Last week, oral arguments were heard in the summary judgement phase of the lawsuit brought by four commercial publishers against the Internet Archive (IA) for its National Emergency Library digital lending program. In prompt fashion, Federal Judge John Koeltl ruled on the motions for summary judgement only five days later. The resulting opinion was a stern rebuke of the Internet Archive and a significant victory for the four publishers.

Controlled digital lending (CDL) is really at the heart of this suit, more so than the focus on the emergency nature of the IA program. In fact, what is striking is that the case hardly mentioned that this was a temporary program in the depths of a lock down that ceased within months of being started. Publishers were not simply seeking to end a short-term program, but are seeking to limit the ability of libraries to engage in new approaches to existing rights that libraries have as established in copyright law.

A gavel and a stack of books with one titled Judgement
Judgement by Nick Youngson CC BY-SA 3.0

This is a situation where bad facts bring about bad case law. The Internet Archive was certainly trying to extend the understanding of existing Fair Use rights under copyright law with the extension of its National Emergency Library, but the practice of controlled digital lending had been in place and practiced by institutions for several years prior to 2020, when the IA pushed out its service to anyone, anywhere, at any time. It is instructive that the publishing community waited until 2020 to act against IA, when the facts were most aligned to provide them a concrete and expansive win. The publishers did not pursue the many libraries that have used much more limited applications of CDL to serve their constituents.

The Internet Archive has already said this decision will be appealed. Regardless of the outcome of the ultimate decision, this was an inevitability as both sides viewed this as a case headed toward higher courts. The value of this stage was ideally to avoid the time and cost that would be involved in a trial. The questions at hand really are not about the facts of the case, which are generally speaking, agreed upon. Judge Koeltl has a reputation for prompt judgements in these types of cases and the turnaround of the judgement shouldn’t be read as reinforcing the obviousness of the decision, as some have done. The legal question is focused on the boundaries between what can be done with physical objects and the circulation of those digital versions. IA was seeking to extend the boundaries of Fair Use exemptions in copyirght in an increasingly digital world. For the moment, those boundaries will remain entrenched, perhaps even more deeply so.

It is worth noting that the ecosystem of digital content that has developed over the past fifty years is different from the marketplace of physical books in several key details, and publishers are anxious to protect this new expansion of IP territory. One key difference between the two markets for the same content — one physical and one digital — is that the world of print and physical items is governed by traditional common law understandings of ownership and transfer, including concepts of first sale and exhaustion. A library has the right to circulate a book, fundamentally, because it “owns” the thing and can do what it likes (up to a point, as proved by this ruling), which includes sharing it with library patrons. The world of digital content exists in an economic realm of licenses and contract law in which traditional notions of ownership do not necessarily apply. Publishers can limit the number of times a digital item can be shared, they can require annual payments to continue access, they can demand any terms they like. In the absence of agreement to these imposed terms, publishers can refuse to sell the content at any price to a segment of the market. The contract-law focused world of copyright for digital content is much more heavily weighted to the benefit of publishers and to the greatest extent possible, publishers would prefer it remain there. It is to every consumer’s detriment that we continue to move ever more in this direction in all manner of our digital lives. In this case, as is in so many aspects, the law is woefully inadequate to address the modern digital ecosystem and how a significant portion of the public interact with content.

The implications of this ruling are potentially profound, and, given the strong lean in the publisher’s favor, they are potentially troubling for libraries and the rights of those who seek to engage with content in our evermore digital and digitized world if the decision stands through the forthcoming appeals. For the significant amount of content that exists in print form and for which there is no publisher-sanctioned digital version available, that content has become effectively walled off from the digital world until it passes into the public domain—essentially for longer than anyone reading this blog is alive. Those who live in close proximity to and have access to world-class institutions with sizable print collections can get access to much of this content. For the vast majority of library users, this will not be the case. Their access will be significantly curtailed, but to paraphrase the ruling, this public interest is secondary to the interests of publishers in exercising their monopoly. This issue is compounded by a significant amount of that content, which exists in the netherworld where contracts were negotiated before digital versions were even considered. The contracts for those works were not explicit in the grant of digital distribution rights, so the publisher withholds the content for fear of running afoul of their own rights to distribute a digital version.

What is most troubling in this decision is its lack of consideration of the fact that the Internet Archive is a library, and the existing rights extended therein, which is a core distinguishing feature of this case. Conducting circulation of content for library purposes to library patrons is one of the foundational aspects of any library. One could remove the “e” from “ebooks” and the same notional harms to publishers argued by the Plaintiffs in this case would still exist. People can and do argue that libraries harm publishers’ sales and profits, because as was quoted in the ruling “It is hard to compete with free”. Yet that is what libraries do, they acquire content and distribute it to patrons free of charge. And conceivably there exists a library patron that might have turned to a bookstore to purchase the item, resulting in this scenario to a lost sale for the publisher. Avoiding the question of what a library is and does as irrelevant in this case seems to me to entirely miss the point. Whether that reader is provisioned with a print book or a digital copy is a matter of form, not function, so long as the publisher is compensated for that sale, which in either case it has been. The Internet Archive is aggressively seeking to extend this argument and will push it up the Appeals Court ladder as far as they can to advance their aims. I am not certain, however, that the court system, as it is currently populated, will be receptive to the IA’s arguments.

It is certainly worth noting that there is disagreement within the library community about the validity of Internet Archive’s approach here. Not all libraries, probably not even a majority would go so far as IA has. While IA has its supporters, there are many who view CDL and IA’s National Emergency Library with a suspicious eye, thinking it was pushing the boundaries of traditionally acceptable behavior too far. That suspicion was reinforced Friday evening.

The judge’s focus on Internet Archive’s CDL activities while disconnecting it from the practice of libraries, was strained. The extensive cases referenced in the decision almost entirely center around commercial distribution of copyrighted content and the commercial benefit to the organization in question (Redigi, Weissmann v. Freeman, and American Buddah) as justification that the infringement is commercial. In an environment when physical circulation was impossible, what was the library to do?  Of course “Buy the book,” would be the response from the publisher and which millions of people did during the pandemic, with publishers’ income skyrocketing. The judge twisted himself into argumentative pretzels to define IA’s work as commercial. In a footnote (Page 16), the judge notes the potential defense under Section 108 of the Copyright Code, but that IA did not “justify its infringing acts under that section”.

Regardless of the outcome of this case on appeal, there remains a strong case there are legitimate uses of controlled digital lending that libraries are well within their rights to pursue. The best example of this is its application for accessibility purposes, when remediation usually involving digitization and secure distribution. This is a well-protected service that libraries can provide in alignment with Section 108 of the Copyright Act, with the Marrakesh Treaty, and as supported by case law, specifically, both in Sony v. Universal City Studios and Authors Guild v. HathiTrust. The decision last Friday also highlighted accessibility as a permissible application of library scanning and CDL. Other applications, such as for special collection materials, likely will remain within the purview of libraries, but these will be edge-case applications. The ruling even went so far as to acknowledge that not all of the actions of Internet Archive resulting from their digitization, such as indexing or display of snippets, are infringing activities and that therefore all of the digital scans might not need to be destroyed.

Another unintended consequence of this ruling will be the costs of interlibrary loan will remain unnecessarily and exceedingly high, even though the technology exists to reduce these costs and to do so in a secure and serviceable way. This ruling will scare most institutions off exploring ways to improve that practice through controlled digital lending.

The cost of ebook licenses in the library community will increase as a result of this decision as it affirms the publisher’s position that the domain of digital content is entirely in the hands of publishers  and that they have nearly unlimited control over that ecosystem when it applies to human consumption of the content. If you’re an author considering a book contract with a publisher, you should reflect on this and demand to see the benefits of those increased license fees. Publishers claim that they are engaged in this fight for the benefit of the authors. If that is truly the case, they then should welcome the opportunity to pass these benefits onto the authors.

This ruling inherently narrows fair use and relegates library use to either the aggressively controlled digital subscription ecosystem, a very narrowly constrained use case set, or the domain of the purely physical. The decision reinforces the notion that digital items should only be held in the world of licensing where ownership rights remain solely in the control of the copyright owner because digital objects are not sold. By emphasizing the distinctions between the physical and digital (even while simultaneously arguing that they are functionally equivalent), this ruling forces us further into a world where we will have to pay repeatedly for the same thing again and again because we rarely actually own anything digital, even if we think we do.


In full disclosure, NISO is advancing a project to create an Interoperable System for Controlled Digital Lending with support from the Andrew W. Mellon Foundation. The project is focused on the practical workflow issues around how CDL is implemented in a library. The developing recommended practice will explore issues such as content sequestration, access control, and workflow for provision of CDL. This project is explicitly NOT focused on the rights issues, the validity of, or selection process around what materials an organization might include in a controlled digital lending service. Regardless of the outcome of this case, there are circumstances where the process of CDL could be implemented, such as for special collections, or accessibility provision, or course materials, where limited duplication and circulation of content is explicitly allowed under copyright law and CDL processes might support that circulation. Because of this case, however, I expect most institutions will be much more circumspect in their application of this technological approach.

Todd A Carpenter

Todd A Carpenter

Todd Carpenter is Executive Director of the National Information Standards Organization (NISO). He additionally serves in a number of leadership roles of a variety of organizations, including as Chair of the ISO Technical Subcommittee on Identification & Description (ISO TC46/SC9), founding partner of the Coalition for Seamless Access, Past President of FORCE11, Treasurer of the Book Industry Study Group (BISG), and a Director of the Foundation of the Baltimore County Public Library. He also previously served as Treasurer of SSP.


16 Thoughts on "Controlled Digital Lending Takes a Blow in Court"

A library has the right to circulate a book, fundamentally, because it “owns” the thing and can do what it likes (up to a point, as proved by this ruling), which includes sharing it with library patrons. The world of digital content exists in an economic realm of licenses and contract law in which traditional notions of ownership do not necessarily apply.

I feel like there’s an important point embedded here that is too often overlooked in library conversations about IP, ebooks, first sale, and copyright — and I think it’s overlooked in part because when you try to talk about it, it seems abstruse and philosophical. But it’s really fundamental.

As Todd points out, a document like a book exists in two legal dimensions — but this is because its existence has two very different dimensions. There’s the book (i.e. the intellectual content itself), which has no existence in time and space but is nevertheless subject to laws that govern its disposition; and the “book” (i.e. the physical document into which the book is encoded), which does exist in time and space and is also subject to laws that govern its disposition — but very different ones.

First sale doctrine governs what may and may not be done with the “book”; copyright law governs what may and may not be done with the book. Too often, I think, our discussion of libraries and ebooks takes it as given that the book and the “book” are the same thing, when they aren’t, either legally or conceptually; this leads to confusion and frustration when (for example) we argue about what interlibrary loan should mean in the context of ebooks. Ebooks, by definition, are not “books,” but books (though “book” versions of them may also exist), and it wouldn’t make sense for the law to fail to recognize that.

This isn’t to say that there aren’t ways to manage ebooks as if they were “books” and thereby distribute and share them in ways that mirror the management of “books” under the first sale doctrine, thus keeping the managers within the bounds of the law — obviously, this is what CDL is meant to do, and I don’t see anything in the Koeltl decision that says that’s not possible. What this decision says is that Koeltl believes IA’s very expansive and aggressive approach to CDL took it outside those bounds. But this isn’t to say that all CDL is out of bounds.

My larger point, though, is that I think we in libraries serve ourselves poorly when we insist on talking about ebooks as if they were “books” and not books, and therefore insisting that ILL and other forms of sharing must look exactly the same in both the ebook and the print book domains.

#Aside. Rick, supporting your point, might note it is a little more complicated even. Using FRBR terms (as we have them) what you are describing are ‘work’ (intellectual content) and ‘item’ (physical book). Using the same model, in between are ‘expression’ and ‘manifestation’, whose definition may have been a little porous. This is where the intellectual content is ‘fixed’ in some particular form, and it is here in the middle that copyright resides. A ‘work’ cannot be copyrighted as it exists only in the ether? The ‘work’ may have multiple ‘expressions’/’manifestations’, which can be copyrighted, and whose variations may also be copyrightable (think illustrations, critical apparatus, etc). And a ‘manifestation’ will have multiple copies (‘items’), which sit on the shelves. Of course, there are other models also, but, yes, it is important to discriminate between senses in which ‘book’ is being used. The intellectual work Ulysses is (differently) realised in multiple expressions/manifestations which each have multiple physical instances. Of course, this has lots of practical implications as we know. When it comes to digitization, one may be interested in all the levels – I just want any Ulysses (work); I may want the Penguin Ulysses with a Kiberd introduction (manifestation); I may want the Ulysses in such-and-such a collection signed by so-and-so (item). From a (library) bibliographic description pov, the description is usually at the manifestation level, so additional work needed to cluster works, and additional item level identification needed to track copies. Etc. #evenmoreabstruse

“But this isn’t to say that all CDL is out of bounds.” – that’s probably right. Only the scenarios wherein one replicates everything and makes everything available worldwide (and at a cost) without permission of the copyright owner. The digital ‘one-stop-shops’ that exist legally at the time of writing all exist because they have consulted the rights owners. So the platform is effectively a licensee of multiple rights owners. Another element nobody seems to have considered is that there is a reason why publishers cannot just go and collaborate and jointly – as a group of publishers – create a one-stop-shop on their own. I’m sure that the applicable laws could also have become an issue in this scenario, but we never reached that point.

I appreciate Todd’s article but there is one point that I want to raise a caution on. Todd suggests that one valid use of controlled digital lending would be to provide access to disabled users. In the USA we do not need to rely on a practice of controlled digital lending to provide access to disabled users. Section 121 says clearly that it is lawful to make accessible copies for disabled users. Period.

If we understand “CDL” to be premised on maintaining an “own-to-loan” ratio, then we need to remember that 121 places no obligation on an “authorized entity,” such as a library, to own or to withhold from circulation a physical copy of a book that is made available in an accessible format to a disabled user. Such a requirement would add a burden to library providing such services, and could inadvertently delay or deter services to the eligible user. It is clear that we need to ensure that only eligible users have access to that specific accessible copy. But that is a matter for authorization and authentication, not controlled digital lending.

When section 121 was rewritten in 2018 to ensure compliance with the Marrakesh treaty, the accessibility and library communities worked very hard to oppose publisher efforts to make it harder for libraries to provide access to eligible users. 

There is no reason for us to add burdens to ourselves, especially in light of Judge Koeltl’s ruling in the Hachette v Internet Archive suit. The judge’s ruling acknowledged that IA’s service’s to the disabled are lawful, but they are not lawful because IA engages in controlled digital lending, they are lawful because they are enshrined in the law.

For reference:

Thank you Mike, for the clarification and the distinction. It is a subtle point, but an important one.

This is an important point and one that was raised at oral argument (though not viewed as persuasive by the court in its ruling), though I’d also point out that fair use also supports making copies for disabled users and can apply even where the strictures of Sec 121 don’t work. As the lawyer for IA attempted to point out in oral argument, in Authors Guild v. HathiTrust, it was fair use (not 121) at issue, and in that ruling the court held that fair use extended to a use case that in many ways goes beyond what CDL allows for: full text copies with no downstream restrictions or similar owned-to-loan controls.

Todd raises good questions here about how far Section 107 applies to libraries beyond what Section 108 provides, but I have to wonder how far he thinks fair use should extend to protect libraries’ digital practices under fair use, especially transformative use as it has been implemented since Judge Pierre Level came up with the idea in 1990. E.g., does he share the view of Jonathan Band that novels, monographs, scholarly journal articles, etc. are written for audiences outside the college classroom context and therefore should be considered as “repurposed” for use in the classroom and can be reproduced freely with no payment to the publishers as transformative use, a view embedded in the ARL’s Code of Best Practices in Fair Use in Academic and Research Libraries (2012). For a review of the path toward expansive fair-use rights, see my article for the Journal of Scholarly Publishing on “Reflections on Copyright Law and Scholarly Publishing over Fifty Years” here:

Of course it’s tactically wise to focus on publishers in your discussion: they can be seen as a sort of disembodied money-grubbing machine. But publishers act on behalf of authors, whose works they are obligated to protect from theft. Do you really think that authors should have to allow free access to their “property” just because it’s socially desirable (if it is in fact)? The Internet Archive’s motivation in providing access in the pandemic was no doubt worthy, but as it involved depriving authors (and publishers as their agents) of legally mandated income from their writing, it must be seen as illegal. There can surely be no fair use provision to the effect free access can be available if it’s deemed socially desirable, or just plain nice.

Publishers have worked hard on the accessibility issue for many years now, and were heavily involved in the discussions that led in WIPO to the Marrakesh Treaty— see… I think a fundamental issue here is that publishers and authors want to work on clearly identified problems (such as disability access), and rely on statutory law and the provisions in the US of Section 108. Folks that want to have more expansive and “flexible” solutions may look to Fair Use as a solution— but FU is not usually a solution for a broad societal need as it depends too much on individual facts. By the way in my view CDL is nothing more than a suggested set of practices that are designed to be interpreted, if challenged in court, as FU. It is a suggested approach to litigation risk assessment.

trial court decision is in southern district NY federal Court, part of the 2nd Circuit in the US… the potential appeal would be to the 2nd Circuit… but this is considered to be an influential court/Circuit especially on IP matters…

“This is a situation where bad facts bring about bad case law.” Isn’t it more about bad marketing and wishful thinking bringing about the wrong expectations? The case law is not bad, it’s actually quite straightforward and clear. The problem here is that Internet Archive did not adhere to conditions relevant to its own non-legally tested CDL invention. Advice: talk to authors/publishers, ask permission, have lawyers (not internet enthusiasts or techno-optimists) craft a robust CDL process, have authors/publishers agree to that, stick to the designed and agreed upon process. Anything else is just reckless behavior. It is still not clear to me why having a digital one-stop-shop and one-online-library-to-rule-them-all would benefit all the other libraries but that’s probably just me. Perhaps some folks like to have 2.6 million library apps on their phone.

My only quibble here would be that when it comes to exercising fair use rights, it’s not only unnecessary but also (I would argue) counterproductive to consult with copyright holders. It’s not copyright holders who get to decide whether a use of their work is fair; the parameters of fair use are defined (however fuzzily) in the law, and the copyright holder’s opinions about the fairness of a given use are irrelevant. To put it another way: a proposed use is neither fairer because the copyright holder approves, nor less fair because s/he disapproves.

If the user does consult with the copyright holder and they reach agreement about permission to use the work in a particular way, then they’ve moved out of the realm of fair use and into the realm of permissions and licenses. The difference between those two realms is critically important. IA is basing the justification for its behavior on fair use, and so for them it would have made no sense to consult with the copyright holders. (The fact that their argument failed is an important, but separate issue.)

It’s an example of what that process can achieve, but maybe more of a cautionary tale than something we might seek to repeat. Courts have not been kind to the Classroom Guidelines. Both the district court and the appellate court in the Georgia State case roundly rejected them as indicia of the outer bounds of fair use. Judge Evans of the district court characterized their restrictive limits as “undermin[ing] the educational objective favored by §107.” Cambridge Univ. Press v. Becker, 863 F.Supp.2d 1190, 1234 (N.D. Ga. 2012). Judge Tjoflat in the appellate court warned that “to treat the Classroom Guidelines as indicative of what is allowable would be to create the type of ‘hard evidentiary presumption’ that the Supreme Court has cautioned against.” Cambridge Univ. Press v. Patton, 769 F.3d 1232, 1273 (11th Cir. 2014). The Section 108 Study Group effort was similarly ill-fated.

” It’s not copyright holders who get to decide whether a use of their work is fair.” – objectively wrong, especially if the suggested fair use includes the reproduction of the work in full. But even if that view was accurate, it would mean the same would apply to the CDL proponents, no? It is then only up to the judge. “To put it another way: a proposed use is neither fairer because the copyright holder approves, nor less fair because s/he disapproves.” – Well, in som cases the suggested use case scenario is so obviously infringing (including unauthorized monetization) that it doesn’t require a rocket scientist to come to the conclusion that it is in breach of existing laws. From reading the court documentation it seemed like the latter was the case here. “IA is basing the justification for its behavior on fair use, and so for them it would have made no sense to consult with the copyright holders.” – it would have saved them time and costs I would think? And loads of disappointed people. The IA folks come across as very intelligent, so I think they took a (less well-calculated) risk here and it backfired. Copyright law is not THAT complicated.

Comments are closed.