Session 2: Confusion Analysis Considered Prescriptively
Introduction: Eric Goldman
Figuring out what’s in consumers’ minds is an enterprise that will never give a clean answer. Multifactor test is a weird mix of consumer- and producer-side factors, all put together. Entire test fails because it assumes one model of search. Network Automation case from 9th Circuit took the standard factors and said they were really about something else—slicing and dicing the test, showing how the factors don’t cohere very well with each other. Seeing therefore a proliferation of bypasses—we don’t need to use the multifactor test: initial interest confusion, post-sale confusion, some of the sponsorship confusion cases. And yet we come back to the likely confusion test. IIC is almost never cited any more and it’s almost never dispositive—empirically a red herring (similar conclusion to Barton Beebe’s w/r/t dilution).
If this group can’t agree on a better test, then there is no hope.
Primary Discussants: Bill McGeveran
Alternate history. What if we had listened to Schecter in the first place and created a dilution remedy early? Maybe liability more tied to source confusion, with greater attention to the consequences if any of such confusion. Secondary cause of action for reputation and the capacity of the mark to stand out—blurring and tarnishment, hopefully with more limitations. Might have avoided deformation of confusion rationale.
Where does the urge to address nonconsequentialist interests come from? Marketers’ desires, which are very strong in a branded culture/economy. There is an economy built on brand value, and law is unlikely to disregard that fact. Any changes in infringement standard or confusion metric will be manipulated/massaged to accommodate that concern. A hydraulic power of their value as assets, as indicated by congressional discussion of TM law which always emphasizes the value in the economy.
Confusion is the boundary limit of the TM right. Figuring out what the TM right is—property, reputation, etc.—is the key battle, and confusion becomes the weak jurisdiction test. So it’s not surprising that confusion stories can be weak in expansive findings and even in core cases, with a low level of confusion accepted as sufficient. To make confusion a higher bar, we could do a lot of things: materiality, search costs, and so on. The very dynamic that leads to current doctrine will lead courts to want to uphold the TM owner’s right for reasons without relation to the harm argument, which can be seen as denying the presence of nonconsequentialist reasons for TM law. Even if courts do try to balance more, we need to ask whether they do it especially well. Materiality may not do all we hope it can.
Courts might be better at protecting rights. In particular, might be better at protecting rights of purported infringers. Should focus on what’s being asked of courts: curtail the rights of third parties. Need a better vocabulary for talking about accused infringer rights, so that expansions of TM harm third parties in ways Congress didn’t intend to do.
Lemley: On the thought experiment: we have less of a problem with bloated false advertising law because we shunted many things into TM, so we might have had less of a problem with confusion if we’d had early dilution. But dilution has its own creep—no attorney wants to tell a client “you don’t have the special mark, you just have a mark.” While Schecter narrowly imagined dilution, there’s no reason to think we would have limited the number of marks covered any more than we stayed limited in ideas of confusion.
McGeveran: one possible difference: you wouldn’t have the dishonesty/incoherence of calling the thing you’re targeted “confusion.” If it had been “blurring” or “tarnishment” you’d at least be calling it something real—IIC isn’t confusion in any non-semantic way.
Lemley: yeah, but we’d be here talking about cabining the expansive definition of blurring.
McKenna: it might have been clearer that this is not about consumers, but the interest of mark owners, which might have put people on guard about the expansion of the right. [RT: though Posner would be saying it was consumer-protective anyway.] If the issue is that courts are protecting value, then fighting over the conceptual basis kind of misses the point.
Dogan: confusion test defines the discovery/record litigators want to build, and helps to shape court’s analysis of facts presented. Theoretically we consider harm, but the places where harm ought to be considered have been watered down too much. So we should think about shaping the test in a way more demanding of plaintiffs to articulate what the harm is either in a particular case or more broadly. Materiality is one way; let’s think of others.
Beebe: mission creep happens a lot—clearly nonfamous marks are routinely found famous. Courts are just offended by misappropriation/effect on personality or reputation of brand. TDRA: some courts have said that even if you show association, you still have to show effect on distinctiveness. (Though I think all these were reversed/dissents—V’s Secret dissent, Charbucks district court; Beebe says Nikepal required a showing of effect on distinctiveness, but it didn’t seem to matter.) Shift of goalposts: show me the harm of association. Maybe this is a way to pull away from hydraulic pressure.
Litman: can we persuade courts that there unauthorized uses that don’t threaten value? As a way of persuading them that we should pay attention to harm caused by use instead of finding harm from use.
Dinwoodie: articulating benefits of third party use is the key to the next step.
McGeveran: courts sometimes see the public as sitting only on one side of the confusion analysis. If we saw it as more of an investment by brand holders, courts might be able to see the public on both sides betters.
Heymann: This is tied to TM as source indicator v. TM as marketing concept (TM as a kind of work of authorship). Borrow from copyright fair use: audience’s interest weighs against protecting the value of the work, after we have first disaggregated the two different conceptions of the TM.
McKenna: virtually everything collapses back into confusion; look at the caveat at the end of every defense—as long as it doesn’t cause confusion! That’s the preeminent policy interest. If you were to look at interests, courts look at the equities as between the two parties, but we might want them to do it more categorically—hard to focus on societal interest where defendant looks to be taking value that belongs to plaintiff. Balancing favors plaintiff by making amorphous interest in society in general weigh against apparently specific harm to plaintiff.
Bone: a brand is partly a source identifier in the TM sense, partly a personality, and partly it’s a product. When I buy a brand, I get that brand of the product—mark becomes vehicle for shaping product through advertising, so consumer gets consumption value from choosing that particular brand. Those are three different functions. If it’s part of the product, way of making product dynamic over time, we are (or should be) out of TM’s domain. TM should be focused on source identification, the first of those functions.
Dinwoodie: different set of lenses for expression may make value of defendant’s use more apparent—by dumping everything in the same pot we might lose that; wants to argue for value of bypasses for specific types of defendant uses.
Litman: what is it that we think consumers ought to be confused about? What, in the best of all possible worlds, would we want to measure? McKenna thinks it’s about buying one product instead of another.
McKenna: confusion about actual source or responsibility for quality (fair to presume harm from that), or plaintiff must show that the confusion is material.
Dinwoodie: interesting that McKenna and Lemley allow the possibility that other kinds of confusion can be actionable on specific proof of materiality; why not just exclude it?
Lemley: he’s willing to allow IIC if you can show an effect on purchasing decision at the end of the day that flows from being deceived, but he doesn’t think it could be shown 99% of the time. Messy bits: what do you do about franchising, outsourcing, etc.? The fact that I bought from a franchisee should be important because there’s a representation that the franchisor stands behind the quality of the goods.
Litman: can you rule aftermarket confusion out entirely? I absolutely might decide to buy a guitar from how it looks across a smoky bar.
McKenna: two questions—what quantum of consumers should be required for that. We think it’s very unlikely harm will follow, so you should have to show materiality.
Dinwoodie: suggests you’re still reluctant to impose prescriptive judgments about what should matter to consumers.
Lemley: there’s a plausible story that consumers shouldn’t care about branded v. generic, but that’s not the role TM should play. Most of the time purchase decisions are not going to be affected by sponsorship, but he won’t take it out of the game preemptively.
McKenna: he thinks the law shouldn’t try to regulate sponsorship either by preventing it or by giving TM owners a right to prevent others from suggesting sponsorship when that’s not relevant to purchase decisions. [What about regulating disavowal of sponsorship? E.g., atroturfing.]
Bone: without a good idea of what responsibility for quality means, a consumer might not think A is involved in making B’s product, but they could believe that A is careful about choosing A’s friends and that A isn’t going to choose a bad one.
McKenna: if this is true, then that might justify a finding of infringement. Still should rule out “they had to license this because use requires licensing.”
Goldman: shifting presumptions depending on the kind of use—maybe that’s something we could learn from Europe which formally channels types of uses into different types of analysis.
My reaction: this discussion of whether we want prescriptive rules about what consumers should care about gets to the Wal-Mart v. Samara issue: whether the game is worth the candle, if it’s so unlikely that harm will actually result. Maybe we shouldn’t allow litigants to try to show that they’re the 1 in 100 who are really affected, given the costs of litigation and the risks of error. This is a big part of my forthcoming article in Penn. You don’t need to have a prescriptive account of what consumers should care about to make this practical judgment, though it ends up bearing a strong relation to a prescriptive account, as Scalia’s opinion in Wal-Mart about what consumers ordinarily rely on (not trade dress) does. Also note that puffery is a similar half-prescriptive, half-descriptive account of what claims consumers (should) rely on. We are completely comfortable with that, not least because it suits the goals of producers to be able to make a lot of claims with no accountability.
Lemley: Wal-Mart allows you to prove that you’re the exceptional product configuration deserving of protection if you show secondary meaning. And (in response to questioning) he thinks that you should probably get the chance to prove that “best” means “best” to your consumers—we’ve taken puffery too far.
McGeveran: back to rules and standards.
Lemley: two questions are intertwined: is puffery per se unactionable/what is puffery?
McGeveran: to have a bypass lane for easy cases, you need both a 1 in 100 determination as in Wal-Mart and also an easy way to figure out what gets into that lane. (And I note that Wal-Mart provides a decision rule: if uncertain whether something is product configuration trade dress or packaging, secondary meaning is required.) And the puffery case law may help you sort given precedents over time.
Lemley: antitrust has a history of developing per se rules for common cases, though we’ve gotten rid of many of them over time. Then we create a bunch of rules within the rule of reason to cabin things.
Heymann: if you always get to prove eligibility, then we face problems like surveys showing that consumers don’t perceive a parody; we might not like that result since we might want a blanket rule protecting parodies. What does source mean? Quality control w/r/t a particular product. Pepsi today will be a same as a Pepsi bought tomorrow. Related brand extensions: with Diet Pepsi we might expect certain characteristics to carry over. Licensing: choosing your friends—but do you expect sound judgment to mean anything w/r/t Pepsi choosing a particular T-shirt company to produce its T-shirts? Why would consumers ever believe that Pepsi was good at that?
Beebe: thinks that Mutual of Omaha v. Novak would come out differently today. Courts would look at the relation of insurer to t-shirt maker, the rule that strength can make differentiation easier.
Litman: note that the Michelob Oily case didn’t reject Novak.
McGeveran: but dramatic improvements in other circuits.
Leaffer: need filters—antitrust has something to teach us—for judicial efficiency. A well-designed puffery rule that generalized claims of superiority aren’t actionable is a good idea. Linedrawing arises, but efficiency demands a filter for irrelevant deception/confusion.
Bone: likes puffery. It’s a normative thing—a level of seriousness. Leaves some consumers perhaps harmed, but the law can’t help everybody, and it’s expensive to try. We might want to adjust it but it’s a good idea. It’s all about false positives and false negatives in the end. When the SCt forces you to prove secondary meaning, sometimes you won’t be able to prove it even though your mark is in fact distinctive. We are generating false negatives in order to protect against false positives, and that’s the same thing we do with puffery. We have to face up to that tradeoff. Post-Wal-Mart, some courts have made it pretty hard to prove secondary meaning in such a case. “Just make you prove it” is not necessarily that easy, given the administrative costs.
Mid-point summary: Mark Lemley
Nobody likes to talk about confusion. Sometimes judges adopt dilution results, sometimes they expand the concept of confusion to include any relationship, sometimes they use pseudo-confusion: efforts to gin up a plausible confusion story to cover the nonconfusion based result (IIC, counterfeiting and post-sale confusion). Jeremy Sheff has an interesting new draft talking about the difficulties with post-sale confusion even in the Veblen goods context. Outside the US we just say same mark, same goods; US uses the fig leaf of confusion. We’re doing something other than protecting against confusion.
Why? Free riding. Unfair competition is an ultimately empty concept—whether it’s unfair is precisely the conclusion that we are trying to understand. We can’t and don’t want to prohibit free riding, except when we do. Why is this particular form of free riding bad? The theories underlying this are effectively copyright stories: a derivative works right to control the mark in an unrelated field: it’s my mark and protecting the right will give me some financial benefit. Without the underlying justification copyright offers, and without the limitations built in by copyright.
Nonconsequentialist possibilities, per Bob Bone. Lying is immoral. Right of publicity broadened from privacy by drawing on TM; now TM is returning the favor by claiming brands have the right to control persona. But why give such a right? Right of publicity lacks good answer, and claim for corporate personality is further removed from constitutive personhood justifications that people like Peggy Radin might give.
Dinwoodie: what would the anticounterfeiting justification be?
Lemley: when consumers might end up with a functionally different product that hurts them. Harder case: communicative counterfeiting, where the only thing imitated is the communication of the existence of the brand itself. At the end of the day the instinct is not a brand instinct at all but some copyright/design patent concept. When fashion people talk about counterfeiting, they talk about batteries and drugs, because that’s a harm story it’s easy to endorse. Veblen harm is harm society might not care much about.
Bone: we don’t always ban intentional lying, but that doesn’t mean the moral principle lacks weight. There are pragmatic reasons we might keep the norm out of law. Sometimes courts may worry about the landscape becoming confusing rather than the consumers becoming confused. People using the same mark on different goods or in different ways—world looks more complicated; natural reaction might be “there’s got to be costs from this!” without necessarily asking “who’s being confused?” Unless court sees good reason to tolerate this confusion, it won’t. First Amendment concerns etc. put some pressure on this, but slowly.
Bently: a registration-based system has similar concerns—first UK registration act didn’t want new word marks (only old word marks with secondary meaning) because the idea that image marks would be clearer. Modern registration system retains something of that with a broader likely confusion test for registration—if you’re thinking about putting something new on the market, why not make them stay further away than perhaps necessary?
Janis: Ferrari case, called post-sale confusion and holding that post-sale confusion extended past potential consumers to the general public—sounds like Bone’s idea.
McKenna: baseline question: do you go in looking for a reason to allow the defendant to do what it’s doing or to prevent it. Functionality: competitive need operates from the presumption that the defendant shouldn’t get to act, which is a very different baseline than a general free market. If we really want order in the marketplace, then we should do away with product categories altogether. One American and no more. Also, order in the marketplace is inconsistent with expanding the types of things we allow to be TMs—8 to 10 layers of protection on some goods. That’s a lot more clutter. On lying: can’t think of a single freestanding regulation of lying. Even fraud requires materiality. It would be weird for this to be the place to decide that lying generally, independent of consequence, is harmful, when the claim most directly about lying is fraud.
RT: there’s a recent set of 9th Circuit concurrences and dissents to the vote to reject en banc review about whether the government can ban lying in itself. The panel invalidated the Stolen Valor Act, but there the government explicitly disavowed a connection to harm, and there will always be harm stories to tell in TM. And it’s a function of post-Lochner lawmaking that the legislature can make probabalistic determinations about likely harm and enforce them. Even in commercial speech contexts.
Litman: in a good person v. bad person litigation, bad person wins, and the worst bad person is a liar; the bad person who lies to judges is going to lose no matter what. Whether or not it’s TM’s business to say whether lying is bad, as a predictive matter judges are likely to slap perceived liars down. If we want to protect people from the distortion caused when the perceived liar loses, it’s not good enough to say that the government can’t prohibit naked lying because real judges will be influenced by perceived honesty.
Bone: materiality doesn’t mean there’s no nonconsequentialist anti-lying norm; there can be pragmatic constraints on enforcement. Materiality aids inferring intent.
Burrell: Courts never seem to inform themselves about how many traders are actually using the mark along with the claimant—TIDE is held to be unique when actually dozens are using it.
Lemley: there is something driving the cases, and maybe we have false consciousness about free riding. But there’s plenty of coexistence/clutter 75 years ago. The difference was marks were used for narrower, more geographically limited things, so interaction may have been less, and as McKenna points out we use multiple brands now on one thing. But we seem to operate today with an underlying assumption that names ought to be exclusive, which has to be ahistorical. Domain names, one and only one mcdonalds.com; that may be driving the idea of uniqueness as a requirement or at least the desirable standard to which we are aspiring. (Maybe we’re reasoning from government IDs; we expect that everyone/every company can be uniquely tracked.)
Beebe: When it comes to goods that convey relative value, Veblen goods, judges do think that everyone has a place and every person should know it, implicit and sometimes all-but-explicit enforcement of social hierarchy. Second nature to judges, there’s something unseemly about this kind of conduct. Noise as an aesthetic issue—a problem of modernity, too many stimuli. Judges might be saying: Self-blurring is one thing, but the conduct of all these riff-raff is too far. Uniqueness of names: it’s quite striking that historically there wasn’t much effort to be unique—New London, New York. It was not until mass production arrived and people are consuming the same things that we start reaching out for intangible and incredibly unpersuasive forms of uniqueness. There are millions of millions of copies of Coca-Cola but somehow we maintain that Coca-Cola is “unique.”
McGeveran: we don’t have to pick just one explanation. Ralph Brown wrote at a time when the brand’s power was not yet dominant and it was possible to be skeptical of the value of the trademark. Judges have now deeply internalized how companies think of the brand as the key asset detached from what the companies actually do. Value doesn’t tell you to whom the value should be assigned normatively, but that’s what’s going on.
Janis: rhetoric of noise/clutter is explicitly articulated by commentators in favor of expansive dilution by blurring cause of action.
Heymann: name as creative output for kids; parallels a sense of property right in the name. Also a greater opportunity to create names than before, rather than choosing a geographic name as would be historically common. Same with personal names; we are used to choosing a screen name and creating ourselves. [Greater social anxiety may be driving some of this—with so much economic insecurity, any advantage whether justified or not needs to be run with, and if you think a name promotes success then you feel positionally disadvantaged if your sister gives her baby the same name.]
McKenna: not obvious that forcing people to use really different names reduces noise. Might be harder to figure out what’s what. We’re trying to explain why someone who looks like a copyist should get off—this is why abandonment has died, because copyists look like they’re trying to get off on a technicality. How politically salient is the idea of a problematic government intervention into the market because consumers can’t take care of themselves?
Dogan: a big problem is the absence of judicial sympathy for interests on the other side. When there’s an articulable interest like parody or nominative fair use, then courts listen. They don’t always get it right, but they can hear the argument. Sponsorship/affiliation, courts don’t see an interest in access to the mark and are willing to act against any misinformation that might occur. [My reaction: This is where the false advertising analogy might be able to help: general commercial freedom to promote one’s own goods and services as the countervailing interest.] How can we change the cases where courts are really getting it wrong?
Leaffer: how many people in this room accept the notion of post-sale confusion as legitimate? A few; why? He believes in a TM property right. Ralph Brown was anti-trademark because he was monopolophobic, a view that is unpopular now because of the Chicago school idea that product discrimination is procompetitive. Post-sale confusion as in Mastercrafters undermines the strong image of the mark, the brand identity.
Grynberg: We are speaking as if, in practice, judges don’t see the burden of proof as being on the TM holder—think there needs to be a good reason to withhold relief. Why do judges think that they have the openness to evolve the law? Do the judges even think they’re expanding the law? Judicial proclamations that they’re doing just that are rare, though you see it in Boston Hockey.
Litman: She doesn’t believe that post-sale confusion is sensible sometimes because of a property right as Leaffer does, but because product distribution chains are really complicated and products may be seeded into the hands of people who will use it. So the moment of sale isn’t always as clear as it can be, and in those unusual cases confusion that happens after the point of sale can affect future consumers’ decision to purchase or knowledge of what it is they’re purchasing.
McKenna: when companies give handbags to celebrities?
Litman: yes, and when someone takes a fake LV bag to school and the catch breaks. People engage in all sorts of complicated marketing schemes and confusion may result outside the chain of sale. People think it’s X and it’s Y and they decide they won’t (or will) buy X.
Dogan: sports teams are given cleats to wear; if you see IU’s team wearing Adidas, maybe you want to buy Adidas, but there’s a slippery slope.
McGeveran: what about the Veblen goods exclusivity issue, when now any prole can get one?
Litman: not a fan of that argument, but as it gets harder to distinguish marketing from placement, you can’t assume the sales transaction is the only one that matters.
McGeveran: once you’ve conceded that the broken catch could be actionable, you’re allowing the claim every time because it’s all hypothetical—it’s the risk of the catch breaking that allows courts to impose liability.
Litman: she disagrees. [I’m with McGeveran.]
Burrell: but people who see the broken catch will think, oh, that’s a fake. Even people who see a genuine LV break may well think it’s a counterfeit because LV is high quality.
Litman: also thinks that a bag produced on the same production line without LV’s authorization should not be actionable, because it has the same quality.
Lemley: consistent with his proposal: confusion doesn’t demonstrate harm, but you could show harm in some other way.
McGeveran: okay, but how? Show that the bag is low quality?
Lemley: no, more evidence on purchasing decisions would be required.
McGeveran: so, survey people about how they’d think about a bag with a broken catch?
Lemley: maybe a showing of what people who had this happen thought. Sounds implausible, but to him that’s evidence that the problem is theoretical and unlikely to move the market.
McKenna: empirical work on effects of counterfeits—counterfeits of high-end goods are net beneficial because of the advertising effect, but lower-end they’re more problematic.
Dogan: materiality rule would discount things that happen in the market that might change a consumer’s perception but are very difficult to trace to a termination in a particular consumer decision. An information-focused rule might be able to think of an articulable harm that matters even if it falls short of materiality.
McGeveran: without administrative costs, the inquiry would look very different. He pressed before on the actual harm requirement because regardless of the abstract normative intuitions, it needs to be reduced to doctrine and will then require some assumptions pointing in one direction or another. Demanding proof might be the same as not allowing relief.
Janis: materiality will be hard to prove [I think this is not a rule for all cases!] and judges will want to find it in some cases.
Heymann: it’s also important to give some content to “material.” Is there a temporal requirement or some other proximity requirement? Will it cause them to buy X when they meant to buy Y? What about sponsorship: I think the TM owner has sponsored this, which may change my view and affect my decision later because it changed how I think about the TM owner? Post-sale confusion? Do we think that I have to be within the group of consumers who’s likely to be making a decision in the short or medium term? Inevitable normative aspects to it.
RT: doctrine has a lot of these answers already.
Dinwoodie: still a normative premise that materiality in false advertising is the same we’re worried about in materiality in TM.
RT: true that some of the concerns of post-sale confusion don’t fit in false advertising materiality terms, but I think that those are not proper confusion concerns. They are actionable, if at all, only as dilution. (And of course I think dilution is an unconstitutional regulation of truthful commercial speech, but you don’t have to agree.) (Jeremy Sheff's paper, mentioned above, sorts types of post-sale confusion, some of which could meet the materiality standard if proven likely and others of which couldn’t but might be dilution, if we knew what dilution was.)