Trademark Scholars Roundtable, Stanford part 2

Session 2: A Forward-Looking Perspective  

To what extent should trademark or unfair competition law reflect consumer expectations or seek to shape or set them?

Introduction: Rebecca Tushnet

One consideration in how we should structure the regime is that casual empiricism is a pervasive and possibly unsolvable problem because courts don’t notice when they’re making empirical claims: SCt cases (even in Tam, on registration as endorsement). Necessary for PTO given volume: TMEP assumptions about functioning as a mark. Heymann: it is indisputably true, as Graeme Dinwoodie has noted (What Linguistics Can Do for Trademark Law), that trademark law must of necessity take certain shortcuts to ensure that litigation does not devolve into a morass of evidentiary issues; true, too, that in a field (law) that depends on precedent for efficient private ordering, there must be certain general rules that can be derived and followed.

So at the outset we should just work on clarifying our understanding of to what extent empirical claims are truly our best guesses about a wide range of consumers and to what extent they are normative claims about how reasonable consumers should act. Cost benefit analysis & consideration of error costs have a role to play in our legal tests: what if we’re wrong about the empirics, how easily can consumers/competition recover from our error?  See this in Qualitex and a bit in Wal-Mart—if we’re wrong about distinctiveness, producers will usually have discrete symbols they can use instead to identify source.  Potential justification for “limping marks” concept as well.

One thing we don’t know yet in law is how consumers learn in response to broad market changes—while a flood of counterfeits seems likely to change consumers’ reliance on branding alone, what pathways are amenable to change and what will remain persistently confused? Much research suggests that demographic characteristics are far more important to consumers’ orientation towards belief than any particular thing they see in the market. [Possiblity of consumer intuitions about right and wrong/free riding, as with promotional goods, encouraging beliefs about the factual state of the world? Motivated cognition—though the research suggests that many more consumers think copying is unfair than think that the trademark owner definitely produced promotional goods.]

But: What messes us up is not what we don’t know, it’s the stuff that we know that just ain’t so: consider dilution and Sprigman et al.’s work about how it’s the surprise that causes people to hesitate—there’s no connection w/ a particular brand, so the response that we see when people confront supposedly diluting marks is not the brand-specific mechanism that we thought that it was.

Many claims made in the reading were (perhaps unsurprisingly) persuasive to me, but striking how little empirical basis there is for so many of them: For example, The Constructive Role of Confusion in Trademark, Alfred C. Yen, makes a great point: our theory of secondary meaning requires consumer learning over time: initial impression is supposedly descriptive, later learning is that the mark source-indicating. The consumer is not necessarily mistaken either time, but responding to market. If that’s what really happens, which we have some reason to doubt!  We all cite the Lee et al study on identifying terms as trademarks by their placement on the package, but what about replicability? That study tells us something that seems intuitively powerful, but how far does it extend? Can it tell us anything about images, fonts, slogans?  In the state of empirical uncertainty, do we have any choice but to use our own intuitions as the model of reasonable consumers’ reactions?

Here’s a hypothesis: Maybe our big doctrines seek to shape producer behavior and only indirectly affect consumer expectations, and maybe that is exactly where we should focus. Examples. (1) Genericity/functionality: we tell producers that they need to develop a signifier that is nonfunctional and not the only thing to call they product. (2) Abercrombie: producers should choose an inherently distinctive term for greater certainty. (3) Free speech limits: don’t bother to try to control certain kinds of speech/convince consumers that your authorization is required because we won’t enforce that.  That might be an approach we could be more confident would actually work than trying to shape consumer expectations because producers are somewhat more likely to change behavior in direct response to legal rules.

Consider the story of keyword advertising: judicial opinion started out w/skepticism and then rapidly came around to endorsement as its utility became more apparent. But how did consumers learn that keyword advertising didn’t always produce only searched-for trademarks? The studies show that they mostly expect that now, but did they always? Was the legal doctrine just the reactions of educated judges and their clerks to the increasingly apparent benefits? [Cf. Heymann on less-educated consumers and the difficulty of judges entering into their perspectives.]  It was producers who pushed back and kept using the tactic because it made so much economic sense, and they were not turned into pirates and counterfeiters; instead they largely won the day.

Consider also the example of private label goods imitating house brands. Several early cases found infringement by house brands that imitated national brands; Yen’s article says that opinions now “generally” state that consumers are not confused because they have learned from prior experience to distinguish private-label goods from their name-brand referents, and I think he reflects the consensus position on this. Which is notable, because, in fact, there are only two significant cases blessing the practice, and one of those cases found infringement by one of the house brands. Yen says consumers learned; Dinwoodie says the same: “consumers in the United States have clearly become accustomed to private label practices in supermarkets. Beside every branded product, there is a store brand lookalike, labeled with the house brand. As a result, following evolving social understandings about marks in such settings, courts have cut back on the ability of large brand owners to prevent the sale of competing store-branded look-alikes.” But was it truly “as a result”?

Why did sellers persist in this technique given the early losses (not to mention the fact that you can’t do it in many other countries)? Talked to May Department Stores’ lawyer.  In general: Copy testing was not part of the process for approving the labeling of a house brand.  Stores took the risk despite initial unfavorable precedents, similar to search engine advertising, presumably because it was likely to prove very profitable if successful, and also because they believed they could successfully distinguish the products from the national brands in consumers’ minds. (Also, some sources suggest, because they believed that most retailers would be too nervous about disrupting relationships to challenge the practice.) Courts eventually accepted that they were right—though notably without too much empirical evidence; note casual empiricism also on display in Conopco and Food Lion cases.

If stores successfully bet on courts’ acceptance of consumers’ ability to distinguish, suggests more consumer perception invariance than we might have assumed—at least w/r/t things consumers already care about. Compare to the TM forfeiture cases allowing universities to reclaim control despite hundreds of years of nonenforcement: consumers can stay stubborn in making associations (or not making them, in the case of house brands), regardless of whether we think they are confused about something relevant. 

Possible responses: (1) commit to real engagement w/ marketing literature; (2) commit to telling producers what we want them to do; (3) commit to normative vision of consumer glancingly informed by empirical intuitions; (4) overarchingly, talk explicitly about which of those paths is important and when. Normative vision may be particularly important for competition/free speech concerns.  But it’s also worth asking claimants like the CYOA plaintiff and the Honey Badger plaintiff to prove up their claims with real empirical evidence rather than relying on proxies, given that there is no history of clear consumer reliance on content of speech to make decisions about whether trademark owners sponsored that speech. I really don’t like the false/misleading distinction in advertising law but there is a point to requiring surveys sometimes as a way of managing courts’ epistemological hubris.

Jeanne Fromer: Relying on empirical evidence requires answering prior empirical questions: what we want to ask for as evidence of “confusion.” Which answers matter? TM as encouraging businesses to follow certain rules makes sense but we have to be clear about the normative goals so our empirics match up to them.

Two different roles: Empirical evidence can help us establish standards and rules and then there are the further empirics we would want in individual cases. Surveys, testimony, observing online behavior are all different sources of empirical evidence: searches originating on Amazon v. searches originating on Google.

How do we get there? Consumer construct. What kinds of confused consumer count? Hasn’t been empirically grounded at all.

Distinctiveness: how grounded in consumer perception do we want to be? Should we require secondary meaning before protecting anything at all? Any well resourced business will be able to generate secondary meaning, while new entrants will find it harder. That perspective’s not anti-empirical but it’s differently empirical: TM law should encourage entrants to choose specific types of marks; we could do a better job of encouraging choice of truly inherently distinctive marks.

Law might become less stable and more fact intensive if we did all this given that marketplaces are dynamic. Changing meaning of words.  Consumer behavior changes over time. Do platforms change the roles of TMs? NYT story on unpronounceable marks on Amazon. One story: everything else is taken; another: trademark semantic content is no longer as important. Fromer thinks one reason is to take advantage of lack of memory: the seller can easily disappear from Amazon and loses nothing by doing so/switching names. Empirics will make law less stable as we have to reevaluate what consumers are doing.

Discussant: Jake Linford: how to measure consumer expectations? One possibility: corpora. Harder to game than 3d party surveys? Could we measure fame by looking for a certain amount of use/a certain frequency? You’d have to hire real empiricists to do it right, not just ask the court to do it itself. Constitutional interpretation folks are trying to convince judges to run with it on their own. No strong commitment to whether this is a good idea or not. Online surveys are getting cheaper; ATurk surveys are at this point no worse than other surveys; survey as a costly screen may not stay true for a while.

Robert Burrell: Consider In ‘n Out Burger: they sell an Animal Style burger. Defendant in Australia started Down ‘n Out Burger w/ a red and yellow arrow pointing down. Marketing played off of In ‘n Out: “Also sell animal style for those of you who want a heart attack on a plate.” Pretty clear that there was no likely confusion; P scoured the internet for evidence of confusion and the most it could find was 7 posts, one ambiguous; another one said “these are pretty close to In ‘n Out” which was strong evidence of no confusion; in 2019 the most recent was 2016. Nevertheless the court found infringement. Why? Court rolled out the idea that confusion can arise when consumers assume a licensing arrangement. Second, court said no high number was required; a few was sufficient. Third, burgers are cheap and people pay little attention. Fourth, imperfect recollection means consumers may misremember the mark. Finally: we must remember that we’re dealing with a registered trademark. The fact that circumstances might mitigate against confusion didn’t matter. Seems that if we want to do better than that, there are a number of things we need to do.

First, the rules that have emerged about how you compare marks for similarity: although it’s supposedly factual, there are rules, e.g., pay attention to the start of words; mechanistic applications should be unpacked. Have to accept that w/in the context of registration, reining TM in will require embracing double identity, accepting the core property style protection where we don’t pretend it’s about confusion. Also, talking about confusion instead of talking about harm is a problem: In ‘n Out has no prospect of opening in Australia any time soon. Registered since 1990s, but only one pop up per year; sold 938 burgers in 7 years.  If we had a real nonuse doctrine, this would have been cancelled. Can’t talk about these issues in isolation. Lost context of consumer response: why are we keeping marks on the register?

McKenna: what default do we have in the absence of empirical evidence? Law is full of assumptions from circumstantial evidence/proxies. Whose burden is it to displace the things baked into the law? More generally, what counts as empirical evidence? Courts rely on proxies for all kinds of elements of a TM case: secondary meaning; likely confusion. Mechanisms about which parts of marks to focus on may or may not be empirical. So we could ask whether these proxies do approximate empirical judgments. The proxies for secondary meaning are really weak. Also possible that these are not proxies, but actually just decisionmaking guides not attempting to get at empirical reality, in part b/c of the instability that would be introduced into the law if it really did respond to consumer behavior. Could we do better at the wholesale level to make these proxies correspond to what they say they represent, or should we be explicit about our desire not to do that? Abercrombie: for arbitrary/suggestive, we could just say we’re not making assumptions about source designation but rather about competitive need.

Dinwoodie: Notable that there are few reported decisions about trade dress, but how do we prove what actually happened in the market? Not a slew of decisions but nonenforcement in the wake of Conopco. Enforcers, juries, private ordering mediate between consumers and their understanding. Maybe rules about the penalties for bringing/not bringing lawsuits may have to be rejiggered to give them the right incentives.

Sheff: Fromer’s point about well resourced firms being able to buy the proxies for secondary meaning: seems like this is by design. Producers are content to undertake the costs of search as long as they got the opportunity to persuade consumers. Has argued that producers take advantage of cognitive bias. The law is designed to reflect those biases, even if not rational. Trusts consumers to find what they want, in a nonrational way; give producers an incentive to set up a market for works for those consumers.

Grynberg: what does it mean that there are weak marks on Amazon? Possible offloading search to platforms. TM work is being done by “Amazon” but the marks there are weak, suggesting that something is going on w/TM value. [He says related to doctrine that allows Amazon to use TMs in search, but it may not be since these seems designed to surface particular sellers when there’s a search on generic terms.] Even if we were to declare the marks on Amazon not marks for failure to function, that wouldn’t change how Amazon works. It’s no longer a TM problem and now a consumer protection problem. Note that Amazon’s control takes some influence away from TM owners but concentrates it in ways that may be a separate problem [and that’s also true of complaints about house brands.]

Focusing on influencing sellers is not in tension with having a normative ideal of the consumer—that normative conception is one way the law has to tell sellers what to do.

Dogan: In ‘n Out is using confusion as a mask for fairness. If we’re thinking about empirical support for features of TM analysis, this is where we need empirical support for the countervailing values. Make salient the effects of not respecting the competing values. Studies by medical researchers showing that especially elderly people rely on color and shape to take medications: that is valuable in not just talking about the values but demonstrating the value to consumers of restricting TM rights in particular ways.

Burrell: Australia is willing to recognize that consumers will more readily perceive differences when marks are stronger, so it was important for In ‘n Out not to argue that it was a well known brand.

RT: just starting to work on house brands but interesting that I haven’t yet found empirical work on copycat house brands as such until after 1995. Before: brand names, other determinants of confusion. After: increasing interest in the marketing literature. Can’t (currently) reject the hypothesis that consumers were never confused; if that was true, there was not really a learning period.

Litman: even after 1995 there were cases that were filed but not litigated when the D changed the label.  Some casual empirical evidence that consumer perceptions changed; when she used to show lookalike trade dress to students, they used to say “how is that possible?” [though it may be relevant that unfairness is one reaction and confusion is another] and now they say “that’s a house brand.” This was the first year that they were unanimous that it wasn’t confusing. The trend over 30-40 years has been pretty steady.

Nonsense brands on Amazon are doing the same work as old style house brand in pre 1995 US—no one remembered the house brand/paid attention to it, it was just the brand that was in Safeway or Kroger or whatever. They weren’t exactly TMs that failed to function, but they did indicate “this is made by the company that makes stuff for this store.”

Lemley: market structure expectations is what we’re really talking about. Should TM seek to preserve existing markets? Should we allow new forms of competition enabled by actors like Amazon?  Suppose the claim is: by having a “store” page on Amazon, that creates the impression that it is authorized by the manufacturer. Consumers may not have a well formed set of opinions about a new process that hasn’t existed before, which might also be true of house brands. TM has to, in that case, serve as a norm entrepreneur one way or another as it reacts to these new forays. Doesn’t mean it can’t change in one direction or another, but the trend he’s seen is to impermissibility. 

Could mean that survey evidence is more useful when consumers have experience w/something than when they don’t. Keyword cases: consumers understood that this was a search result, but were deeply confused about whether it’s an ad or not. [They’re still confused about that!] He thinks consumers got better and better about IDing ads and Google responded by making it harder to tell the difference. Broader point: more reliable evidence when we have experience with a thing.

Silbey: picking winners in an uncertain situation: we can talk about what the default rule should be.

McKenna: many of us have told the story that consumers adapted to private label goods; another possibility is that consumers had no expectations one way or another and they would have adapted to whatever the rule turned out to be. Proxies: had we relied on ordinary proxies like similarity of marks, we might have decided to bar the practice as confusing; sometimes in new circumstances the proxies are especially bad and we should be more suspicious of how well they track.  [A really good example of this is the constant use of the Twitter, FB, Instagram etc. logos and fonts by various people. Traditional nominative fair use doctrine would tell us that the logos/fonts convert the uses from fair to confusing, but no one would be confused about whether Twitter endorsed your local hot dog stand.]

Mid-Point Discussants: Bob Bone: Which side should you err on if uncertain about empirics in a given case? Depends on whether you think there are good alternatives with same benefits for D; maybe on whether there is harm to P, which there often might not be.

Incentives are central to law. We design rules to affect ex ante incentives in contracts, torts, etc. Why not in TM? Maybe they’re not an effective target because of heterogeneity, changeability, etc. Is it easier to affect producers? Maybe they’re more homogenous, stable, but that’s questionable too, though relative to consumers that’s plausible. A moral framework would be a different matter. But given that goodwill is itself dynamic and changeable, how can/should a moral analysis take into account that goodwill might shrink in the future anyway for other reasons?

Jessica Litman: fertile ground to nudge law is found in marketing literature—we don’t need to do all the empirical research ourselves.  Also, focusing on the harm is another way into this question of what the wrong is, if there is any. Nature of incumbency: upstarts lack clout, almost by definition. They can’t get Congress or INTA on their side. Big successful upstarts can’t stand in for the new ones; we can’t either b/c we can’t necessarily imagine who they are. It’s a limit on our ability to advocate for flexible markets.

Secondary meaning: booking.com suggests difficulty with empirical evidence. It’s possible for generic terms to have secondary meaning, and if we’re committed to the policy justification for genericity then we shouldn’t rely on that.

Lisa Ramsey: Trademark WatchDawgs has been doing letters of protest. Sell T-shirts and similar goods; trying to get PTO to pay more attention to failure to function/distinctiveness. First Amendment arguments may also be available to shape TM laws. Need more work on validity stage, to get more marks rejected.  Amazon took down romance books with “Cocky” in the title on the basis of a claim based on a registration that was actually descriptive. More rigor is needed. Should be using common sense to refuse to register.

Sheff: the market is deliberately set up to encourage concentration—it saves consumers from having to think more by encouraging consolidation. Means that not every purchase presents us with the paradox of choice [though it’s not doing a great job on toothpaste]. Only when the alternative serves a significant interest making significant resources available to press back on the expansion of TM claims do we see resistance to rules that increase bigness.

Linford: we don’t ordinarily change liability now at T1 b/c we think the conduct will not be tortious at T2 (e.g. self driving cars may eventually be good, but are they doing harm now?).

Gangjee: reminder that providing detailed information to consumers is often just not possible: they can’t process too much. May need competition authorities to intervene in such cases.

Fromer: defends role of academics in helping little guy. Work w/Congress on TM registration—they tell us they’re interested in the little guy. Can be effective way to get Congressional ear. And sometimes small and large business interests align. Large businesses can be trying to enter the market again and again and again: Target was concerned with fraudulent TM filings; willing to bear the costs to fight it.

McKenna: Amazon’s TM procedures will displace a lot of TM law even if we get the cases perfect, given how easy it is to register. Sheff’s comment: not clear that the system decreases consumer info: we’re awash in signifiers, differentiation, multiple brands, same producer making lots of brands at multiple price points. We might be better off with classic competition.

Sheff: maybe represents some actual segmentation in the market.

McKenna: possible but we all know often there aren’t actual differences. Open Q whether this is any simpler than the alternative.

Silbey: opportunities to push back against Amazon program versus pushback that happened against Content ID? Models to be found?  Are we assuming a policy of equal opportunity/do we want to say that incumbency bias is a problem? Is it bad that having predictable winners and losers based on market share is a thing in TM? If we think so, what would the responses be? E.g., embracing reverse confusion, rejecting IIC, strengthening Tea Rose doctrine, value inherent distinctiveness over secondary meaning, something else?

Grynberg: you don’t need all that much TM law to get an incumbency bias. Ralph Brown was already complaining about incumbency bias when the scope was much smaller. Secondary meaning alone did that. Not necessarily tied to extensions of TM; exogenous forces did the work.

Goldman: TM law already ignores shaping of meaning by intermediaries, e.g. grocery stores who decide where to stock things even if they don’t have house brands. There are pronounced problems w/intermediaries but it’s not just Amazon.

Fromer: EDNY case, 2 weeks ago: unregistered TM; sold goods on Amazon. Chinese individual got a registration using a photo of one of P’s goods & went to Amazon and got a takedown. TM trolling/extortion to the next level. Eradicated P’s business. Sought injunction to get Amazon to reinstate them; injunction granted even though Amazon isn’t a party. Law is not willing to step back fully but motivated parties [not to mention court’s willingness to ignore the terms of the contract w/Amazon!] will have to involve it. TM bar has thought for too long that registrations impose no costs on others; clearer than ever this isn’t true. Amazon is a good place to look for these arguments.

Lemley: To say that TM is a small part of incumbency is not to say it’s not a part. TM law has a role in worsening the problem. The problem is worse now than it was in Brown’s time: every market is more concentrated now than through most of the 20th century: fewer and fewer winners across a wide range of industries. This was not inevitable and Europe is going in the opposite direction: US is now less competitive than Europe, reversing the historic pattern. Blame on antitrust, but also on IP and other tort law wielded against disruptive challengers to incumbents. Another potential step: intermediaries as allies, advocates pushing back against expanding IP. The reason we still have an internet, for as long as we still do, is that a couple of big tech companies lined up ten million people to fight SOPA/PIPA and stop fragmentation of internet. They are dangerous allies b/c their interests may not align; they can be coopted; they are easier targets for legislation. But if you’re a small print on demand company, one way to fight back against brand bullies is to enlist Amazon. The risk is that they cut a deal and back away, but they’re at least provisionally a complicating factor: a big guy with incentives notionally aligned with selling more varied stuff. Also grocery stores!

McGeveran: Ratchet effect of rights accretion may give courts pause: if we have reason to believe that consumers are guessing at what the rule is rather than reporting their own perceptions, that could be reason for courts to be consumer-shaping rather than reactive.

Heymann: could do that with surveys. Ask consumers “why do you think that?”

McGeveran: though there are reasons to question that. Heymann agrees: people don’t always know what their reasons are.

Silbey: TM law isn’t usually thought of as having an equal opportunity to compete component, but maybe it should.

Diamond: what exactly is common sense?

Lemley: poor man’s survey: me and my friends. Patent law flirted with ignoring common sense when you couldn’t find a written connection, and it was a disaster. We do give the jury/factfinder a lot of discretion when we use common sense as a standard, but there are different structural regimes and it does let you weed out lousy claims on both sides w/o requiring a survey.

Dinwoodie: French rulings are almost purely common sense; they don’t tell you much about their reasoning so they are hard to refute.