uncertain how to respond since they may be unsure whether they actually have retained the necessary rights to distribute or reuse their own work.

These takedown efforts to remove content posted by authors can be highly disruptive. In our role as a service provider at Duke, we only receive a few such requests each year, but other online hosts of scholarly content have become targets. In 2017 the commercial publisher Elsevier, a Dutch-owned publishing conglomerate, reportedly issued 100,000 takedown notices to ResearchGate. ResearchGate is a for-profit site, but most of the content is submitted by academic authors to share for free with other researchers and the world. Subsequently, Elsevier and one other publisher, the American Chemical Society, sued ResearchGate for copyright infringement, identifying more than 3,000 articles they claimed rights to. At least some of those articles are likely to be covered by preexisting university open access licenses. At present ResearchGate reports that it hosts 150,000 Duke authored articles. Unfortunately, Section 512 currently contains few mechanisms to address these asymmetries of power and information in the notice and takedown process. For most academic authors I work with, including faculty but especially graduate students, responding to a takedown notice is an intimidating and time-consuming process that most will try to avoid if at all possible.

Second, and perhaps the most important thing I can convey, is how important fair use is for research, teaching, and for libraries that support those functions. Most research is highly iterative, building upon the work of others. Often, for scholarly publication that means one must reference earlier work by copying—whether by simple quotation in a literary critique, by copying charts or graphs in a scientific publication, or by reusing images in a work commenting or criticizing art or history. These are all common place examples of fair use that academic authors and teachers rely on every day. Indeed, I have found that I seldom review a scholarly work that does not rely on fair use in some way. While the courts have consistently found that fair use supports these kinds of core scholarly and teaching uses, other provisions of the Copyright Act can make exercising the fair use right challenging. For scholarly authors who want to share their work through online platforms, Section 512 is one of them.

Section 512 does not explicitly address how fair use factors into the notice and takedown process. The Ninth Circuit’s decision in Lenz v. Universal Music Group Corp., 815 F.3d 1145 (9th Cir. 2016) (the “dancing baby” case) was a welcome development, as it mandated that a rightsholder first consider fair use in order to assert the required good faith belief of infringement when making a takedown request. However, in practice we know that in many instances automated content identification systems are the first method of assessment, and they do not handle fair use assertions well.

This is predictable given the fact-intensive balancing that fair use requires. It is an “equitable rule of reason” that requires careful consideration of several factors. For a recent, ironic, example, YouTube’s ContentID system reportedly flagged a video of a panel discussion of a law school copyright conference hosted by NYU’s Engelberg Center on Innovation Law & Policy. It identified multiple claims of infringement. The videos included several short clips of popular songs, which were necessary for the musical experts on the panel (experts from the well known “Blurred Lines” case) to include in order to explain to the audience how to analyze songs for similarity. Although NYU had a strong fair use claim, ContentID had no way of understanding.

While automated systems certainly have their place, it is important to have adequate processes in place to protect those users, such as academic authors, who rely heavily on fair use when sharing their own research with the world.

* * *

To read more about Hansen’s views on how Section 512 works for universities and libraries as service providers and for his suggestions for how Congress should approach changes to Section 512, read Hansen’s full statement here.

Authors Alliance thanks Professor Casey Fiesler and Corian Zacher for sharing this guest blog post.

* * *

On May 21st, the Copyright Office released a report documenting the results of a study five years in the making. Its goal was to evaluate the current effectiveness of Section 512 of the Digital Millennium Copyright Act, which immunizes internet providers and websites that host user generated content from copyright infringement liability—as long as they provide a system that allows copyright holders to request that infringing content is removed. In practice, Section 512 has enabled websites like YouTube to continue operating without facing excessive legal costs and copyright infringement judgments. In other words, every time you see a “this video is no longer available due to a copyright claim” message, you’re seeing the results of a compromise—because without DMCA 512 takedown procedures, those copyright owners might have just sued YouTube out of existence instead.

One obvious conclusion of the study was that “no potential solution(s) will please everybody” (pg 68). However, the problem is that “everybody” as defined by the Copyright Office study seems to include only two categories of stakeholders: online service providers (OSPs, like YouTube) and large-scale copyright owners like movie studios and record companies. Even in concluding that “the notice-and-takedown system as experienced by parties today is unbalanced” (pg 72) (which we agree is true), nowhere do we see the experiences of another relevant category: content creators who make fair uses of copyrighted content. The Copyright Office virtually ignored these “transformative creators,” along with other kinds of Internet content consumers.

However, these relevant stakeholders are also deeply impacted by the lack of balance in DMCA 512. The fact that platforms like YouTube have entrusted copyright analysis with algorithms rather than human content moderators has resulted in the improper removal of many non-infringing works, particularly those that transform copyrighted content in a way that constitutes fair use—for example, remix videos, or videos that reference movie clips in order to critique a film or educate students through distance learning. The Copyright Office acknowledges that the goal of Section 512 is balancing the needs of all relevant parties, but at the same time, it vastly underestimates the extent that improper takedown notices affect users. The nod to “inaccurate notices” (when non-infringing content receives a takedown request) even focuses on the cost incurred to the OSP—with no acknowledgment of the cost to the non-infringing content creator (pg 148).

However, there is a very real cost to transformative content creators as well, and they should be a relevant stakeholder in policy discussions. Inaccurate takedown notices have a tremendous impact on all creators’ ability to share their work online, and those repeatedly accused of infringing may lose content and accounts as a result. These stakes are particularly heartfelt by small creators, who rely on the reach of platforms to share their content.

In 2017, we collaborated with the Organization for Transformative Works to conduct a survey of fan creators that focused on experiences with and attitudes about copyright, including the DMCA. There were over 2,000 survey respondents, and our qualitative analysis of these answers revealed a number of salient points drawn from real-world experiences with DMCA takedowns. Our conclusion is that between chilling effects and improper takedowns, DMCA 512 procedures have resulted in a great deal of non-infringing transformative content being lost from the internet.

The imbalance here is one of power. Most instances where participants reported that their works were accused of infringing another’s copyright involved fanvids (a type of remix video), generally on YouTube. Participants suggested that YouTube’s application of Section 512 adversely impacts works that constitute fair use because YouTube’s automated system flags content that contains any amount of copyrighted work. After all, if even judges and lawyers reasonably disagree about what constitutes fair use, how could an algorithm make that judgement? Moreover, when a video is taken down because a platform’s automated system flags it, the burden shifts to users to have their work reinstated. Faced with that burden, participants also wrote about how “scary” the process is; even YouTube’s “Copyright School” video suggests that remix video creators should consult a lawyer and emphasizes that they could “get into a lot of trouble” if they file a counter-notice incorrectly. Additionally, fan creators frequently use pseudonyms (as do many content creators) and might be wary of providing their real name on a legal document that would be connected to their account pseudonym, as is required by a counter-notice.

We heard from a number of participants who had their transformative content flagged and decided not to try to file counter-notices, because they were concerned about doing it incorrectly, they were confused by the process, or lacked the time to try to figure it out. However, from the few that did, they were often successful—suggesting that platforms remove content without considering whether the work was a fair use or not, or the work would not have been flagged in the first place. Considering the number of users who simply don’t fight it when their work is taken down, we can reasonably conclude that platforms are frequently removing non-infringing fanworks. Moreover, repeated experiences of takedowns often lead to users abandoning a platform altogether.

Overwhelmingly, the fan creators represented in our survey viewed the lack of balance in the takedown process as stifling their creativity and hindering their ability to share their work. The harm of failure to consider fair use was particularly powerful: Participants whose works were removed explained that they believed that their works constituted fair use, offered examples of uses that fit squarely within fair use by being noncommercial and being used for criticism and parody, and discussed how removal of those works harmed their individual creativity and ability to participate meaningfully in their community.

By assuming that all users are “infringers” or “pirates” and therefore an adversary to the 512 process, the Copyright Office does a disservice to the many content creators who reasonably and legally rely on fair use. Instead, they should be considered a stakeholder along with OSPs and copyright holders, such that their benefits and harms are part of the balance as well.

* * *

Casey Fiesler is an Assistant Professor of Information Science at University of Colorado Boulder, where she studies online communities, governance, and ethics. Her dissertation work for a PhD in human-centered computing (following a law degree) examined the chilling effects of platform design and policy on transformative content creators. She is a member of the OTW legal committee.

Corian Zacher is a May 2020 graduate of the University of Colorado Law School. Her research focuses on technology law and policy.

In March, Authors Alliance submitted comments to the U.S. Copyright Office in response to its request for public input on the meaning of “publication” in the online context. This week, we submitted reply comments to support our position.

In our initial comments, we encouraged the Copyright Office to adopt guidance stating that “publication” occurs when a work is first offered, under the rights-owner’s authority, for viewing online without technological restrictions that prevent downloading or other reuse. This position is supported by the statutory text, the legislative history, the case law, and the policy goals of (1) clear, bright-line guidance and (2) promoting broad use of works by reducing the availability of statutory damages and fee-shifting, which chill use and expression.

Our reply comments explain and expand on the stakes of “forfeiture of rights” under Section 411, a provision of the Copyright Act that provides that a civil action for copyright infringement cannot stand where the plaintiff knowingly included inaccurate information in its copyright application that would have caused the Copyright Office to deny that registration. We explain why this provision and case law underscore the necessity of clear and bright-line guidance, and how the intuitive and unambiguous approach to online publication proposed by Authors Alliance responds to that critical issue. Our comment also explains how a careful examination of the effects of registration errors shows that the stakes are not that high, and most rights can be substantially restored after dismissal due to an invalid registration, with the exception of the already undesirable right to statutory damages.

Authors Alliance thanks the exceptional team at Latham & Watkins for preparing these comments. To read the full comment, click here.