Category Archives: Fair Use

We’re back! (and so is the GSU fair use e-reserves appeal…)

Dave Hansen
Dave Hansen, Director of Copyright and Scholarly Communications

After a few months of quiet, I’m happy to say that the Copyright & Scholarly Communication team at Duke is bringing this blog back to life. Since Kevin Smith left to become Dean of Libraries at the University of Kansas, I have stepped in to take over his old post as Director of the Office of Copyright and Scholarly Communications at Duke. If you’re interested in who I am and what I do, you can check out my new Scholars@Duke profile. I plan to use this blog to cover the same types of issues that Kevin did, especially copyright and publishing, as well as to highlight some of the interesting projects happening here at Duke on those same subjects.

I think it’s only appropriate that my first post is about the Georgia State University e-reserves copyright lawsuit. This blog has more or less chronicled this suit since it was filed in 2008. For anyone unfamiliar, the case is about whether it is fair use for GSU professors to make electronic excerpts of books available to students in their courses.  The plaintiffs Cambridge University Press, Oxford University Press and Sage argue that in most cases it is not fair use. So far, the courts have mostly decided in favor of GSU regarding the specific uses made by GSU and its faculty.

Publishers’ Brief on Appeal

Publishers' SECOND appeal is again before the 11th Circuit Court of Appeals
Publishers file opening brief in their SECOND appeal before the 11th Circuit

On Friday the Publishers filed their opening brief on appeal (their second appeal!) before the 11th Circuit Court of Appeals. The brief isn’t a huge development. But, it makes some fascinating, and concerning, fair use arguments that are worth discussing. On fair use the brief basically argues:

  1. The lower court didn’t balance the fair use factors correctly. It argues that the court placed too little weight on the fourth fair use factor (whether GSU’s uses harmed the market for the works).
  2. That the lower court made it too hard for Publishers to show that their markets were in fact harmed.

Market harm is critical for the Publishers because they don’t have much else left to argue about. Based on what the appellate court said on the last appeal, the first factor (educational in-class use) strongly favors GSU; the second factor (nature of the work) tends to favor GSU in many cases, though the appellate court said this factor doesn’t count for much, and the third factor (the amount and substantiality used) also tends to favor GSU in most instances

Balancing the Fair Use Factors

The Publishers make some good introductory points about how courts should weigh the fair use factors. The brief recites case law saying that the factors should be balanced together, that there are “no hard evidentiary presumptions” as to what types of uses may be fair, and that the district court should not take a “rigid” or “mathematical” approach to fair use (something that the district court was chided for the first time around on appeal).  So far, so good.

But then comes their core argument about market harm and fair use balancing, which I think fails. It starts by asserting that the lower court was too mechanical in its approach. The brief seizes on a statement made by the lower court, that it “estimates that the initial, approximate respective weights for the four factors as follows: 25% for factor one, 5% for factor two, 30% for factor three, and 40% for market harm.” (Dist. Ct. Op. at 14).

I agree that assessing the numerical percent value of each factor is somewhat unusual. What’s really strange, though, is that you would think that the Publishers would be happy with a mechanical approach under which market harm is weighed as 40% of the analysis—in other words, almost outcome determinative.

Market Harm Above all Else?

Apparently almost outcome determinative isn’t good enough. In a section of the brief that contrasts starkly with those opening (and mostly accurate) statements of the law about balancing the factors together, Publishers argue that no use can be fair use if there is any market harm:

“This (literally) calculated choice of numerical weights . . . produced findings of fair use even where the court found market harm under factor four that weighed against fair use. Indeed, the court made clear its resistance to granting dispositive significance to factor four by stating that while factor four would be given “additional weight,” factor three was “critical” because it was “at the vortex of the holistic evaluation required by the Court of Appeals’ Opinion.”

Like some of the other positions the Publishers have raised in this case, this approach might make sense 30 years ago. In 1985, the Supreme Court announced that the market harm factor is “undoubtedly the single most important element of fair use.” But, the Supreme Court has since walked back this approach. It stated clearly in 1994 in Campbell v. Acuff-Rose that “all [factors] are to be explored, and the results weighed together, in light of the purposes of copyright.”

In the first GSU appeal, the 11th Circuit recognized that Campbell is controlling:  “This language [from Campbell] appears to be inconsistent with any single factor being deemed the single most important.”  The 11th Circuit reasoned, however, that in this case because “the threat of market substitution is severe, it is appropriate in this instance to afford relatively great weight to the fourth factor in the overall fair use analysis.” In fact, the lower court obeyed that direction. It gave market harm 40% weigh in each determination, more than any other factor. It just declined to make market harm outcome determinative, (worth 51% or more, in other words) which is what the Publishers wish for.

Takeaways for Educational Users

The weight given to market harm—even the difference between 40% and 51%, if you want to try to attach a percent value on it—is critically important for educational users. We now live in a world where books never go out of print (print-on-demand). License services are proliferating, and one can buy permission to do an increasingly broad array of activities, such as text or data mining. If courts accept that even the tiniest lost licensing revenue in those markets constitutes a “harm” that is judged to be determinative in the overall fair use analysis, then the rights of educators to use copyrighted works is diminished significantly. We’ll have to wait and see what the 11th Circuit does, but I hope it sees through this argument and upholds the decision of the lower court. 

Here we go again: latest GSU ruling an odd victory for libraries


My first thought when I read the new ruling in the Georgia State copyright lawsuit brought by publishers over e-reserves was of one of those informal rules that all law students learn — don’t tick off your judge.  From the first days of the original trial, the arrogant antics of the attorneys representing the publisher plaintiffs — Oxford University Press, Cambridge University Press, and Sage Publishing — clearly put them in a bad light in the Judge Evans’ eyes. Those chickens came home to roost in this latest opinion, especially where the plaintiffs are chided for having filed a declaration about what licenses were available for excerpts back in 2009, even after the Judge told them not to, since that information had not been introduced as evidence in the original trial.  All of that evidence was stricken, and the Judge based her new opinion on the evidence that was before her in that first trial.  I can imagine that the publishers might use that ruling as a basis for yet another appeal, but if they do so, they had better be able to prove that the evidence is genuine and reliable, and to explain why, if it is, they did not produce it at trial back in 2011.

But I have put the cart before the horse; let’s look at the ruling we just received from the District Court.  In case some have lost track, this case was originally decided by a 2012 ruling by Judge Evans that found infringement in only five of 74 challenged excerpts, and awarded court costs and attorney’s fees to GSU as the “prevailing party” in the case.  The publishers appealed that decision to the Eleventh Circuit Court of Appeals, which vacated the trial court holding in 2014, sending the case back to Judge Evans with a set of instructions on how to improve the fair use analysis for these challenged excerpts.  As has been noted many times before, the publishers lost nearly all of the big principles they had wanted to establish in the case; the Court of Appeals refuted most of the publishers’ arguments even as it did what they asked and vacated the first ruling.

Now, using the new fair use analysis directed by the Court of Appeals, Judge Evans has handed the publishers yet another loss.  One wonders how many times they will have to lose this case before they finally learn something about the state of copyright law today.  Still, this loss for the publishers is only the oddest sort of victory for libraries.

The new fair use analysis that Judge Evans uses is carefully designed for situations where the challenged use in not transformative; the non-transformative nature of the use means that the small portions used must be scrutinized very carefully, and it means that the fourth factor — the potential impact of the use on the market for or value of the original — gets extra weight.  It is very important to notice this fact, because it means that this analysis used by Judge Evans will not be applicable in many other situations, especially in academia, where the uses are, unlike with e-reserves, transformative.

Even though both the trial court and the Court of Appeals have held that e-reserves are not transformative, both courts have affirmed that the first fair use factor — the purpose and character of the use — will still favor fair use when that purpose is non-profit and educational.  So throughout this new decision, Judge Evans continues to hold that the first factor always favors fair use.

The analysis of the other factors has changed, however.  For factor two, the nature of the original, Judge Evans does not make a blanket assumption, owing to instructions from the Eleventh Circuit, but looks at the nature of each excerpt.  In most cases, she finds that informational matter is mixed with more individualized scholarly commentary, and the result is that this factor is usually neutral — neither favoring nor disfavoring fair use.  In the few cases where it counts against fair use, it has little impact (the Judge says this factor is only 5% of each decision).

In the same way, factor three now gets a more careful and specific analysis.  The 10% or one chapter rule that Judge Evans used in her first opinion is gone, at the instruction of the Court of Appeals.  Instead, Judge Evans looks at each excerpt and evaluates its appropriateness to the allowable purpose (from factor one) and its potential to substitute for a purchase of the book (market substitution, anticipating factor four).  In many cases, she finds that the excerpts are a very small number of pages and a small percentage of the entire work (so not a market substitute), are are also narrowly tailored to accomplish a specific teaching objective.  In those cases, this factor will favor fair use.

Factor four, which the Judge now believes should constitute 40% of the fair use decision in this particular situation, is where most of the action is in this ruling.  The analysis, the Judge says, is two-fold, looking at both harm to the potential market for the original and harm to the value of the work, which means looking at the importance of the licensing market.  About this latter source of potential value, the Judge says that she must decide “how much this particular revenue source contributed to the value of the copyright in the work, noting that where there is no significant demand for excerpts, the likelihood of repetitive unpaid use is diminished” (p. 9).  The result of this inquiry is that a lot of financial information about each book — its sales over time and the amount of revenue derived from the permissions market — is very important in the fourth factor analysis.  The charts for many of the books that reflect this information make for fascinating reading, and contain information I suspect the publishers would rather not have made public.  This is where it becomes most difficult for libraries to apply the analysis that Judge Evans is using, because the Court has access to information, and time to analyze it, that is not available to libraries as they consider e-reserve requests.  Still, I think it is important to note that the standard the Judge is using in this evaluation is pretty high and it is focused on value to the authors and to users:

[W]e must determine how much of that value (the value of the work to its author and the potential buyers) the implied licensee-fair users can capture before the value of the remaining market is so diminished that it no longer makes economic sense for the author — or a subsequent holder of the copyright — to propagate the work in the first place (page 8, quoting the 11th Circuit).

In other words, this analysis is opening up a significant space in the idea of market harm, which permits potential fair users to diminish the value of the work in question to some degree, as long as that reduction in value is not so steep as to discourage writing and publishing these academic books.  Licensing, in this analysis, is the remedy only for that kind of steep loss of value; it is not a mere right of the copyright holder to obtain all the value from the work that is possible.

Judge Evans applied this complex formula for fair use to 48 challenged excerpts.  It was only 48 because for 26 of the ones discussed in her original ruling she found that there had been no prima facie case for copyright infringement made out, either because the publishers could not show they held the copyright or because there was no evidence that any students had used the excerpt.  This part of the ruling was not challenged, so only these 48 fair use rulings had to be redone.  Bottom line is that she found fair use for 41 of the 48, and infringement only in seven cases.  As Brandon Butler points out in his discussion of the ruling, even that might overestimate the infringement, since it appears that the summary in the decision may list at least some instances of infringement that were actually found, in the specific analysis, to be fair use.

So this ruling, like each ruling in the case, is clearly a disaster for the plaintiff publishers.  Once again it establishes that there is significant space for fair use in higher education, even when that use is not transformative.  Nevertheless, it is a difficult victory for libraries, in the sense that the analysis it uses is not one we can replicate; we simply do not have access to the extensive data about revenue, of which Judge Evans makes such complex use.  So what can libraries do, now that we have this additional “guidance” about e-reserves from the courts?  I think there are two fundamental takeaways.

First, we should continue to do what we have been doing — making careful fair use decisions and relying on those decisions when we feel the use is fair.  While we do not have much of the information used by the Court in this latest ruling, we still do have the security provided by section 504 (c)3 of the copyright law, which tells us that if we make good faith fair use decisions we, as employees of non-profit educational institutions or libraries, are not subject to statutory damages.  This greatly lowers our risk, and adds to the disincentive to sue academic libraries that must surely stem from the GSU experience.  All we can do, then, is to continue to think carefully about each instance of fair use, and make responsible decisions.  We still have some rules of thumb, and also some places where we will need to think in a more granular way.  But nothing in these rulings need fundamentally upset good, responsible library practice.

The second takeaway from this decision is that we should resort to paying for licenses only very rarely, and when there is no other alternative.  The simple fact is that the nature of the analysis that the Court of Appeals pushed Judge Evans into is such that licensing income for the publishers narrows the scope for fair use by libraries.  To my mind, this means that whenever we are faced with an e-reserves request that may not fall easily into fair use, we should look at ways to improve the fair use situation before we decide to license the excerpt.  Can we link to an already licensed version?  Can we shorten the excerpt?  Buying a separate license should be a last resort.  Doing extensive business with the Copyright Clearance Center, including purchase of their blanket campus license, is not, in my opinion, a way to buy reassurance and security; instead, it increases the risk that our space for fair use will shrink over time.

[personal note — this will certainly be my last word here at the Scholarly Communications @ Duke site, and it is fitting that it deal with the Georgia State case, about which this site has seen so much commentary.  For the sake of continuing this conversation and including more discussants, I am also going to post this piece on the new group blog of which I am a part, IO: In the Open.  Apologies if the repetition is annoying to anyone.]

Steal this book?

Last week I was researching a copyright and fair use issue for a faculty member, and needed to see a copy of a book held by Duke’s Rubenstein Rare Book and Manuscript Library.  As I explained the issue and what material I wanted to use to the Rubenstein staff, a researcher sitting nearby listened intently. As soon as we finished, she told me that she was the President of the Authors Guild and that they were suing Google over fair use.  She began to explain to me why Google was wrong, but that the author for whom I was doing the research should be allowed to rely on fair use.  When I introduced myself as a lawyer and copyright specialist for the Libraries, the conversation came to a polite but stilted conclusion.

This week, however, I got a chance to see more fully what that researcher, whose name is Roxana Robinson and who was giving a lecture that afternoon in the Library, has to say about Google, in a column she wrote for the Wall Street Journal called “How Google Stole the Work of Millions of Authors” (behind a paywall).  Ms. Robinson, a novelist and biographer, unfortunately proves what I suspected at the time of our encounter, that her perspective on fair use is based on a preconceived idea about who are good users entitled to rely on fair use (authors) and who are bad, unworthy users (Google), rather than on an understanding of the careful legal analysis of specific uses that actually underlies these decisions.

The WSJ column employs some interesting rhetoric, starting with its title, which is clearly intended to provoke a visceral response.  Many people have noted that the language of theft and stealing is inappropriate when the issue is copyright infringement.  This point is made in great detail in William Patry’s book “Moral Panics and the Copyright Wars.”  As is true for most crimes, the definition of theft includes an intention, a mental state or “mens rea” that is a required element of that crime.  For theft this intention is “to deprive the true owner of [the personal property]” (definition from Black’s Law Dictionary, Seventh edition).  Because of the nature of intellectual property, copyright infringement never meets this definition; that is why the law has a different word — infringement — for the unauthorized taking of someone else’s IP.

So the headline of Ms. Robinson’s column is legally incorrect and intended, I think, to stir up her base rather than to make an argument that could sway the Supreme Court (for more on this point, see the rebuttal published in Fortune “Why the Authors Guild is Still Wrong about Google’s Book Scanning“).

The column also makes a couple of sardonic remarks about quotes that can be found using Google Books.  Here the argument breaks down pretty badly, because both of the quotes Ms. Robinson chooses, one from Shakespeare and one from Emerson, are in the public domain.  Her effort to be ironic seriously backfires here, because her own column is actually proving the utility of the Google Books database in a way that emphasizes its lawful use of PD texts.  Rhetoric has truly overcome logic.

It is worthwhile, nevertheless, to think a minute about the logic structure of the argument that what Google has done is infringement.  Ms. Robinson makes the point that there are many books that were scanned by Google, that Google is a profitable company, and that no authorization for the scanning was asked for or given by the authors of the works that were scanned.  All of this is true, of course, but it does not amount to an argument that Google has infringed any copyrights.  What is missing, at least as I see it, is any notice that the authors have been harmed.  The rhetoric of the column clearly tells us that the Authors Guild, and at least some individual authors who are involved in the lawsuit, are angry.  But it does not explain a fundamental element of any tort action — harm.

The two courts that have considered this case both found that there was no harm done here — no negative impact on the market for or value of the works in question, to use the language that is part of a fair use analysis.  Users cannot obtain any significant portions of books that are limited to snippet views; the AG’s own experts were unable to retrieve as much as 16% of any work using word searches and snippet results, and even that amount of text was randomized in a way that made reading a coherent piece of the work impossible.  The is just no evidence that any sales are lost due to this finding aid, and it is quite possible that sales will be gained.

There is, of course, the question of a licensing market.  But that is almost a silly question.  A market for licensing scans to create an index has never existed, and it is impossible to imagine that any of the authors had such an idea in mind when they wrote their works.  As Judge Leval said in his decision for the Second Circuit Court of Appeals, this is not really even a use of the work, it is a use of information about the work, for which a secondary licensing market simply is not appropriate.  Creating such a market would be revolutionary, and it would do much more harm to the overall environment for books and reading than anything Google could think up.  What the Authors Guild seems to be saying here is that Google should pay us for something we never thought we would or should get paid for, simply because they have a lot of money.  Perhaps when we recognize how weak that argument actually is it becomes understandable that Ms. Robinson relied on overheated rhetoric rather than legal or logical arguments.  But if the purpose of her essay is to convince people that the Supreme Court needs to take the case to right a serious wrong, it falls far short, and is unlikely to convince the nine citizens whose opinion on that issue matters the most.

Copyright MOOCs, new and refreshed

When my colleagues Anne Gilliland and Lisa Macklin and I released our first Coursera MOOC about copyright, called Copyright for Educators and Librarians, we were very pleased with the reaction.  Although our enrollment for that first MOOC was, at just over 10,000 participants, rather low by MOOC standards, we had a higher than normal percentage of completions, and the feedback we got from colleagues was quite positive.

That course ran in the summer of 2014.  In July of 2015, we were able to release a new version of the same course in an on-demand format, meaning that participants are able to start the course whenever they wish and can proceed at their own pace without a proscribed ending point.

The move to on-demand is important because it brought us a bit closer to our overall goal, which has been to provide a form of copyright education that is accessible in the several sense of that word to all of our colleagues in education, especially.  The course is still free, although there is a small fee if the participants want to receive a “verified certificate” of completion.  We began this project aware that the Center for Intellectual Property at UMUC had recently closed, so the education community had lost access to their series of course offering on copyright that carried continuing education credit.  Our hope was to provide an opportunity to learn about copyright that was free to all, but also could be used, through the verified certificates, by those colleagues who want to learn about the subject AND get some form of (less expensive) credit for this professional development activity.

Now we have taken another big step toward that goal, with the release today of our second MOOC, on Copyright for Multimedia.  Like the first course, this MOOC is on-demand, free to take, and relatively short – four substantive modules and an introduction.  In this second course, the modules focus on four different media – data, images, music and film.  It grew out of our awareness how often the questions brought to us focus on different media.  Many of our colleagues seem confused about how copyright “rules” from the print world, apply in an environment rich with diverse forms of expression and communication.  This confusion is understandable, since copyright was born with print technology and continues to adapt only uncomfortably to these “new” media.

When we are asked about what “copyright for music,” or “copyright for film,” looks like, we try to emphasis that the one copyright law in the U.S. is intended to apply without regard to medium of expression.  Nevertheless, it is perfectly true that some provisions of the law are media-specific.  More significantly, the circumstances in which different media are used are often quite different from the more familiar facts surrounding the use and distribution of print.  There is an lawyer’s maxim that says, “change the facts and you change the outcome,” and that is never more true that when we are talking about different media.

Our new MOOC tries to address these differences, and also to further develop the framework for analyzing a copyright issue that we built in the first course.  Now that both MOOCs are available on the Coursera platform, we hope that they will be a continuing resource to improve copyright understanding for our colleagues.

I want to add a couple of personal notes to this announcement of the two-part series of MOOCs on copyright.

First, I want to say what a wonderful experience it has been to work with Lisa and Anne, who are as smart and creative about teaching as they are about copyright, as well as with the online course team at Duke.  I want especially to note my sense of awe at the creative, complex and realistic scenarios that Anne Gilliland can think up to tease out the implications of copyright in different situations; I hope our participants find them as thought-provoking and amusing as I do.

Second, because of the announcement issued today about my new position as Dean of Libraries at the University of Kansas, and thus my departure from Duke, it seems unlikely that I will participate in any more MOOCs in this series.  Our original plan was for three courses, but the two we now have stand alone and, we hope, also work together as a series.  It is now an open question whether there will be a third MOOC in this series, but the process of creating these two has been delightful, and the product, I profoundly hope, useful to our colleagues and to many others.

Google Books, Fair Use, and the Public Good

Note — thanks to several readers who pointed out that I had carelessly misspelled Judge Leval’s name in my original posting.  That error has now been corrected.

On Friday the Second Circuit Court of Appeals issued its ruling in the appeal of the Authors Guild lawsuit against Google over the Google book search project.  The decision was a complete vindication of the District Court’s  dismissal of the case, affirming fair use and rejecting all of the counterarguments offered by the Authors Guild.

As it happens, I was traveling when the decision came down, confirming a troubling tendency of the federal courts to issue important copyright opinions when I am out-of-pocket.  (My wife says that it is not about me, but what sense does that make?)  In any case, that slight delay allows me to benefit richly from the analyses posted by some very smart colleagues.  Here are several great places to read about the decision:

From Brandon Butler of American University.

From Corynne McSherry of the Electronic Freedom Foundation

From Krista Cox of the Association of Research Libraries

From Carrie Russell at the American Library Association

I want to add, or really just pull out from these previous posts, three points that I think are especially important.

First, Judge Pierre Leval, who wrote the opinion, does a nice job of drawing a line from the idea of transformative uses to the public purpose of copyright law.  This is hardly surprising, since it was Judge Leval who wrote the 1990 article that coined the term transformative use and had such an influence on the Supreme Court in its 1994 decision in Campbell v. Acuff-Rose Music.  In this ruling, Judge Leval reminds us quite forcibly that the primary beneficiary intended by copyright law is the public, through “access to knowledge” (p.13) and “expand[ed] public learning” (p. 15).  Economic benefits for authors are instrumental, not the ultimate goal of the copyright monopoly.  Then Judge Leval explains how this analysis of transformation serves those goals, clarifying why fair use is an essential part of copyright’s fundamental purpose.  He tells us that transformation is an answer to the question of how a borrowing from a copyrighted work can be justified.  The court, on behalf of a rights holder, asks a user “why did you do this?”  When the answer to that question is “because I wanted to make a new contribution to knowledge,” that is a transformative purpose.  And, by definition, it is a purpose that benefits the public, which justifies whatever minor loss a rights holder might suffer from the use.  The second step in Judge Leval’s  analysis, asking if the new use is a market substitute for the original, ensures that that loss is not so great as to outweigh the benefit. Thus we have a coherent analysis that recognizes the public purpose of copyright and still respects it chosen method for accomplishing that purpose.

Another important thing we can learn from Judge Leval’s opinion is about the difference between a transformative use and a derivative work.  The Author’s Guild (really some individual authors set up as plaintiffs because the AG has been found to lack standing to sue in this case) argues that allowing the Google Books’ search function usurps a right held by those authors to license indexing of their works.  This is ridiculous on its face, of course — imagine the effect such a right would have on libraries — but the judge does a nice job of explaining why it is so wrong.  The decisions rest heavily on the idea/expression dichotomy that is fundamental in copyright, and stresses that what is presented in the Google Books “snippet view” is more information about books (facts) rather than expressive content from those books.  A derivative work, Judge Leval suggests, is one that represents protected aspects — the expressive content — of the original in an altered form (such as a translation or a movie script).  A transformative use, on the other hand, uses information about the works, as in an index, or uses their content for a different expressive purpose, as in parody or scholarly comment.  This is a difficult distinction to make, as all of us who work in copyright know all too well, and it remains to be seen if the approach outlined above will hold up or prove useful in the full range of situations.  But it is a pointer toward a coherent way to understand a difficult part of the copyright balance.

As an aside, while reading the opinion in this case I was struck by how well the four fair use factors were handled, in a way that showed that the test used by Judge Leval respected all of the factors while essentially applying two basic questions — is the use transformative and does the new work create a market substitute for the original.  In fact, I can suggest three specific passages that are especially exciting, I think, for the application of fair use and the issue of transformation — footnote 21 and accompanying text, which helpfully clarifies the relationship of the second fair use factor to the analysis of transformation; the full paragraph on page 33, that considers the use and misuse of the third factor; and the careful distinction of Google snippets from a case involving telephone ringtones that is found on pages 40-41.  These are discussions that I think will have a significant impact on our ongoing consideration of fair use.

Finally, we should note that the Authors Guild has already indicated its intention to ask the Supreme Court to review this decision.  This is a very bad idea, indicating that the AG simply does not know when to cut its losses and stop wasting the money provided by its members.  The real point, however, is that the Supreme Court is not likely to take the case anyway.  This is not a situation where a fundamental Constitutional issues is involved, as it was in the Campbell case (fair use as a protection for free expression) nor one where a fundamental point about our obligations in the international arena was at issue, as it was in the Kirtsaeng case about the application of first sale to works of foreign manufacture.  In short, this is just a case about a greedy plaintiff who wants to be given an even bigger slice of the copyright pie, which the courts have determined repeatedly it does not deserve.  This is not the sort of issue that attracts the very limited attention of the Supreme Court.  In fact, reading the Court of Appeals’ ruling leaves one with a sense that many of the AG’s arguments were rather silly, and there is no reason to believe they would be less silly when presented to the Supreme Court in a petition for certiorari.

There are some who have argued that there is a split among the Circuit Courts of Appeal over transformative use, which is also a situation that can lead to Supreme Court review.  But that split has always been predicated on the idea that other courts, especially the Ninth Circuit, have carried the idea of transformation too far and departed from the ambit of the original doctrine.  The fact that it is Judge Leval, the author of that approach to fair use, who wrote this opinion, effectively undermines that claim.  In short, this decision closes a circle that outlines a capacious and flexible approach to fair use.  For getting us to this point, I suppose we should thank the Authors Guild for the unintentional support they have provided for a balanced copyright law in the digital age.

Photography, Fair Use and Free Speech

All of us take a bad picture now and then.  You know, one of those pictures of yourself that makes you cringe every time you see it.  Honestly, I don’t think I have liked a picture taken of me since 1995.  But most of us react mildly to a bad picture; we moan a little and move on.  Raanan Katz, however, takes more dramatic action.  Katz, who is a real estate developer and part-owner of the Miami Heat basketball team, has been trying to erase an unflattering picture of himself from the Internet for the past three years.  As is so often the case with this kind of legal action, it has had the ironic effect (often called “the Streisand Effect”) of drawing more attention to the photo than it would have received if ignored.  But it is Katz’s method of trying to make the picture go away that is the reason for this post; he bought the copyright in the picture and brought a lawsuit for infringement.

Those who want to know more about the history of this effort, and see the offending photograph, can read this short story from Forbes magazine.  But I want to focus on a couple of specifics in the holding, which upheld the lower court’s ruling that the use of the photography by the defendants was fair use. There are a couple of items in the Eleventh Circuit’s fair use analysis that are worth attention, as well as an interesting connection with the Georgia State copyright case.

The connection with Georgia State is mostly found in the fact that the opinion in Katz v. Google was written by Judge Tjoflat, who also wrote the appellate opinion in Cambridge University Press v. Georgia State.  The Judge seems to be getting more comfortable with the fair use analysis, although it is also arguable that this one was an easier case.  The GSU opinion actually makes a cameo appearance in the Katz ruling; there is a footnote in which Judge Tjoflat acknowledges Katz’s claim that the trial court failed to do the necessary case-by-case analysis of the challenged works, and cites to GSU for the proposition that such analysis is required.  The Judge dismisses this, however, by finding that the magistrate whose “Report and Recommendation” was adopted by the lower court, had taken account of each of the different uses that the defendants had made of the ugly photo.  He adds that “the district court was not required to write a prolix, unwieldy opinion with 25 separate sections devoted to each alleged instance of infringement.” (FN 3)  One can’t help but read that comment as an oblique reference to the massive decision that Tjoflat and his colleagues confronted in the GSU case; I wonder what Judge Evans, who presumably is preparing a similarly lengthy opinion on remand, might make of that comment?

More substantively, the Eleventh Circuit ruling in Katz raises an interesting point about copyright in photographs.  While discussing the second fair use factor, the nature of the original work, the court finds that the photograph in question is “primarily factual,” a finding that supports fair use, by helping to tip that one factor, out of four, in its favor.  The court explains this holding, while acknowledging that photography may require many creative decisions, this way, “The Photo, however, is merely a candid shot in a public setting, and there is no evidence in the record that Magriso, the photographer, attempted to convey ideas, emotions, or in any way influence Katz’s pose, expression, or clothing.” (p. 9).  This language seems, to me, to suggest that copyright is somewhat weaker, or at least more susceptible to fair use, in photographs that simply  attempt to record events that take place in public, as opposed to deliberately artistic photos.  I wonder how news organizations and photojournalists feel about this.  While it makes sense, I think, it also suggests a difficult line.  If copyright is less protective for such “merely” journalistic photos, is there a point where we should not grant copyright at all, as we don’t, for example, where a photograph merely “slavishly reproduc[es]” a two-dimensional public domain artwork.  The question of how much originality is enough in a photograph, which by its nature is often a record of “facts” such as the appearance of people or the external world, to meet the standard for copyright seems less easy here than it is, perhaps, for other media.  But perhaps the answer is that we will protect most photographs, while recognizing that there may be a lower bar for fair use because of this particular and peculiar nature of the art form.

The other point in this ruling that I found very interesting is the analysis of the fourth factor, where the court found that Katz’s purchase of the rights and registration with the Copyright Office, entirely for the purpose of suppressing the picture, was evidence that the challenged uses did not do any harm to a cognizable market.  In short, because the owner’s purpose was to prevent all use, fair use becomes more likely, since it shows there is no market, because no intent to ever license the work.  This logic casts some doubt on the argument often made by rights holders that they have a right not to allow the work to be used.  That is true, but it is still subject to fair uses, which by definition are not copyright infringement — they do not intrude on rights that the copyright owner actually holds.  While the reasoning seems a bit difficult here, for me it is the most important point in this decision; the market harm factor clearly plays its part in the “safety valve for free speech” function that courts often assign to the whole fair use analysis.  Simply put, fair use gets a boost whenever the issue before the court is this alleged right to suppress; copyright gives the rights holder a lot of power over a work, but it does not convey the right to entirely prevent protected speech, even when (especially when!) that speech is critical of,or distasteful to, the rights holder.

PS — this post was mostly written before I learned of yesterday’s ruling in the dispute over an alleged, and now quite doubtful, copyright in the song “Happy Birthday To You.”  That decision is much more widely covered than the one discussed above, and, because it turns on such unique and difficult-to-establish facts, seems to have less impact on daily copyright issues than the Katz case does.  So while I wanted to acknowledge the decision, and provide a link to the ruling for anyone interested who has not already seen it, I decided to stick with my original plan to discuss Katz today.

Ignore fair use at your peril!

One could be forgiven for thinking of the “Dancing Baby” case as a thing of the past.  It seems a long time ago that a district court affirmed that the music heard in the background (for only 29 seconds) of a video in which Stephanie Lenz’s baby is dancing — the music was Prince’s “Let’s Go Crazy” — was fair use, and told Universal Music that they were required to consider fair use before sending out take down notices under authority of the DMCA.  The video, after all, was posted in 2007.  Lenz sued Universal, arguing that, because they failed to consider even an obvious example of fair use before sending a take down notice for the video to YouTube, they had committed copyright misuse under section 512(f) of the Copyright Act by knowingly misrepresenting the legal situation in such a notice.

There was, in fact, an initial ruling that rejected some of Universal’s attempts to exonerate itself, back in 2010.  Then, in 2013, the judge rejected cross motions for summary judgment, essentially allowing the case to go forward on the misrepresentation claim.  Universal appealed this dismissal, and the Ninth Circuit Court of Appeals issued its decision today.  The Court of Appeals affirmed the lower court ruling, agreeing that the DMCA requires rights holders to consider fair use when sending take down notices and letting the suit go forward to determine if Universal was “willfully blind” to the clear fact that Ms. Lenz’s use was authorized by the law.

The thing about the decision that I really want to bring to readers’ attention is what it says about the nature of fair use.  About a year ago I wrote a blog post about the idea that fair use was an “affirmative defense.”  I pointed out that many positive rights, including free speech, would manifest themselves in court as affirmative defenses, meaning that they would be asserted by a defendant to answer a complaint, but that they were still positive rights — a space for freedom of action, not merely an excuse.  In its new ruling in Lenz, the Ninth Circuit makes this point abundantly clear, telling us that, “Fair use is not just excused by the law, it is wholly authorized by the law.” (p. 11 of the PDF)  In fact, the court is so clear and eloquent on this point that I want to quote a whole paragraph, which is found on page 13 of the PDF and which states the situation regarding fair use and affirmative defenses much better than I did:

Universal’s sole textual argument is that fair use is not “authorized by the law” because it is an affirmative defense that excuses otherwise infringing conduct. Universal’s interpretation is incorrect as it conflates two different concepts: an affirmative defense that is labeled as such due to the procedural posture of the case, and an affirmative defense that excuses impermissible conduct. Supreme Court precedent squarely supports the conclusion that fair use does not fall into the latter camp: “[A]nyone who . . . makes a fair use of the work is not an infringer of the copyright with respect to such use.” Sony Corp. of Am. v. Universal City Studios, Inc., 464 U.S. 417, 433 (1984).

Given that 17 U.S.C. § 107 expressly authorizes fair use, labeling it as an affirmative defense that excuses conduct is a misnomer.

So perhaps it is time to stop scaring people with this language about fair use as an affirmative defense.  As courts keep telling us, fair use is a positive right, an authorized and vital part of the copyright law in the U.S.  We cannot have one half of that law — exclusive rights for creators — without the other half, which is fair use and the other exceptions that protect productive uses.  Without the latter, copyright would be mere protectionism, and, in all likelihood, unconstitutional.

Can this gulf be bridged?

Litigants in court cases often disagree sharply about the law and its application to the facts, so it is probably not a surprise that the briefs filed in the District Court’s re-examination of its ruling in the Georgia State copyright infringement trial should see the issues in such starkly different terms.

If you read the publishers’ brief, the 11th Circuit decision that sent the case back to the District Court changed everything, and every one of those 70 excerpts found to be fair use at trial now must be labeled infringement.  This is absurd, of course, and I don’t actually believe that the publishers expect, or even hope, to win the point.  They want a new ruling that they can appeal.  In my opinion the publisher strategy has now shifted from an effort to “win” the case, as they understand what winning would mean, to a determination to keep it going, in order to profit from ongoing uncertainty in the academic community (and, possibly, to spend so much money that GSU is forced to give up).

On the other hand, the brief from Georgia State, filed last Friday, argues that all 70 of those challenged excerpts are still fair use.  It seems likely that the actual outcome will be somewhere in the middle, and, to be fair to them, GSU does recognize this, by making a concession the publishers never make.  For a number of excerpts where a digital license was shown to be available at the time of the trial, GSU argues that the available licenses were not “reasonable” because they force students to pay based on what they are getting access to, whether or not the specific excerpt is ever actually used.  This is an interesting argument, tracking a long-standing complaint in academic libraries.  If the court accepts it, it would dramatically restructure the licensing market.  But GSU also seems to recognize that this is a stretch, and ends several of its analyses of specific excerpts by saying that the specific use “should be found to be fair if the Court finds the licensing scheme unreasonable, and unfair if the Court finds the licensing scheme reasonable.”  So it seems GSU is prepared for what I believe is the most likely outcome of this reconsideration on remand — a split between fair uses and ones that are not fair that is different than the original finding — probably with some more instances of infringement — but still a “split decision.”

The availability of licenses is one of the interesting issues in these briefs.  The publisher plaintiffs now argue that licenses were available, back in 2009, for those excerpts where the judge said no licenses were “reasonably available.”  They are continuing to try to introduce new evidence to this effect; which is something GSU vigorously opposes.  But those of us who have been involved in e-reserves for a while remember clearly that such licenses were not available at all through the CCC from Cambridge University Press and only occasionally from Oxford.  So what is this new evidence (which the publishers’ brief says was not offered before because they were so surprised that it was being requested)?  It is an  affidavit from a VP at the CCC, and my best guess is that it would argue that licenses were “reasonably available” because it was possible, through the CCC system, to send a direct request to the publisher in those instances where standard licenses for digital excerpts were not offered.  GSU argues that the evidence gathering phase of the case is over, a ruling about licenses has been made and affirmed by the Court of Appeals, and the issue settled.  A lot will depend on how Judge Evans views this issue; so far she has ruled against admitting new evidence.

Another controversy, about which I wrote before, involves whose incentive is at stake.  The Court of Appeals wrote a lengthy discussion of the incentive for authors to write, and its importance for the fundamental purpose of copyright.  To this they appended an odd sentence that says they are “primarily concerned… with [publisher’s] incentive to publish.”  The publishers, of course, hang a lot of weigh on this phrase, and take it out of context to do so.  GSU, on their side, make a rather forced argument intended to limit the impact of the sentence.  Neither side can admit what I believe is the truth here: that that one sentence was inserted into an opinion where it does not fit because doing so was a condition of the dissenting judge for keeping his opinion as a “special concurrence” rather than the dissent it really was.  If I am right, this compromise served the publishers well, since they can now cite the phrase from the actual opinion of the Court; it is seldom useful to cite a dissent, after all.  So the publishers quote this phrase repeatedly and use it to argue that all of the factors really collapse into the fourth factor, and that any impact at all, no matter how small, on their markets or potential markets effectively eliminates fair use.  Authors, and the reasons that academic authors write books and articles, do not appear in the publishers’ analysis, as, indeed, they could not if the argument for publisher hegemony over scholarship is to be maintained.

GSU, as we have already seen, takes a more balanced approach.  For the first factor, they discount the publishers’ attempt to make “market substitution” a touchstone even at that point in the analysis, and focus instead on the 11th Circuit’s affirmation that non-profit educational use favors fair use even when transformation is not found.  The GSU brief fleshes this out nicely by discussing the purpose of copyright in relationship to scholarship and teaching.  On the second factor, GSU discusses author incentives directly, which in my opinion is the core of the second factor, even though courts seldom recognize this.  GSU also points out that the publishers have ignored the 11th Circuit’s instruction, both here and in the third factor analysis, to apply a case-by-case inquiry to those factors; instead, the publishers assert that since every book contains some authorial opinion, the second factor always disfavors fair use, and that no amount is small enough to overcome the possibility of “market substitution.”  For their part, GSU introduces, albeit briefly, a discussion of the content of each excerpt (they are often surveys or summaries of research) for the discussion of factor two, and of the reason the specific amount was assigned, in regard to factor three.

As I said, these differences in approach lead to wildly different conclusions.  Consider these paragraphs by which each side sums up its fair use analysis for each of the excerpts at issue:

The publishers end nearly every discussion of a specific passage with these words — “On remand, the Court should find no fair use as to this work because: (1) factor one favors fair use only slightly given the nontransformativeness of the use; (2) factor two favors Plaintiffs, given the evaluative/analytical nature of the material copied; (3) factor three favors Plaintiffs because even assuming narrow tailoring to Professor _____________’s pedagogical purpose, it is counterbalanced by the threat of market substitution, especially in light of the repeated use; and (4) factor four “strongly favors Plaintiffs,” and is entitled to “relatively great weight,” which tips the balance as to this work decidedly against fair use. ”

On the other side, GSU closes many discussions (although there is more diversity in their analysis and their summations than in the publishers’) this way — “Given the teaching purpose of the use, the nature of the work and the decidedly small amount used, the fact that this use did not supplant sales of the work, and the lack of digital licensing, the use of this narrowly tailored excerpt constituted fair use.”

These are starkly contrasting visions of what is happening with these excerpts and with electronic reserves, as practiced at a great many universities, as a whole.  It will be interesting, to say the least, to see how Judge Evans decides between such divergent views.

Learning how fair use works

How many cases about fair use have been decided in the U.S. since the doctrine was first applied by Justice Story in 1841?  Take a minute to count; I’ll wait.

If you came up with at least 170, the Copyright Office agrees with you.  Last week they announced a fascinating new tool on the CO website, an index of fair use cases.  That index contains summaries of approximately 170 cases, along with a search tool.  The introductory message, however, acknowledges that the index is not complete, so those of you who thought there were more than 170 cases are almost certainly correct.

This index is potentially a very useful tool, and it also raises some interesting questions.  I want to consider the questions first, than discuss how the new fair use index might be used by someone who wanted to learn more about how fair use works (which, by the way, is one of the avowed purposes behind its development).

Coverage is the obvious first question, and, as I said, the CO acknowledges that it is incomplete.  Specifically, it seems heavily weighted toward more recent cases.  There are only 11 cases listed in the index dating to before 1978, and two older cases (1940 and 1968) that are presented in my law school casebook on copyright as important fair use precedents are not included.  So it looks like there are some pretty significant gaps, which one hopes the Copyright Office will address as it continues to develop this tool.

By the way, the issue of continuing development also brings up the question of why the C.O. thought this was a worthwhile investment. It looks useful, to be sure, but there are other sources for similar data, so it is a bit curious that the C.O. chose this among all its potential priorities.

To return to the issue of coverage, it is always important to ask which specific cases were chosen and how they are characterized.  Of the 170 cases, there are 78 for which the result is listed as “Fair use not found,” and 64 in which the C.O. says that fair use was found.  The remaining 29 are listed as “Preliminary ruling, mixed result or remand.”  This last category is rather unhelpful.  For example, the Authors Guild v. HathiTrust case is listed this way, even though it was a strong affirmation of fair use and the remand involved a fairly unimportant issue of standing.  Even more surprising is the fact that this “mixed result” tag is applied to Campbell v. Acuff Rose Music, the “Oh Pretty Women” case from the Supreme Court that is at the heart of modern fair use jurisprudence.  Again, this was a clear fair use win; the case was remanded only because that is what the Supreme Court usually does when it reverses a Court of Appeals’ decision.  So the representation of the holdings is technically accurate, it seems, but not as helpful as it might be in actually focusing on the fair use aspect; while some “mixed result” case genuinely were that — fair use was found on one issue but not for another — many of the remanded cases actually did involve a clear yea or nay about fair use, and it would be more helpful to categorize them that way.

A particularly useful feature of this index is the ability to limit the listings by jurisdiction (the Appellate Circuits) and by topic.  For example, limited to just cases out of the Fourth Circuit, where I reside, I find that the Court has ruled on seven fair uses case and upheld fair use in six of them.  The seventh was one of those genuinely mixed results, where one challenged use was held to be fair and another was not.

If we limit the subject area of the cases to those labeled “Education/Scholarship/Research,” fair use seems to fare better than it does overall.  In that category of 42 total cases there are 18 findings in favor of fair use and 16 rejections.  Of the remaining 8 mixed results, at least two of them — the HathiTrust Case and the GSU case — should be seen as affirmations of fair use, even if the parameters of that use are still unsettled in GSU.  So the impression many of us have that educational and scholarly uses are a bit more favored in the fair use analysis than other types of cases seems to be confirmed.

Things get more interesting when we look just at the Supreme Court in this index, and the issue of how cases are chosen is again highlighted.  The index shows four fair use cases, with one holding in favor of fair use (Sony v. Universal Pictures), one mixed result (Campbell, as discussed above), and two rejections of fair use (Harper v. Nation and Stewart v. Abend).  This last case, Stewart v. Abend, is actually almost never treated as a fair use case; while fair use was dismissed as a potential defense in the case, the real issue involved assignments of copyright and who could exercise the renewal right that existed at that time.  And this case was remanded, just as Campbell was.  So it is odd that Campbell, with its central finding in favor of fair use, is shown as a mixed result, while Stewart v. Abend, where fair use was tangential and there was also a remand, is tagged as a rejection of fair use.  This suggests at least an unconscious bias against fair use findings.

A different listing of Supreme Court fair use cases, on the IP Watchdog site, includes several additional cases — nine, in all — but does not list Stewart v. Abend as one of them. Several of the cases included by IP Watchdog do not seem to me to really focus on fair use, so I am not saying that the C.O. has under-reported the cases.  But the very different lists do suggest that it is a surprisingly subjective undertaking just to identify the cases that should be included in a fair use index.

Finally, the analysis provided in the C.O.’s case summaries needs to be considered carefully.  To take one example, for the recent case of Kienitz v. Scoonie Nation, about which I wrote earlier this year, the short note about the holding ignores the thing that may be most significant about the case — the reluctance of Judge Frank Easterbrook to apply a “transformation” analysis to the fair use question (HT to my friend and colleague David Hansen for pointing this out).  Again, this is not necessarily a problem, and the case summary of Kienitz at the Stanford Copyright & Fair Use site has a similar synopsis, but it is a reminder that these projects are always created by individuals with specific perspectives, viewpoints and limitations.

Even with all these caveats, I think the Copyright Office has created a useful tool, which can be used by those interested to learn a lot about how fair use is applied, especially by looking at the different categories.  The Stanford site, linked above, and especially its own, much shorter list of cases, might usefully be used alongside the C.O. index.  The Stanford descriptions are  very tightly focused on the fair use issue, so reading them in conjunction with the C.O. summaries, with their attention to procedural matters that sometimes obscure the fair use holding, might produce a more balanced approach.

In any case, this new tool form the Copyright Office, and some of the tools that predate it, remind us that the best way to understand fair use, and to become comfortable with it, is too look closely at the cases, both in the aggregate and individually.  This C.O. database offers a statistical perspective, as well as the ability to focus on parody, or music, or format-shifting, while the Stanford summaries emphasize in a few words the core of the fair use analysis.  Both point the interested reader to full opinions, where the analysis can be understood in the context of all the facts.  Combined in this way, these resources offer a terrific opportunity for librarians, authors, and others to dig deeply into the nuances of fair use.

Steps toward a new GSU ruling

It looks more and more like we will get a new ruling from the trial court in the Georgia State case about what is or is not fair use for digital course readings.  The case, of course, was reversed and remanded to the trial court after the publishers appealed the initial decision to the 11th Circuit, with instructions to produce a new opinion consistent with the Court of Appeals ruling.    The publisher plaintiffs then asked the trial court to reopen the record in the case and apply the putative new fair use analysis to a different, more recent, set of readings employed by the GSU faculty.  The University opposed this motion, arguing that what would amount to a whole new trial was not necessary.

Last week, District Court Judge Orinda Evans dismissed the motion to reopen the record and issued an order about briefing the court on what a new analysis of fair use for the original excerpts considered in the trial should look like.  Judge Evans wrote that “It does not make sense at this juncture to spend months, probably longer, on what considerations might govern if Plaintiffs prove they are entitled to injunctive relief by virtue of the claimed 2009 infringements.”  The motion is dismissed without prejudice, meaning that the plaintiffs can renew it at a more appropriate time, although I must admit that I do not see what that would mean if the case is to go forward on the original set of readings.

It appears that once again the publishers have failed in an effort to broaden the scope of the case beyond the item-by-item fair use analysis that has already been done and to possibly reintroduce some of the broad principles that they really want, which have so far been rejected at every stage.  Now Judge Evans has explicitly told them, in her scheduling order, that what is required is “consideration and reevaluation of each of the individual claims” in order to redetermine “in each instance… whether defendants’ use was a fair use under 17 U.S.C.  section 107.”  Her schedule for the briefs is tight, with an end of the briefing now scheduled just two and a half months from now.  Presumably we would still have a long wait while Judge Evans applies revised reasoning about fair use to each of the individual excerpts, but it looks a bit more like that is what is going to happen.