Session 2: Proxies for Distinctiveness; Proving Distinctiveness (and Secondary meaning); Strength of a Mark
Do current doctrinal tests properly assess actual consumer reaction to purported marks? What is and should be the role of consumer surveys in proving secondary meaning (or genericness)? What is the role of other disciplines (linguistics, psychology, marketing) in proving consumer perception in the context of trademark litigation (as opposed to policy formation)? Are the factors identified by current doctrine ideal/helpful? Are current procedural devices such as the presumption of secondary meaning based on time of use, or the concept of incontestability, useful in identifying marks that have reached particular thresholds? What is the relevance of strength of a mark for the scope of protection it receives? Is the concept of incontestability relevant to scope? How does one assess the scope for a composite mark that is distinctiveness overall but clearly comprised of non- distinctive components?
Introduction: Shari Diamond: has an affection for surveys. Two kinds implicated: secondary meaning surveys are among the worst in TM law. Reasons why: low rate that seems to be enough. Distinct from Microsoft scenario where the brand is so strong that people think everything software related is Microsoft related. Control groups can be important where you’re worried about pure guessing/yea-saying; it’s hard to imagine you won’t do better if you have some version of a control. To the extent you get “from one company” you can at least take out the bias of the particular methodology by controlling w, for example, “baking soda” as a control product. Jacoby would say this isn’t a causal Q so you don’t need a control; it’s not just a causal Q but a methodological flaw to not control for yea-saying and guessing.
Output measures: consumer sales, use in publications. Interested in output over input measures like marketing signals. Spending doesn’t mean they got it right. Weak measure at best.
Genericness surveys: Pretzel Crisp case. TTAB cancelled mark as generic; gave weight to dictionary definitions, use by public, including use in media and third parties in food industry, and use by D itself. There were opposing surveys. Fed. Cir. complained about whether the TTAB had treated the mark as a composite mark rather than two pieces put together. Surveys on both sides sometimes trigger “plague on both houses” reaction—contributed to bad reactions to surveys, like Posner’s. Surveyor’s “dark arts.” Here, Fed. Cir. said that TTAB should have paid attention to surveys.
Opposing surveys: Teflon survey: Princeton-Vanguard (registrant): 55% said it was a brand, 39% common name, 9% hadn’t heard/don’t know. Judge in Teflon was surprised how well respondents did distinguishing brand & common names.
Frito-Lay’s survey: 41% said it was a brand, 41% common, 18% not sure.
Look at the methodology. We haven’t resolved the best way to do these surveys. What would have been the right way?
First one: 18+ screened for purchase of salty snacks; tested on understanding of difference b/t common and brand name, and only 64% qualified. Control brands: Cheese Nips, 85% correct, Flavor Twists, 48% correct, Sun Chips, 96% correct. Control common names: Onion rings, 91% correct; Gourmet Popcorn, 72% correct, Macadamia Nut, 92% correct. Pretzel Crisps, 55% Brand name. (Flavor Twists was a related brand.)
Frito-Lay, similar group. Definition of brand/common given and asked whether they understood—2 said no and were excluded. Control: Ritz Bits, 82%, Lucky Charms 87%, iPod 61%, American Airlines 89%, Triscuit 80%. Common: Ginger ale, 72% correct, Automobile, 91%, Potato chips 90%, Newspaper 93%, Popcorn 93%. 41% said Pretzel Crisp was a brand, 41% said common name.
Is it good enough if you exclude 1/3 of the people being tested? Details of survey construction are really important, and to the extent that we don’t have agreement on best way to conduct them we won’t get really good evidence from the surveys.
Bently suggests broad range of common names might focus attention differently v. salty snacks.
Diamond: it’s possible—it’s an empirical question.
Ramsey: results are in the middle!
Diamond: usually in generic area, above 50% is good, but that’s a decision for the court.
McKenna: what’s the margin of error?
Diamond: has to be computed. [My quick lookup suggests for around 250 people, it’s around 2.7% at the .05 confidence level.]
Litman: do we know what affect not training people and just asking them if they knew the difference had?
Diamond: it depends. If you think people will tell you they know but just flip a coin, it would have one result; if there is a systemic bias you might get results. Might be difficult to exclude such a high number.
McKenna: but those people aren’t relying on marks when they buy, if they can’t tell the difference b/t car and Chevrolet.
Rebecca Tushnet: Maybe even outside inherent distinctiveness the concept of TM distinctiveness is not empirical. Tom Lee et al.’s work: putting a term in the TM place on a box makes consumers identify it as a mark, unless the term is generic. But even if you think for efficiency reasons (even setting aside competitive need reasons) we should still be distinguishing types of word marks w/o focusing on their placement, the category of suggestiveness is not empirical, and I now think suggestiveness should simply be eliminated/collapsed into descriptiveness and secondary meaning always required. At the same time, 5 years of exclusive use for a main word mark is probably a decent proxy for the non-empirical idea that someone should’ve shown up and made your use of that main word mark non-exclusive. Interestingly, PTO regularly rejects 5 years alone when it’s dealing w/trade dress, and I think that’s right. But what rules should we have other than word/trade dress?
My obsession w/the consumer who does not know enough to ask, to borrow from the Passover Seder: that is likely to be a big component, and thus particularly hard to test b/c answers so easily distorted when there’s no preexisting opinion. Franklin & Hyman: different answers depending on what stimulus—what they imagined, what they expected, what they saw. Perhaps this is another consideration favoring a less empirical approach.
Incontestability as proxy: increasingly makes sense to me. Admits there’s nothing empirical about it.
Strength and scope: always been a current in the First Amendment/parody cases that strength can decrease likely confusion. I always like to teach with Lollipops and Jellybeans for skate rinks: similarity confusing, but one reason is b/c the mark is conceptually strong for skate rinks but lacked marketplace strength. Affiliation confusion fights back against that insight—it admits that consumers will distinguish from the well-known brand but claims that consumers will interpret the difference in a particular way that means they’re confused. It’s a way to put a fundamentally nonempirical concept on the TM owner’s side.
What about strength of D’s mark? House mark arguments are often made.
Fromer: composite mark issues matter—Pretzel Crisps bag has a particular appearance; might come to mind for consumers even if they’re only being asked about the word mark. Goes back to the Lee et al. study. Not even clear what we’re pinning down in the questioning! Trademark claiming: 2D in terms of having a mark and a class of goods/services. Surveys ask about symbols only, but take advantage implicitly of product category—petitioner’s survey may have been using lots of salty snacks and therefore priming consumers to think about the many salty snacks out there.
Is there a way to use metastudies to help here? Courts have a duty to understand why they get different results, and metastudies have been useful in that regard to explain why different studies come out differently.
Role of empirical work in what TM law/policy should look like? Surveys are presenting case by case empirical work. Nonempirical line of Qs is empirical too in some sense b/c we have to decide when we want to make marks incontestable and what that will mean, and whether we want a category of arbitrary marks. May not be case by case but we have to figure out where and why to draw lines, some of which will be normative but not all.
McKenna: surveys directed at words—Gucci v. Guess case, and horrible survey accepted by court as evidence even w/net 6% confusion. General challenge w/non word marks is we don’t have good system for forcing the P to articulate its claim. So then it’s hard to figure out the right control—enormous amount of gaming.
In principle, 5 years makes sense, depending on 2 assumptions: (1) PTO’s evaluation of whether substantially exclusive use has occurred, given incentive of applicant, and (2) what happens if PTO is wrong—can parties fight?
Affiliation confusion/house marks. Affiliation confusion is a way of putting nonempirical concept on side of mark owners. See courts sometimes say that house marks help w/straightforward confusion, but affiliation confusion can make role of house mark irrelevant/complicates the role of other factors in the assessment.
Andrew Moshirnia: Empirical weaknesses of both surveys: On exclusion of respondents. In a robust study, exclusion should happen at the end, not the beginning, and then justify why you reduced the power of your survey—people didn’t want to create a record b/c they were scared about what those excluded people would have said. Complicates metastudies by sculpting population.
Priming/cuing: pretest effect. Normally if you apply a filter, you want the application to be disguised so you’re not cuing your subject to understand what you want them to do. If you’re going to apply a pretest, create a randomized order. [But if it’s something they don’t ordinarily think about, training them on other examples seems like the best thing to do.] If we’re demonstrating proper behavior we want them to exhibit, they want to please us. [But they won’t know what that is, will they, other than “classify what you think the answer is”?] Better: Pure control w/pretest, pure experimental w/pretest, pure experimental w/no pretest. That increases # of subjects by 50%, so it’s more expensive.
[he says yes, they need to know how to do the task, but he’d do the filter last; maybe change up what examples you train them on; maybe add a query on the amount of confidence, allowing you to work backwards to deal w/lack of knowledge. Is the training itself the qualification, so people who fail are screened out? Diamond says the problem w/follow ups is that it’s consider a toggle: either the mark is generic or brand name. Not sure what the confidence level would buy you. Litman says: would tell us how much people are just giving us an answer just b/c they want to give us an answer. Diamond says: controls help w/that too. Large literature: when you give people forced choices, they express positions but if you give them a don’t know option they’ll say don’t know even when they do have opinions. Mosnirnia: confidence can help you identify people who really did want to say don’t know but want to please you. You can’t ask them at the original question—you have to do it afterwards, but it’s a good way to deal w/a pretest effect]
Easier way of dealing with it: Strength of feeling of subject after they’ve made their decision—how confident are you in that assessment? The logic often is priming will have a larger effect on people who are unsure.
Diamond: never seen it done in litigation, lots in academic studies. Cost is the gorilla in the room.
Moshirnia: if both sides are deliberately shaping data, meta-analysis will be difficult. Need to be able to regroup people. But both parties do odd things to their data. No exclusion is bad—asking “do you understand?” leads people to shut up. Filtering first is also a big statistics no-no.
Ramsey: Fine w/quieting title as to first to use but not about future competitors—it costs money to challenge, and if you’re not currently a competitor there’s nothing to do. PTO is the gatekeeper but it’s not doing any substantive examination of incontestability, and that is a problem.
Linford: have same questions about incontestability—do we have the same sort of confidence about giving incontestability for 5 years of uncontested use after registration as we do about presuming distinctiveness from 5 years of exclusive use?
Still wondering what to do about consumers who haven’t thought concretely about source significance—even for them, marks may be communicating in ways they are recognizing, but not recognizing consciously. Less worried about those who can’t articulate source significance if there is a way to figure out that they are detecting it/attracted to it [how do you determine that they’re detecting source significance v. detecting some other feature they desire?].
Parody: strong mark = more effective parody. He worries about that b/c there’s something different than run of the mill likely confusion case. At least if you buy Souter’s idea that it says 2 things at once: I’m not the original/I’m the original. Dual communicative aspect won’t be the same for close follower/ordinary copier. Maybe a strong mark is hypersalient so consumers notice small differences, but he’s not convinced the parody cases teach that.
Beebe: Concept of TM strength v. affiliation confusion. We’d [Beebe & Hemphill] like to think of strength as increased consumer sophistication w/r/t the signifier but also w/r/t the products w/which the brand is traditionally associated. The increased strength may result in decreased likelihood of affiliation confusion b/c of increased consumer knowledge. Population of individual consumers aggregated: as strength increases, a few percentage points of consumers are no longer confused with each increase in strength, and maybe just enough that there’s no longer justification for intervention in the form of an injunction.
Bone: Not sure what it means to go from empirical to normative—if normative theory is dependent on some empirical results, then we still need empirical results. Genericity is an area where we might want to think more normatively. Canfield approach—chocolate fudge soda—when we don’t really want to survey b/c we don’t know what to ask. Ask: Is that mark/symbol really going to be very useful for designating a particular product class? We will make mistakes, but we can often tell w/o surveys.
Might think about percentage of population who uses the mark; might also think of vividness—are there many similar marks out there? Ability to call to mind quickly when I see something similar? Branding/the more meanings it has, the stronger it is?
Litman: incontestability: the statute separates the quiet title from the incontestability. Congress adopted it way back when as an incentive to register, which Congress thought we’d need here b/c of the history of success in litigating unregistered marks. My guess: it’s not doing that work, ever; when you want to sell stuff in countries other than this one you have ample reasons to register and incontestability isn’t on the radar. Do practicing TM lawyers really value incontestability? Her guess: may give them as much trouble as it does joy.
Mid-Point Discussants: Michael Grynberg: human tendency to simplify is present even if that means offloading decisions to experts, linguistics or otherwise. Disheartening to read about how badly incontestability is done now, b/c concept has appeal. But changes have problems of their own. Suppose descriptive/suggestive line doesn’t make any sense. But then we have lots of fights about suggestive/arbitrary line that we don’t have to worry about today. Fromer says Apple isn’t arbitrary for computers, Fromer’s reasons are plausible: so now what? Another possibility: get rigorous about secondary meaning.
Reasonable consumer exercising ordinary care: is a consumer really exercising ordinary care if he perceives possible affiliation b/t Jose Cuervo and Maker’s Mark based on use of a red wax seal? Might be a way to deal w/various meanings of distinctiveness by changing standard/considering materiality.
Litman: Likes Bone’s idea of a court appointed expert. Would improve quality/probativeness of surveys.
Omri Ben-Shahar and Lior Strahilevitz have a paper suggesting that courts adopt the survey mechanism to interpret contract language. Grass is always greener in the other field!
Our proxies for distinctiveness are stuck in the early 20th c. You gain distinctiveness by selling products to consumers for a certain amount of time; it doesn’t disappear quickly. But today we have instant secondary meaning, as w/Chrysler’s “Imported from Detroit,” but courts don’t have a way to deal w/that. We also have secondary meaning that dissipates in part b/c distinctiveness is happening not just through sales but through social media/products getting 15 minutes of fame. Plus, different TMs have different flavors not captured by linear strength. Affiliation confusion: Tiffany guitars wouldn’t confuse anyone; Starbucks coat hangers. But Orville Redenbacher snow shovels—she’d believe there had to be a relationship—it’s not that the term is inherently distinctive. It has a kind of secondary meaning that is different from Tiffany and Starbucks. That gives her concern not just that our categories are old and outdated, but that they’re causing us to miss something.
McGeveran: do these need to be empirical questions? Even if we think that the particular flavor we’re looking at is a Q of fact and not of law, is it a Q of fact that must and should be determined in a form of empiricism as broad as the survey? Are there more times when judges should feel they have the opportunity to make factual determinations w/o very expensive/high admin cost evidence? Geog. misdescriptiveness refusals aren’t adjudicated w/surveys. How do we account acceptance of that other than history?
McKenna: those are almost always about registration, which uses surveys less. Why are we so willing to avoid empiricism in registration?
Affiliation confusion becomes a black box for so much of this. Beebe says strong marks might decrease affiliation confusion. Super-strong marks: consumers more aware of differences in visual presentation. That doesn’t mean keen awareness of range of goods and services on which they’re used. Brand extension literature operates on assumption that the source is in fact the same, but at least it talks about similarities of goods and services in possibly useful ways. Need more empirical literature on strength and effect on consumers.
Leaffer: Conglomeratization of American life in 20th c. to today, affiliation confusion exploded. Consider the importance of subsidizing the TM office in requiring all these submissions.
Burrell: 1920s assumptions about how consumers behave—we need to test more generally how consumers behave before we can think about how consumers might be confused. We don’t even know what nonconfusion would look like.
Dinwoodie: role of excavation both history and normative factors embedded in distinctiveness analysis to free courts from fossilization that has occurred. The nuance in the doctrine doesn’t match how the courts talk about it.
Ramsey: Defending Abercrombie. So many judges aren’t grounded in TM. Lawyers think it’s “soft” and they can step in and litigate; RT’s paper shows that courts and litigants just screw up incontestability. Doctrinal tests force parties to talk about important issues and hopefully the judge will address them.
Dogan: when what the D does has social value, the courts have the tools to allow the use. We can push hard on this area where there are so many feedback loops: room for advocacy and scholarship on how courts should think about the relationship between risk of affiliation confusion as measured against social value that comes from informational uses of TMs.
McKenna: we’ve seen courts feel increasingly free to pick factors, but we haven’t had a methodology for picking factors. They may be clustering in particular ways. Also true of different types of confusion. I read cases that could have been presented as IIC but just weren’t; no rhyme or reason. Same w/sponsorship or affiliation, e.g. the Cracker Barrel case. The court says they aren’t confusing and doesn’t contemplate possibility of co-branding.
Beebe: most effective defense of Abercrombie is competitive need. Is it fair to say that the factors from Zatarain’s—use by competitors, dictionary definitions, general competitive need—compare to factors in the European or British approach?
Bently: court of justice drew a distinction b/t the reason why we have an exclusion and the test. The reason is so traders can use descriptive terms either to describe products/services or as parts of composite marks. They put a strong emphasis on public interest underpinning purpose of exception. But what then is the test? Whether the sign is being used or is being capable of being used in a descriptive way of some characteristic of product/service. Doublemint: doesn’t have to be core/essential characteristic; can be peripheral. That means descriptiveness has quite a capacity to exclude what US calls suggestive marks under Abercrombie. UK registry held that “sushi” for chocolate in the shape of sushi was not descriptive; inconsistent w/CJEU holding that it must merely be descriptive of a quality of the goods.
Fromer: Apple for bananas: what is it on the Abercrombie spectrum? Level of generality issue: are apples and bananas substitutable?
Thinking about corpora as a substitute for dictionaries, as Linford suggests: they are useful for large-scale data. But computational linguistics aren’t there yet for really understanding the corpus other than on a case by case basis for an individual litigation.
Bone: imagination test: I don’t even know what that is. I don’t know why dictionaries are important. Competitive need makes sense to him for reasons we might want to demand a showing of secondary meaning. Ultimate standard might be anticompetitiveness, then develop categories where that’s less costly than doing case by case analysis. That’s why suggestive should be thrown out—it’s not doing anything helpful in this regard.
Grynberg: if you get rid of suggestive, what is Penguin for refrigerators? [RT: I’d say don’t get rid of it, just make them show secondary meaning.] Consumers who have high need for cognition may like it a lot.
Bone: sure it’s suggestive, but what does that have to do w/TM? I can tell a story, but the story is about anticompetitive effects.
McKenna: automatic assumptions about source by consumers as a story?
Bone: no, a different starting story: why not require secondary meaning for everything?
McKenna: empirical presumption about what they’ll think.
Bone: just-so story.
Dogan: Might not be a problem to grant protection to suggestive marks if we’re really strict about what suggestive means. Breadth of protection is also important. Suggestive = plays linguistic function w/r/t feature of product. Courts that say that suggestive = inherently strong make for harmful consequences. We should presume that suggestive marks w/no commercial strength are weak.
Burrell: we never truly ask people to prove distinctiveness. We only ask some of them to prove use for a while + expenditure of enough money on the mark. Versus you get your monopoly straight away.
Dogan: to the extent people are concerned about automatic protection for suggestive marks, those concerned are softened if we make the threshold higher & allow protection only for marks unlikely to have significant adverse effects on other players [see also: scope]
Dinwoodie: why waste time on proof when we’re pretty sure—he still sees that as valuable. [I do too, but I’d put the line for pretty sure at arbitrary rather than at suggestive.]
McGeveran: doesn’t agree about Orville Redenbacher v. Starbucks—Williams Sonoma will sell shovels and popcorn. Once something is recognized as a brand, not just as a mark, it’s probably empirically true that many consumers might think that recognizably brand thing is slapped on is use as a source ID, or at least as affiliation. We could decide not to protect those things b/c we don’t care as much about affiliation/endorsement confusion as the current structure does, but as a descriptive matter he would expect a perfect survey to find that many people think the Orville Redenbacher shovel came from “the Orville Redenbacher people.” Why are we even asking about distinctiveness in this setting? It will always be distinctive once it passes the threshold of being a brand.
Litman: I wouldn’t assume Perdue cologne was co-branded by chicken co. But I spend a lot of time trying to weed out of my students the notion that TM distinctiveness is the same as other kinds of distinctiveness, but that’s what’s going on. The original Starbucks design, if I saw that on anything I’d assume that was affiliated w/Starbucks, even for toilet cleaner. That’s got something to do w/how distinctive it is not in the TM sense but rather in how unusual it is [what Beebe calls differential distinctiveness?].
McGeveran: yes, but don’t then use distinctiveness as “do you think this tells you something about where this came from.” Increasingly the empirical Q of “could this be a Julie Andrews snow shovel” will be answered, “sure, why not.” If we hinge distinctiveness on consumer perception we’ll have increasingly extensive brand extensions.
Dinwoodie: this is really an argument about scope/enforcement, really about dilution. What range of goods you can cover.
Litman: original Starbucks design is probably not famous; would play badly in the Midwest. But it is very unusual.
Dinwoodie: speaks to whether fame is proper requisite for dilution.
Litman: but I’m talking confusion—there’s no reason to put her on your product unless you’re trying to say “this comes from the same place as Starbucks coffee.” In the sense of being affiliated with.
Dinwoodie: maybe affiliation is a dilution cause of action.
Ramsey: But what about the Orville Rockenbacker shovel? Consumer would probably think it’s parody. Unrelated goods—you really need identity or virtual identity.
Linford: stronger anticompetitive = generic; weaker anticompetitive effect = descriptive. Competitive need is the first order question we’ve been assuming w/suggestive marks. We’ve been using imagination test as proxy for lack of competitive harm. Maybe courts should say expressly that they’re not trying to categorize but trying to think about what the competitive harms.
RT: Concerned about courts that call suggestive marks strong too, but interestingly I think that’s increasingly less common than it was 10 years ago (but would need to do more empirical work to be sure). Conceptually related to the idea of whether incontestability has any effect on strength?
Want to bring in descriptive fair use and courts’ confusion about the problem of descriptiveness in that context: Car Freshner—court of appeals dodges the real issue; if D’s use is descriptive, P’s use is sure as hell descriptive. Being more aggressive about melding suggestiveness into descriptiveness might have helpful consequences or troubling ones for descriptive FU. See also the Victoria’s Secret case about “delicious,” where the 9th Circuit says that use of the term to describe the wearer would not be descriptive use. Whaaaaat? Finally: consider treatment of retrievers/dog images by PTO early in internet era. Thought the idea of the retriever dog was highly descriptive in the context of internet search; probably isn’t any more; probably didn’t do much damage by requiring retriever image users to wait, but other lessons could be taken from this episode about the ability of the PTO to predict uses.]
McKenna: Always driven me crazy for courts to say there are 2 components to strength. If it’s empirical, it’s conceptually impossible that use by a different party for different goods and services can disrupt the strength of the connection b/t my mark and my goods. Penguin for refrigerators doesn’t affect Penguin for hot dogs. It’s not a nonsense question if you’re giving TM owners an incentive to pick something at the higher end of the spectrum.
All of us have gone to the really high end of the spectrum for “brands.” But there are also run-of-the-mill TMs where they are measuring how likely the mark is to cause consumers to think that use in another field is from the same party.