The Trademark Singularity: Part 1
Well, trademark use is all the rage in trademark thinking and commentary now, and well it should be. It goes, after all, to the heart of the endeavor, and I have written about it here in many different guises over the years.
Now Mark McKenna, one of my favorite guyses, has published an article in the Iowa Law Review on the topic. It is billed as a response to this earlier — and, it is fair to say, sensationally influential — piece by one of my other faves, Alexandra Roberts, in the same journal last year, but the lazy guys among us would do fine starting with either one of them because Mark does a great job of explaining how we got here, to wit:
Back in 2007, the Iowa Law Review published a dialogue between Graeme Dinwoodie and Mark Janis, on the one hand, and Stacey Dogan and Mark Lemley on the other. The topic was “trademark use,” and the question was whether such a doctrine really exists. Dinwoodie and Janis said no—that while only commercial use of a trademark can be considered infringing, there is no threshold requirement that the plaintiff prove that the defendant has used the mark in some particular “trademark” way. Dogan and Lemley said yes —that some “uses” of a mark simply don’t trigger liability, and a court needs to determine, as a threshold matter, whether the defendant has used the mark “as a brand.” I was not a disinterested bystander on this question. When I wrote, responding to both papers, I argued that trademark law does, and must, impose liability for only particular kinds of uses of a mark—uses “as a mark.” But because “trademark use” can only be determined from the perspective of consumers, I argued, the question of whether a particular use qualifies inevitably collapses into the likelihood of confusion analysis (and therefore isn’t a separate, threshold question.)
The impetus for this now well-cited exchange was a series of cases in which the issue was whether a party that delivers keyword advertisements infringes a mark owner’s rights by selling to competitors the right to have their ads triggered by keywords that correspond to the plaintiff’s trademark. In the years following the trademark use debate, courts largely sided with Dinwoodie and Janis (or, I think more accurately, with me)—rejecting the contention that trademark use is a separate, threshold requirement. That doesn’t mean that courts typically have found keyword advertising uses to be infringing. In the main, courts have held defendants’ sales of keywords for advertising non-infringing, though they usually have reached that result by finding that the defendants’ conduct was not likely to cause confusion.
Still, for most trademark lawyers, the idea that there’s no such thing as “trademark use” always seemed wrong. In fact, trademark use is everywhere in trademark law—in cases dealing with acquisition of common law rights, in priority disputes, in infringement cases, and in cases involving a variety of defensive doctrines. And, as Alex Roberts details in her excellent article Failure to Function, even more pervasively in the registration context.
Bottom line for Mark, at this point: Trademark use is a thing, of course, but judges understandably have a hard time separating it from likelihood of confusion, even though both concepts exist in both the statutory and common law definitions of trademark infringement. And Alexandra imports the dilemma to the USPTO in the context of trademark registration, which really raises the stakes — for two reasons: (1) As the Supreme Court reminds us, trademark registrations are really important; and (2) judges, and even more so juries, really really think they’re important.
Frankly, the discussion gets very esoteric very fast, and of course that’s what law professors do (get esoteric). And I, in contrast, am a simple suburban trademark lawyer. But I do want to consider one aspect of the discussion here, and it is embodied in this passage in Mark’s article. I have added emphasis:
Roberts is certainly right that there are no clear rules for determining use as a mark. It seems unlikely, however, that the Trademark Office could do much better given modern law’s functional understanding of the concept. As Roberts acknowledges, use as a mark is ultimately determined from the perspective of consumers. Since it is not feasible in the registration context to test that understanding with survey evidence, the best examining attorneys can do is rely on rules of thumb. Roberts suggests that Office could look to empirical studies to learn at the wholesale level what sorts of clues consumers rely on to draw source conclusions. But as it turns out, the features that literature emphasizes correspond reasonably well to those the T.T.A.B. has highlighted. Specifically, the empirical research suggests that consumers focus on features like shape, color, prominence, font, and marketing context in drawing conclusions about whether particular words or other indicators serve as branding elements. Those are the same sorts of contextual clues on which the PTO focuses, as Roberts notes.
The problem, then, doesn’t seem to be that the law hasn’t identified the right proxies. It’s more that proxies are only proxies, and functional evaluation resists predictable application. So it is with much of modern trademark law. The more holistic and robust use as a mark doctrine that trademark law once employed was tractable precisely because the law used to have a much stronger formalist dimension.
Here’s what I bolded, again: “It is not feasible in the registration context to test that understanding with survey evidence, the best examining attorneys can do is rely on rules of thumb.” Hold that thought, and stick with me while reading this next:
Given the empirical evidence demonstrating that contextual clues have a strong effect on consumer perception of source indication, there’s an obvious logic to making the Abercrombie categories (which are based on assumptions about consumer reaction that aren’t empirically supported) less determinative.
[N]either the concept of distinctiveness nor the Abercrombie methodology are motivated solely by assumptions about consumer understanding of particular types of terms. Descriptive terms (and other terms assimilated to that category) were once not considered trademark subject matter (categorically), and that wasn’t because courts assumed those terms could never indicate source. Indeed, unfair competition actions were available to parties using descriptive terms when they could prove the terms did, in fact, indicate source, and that others were using the terms to divert customers. Descriptive terms were excluded from trademark proper because they didn’t automatically indicate source, and because others had potentially legitimate reasons to use those terms. The prospect of legitimate use by competitors remains a concern with respect to these terms, and as a result, there remain good reasons to discourage claiming of descriptive terms even if they could be used in a way that indicates source.
And now I have slalomed my way through Mark’s article and completely ignored his important point about how, in the registration process and all that follows is, there is a risk of a disconnect between a plain word mark granted registration based on a specimen of use that may show not-so-plain use at all, and how courts deal with that possible dichotomy when asked to enforce that trademark. This builds on the influential work by Rebecca Tushnet cited in Alexandra’s piece, to wit:
In a 2017 article in the Harvard Law Review, Rebecca Tushnet recommends paying the registration process “renewed attention” as an opportunity to add necessary structure and increase consistency within and across federal trademark law. While the USPTO regards registration as substantive, Tushnet argues, courts often treat it as merely procedural. Her hypothesis suggests courts both undervalue the work of examining attorneys and the role of registration and, at the same time, overinflate the importance of the result of that process. By ignoring the details and treating registration as binary, courts often “round up,” assigning broad, robust protection to matter that might have gained registration based on a much narrower interpretation of the mark’s scope and the mark owner’s rights. The possibility of a registered mark achieving incontestable status further intensifies this effect. . . .
If the USPTO and courts were to pay use as a mark the attention it deserves, a more radical proposal arguably follows: courts could limit the scope of registered trademark or trade dress protection to the uses identified in specimens submitted to and accepted by the USPTO in connection with registration. After all, the scope of registered rights is already limited to those goods and services identified in the application.
Well, what this demonstrates, it seems, is how all roads lead us back to a sort of trademark Singularity.
So: It is axiomatic that trademarks only exist in the context of consumer perception — i.e., secondary meaning. Let us call this the Perception Axiom. But still, empirical “measures” (I will not say “proof”) of consumer perceptions regarding trademarks do not define the entire set of valid conceptual criteria for determining whether a trademark (a) exists and (b) is strong (Mark’s “formalist dimension”). And while these criteria may be viewed nonetheless as proxies for consumer perception, which seems to be unavoidable because of the Perception Axiom, we also acknowledge that (whether “more or less determinitive”), the Abercrombie categories are based on assumptions about consumer reaction that “aren’t empirically supported.”
So… what do I, reductionist-caveman-lawyer-like, see here? Well, let me add a few more points: As Mark and Alexandra and anyone else working in trademarks will tell you in a jiffy, empirical proof of consumer perception in the form of surveys is, conceptually and practically, a hot mess. The word is out that the cases say everything and anything about the presence, absence, admissibility, weight of surveys — not least in the context of genericness. The harsh truth is whether you spend $50,000 for $500,000 for a survey, judges can and do use or toss them to get where they want to go.
And where do they want to go?
It is not feasible in the registration context to test that understanding with survey evidence, the best examining attorneys can do is rely on rules of thumb
They want to employ rules of thumb just as the PTO seems to do — rules of thumb well guided by masterful decisions such as the one in Abercrombie. When it comes to trademarks, judges either (a) want nothing to do with them (and thus are never imposed upon to do so by the spin of the mythical “wheel” of judge assignment) or (b) are every bit as subjective, artistic and, well, capricious as anyone else about the damned thing.
That may be just fine, though it leaves us having to rely on a fiction: That the considerations expressed in Abercrombie are, really, just proxies for what, if our information were good enough, empirical proof would reveal to be consumer perception.
Which is just not true. Consumer perception, both atomically and in the aggregate, is at least as much of a confused mess concerning distinctiveness as your average trademark-curious judge, not least because it is affected by consumers’ own understanding of the law of trademark itself. That is absolutely a “contextual clue.”
Call it the Heisenberg Trademark Principle. Can I prove this is so empirically? No, but my three decades of prosecuting, litigating and trying cases about trademarks make my famously febrile figuring a fine proxy for that proof! It’s true that, unlike an Article III judge, the President didn’t appoint me to decide it, of course, yet I do so almost every day, just as everyone else does.
But didn’t I say something about some kind of trademark singularity? I did, because what did we say about surveys and the PTO? That they are not chocolate and peanut butter: They don’t go together. They can’t. I could do a whole separate blog post on that question, and I think maybe I will, but let’s take that assertion by me as an axiom, too. So we are back to employing proxies, or if you will other proxies, for consumer sentiment (“other proxies” because a survey is not empirical proof of “consumer sentiment” but a purported representation of it). Rebecca and Alexandra, more than Mark, argue that we should at least employ a lot more of them, and hearken to a greater appreciation for “context.”
This is not a law review article, however, as should be obvious in ways too painful to elucidate. That fact matters most, right now, in the sense that this post is getting too long. So: In my next post, I will discuss why “context,” while a necessary corollary of the Perception Axiom, presents numerous Big Problems in trademark registration and theory; why “all this” ends us getting us right back to the Booking.com descriptive / generic / acquired meaning / consumer perception / policy conundrum; and a third thing, pursuant to the Rule of Threes. Oh, yeah, this is the third thing: The Slants.