Java-Android: A Critique of the Federal Circuit’s Fair Use Analysis Part 2

Turning Tables Is Fun, but It Doesn’t Make the Tables Go Away

This is my third blog post about the most recent Federal Circuit decision in the Java-Android case, and the second about the Federal Circuit’s fair use analysis. My first post looked at the Federal Circuit’s holding that fair use is a question for the judge, not the jury, and I suggested that the holding was not obviously wrong and might strengthen fair use in the long run. My next post was about the first two fair use factors, and made the following main points:

• While we can argue about when placing a work in a “new context” qualifies as a “transformative use,” no one would say that the context must be “novel”—i.e., never done before—to qualify. Yet, that’s what the Federal Circuit held.
• Regarding the idea that uses are only “transformative” if they imbue the underlying work with new meaning, how does computer code “mean”? Is it basically impossible to transform computer code for fair use purposes? Well, even if that’s true, perhaps it’s only fair, since computer code does poorly under the second factor.
• But wait, after conceding that the organization of 37 API packages isn’t very expressive, the Federal Circuit then deprecates the second factor so that it all but falls out of the analysis. This is wrong. The same rules apply to all categories of copyrightable works, and computer code doesn’t need or deserve special treatment.

I would have weighed the first factor against fair use, but less heavily than the Federal Circuit, and I would have weighed the second factor in favor of fair use, instead of ignoring it, as the Federal Circuit did. Let’s move onto factors three and four…

“All ur droidz are belong to us!”
(By Oracle Corporate Communications – originally posted to Flickr as Larry Ellison on stage, CC BY 2.0, https://commons.wikimedia.org/w/index.php?curid=8812148)

Factor Three: Amount and Substantiality of the Portion Used

This factor requires us to look at not only how much of the copyrighted work was taken, in relation to the copyrighted work as a whole, but also how important the portion taken was to the copyrighted work as a whole. We are warned that this isn’t meant to be an exercise in mathematics, e.g., 4% is OK but 5% is too much.

Is Organizing APIs an Art or a Science?

In this case, what was the copyrighted work? Java 2 Standard Edition, which is (as I understand it) the entire set of APIs. It consists (we’re told) of 2.86 millions lines of code. In terms of copyright, it also includes how those API libraries are organized. What did Google take? The way 37 API packages were organized. It took an abstraction of a portion of some of the API packages.

How important is that to Java 2 SE? Over time, of course, that organization became very important because it became very familiar to developers. And that’s why Google took it. But when should we measure importance—at the time of creation, or sometime later? In the previous appeal, we learned that functionality is measured at the time of creation. Shouldn’t the importance of the organization to Java 2 SE also be measured at the time of creation? This make sense, because otherwise “importance” would rise and fall based on how a work’s audience treated the work over time. Normally, the importance the author and the audience place on a particular part of a work will coincide. The organization of API libraries of a popular platform would seem to be an exceptional case. Well, that’s software for you.

So, let’s go back to when somebody at Sun Microsystems was organizing the Java APIs. They have to be organized somehow, so any organization is going to be important. But what we’re interested in is a particular organization—that’s what’s expressive. So, what is so expressive about the particular way the Java APIs were organized? At that time, what was so important about that particular way of organizing the APIs? Is the organization the “heart” of Java 2 SE? Or is more like a box we put the APIs into?

Before you say something like, “It was an especially intuitive organization,” or “It was very elegant,” or “It was very efficient,” remember that intuitiveness, elegance and efficiency are not protectable expression. Indeed, what we’re most interested in are the un-intuitive, inelegant, inefficient aspects of the organization, which bear the stamp of human originality.

No, the particular way the APIs were organized was not important to Java 2 SE, at the time they were so organized, beyond being a useful way to organize them. I think that’s obvious. And Google didn’t even take the entire organization, so the third factor should have weighed fairly heavily in favor of fair use.

It’s All Relative: Turning the Tables on Google

It wasn’t obvious to the Federal Circuit. It found the third factor was either neutral or perhaps weighed against fair use. It reasoned that, because Google needed only 170 lines of code to “write in Java,” anything above this amount is too much, and Google took 11,500 lines, oh my! This reasoning is based on three mistakes. The first mistake is the fixation on 170 lines, which I believe to be inaccurate. Second, the Federal Circuit looked at “importance” from Google’s point of view, instead of Oracle’s, based on its purpose for the taking (from the first factor). Third, and most important, it held the amount permitted to be taken under the third factor depended on how much is necessary to achieve Google’s purpose.

The Parties Didn’t Stipulate to the 170 Lines.

It was undisputed that Google (indirectly) copied 11,500 lines of code from Java 2 SE in order to make Android. As we’ve discussed, these lines are merely reflections of the organization of the 37 API libraries. Google not only admitted to copying 11,500 lines of code; it emphasized the fact. Why? Because they were only a tiny fraction of the 2.86 million lines of code in Java 2 SE, which is about .4% of the total. I mean, we’re not supposed to let mathematics dictate this factor, but it’s not as though it completely falls out of the analysis, either.

But the Federal Circuit saw those 11,500 lines as something should be ashamed of, because it only needed 170 lines “to write in the Java language.” That’s 11,300 more than necessary. For shame!

The Federal Circuit said the parties “stipulated” to the 170 lines. That’s not accurate. The parties stipulated that you needed at least 170 lines to “write in Java.” But it was left up to the jury to decide whether you needed more than that. Don’t believe me? Check out jury instruction no. 32.¹“It is established that 170 lines of code at issue are technically necessary to use the Java programming language. Those 170 lines of declaring code are listed in Trial Exhibit 9223. Because that declaring code is necessary to use the language, it is established that Google’€™s use of the declaring code in Trial Exhibit 9223 was a fair use. It is for you to determine the extent to which other additional declaring code beyond those lines identified in Trial Exhibit 9223 either was or was not necessary for use of the Java programming language.”

How many lines of code are necessary to write in the Java language is one of those “historical facts” about which the Federal Circuit says it defers to the jury. What can we assume about the jury’s finding on this issue, given its overall finding of fair use? Do you think the jury found that only 170 lines were necessary? I think we can assume the jury found that much more than 170 lines were necessary to “write in Java.”

Parody Is a Transformative Use, But Not All Transformative Uses are Parodies.

You might wonder why we’re so concerned about how many lines you need to “write in the Java language.” That has to do with another of the Federal Circuit’s mistakes. It held that how much you’re allowed to take from the underlying work depends on why you’re taking it. As it happens, by the time you get to the third factor, you’ve already fully explored the “why,” because it’s the same thing as the purpose for the taking you explored in the first factor.

Small problem: the only legitimate purpose is a transformative use. There’s a good deal of legal authority supporting the idea that the third factor doesn’t weigh against fair use so long as you took no more than what you needed to carry out your transformative use. The Federal Circuit reverses this formulation: if you take more than “necessary,” the third factor weighs against you. I am aware of no legal authority for this reversal, and the Federal Circuit cites none.²The Federal Circuit cites Seltzer v. Green Day, but that’s a “you can get away with anything if your purpose is transformative” case if there ever was one.

However, as a general principal of equity³Also, to the extent Campbell expresses a general rule for all types of transformative uses, its reasoning also strongly suggests such a rule goes both ways., I’m prepared to accept that what’s sauce for the goose is also sauce for the gander. So, allow me to question the whole idea. The idea that evaluating the third factor depends on the transformative purpose of the first factor is said to originate in Campbell v. Acuff-Rose Music, the 2 Live Crew-“Pretty Woman” case. That case involved a song that replaced the original lyrics to “Oh, Pretty Woman” with raunchy lyrics. The Supreme Court found 2 Live Crew’s version to be a parody of the Roy Orbison version—an indirect parody, but a parody nonetheless. It was thus a “transformative use”⁴Campbell also legitimized Judge Leval’s concept of transformative use.. Turning to the third factor, the Supreme Court had to acknowledge that 2 Live Crew took the entire musical composition, and thus, the “heart of the work.” But the Supreme Court sort of excused that extensive taking because parody only works if the audience can recognize what’s being parodied. But it cautioned that this shouldn’t become an excuse for wholesale copying where the parody was barely noticeable.

There is no question that, under Campbell, a finding of parody greatly extends how much you can take under the third factor. It is also true that Campbell describes parody as a species of transformative use. But we must also acknowledge that parody is a very, very special kind of transformative use, and not everything in Campbell relating to parody extends to all transformative uses. Parody has a long history that pre-dates copyright law. Unless you want to cut off this long tradition, you must give it room to breathe; otherwise, the other fair-use factors might overwhelm it. It’s not just the third factor, either. The second factor drops out because only highly expressive works are worth parodying, and the fourth factor drops out because, if done correctly, no one would want to buy the parody instead of the real thing.

For good reason, then, does parody come with a strict limitation: it must comment on the underlying work. This is why The Cat Not in the Hat wasn’t a fair use, even though it purported to parody The Cat in the Hat. It told the story of the OJ Simpson murder case, as though a deranged Dr. Seuss had written it. It commented on the murder case, but not on The Cat in the Hat, so it wasn’t legally a parody.⁵In literary terms, “parody” would encompass works that comment on the underlying work and those that use the underlying work to comment on something else.

Parody, then, is a special case. What applies to parody doesn’t apply to other forms of transformative use. Transformative use still pushes the first factor toward fair use, but the reason for the third factor isn’t diminished thereby. We are simply less concerned with trivial⁶But still greater than de minimis. takings than we are with significant takings. If you’re going to take the entire work, you’d better have a very good transformative reason, but not because your very good transformative reason will change our analysis of the third factor, but because the third factor is going to weigh against you, so the first factor had better weigh heavily in your favor.

This is all the more urgent when courts start taking an expansive view of transformative use (which, the Federal Circuit, somewhat to its credit, didn’t). Imagine a minor transformative use that just so happens to require taking the entire underlying work. Do we just ignore the third factor? Under the Federal Circuit’s logic, we should. I don’t think that’s right.

Writing in Java Wasn’t Google’s Purpose. (And What Happens if There Is No Transformative Use?)

OK, fine. Let’s just accept the Federal Circuit’s view that the transformative purpose found in the first factor sets the baseline for the third factor; i.e., the third factor weighs in favor of fair use if you take no more than necessary to carry out the transformative purpose, and it weighs against fair use if you take more than necessary. What if there was no transformative use, as the Federal Circuit found? That really ought to show why this idea is a bad one. If there’s no transformative use, then logically the baseline is zero, and any taking would cause the third factor to weigh against fair use, no matter how small or insignificant.

I think this is why the Federal Circuit latches on to the number of lines “to write in Java.” It regards writing in Java as a kind of default purpose. In a way, the Federal Circuit was doing Google a favor, since the alternative was to find no purpose whatsoever and thus set the baseline at zero, but the difference between zero lines and 170 lines, out of 2.86 million lines, isn’t statistically significant.

But at trial, “writing in Java” was a separate basis for the jury to find fair use. Oracle had to admit that, because it gives away the right to program using the Java language⁷Not the same as Javascript, by the bye., to the extent you have to use the organization of the API libraries to do so, that’s a fair use. Oracle’s admission was limited to 170 lines. Google argued for more. The trial court said it was up to the jury to figure it out. Here’s the thing: if the jury found that all 11,500 lines that Google “took,” then the fair-use analysis is over, and Google wins. We never find out, of course, what the jury found. For all we know, its overall finding of fair use was based entirely on Jury Instruction No. 32.

The Federal Circuit thus makes several mistakes. First, it assumes the parties agreed on the 170 figure, when they didn’t. Second, it assumes that “writing in Java” was Google’s purpose, when it wasn’t really. Third, it assumed that the baseline for taking too much is based on Google’s transformative use.

Important to Whom?

I’d like to suggest another reason why basing the third factor on the first factor is a bad idea. It requires us to look at “importance” in from the infringer’s point of view. The Federal Circuit makes this quite explicit:

Even assuming the jury accepted Google’s argument that it copied only a small portion of Java, no reasonable jury could conclude that what was copied was qualitatively insignificant, particularly when the material copied was important to the creation of the Android platform.

In other words, Google wouldn’t have taken it if it weren’t important. But this is contradicted by the third factor’s actual statutory language (emphasis added):

(3) the amount and substantiality of the portion used in relation to the copyrighted work as a whole.

By the Federal Circuit’s logic, any taking will be “important” for third-factor purposes because the defendant took it. The third factor will always weigh against fair use. The Federal Circuit then compounds the error by reminding us why Google took the organization of the 37 API packages in the first place: because they were well-known and Google didn’t want to discourage potential developers by making them learn a new organization. But, as discussed above, the importance of that organization has less to do with originality than with happenstance.

Mangez le Monge

Although the “take no more than necessary” idea arises with Campbell (incorrectly, in my view), the Federal Circuit cites Monge v. Maya Magazine for the proposition. But Monge not only is completely distinguishable, but suggests that we have the relationship backward: the first factor doesn’t inform the third, but the third enlightens the first.

Monge is an unusual case. It involved six photographs taken of a secret celebrity wedding. The photographs were taken on the celebrity’s camera⁸Though not by the celebrity herself, which raises the question of how the celebrity claims ownership in their copyright. But, whatever., and the memory stick got left in someone’s car, that someone tried to extort the celebrity, and when that failed, he sold the memory stick to the defendant magazine. The magazine published all six wedding photographs.

When sued for copyright infringement, the magazine said all it was doing was news reporting. The Ninth Circuit said that it totally respected news reporting, and it was OK to blow the lid on the celebrity wedding. But it didn’t see how reproduction of the photographs were really news reporting: once you tell the world about the secret wedding, you’ve done all you need to do. When the Ninth Circuit turned to the third factor, it questioned why the magazine published all six photographs, if its purpose was news gathering. Surely, a single photograph would have sufficed, if the magazine was worried about verifying the secret wedding? Indeed, the magazine could simply have checked public records about the marriage.⁹The analysis of the third factor included the ludicrous argument that the magazine’s taking should be measured against the total amount of photographs on the memory stick, which, like everyone else’s memory stick, is just a haphazard collection of pictures. Unbelievably, this argument convinced one judge. It’s not just that the magazine took 100% of all the photographs, but that taking more photographs than “necessary” belied their stated purpose.

But read more carefully. What comes out is that using all six photographs belied the magazine’s asserted purpose. It wanted to do more than just report on the secret wedding. It wanted the photographs because of who, where and how they had been taken.¹⁰Consider: if a photograph had been blurry or out of focus or too dark, would the magazine have used it?

Even if the magazine had used only a single photograph, it still would have taken the entire thing, and the third factor still would have weighed against fair use. It’s just that the first factor would have been taken more seriously. We might posit that relative takings go to the first factor, but absolute takings go to the third factor. We might further posit that it’s sometimes OK for the third factor to weigh against fair use, and yet still find fair use overall, if the first factor is strong enough (and that you can find the third factor weighs in favor of fair use but still find no fair use).

In sum, the third factor should have weight significantly in favor of fair use, and yet somehow, the Federal Circuit found “the third factor [to be], at best, neutral in the fair use inquiry, and arguably weigh[] against such a finding.”

Factor Four: Effect Upon the Potential Market

By my scoring (not the Federal Circuit’s), this is a close case so far. Factor one weighs pretty heavily against fair use. Factors two and three weigh heavily for fair use. But, as much as I feel a need to defend factors two and three, I’m not about to say they’re more important than the first and fourth factors. Whereas I can see factor one or four by itself carrying the day, I can’t say the same about factor two or three.

According to the Federal Circuit, harm to the potential market is one of those “historical facts” that it was going to defer to the jury about. It forgot to add: except this one.

To its credit, the Federal Circuit declined to take the easy way out. It could have followed Ninth Circuit precedent that market harm is presumed if the use is commercial and non-transformative.¹¹This is a fairly common presumption. As it noted, the Supreme Court has made it fairly clear that there is no such presumption.¹²The presumption isn’t crazy, but it does undermine the balancing nature of the fair-use analysis. It allows the first factor to dictate the fourth factor. While there is going to be interplay among the factors, such interplay shouldn’t be mechanical.

Instead, the Federal Circuit constructed a legal box from which fair use could not escape. It emphasized that (1) the fourth factor was not limited to actual markets but included potential markets (non-controversial—it’s in the statute), (2) it was not limited to the copyrighted work itself but could include works that derive from the copyrighted work, and (3) it was not limited to markets that the copyright owner had actually entered, or had planned to enter, or even had given passing thought to enter, but extended any hypothetical potential market.

Under this logic, it is difficult to imagine the fourth factor ever favoring fair use, if there is any commercial purpose present at all. It just creates the presumption in favor of market harm but in a different way. The fact that the secondary work found a market would be, by itself, evidence that a hypothetical market did exist. That the defendant was clever enough to find and exploit the market, and the copyright holder was perhaps utterly incompetent at basic business, would be irrelevant.

A Right to Say No?

If you look at the main legal authority for this logic, Worldwide Church of God v. Philadelphia Church of God, it’s actually even weirder than the Federal Circuit says. Worldwide Church involved a religious work that the plaintiff used to publish but had withdrawn from distribution, in large part because the church’s doctrine had evolved to the point that the work was no longer current. However, a splinter church evidently believed the work was still relevant and began publishing verbatim copies of the work. When the original church sued the splinter church for copyright infringement, the splinter church convinced the trial court that its verbatim copies were fair uses.

The Ninth Circuit reversed. Most important for our purposes, it held that, although the churches didn’t compete with each other for money (at least, not directly), they did compete for parishioners. It also held that the original church’s failure to do anything with the copyright for ten years was irrelevant (though the original church had plans to publish an “annotated” version). The original church had the right “to change its mind” and not publish. For this, the court cited a case involving J.D. Salinger’s unpublished letters. In that case, Salinger had decided against publishing his letters, at least for the time being, but those letters would surely be worth hundreds of thousands of dollars when he did choose to publish them.

What drove both Worldwide Church and Salinger was something rather more than just the right to change one’s mind. In Salinger, it was the right of first publication, a venerable right that still finds purchase under modern copyright law. In Worldwide Church, it was the right for a church to control its own ecclesiastical message. Is there a difference between choosing not to participate in any market and wishing to participate but being either unable to (e.g., for lack of resources) or really bad at it?

The Federal Circuit’s use of Worldwide Church suggests the following (problematic) logic is at play. Even where the copyright holder has abjured the enjoyment of a market, the fourth factor still weighs against fair use, so long as there is some hypothetical market. Thus, when the copyright holder does try to take advantage of a market, the fourth factor weighs all the more heavily against fair use. I’m not sure this makes sense, though. Worldwide Church (and Salinger) make much of an affirmative right to not participate in a market, i.e., a consideration related to but independent of the main fourth-factor considerations. If the Worldwide Church was still publishing the work but not in sufficient quantities to keep up with the demand of true religious adherents, and if it refused to license the right to anyone else, the court might have been more sympathetic to the defendant’s wholesale copying. In J.D. Salinger’s case, the value of the letters lay almost entirely in their secrecy. Once they were published, their value would be diminished, so it was crucial to protect Salinger’s right of first publication.

Markets, Profits and Competition on a Spectrum

Still, it makes sense to see actual, potential and hypothetical markets on a spectrum, together with the copyright holder’s ability and desire to meet that demand while also maximizing profits. Consider this spectrum:

• Secondary work competes directly with primary work and directly hurts its profits.
• Secondary work competes directly with primary work but fails to dent the primary work’s profits (perhaps because the copyright owner has maxed out its capabilities and has refused to license).
• Secondary work competes in a market that the primary work was actively planning on entering.
• Secondary work derives from the primary work, and it thus competes in a slightly different but related market.
• Secondary work competes in a market that the primary work decided not to enter, although it would be profitable (the Salinger situation).
• Secondary work competes in a market that the primary work decided not to enter because it would be unprofitable.
• Secondary work competes in an utterly unrelated market that was never considered by copyright owner.

If fair use is a balancing of social utility against private rights—something the Federal Circuit rejects, remember—then the public is served by competition and when copyright works are introduced into new markets, but the copyright owner is entitled to profits it would have made had it been able to fully exploit its copyright as it intended. Further we wish to encourage licensing, which naturally only happens when the copyright owner does not wish to compete in a market. But even under the Federal Circuit’s preferred view of fair use—i.e., how it affects the incentive to create—this gradient makes sense. One is surely incentivized to create by markets she envisions at the time of creation, and one is surely not incentivized by markets she couldn’t imagine, and somewhere in between are markets that were vaguely foreseeable at the time of creation.

Java’s Forays into the Relevant Market Were Tenuous and Unsuccessful—but There Were Forays.

Oracle did license Java 2 SE for mobile phones. And those mobile phones were commercially unsuccessful. I mean, they bombed. The Federal Circuit also latched onto the idea that Amazon had used Android’s free price to negotiate a steep discount for the use of Java in its next generation of e-readers (i.e., the Kindle). Only, that’s not quite what the jury heard. What the jury heard was Safra Catz of Oracle say that Oracle gave Amazon a steep discount, not why.¹³What Amazon was thinking would be hearsay. Curiously, a hearsay objection was sustained around the time of this testimony, but I’m not sure if that’s related. For all we know, Oracle caved in unnecessarily, or caved in for reasons other than, or in addition, to competition from Android. True, the jury certainly could have inferred from that testimony that Android was harming the market for Java in e-readers. Indeed, it’s the most likely inference, just not the only reasonable one. But we also know that the jury did not make that inference.

The Federal Circuit puts together a compelling story about market harm. And it’s one that the jury could have inferred. But the jury did not. The jury was present when Safra Catz gave her testimony, and maybe the jury didn’t believe her. Maybe the jury thought she was exaggerating. Maybe the jury thought she was making stuff up. Making credibility determinations is almost the jury’s number one job.

In the end, the Federal Circuit found not only did the fourth factor weigh against fair use, but that it weighed heavily against fair use.

Allow me to dissent. An example of a fourth factor that weighs heavily against fair use is direct competition with solid evidence of actual market harm. What we have here is some direct competition in minor markets, and indirect competition in a derivative market. Further, we have Oracle’s inability to turn Java into a mobile OS—it might be harder than it looks. If I weren’t inclined to defer to the jury, I would find this factor weigh slightly against fair use. But the jury clearly thought there was no market harm, and I’m not as eager as the Federal Circuit to throw that finding away.

In sum, I would find the fourth factor to weigh slightly against fair use if it were entirely up to me, but I’d find it to weigh it in favor of favor use if I felt I had to give the jury any deference. There was a market for Java 2 SE for mobile operating systems, but Oracle couldn’t figure out how to exploit it. There was evidence of market harm, but it was not very reliable. The Federal Circuit found the fourth factor weighed heavily against fair use because it thinks any potential market, no matter how remote or unlikely, is a strong as any other potential (or actual) market, and because it defers to jury right up until it doesn’t want to any more.

Punching Google in the Mouth Isn’t Really Policy

There’s a funny little theme that runs through the Federal Circuit’s opinion. It takes delight in turning Google’s arguments against it. That is to say, it takes arguments that defendants rely on for fair use and uses them—more or less consistently—to rule against fair use. It’s always funny to see engineers hoisted on their own petards. This is especially so if the engineers work for Google, and you have come to see Google as emblematic of everything that’s wrong with technology/copyright law/privacy/whatever.

But one might pause and consider that, in turning these rules against fair use, these rules are strengthened. Yes, the Federal Circuit was able to say that Google “took” 11,330 more lines than it needed, but the general rule that the third factor favors fair use when the defendant takes no more than necessary will nearly always favor the defendant. More harmful is what is happening to the second factor, which benefits the vast majority of those who rely on copyright. Most copyrights we are concerned about go to highly expressive works. Their owners will not thank the Federal Court for weakening the factor that weighs in favor of highly expressive works.

The final score for me:

• First factor: Against fair use, but not heavily so.
• Second factor: Strongly in favor of fair use. This factor matters, but not as strongly as the first and fourth factors.
• Third factor: Strongly in favor of fair use. This factor matters, too, but not as much as the first and fourth factors.
• Fourth factor: Let’s split the difference between deferring and not deferring to the jury and call it a push.

By contrast, the Federal Circuit found the first factor weighed heavily against fair use, the second factor weighed in favor of fair use but was accorded minimal or no weight, the third factor was neutral but leaning against fair use, and the fourth factor weighed heavily against fair use.

Final conclusion: this was a close case, but I think the case for fair use is just a touch stronger than the case against. If you want to say that the case against fair use was a touch stronger, I won’t really argue. However, there are a number of “rules” that the Federal Circuit relies on that are either straight-up wrong or at least very questionable. In particular:

• Whether the rubric of “transformative use” really works with computer code.
• That “new” meaning or expression must be “novel” in the patent sense (this is a terrible idea).
• Whether we can just ignore the second factor when it pleases us.
• Whether computer code needs special protection from the second factor.
• Whether the baseline for “taking too much” should be based on what is necessary to carry out the transformative use.
• Whether the importance of a portion is measured at the time of creation or at the time of taking.
• Whether the importance of a portion is measured relative to the underlying work as a whole, or the secondary work (the underlying work, obviously).
• Whether the “right to say no” is a separate aspect of the fourth factor, or a baseline market harm.
• Whether markets and the copyright holder’s exploitation of markets should be viewed on a spectrum.

Thanks for reading!

Rick Sanders

Rick is currently General Counsel for Software Freedom Conservancy. Previously, he has been practicing law as an intellectual-property litigator since 2000.