Commemorating 50 Years of the Copyright Act, part 3

STLR Panel 2: Litigating Fair Use in Copyright

Zahr Said: Substantial similarity is confused. Sedlik is a good example. Laudable for concurrences to recognize need for reform in this crummy, confusing test, but that’s not reflected in the case itself.

First, because the court’s diagnosis of what went wrong is mistaken. Second, the case didn’t go wrong: it was correctly decided. Third, this is a really unusual case, not a run of the mill © case.

Court confused copying and substantial similarity. It made a number of errors along those lines, ignoring intermediate copying—“it’s the photo tattooed on an arm.” P’s counsel encouraged the conflation of the photo and tattoo, even though Ds only conceded that the reference photo Kat von D used was 100% identical to the Sedlik photo. This matters b/c it’s calling for reforms, and the court said it was representative of a too defendant-favorable standard, and that’s wrong. If you treat the unusual as the common, we’ll be mired in further doctrinal mud.

Court thinks the outcome is wrong for the right reasons. But it was right! (I agree and filed an amicus focused on this.)

The timing was also weird—Warhol came down in the middle of it and the court had to revise its analysis. What to do: don’t build reforms on a landfill! Change the language; stop relying on Nimmer’s treatise and use Goldstein on infringement instead—precise, clear, elegant. Clearer thinking.

(or?) Dispense with the test! That’s what the concurrences were thinking about doing anyway.

Daniel Goldstein: Represented blind scholars & National Federation of the Blind in HathiTrust. Kudos to Peter Jaszi. Early commercial ebooks were inaccessible; not interested in the market. University of Michigan, by contrast, considered accessibility key to the digitization of their library, even before Google got the message. Librarians are the best. Intervened as a defendant in the litigation, and university didn’t oppose even though he’d sued universities for ADA violations in the past; publishers didn’t oppose, probably because they didn’t want to lose the first motion in the case.

The romance of the case was important; lots of excitement about blind researchers being able to access works for the first time. And the fair use argument was strong: dicta in Sony about accessible books for the blind being fair use; Chafee amendment; etc.

The publishers wanted to impound the entire library; the National Library Service can only generate about 2000 accessible books/year but publishers never suggested just giving it a copy. ADA was also relevant as a clear & comprehensive national mandate for accessibility/against discrimination against people with disabilities—universities have to offer the same experience when feasible and this was a way for schools in HathiTrust to make the research program available to blind/print disabled students. Why didn’t the publishers ask for the digitization program to pause during settlement discussions? Not sure—seems like a bad call. Their arguments consisted of saying that the Chafee Amendment/Sony dicta was only about making a single book for a single blind person accessible and that universities weren’t authorized to make accessible copies.

Corynne McSherry: Public Resource v. ASTM: Teeny nonprofit that makes government information available online in an accessible way. Conflict with ASTM over building standards. National Archive has a copy of every standard incorporated by reference in the Federal Register; or you might find a copy in the library; or you might buy a copy. Public Resource decided to change that. Eventually won on fair use; thinks this should have been decided on copyrightability. The law that governs us cannot be copyrightable. Fair use is more complicated.

Also, all the creative work in standards is done by volunteers who want to create good standards for their industries; the work done by the profit-seeking entities is not creative. © is not needed.

Fair use is bearing too much weight. Rights holders think fair use is too big, but the expansion of copyright is what creates the need for fair use (as Jessica Litman wrote). But it shouldn’t have to bear so much weight. 102 should have prevented copyrightability of standards enacted into law. We should also have digital first sale, but we had to use—and lose—on fair use instead.

Samuelson: very interested in fact issue v. legal issue in fair use cases. Court said the overall fair use inquiry was mostly law, but what is it that can be found as a fact? Although fair use isn’t tried to juries very often, and Sedlik is an outlier, as is GvO, that is something that not many practitioners have focused on.

Goldstein: in HathiTrust, there was good evidence that publishers had no interest in creating a market for the blind/print disabled, and they donated books to the National Library when they did. That was the key factual part from his POV; the rest was law and policy.

Sprigman: Restatement of © picks up on this—the parties’ interests are relevant. Emphasized in SCt Public Resource case—they resolve it on protectability; some affected parties will be willing to roll the dice on fair use. The less bold among us would have to think twice before using the law that belongs to all. Fair use isn’t just framed as a defense, it’s case-specific.  Maybe the Court has signaled to lower courts that they should be relying on copyrightability.

McSherry: UpCodes cases—ASTM keeps pushing the idea that commercial uses are fundamentally different. They’ve mostly lost in dct, but not entirely. It shouldn’t matter: it’s the law! Make the law more available!

Said: it’s troubling to give the jury an idea/expression instruction unless they have some understanding of the caselaw—they’ll interpret it according to lay understandings. That’s led her to worry about transformativeness. Sometimes it’s a factual question, but Kelly v. Arriba Soft is not the kind of case where a jury would have been able to figure that out.

Fisher: 107 makes explicit that the four factor list is nonexhaustive. Puzzling how few cases press a fifth factor. E.g., shouldn’t it matter for social justice that we provide access to the blind? We’ve known since Rousseau at least that the rule of law requires accessibility. We don’t need to frame that in constitutional terms. [I teach that courts find a way to stuff considerations like this into the four.]

Said: why isn’t that public benefit?

Fisher: that’s controversial as an extension of factor 1. Breyer is willing to do it in GvO, but Sotomayor is not in Warhol. It would be a way to address ambiguity if we had a more sensible allocation of what you were trying to show. [I’m not sure that would help with ambiguity. It’s not the numbering that’s the problem.]

McSherry: we’re always framing it one way or another, including “consider four factors in light of the public interest in …” She doesn’t like good faith as a fifth factor.

Goldstein: could put all his considerations in factor one b/c they were about the accessibility purposes of the use; not transformative, but still a fair use.

Said: Trevor Reed, Fair Use as Cultural Appropriation—interesting argument about factor two, considering whether the work was being appropriated from an indigenous culture—judges are sociologically conservative (but maybe can be invited).

STLR Panel 3: The Role of Fair Use in Copyright

Trevor D. Lohrey: Should we have factors?

Christopher Sprigman: predictability is an issue. Leg history frames fair use as an equitable rule of reason. Court is supposed to be doing something synthetic, not considering factors like in a “test.” A platform for thinking about the individual case with attention to ©’s progress-promoting purpose. Predictability is often a virtue, but not always—like patience. I don’t think we should be looking for a rule.

Laura Heymann: doesn’t fault judges for going through factors in an atomized way. The synthesis is challenging. It’s really about giving more weight to the “fair” in fair use. But judges don’t want to be unmoored.

Peter Henderson: we have examples of conflicting holdings in what looks like the same case. We don’t want to have bright lines.

Heymann: consider folk knowledge that develops around areas of law. “You can copy 300 words or less.” That suggests that people want those kinds of sub-rules.

Henderson: you can also see a desire for that in AI training—don’t output more than 300 words w/o a license, etc.

Barton Beebe: Appellate reversal rate for fair use cases is about 1/3, which is almost exactly the same as for other civil litigation. The dissent rate is 1 in 10, almost exactly the same as for other civil litigation. The hard cases are hard, which is why they’re being litigated—that’s why there’s unpredictability in the individual litigated case; fair use is not that special.

Heymann: though there also effects on nonligitation decisions.

Lohrey: change over time?

Sprigman: The use of the “purloined” manuscript in Nation: a journalist would think that’s how the system works, while a lawyer thinks it’s bad; a lawyer thinks it’s good to keep secrets for your clients while a layperson might object. O’Connor failed to see outside of her bubble. Leval and others helped erode the idea of “good faith.” Fair use is about market competition now. Restatement focuses on how the factors can aid in assessing whether the defendant is undermining fair competition. That’s something that takes time and has slowly matured.

Heymann: shortcut for judges can be “is this tech analogous to tech we’ve encountered in the past?”

Sprigman: what Congress meant to do was to take a snapshot of judicial creativity in fair use, but in a way that didn’t freeze it and rather spurred that creativity. In some areas of the Act Congress said very distinctly what it wanted and in others it approved continued judicial innovation—to keep development going. So when people complain we’ve gotten away from the purpose of fair use, I want to ask: what did you think that purpose was?

Lohrey: is fair use now too market-focused and about battling economists?

Sprigman: your answers depend on your priors. He thinks it’s a good development b/c he came from antitrust where those issues are asked every day and there are tools. If the bete noire of © is substitution, he thinks that the question of substitution can be answered much more accurately than a general fairness inquiry.

Heymann: that devolves into fights over market definition, which seems pretty hard. Does milk compete with Coke? [Sprigman says it’s surprisingly difficult to answer that question—which to me doesn’t make the point he thinks it does. I’m not sure that you can solve the difficulties by switching the terms of debate—you just get different difficulties that you might only notice after a bunch of cases.]

Sprigman: hypotheticals can help—it’s theoretically tractable to ask whether the market for one work affects the market for another. In fairness, we have to figure out whose interests count and for how much.

Henderson: you still have to choose your models, components of your models—it constrains degrees of freedom but not infinitely. In the latest evolution of market dilution theory, some of those models aren’t really couched in anything particularly data-grounded. [E.g., the Court’s claims about derivative markets in Goldsmith.] We will often not have data, especially about potential future market harms.

Beebe: one hypothesis—when a court finds fair use it will find that it’s a question of law, and no fair use will be a question of fact.

Heymann: you see some cases saying that there’s no market because the corporation could have exploited it but didn’t and that means there must be none, whereas cases like Monge say that the © owner might change her mind. [I think these can be made more coherent by thinking about the idea of corporate rationality versus individual caprice. But I admit that this doesn’t cover cases like the Seinfeld Aptitude Test case.]

Beebe: don’t ask “is this conduct fair use”; instead ask “should this conduct be fair use?” There is no “is” in fair use in cases of this nature. It’s all “should.” There’s no preexisting rule that we apply to the facts.

Sprigman: many opinions leave lots unresolved. AI cases: One of the Cal. cases finds fair use might give the rule “don’t use pirated datasets”? The other Cal case says: “this is terrible—this shouldn’t be fair use but it is, b/c Ps are so inept. I know in my bones that market dilution harm is happening.” But competition for an industry—w/a class of human authored works—is not something we have experience with. As a matter of © law I think the answer is that © shouldn’t recognize that as a theory of harm, but we can think about how to address it as a different kind of harm—labor law, tax law. © is a poor vessel for that.

Beebe: A human supremacist view: more and more stuff is not the goal. The processes of creativity are important to human flourishing, not the consumption of stuff. The solidarity of sharing it with others. Humans over machines and corporations! Fair use should be resisting both, not just machines.

Ochoa: if fair use is a “should,” are we getting things systematically wrong?

Sprigman: thinks general fairness concerns interesting but not central; fair use law holds up pretty well. Mistakes always exist and don’t seem directional or pernicious.

Beebe: Clarification: Lots of fair use questions aren’t that hard. 80% of the time fair use is an “is.” District judges don’t have time for a lot of “should.” We need is or it will be chaos. But for LLMs/AI training, there is no “is.” The hard questions are at the edges.

Sunder: don’t romanticize what our © has wrought—women as naked models/objectified subjects, not creators. Hollywood gives us sequel after sequel, movies based on commercial products like Lego and Labubu. The cheerleading uniform! The human creator supposedly fostered by this law deserves a closer look. Possibility of hybrid innovation/cyborg creativity—good with human flourishing, not supremacy.

Beebe: fair! Romanticism: thinks of American romanticism of mid-19^th century, very folk-oriented, not the heroic muscular unusual author. But fair to point out that lots of people were not enjoying the mid 19^th century!

Sprigman: Readers should be important in our theories—the people shaped by creativity as well.