Using Experts to Prove Software Copyright Infringement

[UPDATE: It turns out that my initial thoughts mirrored EA's here, and that Antonick filed a reply brief. It's interesting enough that I took a closer look at the initial briefing (and at the District Court), and I've updated/edited my post below.]

I ran across an interesting cert. petition today that I thought I would share and discuss. The case is Antonick v. Electronic Arts, 841 F.3d 1062 (2016), and the petition (filed by David Nimmer, Peter Menell, and Kevin Green) is here. The case is interesting because it is about software copyright infringement, a topic near and dear to my heart on which I've written and blogged several times.

It's also topically relevant, because it is about Madden Football, one of the more popular sports video game franchises (it's probably the most popular, but I didn't do a search to find out). Antonick was an author of the original game, dating all the way back to the Apple II (!), but had a license that he would be paid for any derivative works. And so the question was whether his code was incorporated into newer versions of the software published for Sega and Super Nintendo.

The problem was that nobody could find the all source code for any of the versions to compare, and the graphic displays were not admitted into evidence. There were snippets, drafts, and binary data files. Using these, "Antonick's expert, Michael Barr, opined that Sega Madden was substantially similar to certain elements of Apple II Madden. In particular, Barr opined that the games had similar formations, plays, play numberings, and player ratings; a similar, disproportionately wide field; a similar eight-point directional system; and similar variable names, including variables that misspelled 'scrimmage.'" Based on this and other circumstantial evidence, the jury decided for infringement and the plaintiff.

But the District Court reversed the judgment as a matter of law. The Court ruled that the jury could not decide infringement because it did not have the source code in evidence to compare, and that the expert's testimony was insufficient to show infringement.

And here is where the interesting legal issue comes into play: what is the role of expert testimony? I'll discuss more after the jump, but here's a teaser: I think the expert can play a role, and while that is the focus of the "legal" issues in this case, I am not sure that's what's driving the opinion. In other words, my sense is that the cert. petition's claim of a circuit split is in law more than it is in practice. That may be enough for a certiorari grant; I tend to think that Antonick got a raw deal here, so if his lawyers can convince the Court to take this case, more power to him. That said, my gut says that, perhaps through no fault of his own other than waiting too long to sue, the plaintiff just didn't have enough evidence here--and if they did, they couldn't convince the District or Appellate courts of it.

We start with the cornerstone case in this area, Arnstein v. Porter. This was a music case, but it sets out a two-step process used in virtually every copyright case. First, determine if there was "actual" copying - did the defendant copy from the plaintiff. Second, determine if there was "illicit" copying - did what was copied constitute so much similarity that the intended audience believes something was wrongfully taken from the plaintiff.

The first step is generally objective. Experts are used for the analysis, and we consider similarities even in those things that are not copyrightable, like copied facts. We're only trying to tell whether the defendant copied or created it independently. The second step is subjective, and the court says experts have no place in telling juries what the important parts of the work are.

The 9th Circuit, where Antonick's case takes place, has adopted this test generally, though it gives a different name: the extrinsic test, which is objective, and the intrinsic test, which is subjective. There are some other differences, but they are not terribly important for the core dispute here, namely when you can use an expert.

Using this framework, in Antonick, the 9th Circuit makes clear that "Antonick's claims rest on the contention that the source code of the Sega Madden games infringed on the source code for Apple II Madden. But, none of the source code was in evidence. The jury therefore could not compare the works to determine substantial similarity." As the district court put it, "the jury had no evidence of Apple II Madden or Sega Madden as a whole to enable it to make this subjective comparison."

Finally, the court ruled that, "our law is clear that expert testimony cannot satisfy a plaintiff's burden of proof under the intrinsic test, which 'depend[s] on the response of the ordinary reasonable person.'"

It is here that the cert. petition cries foul. It argues that the Arnstein v. Porter rule (and the related 9th Circuit rule set in a case about McDonald's characters) were simple enough to understand for those works, but that it is a real problem to allow juries to try to decipher computer software without expert witness help.

I agree with the plaintiff's position; experts should be allowed to explain the code. But in the context of the case, and in the context of law on the books v. law on the ground, I'm having a harder time.  Let's start with the context of the law on the books. Despite the nominal rule that the 9th Circuit disallows experts, the case law and practice is to the contrary. The famous case of Apple v. Microsoft makes this clear: "Using analytic dissection, and, if necessary, expert testimony, the court must determine whether any of the allegedly similar features are protected by copyright....Having dissected the alleged similarities and considered the range of possible expression, the court must define the scope of the plaintiff's copyright...and set the appropriate standard for a subjective comparison of the works to determine whether, as a whole, they are sufficiently similar to support a finding of illicit copying." As Antonick's reply brief points out, experts were allowed in Oracle v. Google and no one blinked. And nobody objected in this case.

In other words, expert testimony can and should be used to understand computer software, just as the plaintiffs would have it here. However, once that understanding is set, the jury must decide whether the remaining similarities constitute illicit copying. I have used this methodology in my own cases in Ninth Circuit courts. I have never had a problem getting expert testimony admitted to show copying or aid the fact-finder in understanding the code where there was competent evidence available for comparison. It turns out that Antonick, doesn't disagree with me. But he argues that this difference between law and practice creates uncertainty and bizarre ways that expert testimony comes in. This is a fair enough point, and may be enough to get the attention of the Supreme Court.

But that leads us to the context of this case. The expert testified to the jury that the source code was copied and similar. But there was a significant dispute about how the expert could have known that without having the full source code. The district court opinion, which is unpublished, makes it clear: the expert said that a lot of the features were the same, plus the misspelling of the one variable. But they could have created those similarities using their own re-written source code. Indeed, in a clean room, they may have misspelled the "scrimage" from a display screen rather than by copying code. And any source code copied could have been miniscule compared to the remaining software and/or unprotected elements. Further, the court had already examined the two programs and found much unprotectable, such that infringement required virtually identical copying. But the expert could not opine on virtually identical copying. We don't know any of this because there was, in the eyes of the court, insufficient evidence.

A couple points on this. First, I'm not saying that this is actually what happened. I wasn't in the courtroom, and I have not read the exhibits. I'm saying that the trial court thought this, and the appellate court disagreed. Second, note that my skepticism here does not come from an anti-software bias; I'm generally of the view that structure, sequence, and organization of code is protectable (and have said so in my writing). But you have to have some structure, sequence, and organization to show as your copyrighted work, and to compare against. In this case the court said there was neither.

Thus, there may well be a circuit split where--in practice--the use of experts in helping the jury understand software would make a difference depending on the circuit. But I'm still not sure that's  this case. I see this case as a failure of evidence case, indeed a best evidence rule type of case that comes out the same way in every circuit. And, in reading the briefing at the district court level, those were some of the primary arguments on each side; the expert issue was a secondary. The court has ruled that where the underlying evidence is unavailable for examination by the jury (or anyone, really), an expert may not simply opine that there was copying. Indeed, the Ninth Circuit makes this clear: right after discounting the expert testimony, it implies that lay testimony about what the source code looked like might have been useful: "the lay testimony was about how the games appeared, not how they were coded — and Antonick does not assert a copyright interest in Apple II Madden's audiovisual appearance, only in its coding." Antonick will surely argue that's the evidence he, in fact, presented. If so, then he got the short end of the stick on this one. But I believe that if he had the full source code and his expert made the same testimony, the court rules in his favor on identical facts. I'm just not convinced, in practice, that this was an expert issue.

In short, this is an interesting case. It is a credit to Antonick and his counsel that the case progressed as far as it did. But proving copyright infringement of computer software source code without the actual source code to compare is next to impossible. While the Ninth Circuit is stricter with experts than other circuits, the differences are surmountable...until they aren't.

[Update Nov. 6, 2018: The Supreme Court denied cert. a week after I wrote this. The timing is such that I'm certain this had no effect on it, though it's pretty self-important of me to think that it would have]