Third Trademark Scholars’ Roundtable, Bloomington, Indiana
The Role of Confusion in Modern Trademark Law
Graeme Dinwoodie and Mark Janis: Welcome and Introduction
Janis introduced the topic: confusion seems obvious but that’s why we’re focusing on it. Dinwoodie suggested that there is also ferment in European law and there may be some opportunity to translate scholarly thinking on the topic into legal outcomes.
Session 1: Methodological Perspectives on Confusion
Introduction: Jessica Litman
Overview of history, with big holes in it (Bob Bone’s paper covers a lot). 19th century cases: confusion over source was implicit in technical TM cases and explicit in unfair competition. It was very much “I know it when I see it” through the mid 20th century. Likelihood of confusion first showed up in statute in late 19th c. Precursor to 2(d) prohibiting registration of any mark likely to cause confusion with mark previously registered or previously known. Lanham Act draft initially mentioned confusion only in registration section but called the actionable thing “infringement.” Hearings made clear that people considered infringement to be confusion of source. Initially defined infringement in the definitions section involving deception/confusion as to source, then changed it to what became §32. Chief reason: concern about how the innovation of incontestability might expand the scope of TM rights, to rebut concern about people owning rights.
After that: mission creep for the multifactor confusion test. By 1965, we see district courts applying the test to products that are related but not identical, and by mid-1980s everyone’s using the test even for cases involving competing goods. For district courts: seemed to guide discretion without really constraining it in any way—test is a wonderful tool for persuading yourself that your initial intuition is correct. Courts of appeals: gave them something concrete to review and reverse. Inwood v. Ives came down in 1982, while all this was happening—we put the contributory infringement part in the casebook, but it’s worth noting what the case actually held—district court found noninfringement and functionality; court of appeals reversed on an “I know it when I see it basis,” but SCt reversed because court of appeals hadn’t found clearly erroneous factual findings. Great thing about multifactor test treated as mixed question of law and fact is that you can reverse a district court you think has erred without having to find clear error.
Primary discussants: Mark McKenna
Literature has really come to terms with enormous conceptual change in the 1920s-1940s; before that there was tendency to tell story of TM as historically consistent view of confusion as always a problem with slow change, until things went off the rails in the 1970s and 1980s. Applying TM to noncompeting goods required a real rethinking of premises; that’s where we start seeing “confusion” supplant “deception” as the right word. You don’t see deception any more after the 1920s-begun shift, and then “likelihood of confusion.” The case that was made to expand TM rights involved the reputational interests and market preemption of mark owners; tied to consumer interests [see Mark's comment below for more on that relation].
Reputational harm confusion: confusion resulting when consumers think plaintiff stands behind defendant’s goods. It may be that this was consistent with consumer expectations with some very related goods, but proponents were consciously trying to get out in front of marketing practices of changing. Only some companies were expanding, and only to closely related goods; but proponents believe in market preemption more generally. Courts start embracing likely confusion as a non-empirical concept. They aren’t requiring proof of likelihood if it’s implicit in the posited set of circumstances, and they aren’t requiring proof of harm, which is assumed to follow when people think the parties are related. They know what will happen if there’s an empirical test—a modest broadening of rights, but not the big expansion desired. Courts are willing to presume the harm. So expansion has been tied to a non-empirical claim, both at the level of whether confusion results and whether harm follows from confusion.
McKenna’s position: modern doctrine reflects search costs rhetoric pretty closely and that’s the problem. Search costs was an explanation for consequences of lack of protection for competing goods. Ironic, because this explanation is not developed until well after it’s clear that TM is far broader than that. Posner/Landes do not explain why search costs theory applies to noncompeting goods; assume that it self-evidently extends to all other aspects of TM law. Search costs makes the most sense w/r/t competing goods. Over time, search costs and confusion have been conflated. Search costs were thought to be the consequences of confusion—but courts have started to use the language of search costs to describe confusion, and any form of search costs can be used to find confusion.
Likelihood of confusion has always been nonempirical; should it reflect research findings? Some research goes to the question of when and whether confusion is likely. A lot of that research isn’t particularly useful because it doesn’t tell you very much about what kind of confusion they’re testing. They’re often asking consumers whether they associate two things, and then call that “confusion.” Not very robust. Another set of research is about reputation: consequences to mark owner if people think there’s a relationship—TM assumes this will happen as a matter of course but empirical literature tells us it happens rarely, and only with very related goods. Even if consumers believe there’s an association, consumers easily segregate reputational consequences. Literature strongly supports the idea that uses in other markets can have a preemptive effect—may prevent brand extension or shape meaning of brand making it more difficult to move it in another direction. This raises the question of whether this is normatively the kind of harm we should respond to—keep people out of the mark owner’s way so it can expand anywhere it wants. Let’s not talk about harms as if they were self-evident from confusion.
British history until 1994, when it became European. Contrasting history of expansion of confusion.
Other dynamics where notion of confusion contracts/willingness to ignore certain kinds of confusion. Misleading marks: inherently deceptive marks—19th c. courts were very quick to refuse relief where there was any sort of deception inherent in the mark—moralistic. If TM owner suggested the goods were patented when they weren’t, plaintiffs weren’t entitled to relief. Within the registration system we’ve always precluded registration of deceptive marks. Gradual increase in the threshold there—now require a serious likelihood of real deception as to the character of the goods before prohibiting registration. Narrowing of that concept of confusion. Why does that happen when we’re expanding the concept of confusion in other ways?
Willingness to ignore real confusion: where TMs are assigned and ownership transfers, either compulsory or voluntary. Many postwar cases involves TMs owned by the enemy, given to a UK-based establishment. After the war the old manufacturer wanted the market back; the courts would ignore the real confusion because they said the ownership shifted to the UK mark owner. Modern doctrine tolerates confusion generated by assignment—we want to regard property ownership/transfers as a trump. Sometimes we assume that the average consumer knows that the mark doesn’t indicate association with particular person, but association with an arbitrary source of the goods—the average TM attorney, not the real consumer.
Keep those contrasting stories in our minds when we consider expansion of confusion.
UK: more structured requirements: confusion requires the marks and the goods to be similar, at which point you then ask whether there’s likely confusion. This cabins likelihood of confusion compared to the US’s expansiveness. When we do get to our likely confusion analysis, we say something like: likely confusion means risk public might believe the goods/services come from the same or from economically linked undertakings. In that wording there may be a narrower notion of likely confusion than sponsorship/affiliation. But we haven’t really had any exploration about the breadth of economically linked undertakings. That could be very elastic.
Intent: not a consideration in European law. But English courts are much more likely to say that intent is evidentially relevant to likely confusion, though formally it’s not relevant. Infringement analysis is a mirror of registration analysis, and intent is not thought relevant when someone applies to register a later mark. In registration you’re just concerned bureaucratically whether the marks/goods are similar.
Free riding falls easily within the dilution concept in Europe, so there’s less temptation to expand the concept of confusion to cover free riding.
Max Planck Institute recently reviewed the Community TM. 200-page report covers likely confusion in about a page; it’s not particularly controversial. Pharma cases: the disconnect between how the pharma regulatory authorities behave and the TM authorities—TM allows more similar marks onto the register, but regulators are worried about people taking the wrong drugs. Institute suggests greater attention to the regulatory approach in the TM system w/r/t drugs because of the potential harm. Raises possibility of treating different fields of activity differently.
[US pharma regulators have their own name approval processes too, which played a role in one of my favorite cases, involving the TMs TRAVATAN and XALATAN. Pharmacia Corp. v. Alcon Laboratories, Inc., 201 F. Supp. 2d 335 (D.N.J. 2002). There is a good paper to be written, discussants agreed, about field-specific regimes for approving or disallowing names on the basis of similarity to an existing name in the field—DDMAC does it for drugs, and there are similar regimes for wine though Lemley says they receive no deference from the courts.]
The US is more empirical than Europe. The registration mentality leads to a nonempirical, routinized process.
TRIPs: says we have to give rights in cases of likely confusion with similar marks; in a developing country no one would think that you could change the law to a harm-based analysis.
McKenna: maybe likely confusion isn’t opposed to a harm test. We would want a focus on confusion of a certain type. Confusion about what? You can be confused about a huge range of things; specifying the type of confusion that is likely to cause harm makes sense.
Bently: likely confusion within TRIPs could mean a whole bunch of things, but if you had a very narrow concept of likely confusion there’s a good change the US would take you to the WTO on behalf of its TM owners.
McKenna: that makes it more important for the US to act first.
Lemley: Bently’s description: confusion caused by defendants is actionable, but confusion caused by TM owners is excused.
Lemley defends search costs model. Invocation of the term search costs alone can be divorced from a real context. Talking about meaningful consumer search costs is a way of trying to tie likely confusion to harm—why it has a consequence I should care about. It’s not that any confusion is problematic, but that we need to know whether defendant’s conduct is likely to confuse consumers in a way that causes harm. That’s an empirical question, though the US hasn’t treated it as such. Once you do that, it turns out that confusion is less often harmful than you think. Asking the empirical question reinforces rather than undermines the idea that we should care about search costs.
Robert Burrell: We ignore confusion in other cases. Australia: cases of market flooding, where the reputation of the junior user overwhelms that of the senior user; senior wins. Ignore confusion when there’s a perception of relation between a performer and another performer of the same work. Record consumers encountering a cover version often assume authorization by the performer instead of the copyright owner; we ignore that. We ignore evidence of confusion when the goods are so far removed that we don’t want to give the owner a remedy—“Big Mac” on bottles of wine—evidence of confusion, but held to be only an “erroneous assumption.” [We tend to do that in the US only with false advertising!]
Psychologists: if you ask people if TMs are confusing, people are good at figuring out what would cause delayed reaction times. If you give them a list of factors and ask them to evaluate whether there’s confusion, people become much less good at figuring that out. [RT: Verbal overshadowing!]
Bob Bone: Took McKenna’s argument to be more explanatory than normative—focus on search costs leads courts to look for confusion in the abstract, which then expands. He’s not sure the story is right. Search costs seem like a relatively recent idea; not a prevalent way of talking though it happens, and the focus on confusion dates further back. Other story: for variety of reasons, we detached confusion from harm, which was reinforced by the search cost rationale unhinged from its normative basis.
Bone thinks we’ve ignored nonconsequentialist rationales in TM normatively (though many are empty) as well as descriptively. Intent, secondary meaning strength are significant factors but marketing channels, bridging the gap are not—why are the former salient to the test? They tie directly into nonconsequentialist ideas (intent, anti-free riding, even appropriation of identity), and they shift us away from the need to do any kind of empirical inquiry.
Stacey Dogan: How does the Arsenal case fit into Bently’s story?
Bently: Double identity rules: identical mark on identical goods, there is no need to show likely confusion.
Dogan: swallow the rule? Or at least make the US story look more like the European?
Bently: because it’s so easy to register, that is a possibility. TM owners aren’t worse off in Europe, but the problems aren’t so much in the idea of likely confusion, but registering without any proof of use (though five years of nonuse can get you in trouble)—one big issue is when you can get a mark off the register.
McKenna: he’s happy to say courts weren’t motivated by search costs when they developed these doctrines. They were happy to talk about confusion and then assume the harm; they had a clear idea of the harm and just wanted to assume it happened without needing proof. There is a way to articulate search costs in a way tied to harm, but it’s so watered down that all we think is that confusion is harm, and this concept has infected too much of our reasoning.
Laura Heymann: connection between confusion and reputation: related to expansion of TM from source identifier to brand with personality. That expansion permits greater capacity for the sense that there are reputation-related injuries to the brand as entity/person, rather than reputation for particular goods. These harms sound more in defamation than in impersonation.
Marshall Leaffer: notion of the brand goes w/vast economic changes of 1960s; conglomerate enterprise led to the notion that one company is involved in any number of endeavors.
Bill McGeveran: what do we mean by “empirical”? Some of us are talking about what we’re measuring—to what extent is the thing we’re interested in a thing in the world we could go out and find, such as something anchored in consumer perception? Nonconsequentialist harms could be nonempirical in the sense that you wouldn’t need to measure the world (intent of defendant). Other issue: does the law require proof of it in the sense of surveyors or does the law engage in presumptions/assumptions to decide if the harm actually happened. You can have an empirical question and either measure it or decide what’s probably true (presumptions).
Morality/brand hijacking: could do the empirical/nonempirical analysis here—the problem might be consumers’ changed perceptions, or it could be the mere assault on the meaning regardless of whether it succeeds.
Dinwoodie: we’ve become too obsessed with empiricism because you can find people confused about anything. Need to ask whether we care about 10%, and about the type of confusion they’re suffering. If you were to limit the scope of rights in a confusion based system, we’d have to move away from empiricism.
McKenna: we’re nonempirical on the harm side. We should care less about empirics of likelihood, but the expansion came from indifference to the empirics of harm. The reputation question could be tied specifically to a market preemption idea. Market preemption could be about (1) occupying a market space through priority or (2) using the mark in a way that makes it more difficult for consumers to accept expansion into a third area. He’s most interested in (2); if you believe mark owner should have right to determine brand trajectory, (2) is tied to that.
Dinwoodie: empirics at macro and micro level. Injunctive nature of TM cases leads us to create a series of presumptions about how you assess the need for a particular form of relief.
Lemley: Theoretically in order to demonstrate plaintiffs’ losses, we ought to have to engage in specific harm and causation inquiries. It’s not impossible, but hard; if it’s so hard to demonstrate harm that we rarely get money damages based on plaintiff’s losses, that fact tells us something about whether harm is really following confusion.
Bone: if you care about consequences, then the question is when you infer the existence of particular facts. If we take Heymann’s point seriously about corporate personality, and think about it from a publicity rights perspective, you can think about trade secret as protecting corporate privacy. Is this utilitarian (protecting it for the benefits flowing from it) or moral? What kinds of moral rights do corporations have? There are judges out there who, without necessarily confessing it, think that defendants have taken the corporation’s personality—it’s not free riding but taking identity. And if we only talk about free riding we’re talking past this idea.
Dogan: exacerbated by blending of 43(a) and right of publicity claims in false endorsement cases—leads to blurring that may bleed over into the corporate context.
Litman: Elvis Presley is not meaningfully different from Procter & Gamble today. They’re both brands.
McGeveran: Haelan and other right of publicity cases are happening in the Second Circuit at the same time as all these Lanham Act cases. Interestingly, courts used to be very protective of personal names for defendants for these moral rights reasons, but trend lines have been different there. (Again, some “people” are more equal than others.)
Dinwoodie: causes of action interacting with each other. In Europe we have double identity, likely confusion, dilution, and unfair competition. Each cause of action should operate to create structure in the overall analytic system and a safety valve, but they can act as a mission creep reinforcement.
Bently: early days on initial interest confusion in Europe. The Google case; Ochs-Ziff case accepts the concept—lots of American law is coming through Dinwoodie & Janis to European law.
Lemley: there must be some limit on double identity, otherwise the Globe could be sued for mentioning the Daily Mail in an article—functionality?
Dinwoodie: yes, and Google France suggests that maybe confusion (generally and liberally defined) is the limit in such cases, though they’ve added a bunch of other factors such as advertising.
Bently: so will that broad and general idea of confusion in double identity then move into likely confusion cases? European cases are hard to predict in that way.
Dinwoodie: for all the flaws of the European system, it is highly structured and mechanical. Could be beneficial in preventing overspill from one cause of action into another.