Very interesting case with a lot to chew on. Among other things, in the course of holding that buy-a-lexus.com and buyorleaselexus.com are nominative fair uses of the Lexus mark, Judge Kozinski rejects earlier panel precedent holding that the burden of proof for nominative fair use is on the defendant. Under KP Permanent, he points out, the burden of proof of likely confusion is always on the plaintiff. Therefore, a defendant “need only show that it used the mark to refer to the trademarked good,” after which the burden “reverts” to the plaintiff to show that the use was not nominative. Well, sort of: the burden reverts to the plaintiff to show “a likelihood of confusion.” And the nominative fair use test “evaluates the likelihood of confusion in nominative use cases."
See, the problem is that third factor of the nominative fair use test, which Kozinski here restates as whether “defendant falsely suggested he was sponsored or endorsed by the trademark holder.” Since we know that in the Ninth Circuit the nominative fair use test replaces the standard multifactor likelihood of confusion test, and Kozinski explicitly says that it was error for the district court to use the multifactor test and the nominative fair use test sequentially, how are we supposed to know whether this occurred?
As in Abdul-Jabbar, the Tabari court made clear that something less than explicit claims of official authorization would satisfy the third factor. With respect to domain names, the court explained that “naked” use of a trademark, or a trademark plus a geographic indicator, as a domain name would not be nominative fair use. Though the opinion didn’t use the term, the reasoning is apparently that the necessary implication of trademark.com, e-trademark.com, trademark-usa.com and the like is that the site comes from the trademark owner, whereas no such necessary implication comes from more contextually ambiguous domain names such as trademarkdealer.com. But this rule only works if courts are very confident of their ability to assess context and also very confident that consumer understandings of domain names remain static—as Judge Kozinski evidently is.
Also recall that the alternative to explicit falsity/necessary implication in false advertising is proof through a consumer survey. Will a survey now suffice to rebut a nominative fair use defense? Can a court eyeball the situation? Given that the usual key factors in the multifactor test are pretty bad guides for figuring out whether a referential use is confusing, what might we substitute other than our expectations about how consumers think?
I don’t think Kozinski is currently wrong about how consumers use domain names. (Judge Fernandez’s concurrence notes just how many factual claims the opinion makes; the question of the threshold for adoption of common knowledge into legal rules is an interesting one, and quite relevant to many Lanham Act cases, but I won’t go into it further here.) But this ties into what I think is the basic problem here. Judge Kozinski’s version of nominative fair use repeatedly conflates the normative with the empirical. Nominative fair use precludes plaintiffs from showing likely confusion with the multifactor test, including with evidence of actual confusion or consumer surveys. It preserves space for competition and free speech and, not incidentally, deters trademark owners from attempting to exercise certain kinds of control in the first place, thus deterring them from trying to change consumer understandings of the meanings of certain uses. But this has to be a normative commitment if it’s going to work in the long term, which is why there is so much trouble with that third factor.
So what exactly happens when a defendant shows that its use is referential? Apparently a plaintiff then has to show that the use is likely to confuse by showing that the use, while nominative, is not necessary (with necessary being defined broadly), uses more of the mark than necessary and/or suggests source or sponsorship. Even aside from the factor-three problem, this raises various questions. For example, what if the plaintiff shows that the use is not necessary even under a broad definition of necessary. Because the nominative fair use test is a substitute for the regular confusion test, does the defendant automatically lose, or do we then hold a confusion inquiry?
Judge Kozinski gives a (disturbing) hint of the answer when he discusses using more of a mark than necessary. Notably, Judge Kozinski stated that doing so “might” cause confusion, a lower standard than “likely” confusion, and yet he still seemingly presumed that flunking the nominative fair use made confusion likely: “use of the stylized Lexus mark and ‘Lexus L’ logo was more use of the mark than necessary and suggested sponsorship or endorsement by Toyota....[T]hose visual cues might lead some consumers to believe they were dealing with an authorized Toyota affiliate.”
Another puzzle: the opinion goes out of its way to emphasize the availability of dilution claims against nonconfusing websites (though federal dilution isn’t available against a nominative fair use), and even calls out the totally unrelated peopleofwalmart.com as potentially dilutive (even though that seems like a classic nominative fair use). I’m not sure how Judge Kozinski squares his support of dilution with his position that enjoining Tabari’s use of the domain names violates the First Amendment because it interferes with commercial speech that is neither false nor misleading. Even if peopleofwalmart.com is commercial speech, which I doubt he thinks is true, the domain name and the website seem … true and not misleading.
Finally, there's a remand, though what it's for is hardly clear. The opinion might be read to hold out the formal possibility that a proper application of the nominative fair use test, with all the uncertainties discussed above, might legitimately result in a narrower injunction against certain uses of lexus in the Tabaris' domain names, and yet the discussion also nearly compels the conclusion that their domain names are exactly the types of domain names that ought to pass the nominative fair use test. (E.g.: "Outside the special case of trademark.com, or domains that actively claim affiliation with the trademark holder, consumers don’t form any firm expectations about the sponsorship of a website until they’ve seen the landing page—if then.... So long as the site as a whole does not suggest sponsorship or endorsement by the trademark holder, such momentary uncertainty does not preclude a finding of nominative fair use.") My best interpretation is that a remand is required to figure out if any prospective relief is available with respect to the site content, not the domain name, since the content too is entitled to a nominative fair use analysis.