Maybe We Could Do Something About It?
For all of copyright law’s technicalities and niceties, it can come as a shock to learn that we lack a clear understanding of infringement. Often, this isn’t a problem. If you’re photocopying a book, putting a digital photograph on your webpage, or pirating software, we don’t need a theory of infringement. But copyright covers much more than this—what is called “literal” infringement—and we don’t have a good sense of what.
A comparison to patent law might be instructive. In patent law, to infringe a patent, the accused product must have or do every element of one claim of the patent. It doesn’t matter if you’re aware of the patent or any products made in accordance with the patent. But your product must still have or do everything that’s in the claim. Thus, patent infringement has some pretty sharp borders.
For all that, there’s a little fuzziness to those borders. For one thing, the claims are described using language, and language has its own built-in imprecision. For another, patent law effectively expands the scope of the claims through something known as the “Doctrine of Equivalents.” This has resisted clear definition, but the purpose is to prevent infringers from gaming the system by making trivial changes to their products to avoid infringement. This much is certain: the grey area created by the doctrine is very small.
By comparison, copyright’s version of the Doctrine of Equivalents—substantial similarity—is very broad. This is what the leading decision on the subject has to say (here talking about fabric designs):
No one disputes that the copyright extends beyond a photographic reproduction of the design, but one cannot say how far an imitator must depart from an undeviating reproduction to escape infringement.
However, the ordinary observer, unless he set out to detect the disparities, would be disposed to overlook them, and regard their aesthetic appeal as the same. That is enough; and indeed, it is all that can be said, unless protection against infringement is to be denied because of variants irrelevant to the purpose for which the design is intended.
But this only represents one kind of fuzziness for copyright law, which might be thought of as a spreading out from a core center (and, yet, it might not grow appreciably thinner at the edges). There is another axis of fuzziness: abstraction. Copyright covers not just literal expression, but also structural expression. In a novel, this would take the form of plot and well-developed characters. In a design, this would take the form of shape, arrangement and maybe color. You can’t (necessarily) avoid copyright infringement simply by re-writing someone else’s novel, even in your own words; or by appropriating a design but with different colors.
Abstraction has a limit because copyright doesn’t protect ideas (which is patent’s job). The inflection point is hard to define, unfortunately. As that same leading decision explains:
it is well settled that although the “proprietor’s” monopoly extends beyond an exact reproduction of the words, there can be no copyright in the “ideas” disclosed but only in their “expression.” Obviously, no principle can be stated as to when an imitator has gone beyond copying the “idea,” and has borrowed its “expression.” Decisions must therefore inevitably be ad hoc.
There must be some relationship between these two axes of the scope of copyright protection, which we might call similarity and abstraction, because courts generally talk about them in the same breath. (The exception is software code, where abstraction is built in and there’s a heightened need to avoid protecting the unprotectable.) Is it the case that, the more abstract, the more similarity you need? Maybe. That seems natural enough, but even the software cases (of which there are very few) don’t say so explicitly.
We’re not done, yet. That’s because copyright only protects what’s original13This is NOT the same thing as novel. It just means that the expression originated from your human brain, not from somewhere else.. That means it doesn’t protect facts (you didn’t make them up, right?14Unless you did make them up, e.g., the “facts” about quidditch, which is an entirely made-up sport. Those “facts” are protected by copyright. There’s even a case about it.), and it doesn’t protect “scènes à faire,” which are conventions, tropes and techniques in common to all creators of a given field.
So, by comparison to patent law, copyright law seems wide open and wild. But there’s something that constrains copyright law. It is not enough to prove that the two works are sufficiently similar. The copyright owner must also prove that you copied. It’s a little like trying to figure out who cheated on a test. It’s not enough that X’s and Y’s answers are similar—that could really happen—but that you saw Y maybe sneaking a peak at X’s answers.
Copyright law struggles with this basic premise of copyright law because it throws copying into the same mixer as substantial similarity. You’ll often see courts say that, if you can’t prove copying by direct evidence, you can prove it indirectly by proving access and substantial similarity. Which doesn’t make sense for a moment. Let’s consider a hypothetical:
Hypothetical: Me in the Library with a Book, a Pen and Some Paper
Let us say that you see me in a library, with a book open, and I’m scribbling on a pad while clearly reading the book. I do this for more than a de minimis period of time and number of pages. Do you have enough evidence to conclusively establish that I was copying the book?
No. You don’t know enough about what I’m writing. Am I copying the book verbatim? Am I taking highly detailed notes? Am I writing a response to the ideas in the book, quoting only as necessary? Am I really good at multi-tasking, and what I’m writing is just a letter to my mother?
That last possibility sounds silly, but it’s important theoretically. It’s not enough to show access. You still have to take a look at my writing to make sure that it has something to do with the book. You don’t need much—you did see me with the book, after all. Just similarities in theme, or maybe just some recognizable phrasing, would be enough to turn mere access into meaningful access.
OK, now imagine that you didn’t see me with the book at all; in fact, I’ve never even been seen in the library. But you’d heard that I might’ve copied the book in a legal pad. You’re allowed to read my legal pad. How much similarity between my writing and the book would you need before you concluded that, yes, I had copied the book? Just similarity of themes and some phrases wouldn’t be enough. Themes are general enough that they could have come out of my own head, or maybe there’s been enough discussion about them that I was influenced by an entirely different body of literature. The same thing with phrasings. But what if my writing is verbatim of a significant portion of the book? At some point, mere coincidence is no longer possible, and the only explanation is that I copied the book.
These two extreme examples give you an idea of what is sometimes known in copyright law as the “inverse ratio” rule. The more access, the less the need for similarity. And the more similarity, the less the need for access. This should not be controversial.
But—here’s the hard part—you have to keep in mind what access plus this type of similarity are proving. They only prove that I did something legally meaningful with someone else’s copyrighted work. They prove only “copying,” which is just one half of proving that what I wrote was illegal, that it infringed someone else’s copyright in that book.
Let’s assume that you saw me with the book, but my writing are notes I made from what I read. They are not verbatim notes, however. Have I infringed copyright? We take notes all the time! Surely that’s OK?
Well, it might matter how detailed the notes were. Just as you can infringe a novel by copying its story (but not its themes), you could produce such detailed notes that you infringe the book. I don’t know why I’d do that, but it’s possible. If, on the other hand, my notes were just normal notes, that any student might take—i.e., I’m just jotting down the main points—that’d be OK.
But wait! Even if my notes were detailed, doesn’t it matter what I was writing down? What if my notes consisted mostly of factual matter I had gotten from the book, and it so happens the book has a lot of factual matter? We know nobody owns published facts, so even my detailed notes are probably OK.
Two Distinct Steps, Two Kinds of Similarity
There are—there must be—two steps to determining whether an act is an act of infringement: one to show whether I copied, and the other to whether what I copied was… how do we put this? Wrongful? Too much of the wrong stuff? A crucial corollary of this: Some copying is actually 100% OK and legal. When you take notes, you are copying (in a copyright sense), but it’s legal copying.
Confusingly, these two separate steps involves different types of similarity: a similarity that shows you did something meaningful with the copyrighted work, and a similarity that shows you took too much of the wrong stuff. Academics, and sometimes even courts, call these types of similarity “probative similarity” and “substantial similarity.” The terrible nomenclature aside, these two steps, and these two types of similarity, are essential to a consistent and rational understanding of what it means to infringe copyright.
There are other consequences of this two-step analysis that might take some getting used to. For probative similarity, we don’t care if the similarities involve non-protectable elements. We’re just interested in whether you had access to the work and copied it. If you copied a bunch of facts that are only found together in a certain book, that’d be strong evidence of copying. But it wouldn’t be enough to satisfy the substantial similarity step, where we do care very much about what’s protectable and not protectable, because you’re legally permitted to copy facts.
For probative similarity, we’re willing to listen to experts, who can tell us whether shared elements are markers of access or just coincidence or possibly derived from other sources. But for substantial similarity, we’re interested in how the similarity “plays” for the intended audience, and for that, we need a jury.
For probative similarity, we can dissect the works down as much as we want—or not—because we’re just trying to see if there’s enough there to support meaningful access. But for substantial similarity, we look at the works as a whole (though we still have to somehow filter out unprotected elements while still recognizing that a combination of unprotected elements can amount to protectable content15This “filtering” issue is a whole other problem.).
Problem with the Approach: Only One-Half of One Circuit Follows It.
Do I need to tell you that courts don’t recognize these steps? Well, the Second Circuit does, and only half the time. You can actually test this by looking at every case that has cited Laureyssens v. Idea Group, which is the leading opinion for this two-step, double-similarity analysis. Conclusion: no court outside the Second Circuit really uses this test, and the Second Circuit is internally split about it.
Most courts do use a two-step process, but they describe it incompletely: (a) access, then (b) substantial similarity. They don’t look at probative similarity in connection with access. And they sometimes consider elements of probative similarity—inverse-ratio rule, expert testimony, non-protectable elements, for example—in connection with substantial because there’s nowhere else for them to go. For the most part, the error is harmless. The court ends up looking at probative similarity, after a fashion, by looking at substantial similarity, and access remains important. The analysis is necessarily squishy, but one can’t say it routinely leads to wrong decisions.
The Ninth Circuit, however, has its own test. It goes like this: infringement is proved by showing (a) access, and (b) substantial similarity (just like other courts) but with this important twist: substantial similarity has two different aspects: an “extrinsic” test and an “intrinsic” test. The “extrinsic” test is said to be “objective.” It “requires an analytical dissection of the work and expert testimony”16“Requires?” and it can include a comparison of non-protectable elements. Sound familiar? The intrinsic test is said to be subjective, and asks only whether a jury would find the “total concept and feel of the works to be substantially similar.” Also sound familiar?
The extrinsic test is (or very nearly) the same thing as probative similarity, and the intrinsic test is the same thing as substantial similarity. But here’s the problem: the extrinsic test is separated from its mate, access. Instead, it looks like it were just a flavor of substantial similarity. Unlike other tests, which just mix up the different types of similarity, this test keeps the different types separate, but in the wrong places.
Again, generally, the effect isn’t very much of a problem. Most defendants vigorously contest access, which is usually the weakest part of the plaintiff’s case. Thus, when judges and juries consider access, they usually consider it in the context of probative similarity, even if the letter of the rule would keep them separate. It’s fairly natural, I think, to wonder about something like probative similarity (even if you don’t know what it’s called or quite what it is) when analyzing access. The problem of separating probative similarity from access arises only if access were conceded, which almost never happens.
Which Never Happens. Except Sometimes it Does.
This brings me to the “Blurred Lines” case, Williams v. Gaye and the Ninth Circuit’s recent decision to uphold the jury verdict of copyright infringement. (Update 7-16-18: The Ninth Circuit amended its decision on July 11. It did not change its ultimate ruling but made at least one significant analytic change.) The facts of this case might have increased the tension inherent in the Ninth Circuit’s treatment of access and similarity past the snapping point.
The problem is that, not only did Pharrell Williams and especially Robin Thicke concede access, but they had to testify that they were specifically being influenced by “Blurred Lines.” I don’t think I’ve seen this fact pattern before. Artists talk about their influences all the time, but rarely do they go around saying they were influenced by one specific song. Not for legal reasons, necessarily. Then again, most artists aren’t famous enough to be interviewed by major news outlets multiple times.
Under the “correct” Laureyssens view, all the proof of access in the world shouldn’t have made that much of a difference. All it does is reduce the amount of probative similarity that would need to be shown to prove copying. But, once copying is proven, that’s the end of that element. It’s time to turn to substantial similarity, which is a completely separate inquiry with nothing to do with access.
But under the Ninth Circuit’s test, this proof of access took on outsized—perhaps even overwhelming—importance. The “inverse ratio” rule isn’t limited to the element of copying (as it should) but is applied to similarity. Remember that the extrinsic test (the equivalent of probative similarity) and the intrinsic test (the equivalent of substantial similarity) are lumped together as “substantial similarity.” Thus, strong proof of access means that not much proof of (real) substantial similarity is needed. Remember, too, that (real) substantial similarity is where we start to filter out (or otherwise deal with) non-protectable elements. If this is your legal framework, and you have Robin Thicke telling the world how much he wanted to make a song like “Blurred Lines,” then the jury’s verdict is justifiable. As the Ninth Circuit put it:
In addition, the verdict was not against the clear weight of the evidence because there was not an absolute absence of evidence of extrinsic and intrinsic similarity between the two songs.
(Emphasis added.) In other words, the proof of access (and influence) was so strong that there needed to be no similarity at all to overturn the jury’s verdict17The fact that this is a jury verdict, and the defendants didn’t preserve certain avenues of appeal, enters into the analysis here. On appeal, the defendants had to show the jury had basically lost their minds in finding copyright infringement—that their verdict utterly lacked evidentiary support. No matter how you feel about the verdict, we all have to admit that there was at least SOME evidence of similarity.. (Update 7-16-18: In the amended opinion, the Ninth Circuit ditched all mention of the “inverse-ratio rule”!)
The problems don’t stop there. Consider the degree to which the court’s opinion, and the trial testimony, is dominated by experts.18Including one I’ve worked with before. Under both the extrinsic test and probative similarity, these experts are necessary only to establish copying (in conjunction with access). Once copying is established, their testimony doesn’t go to the intrinsic test (or real substantial similarity). Can you blame the jury, though, for thinking the trial was basically a “battle of the experts,” and once they decided they believed the Gaye family’s experts more than the other experts, their decision was pretty much made? The Ninth Circuit’s way of looking at similarity all but invited them to.
Consider also that Robin Thick was a terrible witness, and juries hate terrible witnesses, and his terribleness probably hurt his case. But consider also that his testimony was necessary only for proof of access, which only goes to copying, which was all but conceded (and would have been conceded except that the extrinsic test was separated from access).
If You Hate the Verdict, How Would You Change the Law?
As I opined back when the jury announced its verdict, this case really won’t have much of an effect on songwriting. Maybe songwriters will be more circumspect about how they describe their specific influences on a specific song, and that would be too bad. But the fact pattern behind the jury verdict is too rare to make a meaningful difference.
But if the Ninth Circuit’s decision to uphold the jury verdict does make a difference, it should be that the Ninth Circuit ditch its extrinsic-intrinsic framework in favor of that provided by Laureyssens. And then other Circuits could follow suit. And then we could have some consistency about what it means to infringe copyright.
Or, if it suits you better, we can have a more general discussion about the role of juries in civil cases. By coincidence, just days after the Ninth Circuit upheld the jury verdict in “Blurred Lines,” the Federal Circuit decided to completely ignore a jury verdict and substitute its own verdict.19The case is called Oracle America v. Google. You might have heard of it. Should juries be called on to decide super-complex issues like copyright infringement, antitrust and patent infringement? I’ll just drop this bit, though: The guys who wrote the Bill of Rights thought civil jury trials were important enough to give them their own Amendment. But times change. The problems the Seventh Amendment were meant to solve are perhaps no longer problems. And on matters where juries are especially valuable—assessing norms of behavior20Think negligence actions, or other legal questions of what’s “reasonable.” and credibility of witnesses, the court could always impanel special juries to listed to specific bits of testimony and answer specific questions. Cases would become less the “morality plays” so beloved by trial lawyers and more technical. And the reasoning of courts would become more transparent—juries tend to be “black boxes.” But whatever your reasons, you’d need a Constitutional amendment.
This decision also gives us a chance to examine not only how we analyze copyright infringement, but how we think about it metaphorically. Unlike patents, which can be infringed by complete accident and in complete ignorance of the patent, copyright has the copying element. You can copy “subconsciously” (i.e., the influence is buried deep in your consciousness but is dictating your output) but you still need have ingested the underlying work and copied it. Copyright litigators often call this “theft,” and that infringers are “thieves.” Under this metaphor, the result of “Blurred Lines” is just (and not just justifiable), and its application of the inverse-ratio rule makes sense. Williams and Thicke “took” from “Blurred Lines,” and the more they intended to take—the more they looked like thieves—the less they had to take to be in the wrong.
But now some folks are discovering the virtues of the other way to look at copyright: as a kind of trespass. The metaphor is that, regardless of access, all copyrighted works have a boundary beyond which there is no substantial similarity.
The truth is that you need both metaphors to understand copyright infringement, which is why we separate the copying (taking/theft) part from the substantial similarity (trespass) part. Otherwise, you’re stuck with an awkward qualification. You can steal, so long as you don’t steal the wrong things. Or, you can trespass, so long as you have no idea where you are.
What if the jury had been instructed as they would have been under Laureyssens, and what if the defendants conceded copying, so that Thicke’s and Williams’ testimony had been truncated, and the experts didn’t even have to testify? Would the jury still have come out the same way? I’m thinking not. If nothing else, we would be on much firmer ground to reverse a verdict of infringement.
The main argument against the “Blurred Lines” decision is that it will chill creativity because the jury found copyright in a style, which should have been treated as a non-protectable idea or scène à faire. I’ve already explained why I think that’s overblown—the jury verdict was a perfect storm of unlikely events.
All the same, the case exposes a tension and another place we might re-think copyright law. I would suggest that a cleaner separation between copying and substantial similarity is also a more fruitful way of discouraging juries from finding copyright in things that can’t be protected by copyright. For one thing, the inverse-ratio rule would be placed in a box, where it belongs. For another, the jury can think more clearly about that inflection point between expression and idea, and the defense can focus on it without distraction.
The thing is, the “Blurred Lines” jury was properly instructed about the difference between idea and expression (see especially instruction nos. 26 and 30). Perhaps if the instructions specified “style” as a kind of idea… but that sort of case-specific jury instruction is almost never even proposed, let alone permitted.21I’ll say it again. The craziest thing about litigation—which is pretty much crazy from beginning to end—is how rushed and rote jury instructions are handled.
No, if we want to fix something the “Blurred Lines” verdict shows to be broken, let’s just follow Laureyssens and keep copying separate from substantial similarity.
The good news is that the Ninth Circuit might be moving that way. In the Ninth Circuit’s Rentmeester v. Nike decision (involving the extent to which Michael Jordan’s pose in a photograph could be protected by copyright, which I discussed here), the court expressly distinguished between copying and similarity, noted the confusing nomenclature of using “similarity” in connection with two different concepts, and cited Laureyssens!22And it was cited for something similar in 2016 and 2003, though the 2003 cite seems not to have gained any traction. (See pages 8–9 and footnote 1.)
Alas, I don’t think this case is a good one to change the law in the Ninth Circuit. The intrinsic test/extrinsic test was incorporated into the jury instructions, and I don’t believe Williams and Thicke objected to it. Perhaps for that reason, Williams and Thicke didn’t make this argument on appeal. Which is too bad, because it’s pretty rare that separating copying from substantial similarity matters, so this would’ve been a good time for someone to raise it.23Not that I blame lawyers for, you know, following the law of their circuits. All we can hope for is that lawyers outside of the Second Circuit24And, well, the Second Circuit, too. insist on the distinction and that, at some point, it matters enough to be the basis of an appellate ruling.
Until then, thanks for reading!