Tuesday, August 06, 2013

Keller v. EA: visual elements mean game isn't protected by First Amendment

In re NCAA Student-Athlete Name & Likeness Licensing Litigation (Keller v. Electronic Arts Inc.), No. 10-15387 (9th Cir. July 31, 2013)

Short version: replay of Hart v. EA.  Blech.

Tyler Ochoa has another summary, with commentary, up at Eric Goldman’s blog.  He notes the undeniable medium discrimination at work, and also points to a peculiarity of the way the courts are forced to deal with Rogers’ right of publicity analysis—since Rogers isn’t just a Lanham Act case—to pretend that there is no circuit split: “[t]he Keller court bizarrely suggests that the Rogers ‘wholly unrelated’ standard was merely a federal court’s Erie prediction about a matter of state tort law, rather than a federal First Amendment limitation on state tort law.” 
By adopting the transformative use test to balance the First Amendment considerations because that’s what Comedy III did, Prof. Ochoa notes that “[t]he Keller court thus (somewhat strangely) abdicated its role as a federal court to construe federal law, and instead deferred to a state court’s view of federal law.”  I’m in full agreement with these points—as to the last, here’s how the court summarizes its decision: “Under the ‘transformative use’ test developed by the California Supreme Court, EA’s use does not qualify for First Amendment protection as a matter of law because it literally recreates Keller in the very setting in which he has achieved renown” (emphasis added). 
Like the court in Brown v. EA, the court here spends a lot of time on the game’s aspirations to realism; but where realism confers First Amendment protection in Brown, it strips that protection in KellerComedy III, the court held, gave “at least five factors to consider” in determining transformativeness.  (Which on its face sounds like a pretty unpredictable and chilling test for artists.  But of course medium discrimination/discrimination against visuals will get us most of the way to a result, so it’s only going to chill visual/audiovisual creators.  Yay?) 

(1) If the celebrity likeness is one of the “raw materials” from which an original work, it’s more likely to be transformative than if the depiction/imitation of the celebrity is the sum and substance of the work.

(2) A work that is primarily the defendant’s own expression is protected, as long as the expression is something other than the likeness of the celebrity.  (The visual again!  Even a laser-focused biography would clearly be protected as “primarily the defendant’s own expression.”)  This factor evaluates likely purchasers’ primary motivations, as inferred from the work itself (I added that last bit, but it’s not as if courts ever consult anything but their own intuitions about this)—whether the purchaser wants a reproduction of the celebrity or the artist’s expressive work.  (How would this even be tested?  If Andy Warhol is protected because he’s famous and thus people want his work specifically, as per Comedy III, then surely EA’s games have similar status among the relevant audiences.  But I suppose the court presumes that videogame players can’t be interested in aesthetic/haptic experiences, though I have no idea what the court thinks they are interested in, other than an undefined “realism.”)

(3) To avoid quality judgments, the court should conduct a quantitative inquiry and ask “whether the literal and imitative or the creative elements predominate in the work.”  (1 unit of realism = x units of artistic skill?)

(4) In close cases, the court should ask whether “the marketability and economic value of the challenged work derive primarily from the fame of the celebrity depicted.”  (Why is this different from (2)?  It is certainly no different in the evidence the courts use to determine the answer—their own evaluation of the appeal of the work.)

(5) “[W]hen an artist’s skill and talent is manifestly subordinated to the overall goal of creating a conventional portrait of a celebrity so as to commercially exploit his or her fame,” the work is not transformative.

The court ran through the key California cases: Comedy III, Winter v. DC Comics (transformative because the Winter brothers “are but cartoon characters . . . in a larger story, which is itself quite expressive,” which the Keller majority specifically calls out as important but never mentions again); Kirby v. Sega; and No Doubt v. Activision (holding that appearing in a context of a game with many other creative elements didn’t make the avatars anything other than exact depictions of No Doubt’s members doing exactly what they did as celebrities).  The 9th Circuit also has Hilton v. Hallmark Cards, finding that a greeting card showing Paris Hilton working as a waitress wasn’t necessarily transformative.

Using those cases, the court concluded that EA’s use wasn’t transformative enough to entitle EA to the defense as a matter of law.  As in No Doubt, “users manipulate the characters in the performance of the same activity for which they are known in real life,” and the context was similarly realistic.

The dissent believed that the overall context of the game, in which Keller’s avatar was only a small part, showed transformativeness, but the majority didn’t agree: just because there were other creative elements in the game, that didn’t transform the avatars at issue into anything more than exact depictions of the players doing what they did as actual players.  “[T]he fact is that EA elected to use avatars that mimic real college football players for a reason. If EA did not think there was value in having an avatar designed to mimic each individual player, it would not go to the lengths it does to achieve realism in this regard. Having chosen to use the players’ likenesses, EA cannot now hide behind the numerosity of its potential offenses or the alleged unimportance of any one individual player.”  Nor did the ability to alter avatars matter; the avatars weren’t “fanciful, creative characters” or “portrayed as . . . entirely new character[s].”  (So much for biopics, eh?)  No Doubt was the best evidence of what California’s highest court would do, and the case was so similar that it controlled, given the court’s obligation to follow state courts on a matter of state law.  (As opposed to the First Amendment controlling, I suppose.)

In footnotes, the majority also rejected EA’s argument that No Doubt was distinguishable given the allegedly breached license agreement in that case; the No Doubt court didn’t rely on breach of contract in its right of publicity analysis.  Plus, Keller argued that EA contracted away its First Amendment rights in a licensing agreement with the NCAA that purportedly prohibited the use of athlete likenesses. This was for the district court on remand if a factfinder determined that EA’s use was transformative.  

Separately, the majority rejected the dissent’s argument that Keller was distinguishable because “an individual college athlete’s right of publicity is extraordinarily circumscribed and, in practical reality, nonexistent” because “NCAA rules prohibit athletes from benefitting economically from any success on the field.”  The NCAA rules are unfair, “but setting fairness aside, the fact is that college athletes are not indefinitely bound by NCAA rules…. EA’s use of a college athlete’s likeness interferes with the athlete’s right to capitalize on his athletic success once he is beyond the dominion of NCAA rule.”

The majority also disputed the argument that its rule would threaten all realistic depictions of actual people.  Comedy III “requires an examination of whether a likely purchaser’s primary motivation is to buy a reproduction of the celebrity, or to buy the expressive work of that artist,” quoting McCarthy.  (So what kind of evidence does EA get to put in on this?  Does it get to show that, like Andy Warhol, it has a reputation that drives sales of new EA games?  Does it get to show target consumers a crappy game featuring Keller’s likeness and an EA game and ask if consumers care about the difference, thus almost certainly demonstrating that the primary motivation for picking a game is the quality of the game and not the fact that it has Keller’s likeness in it?  Despite the total absence of evidence on these issues in previous cases, it seems to me that if we’re going to pretend that it’s the selling power of the celebrity identity that matters, one might at least empirically investigate whether that celebrity—or even the aggregate identities of the players—is actually affecting consumer choices compared to other elements of game quality.  Of course, “primary motivations” can be difficult to disentangle; Keller’s lawyers will probably suggest a different counterfactual/control, such as a generic football game produced by EA, though I imagine that even then they might be disappointed by the results.)

Anyhow, says the majority, this “primary motivation” test (which incidentally is pretty hard to distinguish from the execrable Doe v. TCI) “leaves room for distinguishing between this case—where we have emphasized EA’s primary emphasis on reproducing reality—and cases involving other kinds of expressive works.”  (Like biographies?  Or paparazzi photographs?  The court is simply unwilling to recognize that video games might have artistic value of their own; as with every other form of art, one way that such artistic value can come is from a style recognized as realist.)

The court then rejected Rogers for the right of publicity, since Rogers was designed to protect consumers from confusion, whereas the right of publicity “protects a form of intellectual property [in one’s person] that society deems to have some social utility.”  (Which society?  Which legislature explicitly decided that the right had utility applied to artistic works?)  Acquiring notoriety may require a lot of investment to generate commercial value, which is then allocated to the celebrity: “The right of publicity protects the celebrity, not the consumer.”  (As Professor Ochoa also noted, it’s very hard to understand why this is an argument for more limited First Amendment protection against right of publicity claims, and the court doesn't try to tell us.)  EA appropriated Keller’s talent and years of hard work on the football field.  (Like biographers, sports photographers, etc.)  True, Rogers also evaluated a right of publicity claim, but it was just guessing what Oregon courts would do, whereas the Ninth Circuit knows what California courts would do.

The majority also addressed two other defenses: one is for common law right-of-publicity claims for the “publication of matters in the public interest.” For statutory right-of-publicity claims, the “use of a name, voice, signature, photograph, or likeness in connection with any news, public affairs, or sports broadcast or account, or any political campaign” is exempt.  Both defenses protect only the act of publishing or reporting.  California courts have held documentaries protected, as well as baseball players’ data, verbal descriptions, and video depictions in game programs and on a website.  And posters portraying football victories were also protected.

Here, however, unlike those cases, “EA is not publishing or reporting factual data.”  But somehow the video game is still realist.  There is in fact a particular conception of realism that works without facts—realism as ideal type (e.g., the Pilgrim’s Progress or Audubon’s bird portraits)—but it’s completely inconsistent with the concept of realism as accuracy to specific individuals to which the court appealed when finding EA’s use nontransformative. 

Discrimination against video games allows the court to have it both ways: “A video game is a means by which users can play their own virtual football games, not a means for obtaining information about real-world football games.”  (Note the subtle shift from information about the players to information about the games.)  True, EA incorporated actual player information into the game, which is the basis for the suit, but “its case is considerably weakened by its decision not to include the athletes’ names along with their likenesses and statistical data. EA can hardly be considered to be ‘reporting’ on Keller’s career at Arizona State and Nebraska when it is not even using Keller’s name in connection with his avatar in the game. Put simply, EA’s interactive game is not a publication of facts about college football; it is a game, not a reference source.” The state law defenses didn’t apply.

Likewise, the majority rejected the dissent’s argument that the First Amendment should control because “the essence of NCAA Football is founded on publicly available data,” as in the fantasy baseball games protected in C.B.C. Distribution.  But there was a “big difference” here: pictures!  Fantasy baseball just used names, performance, and biographical data.  NCAA Football used virtual likenesses.  (I was under the impression that both the statutory and common-law rights of publicity purported to cover both names and likenesses on the same terms.) 

“It is seemingly true that each likeness is generated largely from publicly available data … but finding this fact dispositive would neuter the right of publicity in our digital world. Computer programmers with the appropriate expertise can create a realistic likeness of any celebrity using only publicly available data.”  Judge Kozinski’s White dissent remains prophetic: “But what does ‘evisceration’ mean in intellectual property law? Intellectual property rights aren't like some constitutional rights, absolute guarantees protected against all kinds of interference, subtle as well as blatant.  They cast no penumbras, emit no emanations: The very point of intellectual property laws is that they protect only against certain specific kinds of appropriation.” 

But, the majority continued, what if EA created a virtual likeness of Tom Brady using only publicly available data? Would it have free reign to use that likeness in commercials?  Of course not! Therefore EA loses in this case.  (You don’t need me to point out that adding “in commercials” changed the example entirely and proved nothing, right?  Also, if I’m not mistaken, this is the circuit that found a claim in the use of the name—not the image—“Lew Alcindor” in an ad; does the court now think that use of nonvisual data alone in an ad is protected by the First Amendment, since it’s the virtual likeness and the biographical data that makes the difference?)

Judge Thomas dissented, largely agreeing with Judge Ambro’s Hart dissent, down to accepting transformativeness as the proper test.  “Because football is a matter of public interest, the use of the images of athletes is entitled to constitutional protection, even if profits are involved.”  The dissent cautioned that Comedy III didn’t really set out five factors, but engaged in a “more holistic examination of whether the transformative and creative elements of a particular work predominate over commercially based literal or imitative depictions.”  Too much deconstruction misapplies the test, which is what happened here.  The key question is whether the overall work is transformative, “and whether the transformative elements predominate, rather than whether an individual persona or image has been altered.”

NCAA Football was a work of “interactive historical fiction.”  Players role-played as college football players.  Gamers have many options for how they can behave and alter avatars.  “The athletic likenesses are but one of the raw materials from which the broader game is constructed. The work, considered as a whole, is primarily one of EA’s own expression.”  The marketability and economic value of the game came from its creative elements, not pure commercial exploitation of celebrity image.  (For fairness: I also wonder how the dissent knows this.  The fact that the majority and the dissent know completely different facts about the sources of the economic value of the game indicates that the word “economic” is a screen for a different concept altogether, something about artistic value and/or desert and/or acceptable versus unacceptable free riding.)  “The game is not a conventional portrait of a celebrity, but a work consisting of many creative and transformative elements.”

No Doubt was distinguishable—the literal representations there couldn’t be transformed in any way, but were created by using motion capture photography.  Character immutability was a key distinguishing factor.  Initial placement of realistic avatars in the game wasn’t enough to overcome the First Amendment’s protection, as No Doubt held, “even literal reproductions of celebrities may be ‘transformed’ into expressive works based on the context into which the celebrity image is placed.”  

The dissent wouldn’t punish EA for its realism and “for the skill of the artists who created realistic settings for the football games. That the lifelike roar of the crowd and the crunch of pads contribute to the gamer’s experience demonstrates how little of NCAA Football is driven by the particular likeness of Sam Keller, or any of the other plaintiffs, rather than by the game’s artistic elements.”

The dissent also pointed out that a “balancing” inquiry also requires consideration of the strength of the interest on the other side of the scales: not all right of publicity claims implicate the same concerns.  Here, NCAA Football was different from the traditional subject matter of right of publicity cases, both quantitatively and qualitatively.

Quantitatively, the sheer number of virtual actors involved made a difference.  The game’s cast of thousands gained particular significance in determining “whether the source of the product marketability comes from creative elements or from pure exploitation of a celebrity image.”  There was no evidence that Sam Keller possessed any personal marketing power, as distinguished from the appeal of the game’s creative aspects, and in any event, the sheer numbers involved “inevitably diminish the significance of the publicity right at issue.”  In the key California cases, the celebrities were the only subject matter, or they were pivotal characters—they were “central to the production,” and contact with consumers was “immediate and unavoidable,” whereas one could play NCAA Football thousands of times without ever encountering a given avatar. 

Qualitatively, NCAA Football was based on publicly available data.  True, EA solicited and received information directly from colleges and universities, but that information was hardly proprietary—it was statistics for players, found all over the place.  This was indistinguishable from C.B.C. Distribution.

The dissent also addressed whether “any use of a likeness founded on publicly available information is transformative.” The hypothetical virtual Tom Brady ad, the dissent reasoned, was inapposite because (1) commercials aren’t interactive, and (2) Brady had established marketing power, as plaintiffs didn’t.  (Sigh.  This is what you get when you fail to distinguish commercial from noncommercial speech—a bunch of other distinctions that don’t work very well.) 

Here, the structure of the game wasn’t founded on exploiting individual publicity rights—the players were unidentified by name, though one could easily put names to stats.  Anonymity bore on whether the product’s marketability was driven by an individual celebrity, or by the game itself.  (One might distinguish between marketability driven by realism and marketability driven by desire to possess some artifact representing the celebrity avatar, though again that requires a level of psychological inquiry I doubt courts can handle.)  The lack of names also distinguished No Doubt, “where the whole point of the enterprise was the successful commercial exploitation of the specifically identified, world-famous musicians.”  Anyway, college athletes’ rights of publicity are heavily restricted.  This is unfair, but the NCAA’s relationship to student athletes was beyond the scope of the appeal.  Given the quality of the right at issue—tiny—the First Amendment balancing should favor EA.

The majority responded that EA’s use of realistic likenesses demonstrated that it sees “value in having an avatar designed to mimic each individual player.”

But the same is true of any right of publicity case. The defendants in Winter saw value in using comic book characters that resembled the Winter brothers. Andy Warhol—whose portraits were discussed in Comedy III—saw value in using images of celebrities such as Marilyn Monroe. In those cases, the products’ marketability derives primarily from the creative elements, not from a pure commercial exploitation of a celebrity image. The same is true of NCAA Football.

As the dissent pointed out, “[t]he logical consequence of the majority view is that all realistic depictions of actual persons, no matter how incidental, are protected by a state law right of publicity regardless of the creative context. This logic jeopardizes the creative use of historic figures in motion pictures, books, and sound recordings.”  (In practice, I expect the dissent will be wrong, given the persistent medium discrimination in which courts engage—but it could be right about motion pictures!)

No comments: