The Happiest Ending for the “Stairway to Heaven” Case

Wherein Some Bad Old Mistakes Are Fixed, but Led Zeppelin Need Suffer No Longer

The last couple of times I blogged about the “Stairway to Heaven” case, I was defending the Ninth Circuit’s decision to send the case back down to the trial court for a re-do, even though Led Zeppelin was likely to prevail again.¹OK, but in that last one, I was starting to agree it wasn’t worth saving. It was too bad Led Zeppelin would have to go through the trial all over again, but the Ninth Circuit was moving away from one of copyright law’s more pernicious mistakes: thinking that “copying” can be proven by the defendant’s “access” to the work and the “substantial similarity” between the two works.

What’s so pernicious about this idea is it mixes up and mixes together two distinct concepts: factual copying and wrongful copying.

What’s exciting is the Ninth Circuit found a way to fix this mistake, jettison the “inverse-ratio” rule, have a grown-up conversation about “scope of expression” (especially in the context of songwriting), all without making Led Zeppelin suffer. Instead of sending the case back down for a re-trial, the entire Ninth Circuit (as opposed to a normal three-judge panel), known as “en banc,” just stepped in and fixed everything.²Actually, I think the Ninth Circuit is so large that even “en banc” doesn’t include every single judge.

I think the amici deserve a lot of credit for getting the Ninth Circuit to take these issues seriously and giving them the necessary perspective.

(Representation of the Ninth Circuit’s journey to enlightenment.)

Solved: The Mixed Up Framework for Infringement

It’s easy to forget that copyright law protects against copying, not merely similarities. It’s not enough to say (for example), “OMG, that song reminds me of this other song.” You have to first prove that the writer of “that song” copied from “this other song.”

Now, sometimes that songwriter just up and tells a reporter, “I was like, damn, we should make something like” that other song. In that case, well, sure, factual copying has been proven. Or, at least, mostly proven, so long as the songwriter doesn’t go on to say something like, “Then Pharrell started playing a little something, and we literally wrote the song in about half an hour.”

Normally, stuff like that doesn’t happen. You have to prove copying circumstantially. Traditionally, this means proving “access” and “substantial similarity.”

And immediately, our problems begin. First, “substantial similarity” isn’t about copying. It’s about wrongfulness, about taking what you shouldn’t from another work. Or, to reverse it, about taking what you’re permitted to take from another work. When someone sues because his and your TV shows are both about a family patriarch who runs a successful and glamorous music label and must contend with both family and business drama, and you win because no one can own the concept of “a family patriarch who runs a successful and glamorous music label and must contend with both family and business drama,” that’s substantial similarity at work (you can call it “substantial dissimilarity” if you want). You aren’t liable for copyright infringement even if you admitted you were influenced by his TV show. Substantial similarity cares about what is similar and whether those similarities are protected by copyright law.

Factual copying doesn’t get into those weeds. It just wants to know whether you copied. It doesn’t care if your copying was wrong or not—that’s for later. Thus, the similarity between your TV show and that guy’s TV show would be admissible to show copying, even though it ultimately isn’t enough to show wrongful copying. We just want to know whether you were at least influenced by the guy’s TV show.

That leads us to the other problem, access. As in, how much access is “access.” From the TV show example, you can see why access is an important component. If the guy’s TV show never aired and you never saw it (or any “treatment” of it), you can’t have copied it. Obviously. But let’s say the guy’s TV show did air, but only locally, or briefly, or on obscure cable channels. Assuming you deny seeing the show (or just can’t remember), is that enough “access”? Seems kind of thin, doesn’t it? Is that enough to make you think, “Yeah, he probably got the idea from that other guy”?

But a lot of courts and juries seem to think access as a threshold issue, not a quantitative one. That’s what seems to have happened in the “Dark Horse” case, where the copyrighted song got some radio play but not a lot, there was no further proof that any of the defendants ever heard it, and there was only one point of similarity.³As it happens, that jury verdict was just reversed, but on other grounds. The jury was instructed that any amount of access could, without more, be enough to prove copying.

Solved: “Inverse-Ratio” Rule

This is where the bugbear “inverse-ratio” rule comes in—or, at least, is supposed to come in. The more access you prove, the less factual similarity you have to prove. And the more factual similarity you prove, the less access you have to prove. The rule isn’t quite as symmetrical as that, though. Imagine you photocopy a book. I don’t need to know you had access to the book to prove copying because there’s no way you would independently create an exact copy of that book—some similarities just can’t be merely coincidental. But the converse isn’t true: just because I saw you with the book (rock solid access) doesn’t mean everything you ever write from then on out copies the book.

Coming back to the 9th Circuit’s en banc decision in the “Stairway to Heaven” case, the court did two important things regarding how to prove copyright infringement:

1.  It established, once and for all, that factual copying and “substantial similarity” are two different elements that both must be established to prove infringement. And it further clarified that these two elements look at similarity between the two works differently. With factual copying, similarity is relevant to establish a connection between the accused infringer and the copyrighted work, and thus we don’t care about what is and isn’t protectable at this stage. With substantial similarity, we care very, very much about what is and isn’t protectable because similarity is relevant to prove whether what was taken from the copyrighted work was legal or not.
2. It ditched the inverse-ratio rule.

OK, this last one requires some unpacking, though. As I mentioned, the inverse-ratio rule is justifiable if it’s limited to proof of actual copying. But if you let it out of its cage, it wreaks havoc. The Ninth Circuit considered limiting the inverse-ratio ratio rule to proving actual copying but decided against even that. I think there are two reasons for this. First, it’s too easily misunderstood and dangerous. Second, it’s misleading.

It’s misleading because it implies a mathematical precision to the determination of whether actual copying has occurred, like this: If access + similarity > 1, then copying has been proved. I myself have always just seen “inverse-ratio” as a mathematical metaphor, and like all metaphors, it has its limits. As I’ve mentioned, it’s not even symmetrical: in the mathematical formula, similarity can equal 1 (e.g., the photocopied book), but access never can (e.g., seeing you with the book). But the Ninth Circuit thinks even as a metaphor, the inverse-ratio rule can do, and has done, too much mischief.

Instead, juries will just hear evidence of access and “probative” similarity and come to their own conclusion about whether factual copying has been proven. So instead of a metaphor of a balance, the metaphor is a cookpot: stuff goes in, you stir it while cooking, and see what you have. Some further guidance to the jury might help, e.g., access alone can never prove factual copying, but similarity alone sometimes can.

A final note about this: It’s really, really unfortunate that we have to use the word “copy” to describe this. First of all, not everything that constitutes “copying” for legal purposes looks like “copying” in common parlance. “Copying” in the legal sense means doing any of the things copyright law says is exclusive to the copyright owner and its licensees: reproducing, adapting, distributing (“copies” of, ugh), publicly performing or publicly displaying.⁴OK, there’s a really obscure sixth one. Let’s not worry about it. Second, it implies the kind of “copying” you were told not to do in school: i.e., having the work at hand and intentionally, well, copying. It can be a lot broader than that. There is such a thing as unintentional copying, e.g., a catchy tune you heard somewhere and started to believe you made up yourself. When Robin Thicke told the world that he wanted to do a song like “Got to Give It Up,” that was enough to prove “copying,” even though he and Pharrell Williams only had the song in mind, not right in front of them. They were, at most, influenced by “Got to Give It Up.” And, remember, it was completely possible, even then, that whatever Williams composed in that moment⁵Alleged moment, since Thicke recanted everything, but not everyone believes him. was not illegal. You can totally be influenced by something and not infringe its copyright. The key is keeping those two concepts straight.

NEW Problem: Keeping the Different Similarities Separate

It’s great that the Ninth Circuit finally recognizes “different similarities.” I.e., that similarities between the two works serve different functions. As an unhappy corollary, different types of similarity will be relevant to these different purposes. And if that sounds confusion to you, just imagine how it’s going to be for the jury.

Here’s what I’m getting at. Substantial similarity (function: is what was taken wrongful?) will be proven by similarities of protectable elements, but factual or probative similarity (function: did accused infringer “copy” the copyrighted work) can be proven by non-protectable similarities. Remember, with factual/probative similarity, we’re just trying to draw a connection between the accused infringer and the copyrighted work, but we’re not (yet) arguing the accused infringer took what he or she shouldn’t have. Thus, the similarities that will be highlighted for the jury for purposes of proving “copying” will be broader than those highlighted for purpose of proving substantial similarity.

But this risks cross-contamination. Will the jury be able to “forget” the non-protectable similarities when considering substantial similarity?

Avoiding this will require careful planning by defense counsel AND some very carefully-written jury instructions. Since jury instructions tend to be terrible anyway, I suspect the burden will fall mostly on defense counsel.

Unsolved: The “Intrinsic” and “Extrinsic” Tests

Before we move onto some other exciting parts of the opinion, let us pause for a moment and wonder:

What about the “extrinsic” and “intrinsic” tests?

These two “tests” are supposed to be separate elements for proving “substantial similarity,” as that term has been used and misused in the past. Under the extrinsic test, you’re allowed to dissect the works into components for the sake of comparison, with the aid of experts, if you choose, and you’re allowed to consider any type of similarity, protectable or not. Thus, it sounds like a kind test for “probative similarity,” i.e., drawing the link between the accused infringer and the copyrighted work. But it has been made part of “substantial similarity.” The “intrinsic test” is just a jury’s gut reaction about whether the works, taken as a whole (no dissection!), are similar enough.⁶Fortunately, the “intrinsic” and “extrinsic” tests are not much used outside of the Ninth Circuit.

There is no place for these “tests” under the framework announced by the en banc Ninth Circuit in the “Stairway to Heaven” case. This, I would argue, is a good thing. But they were not explicitly rejected by the court, and they have such a long history in the Ninth Circuit that I fear they won’t go away immediately. They will still seem “good law.” I fear that when district courts will try to force these tests into the framework set forth here, they will make a mess of it. Because the tests partake of both actual copying and substantial similarity, but are described as components of substantial similarity, I fear that district courts will end up collapsing the distinction between actual copying and substantial similarity that the en banc panel has worked so hard to make. Other mis-shapen creatures are also possible.

This en banc opinion, though clearly as comprehensive as it could be⁷Remember, an appeals court can only address the questions put before it. Anything else and it’s “making law,” which is for legislative bodies., does not address what might be called the “filtration” or “dissection” problem that the intrinsic-extrinsic tests were meant to address. Works are meant to be compared as a whole because that’s how they present in the marketplace, but it’s also legal to take unprotected components, which requires dissection.

Well, actually, maybe it kind of does. I’ll try to explain next time.

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.