Wednesday, July 29, 2020

The McCarthy Series: U.S.P.T.O. v. Booking.com: What the Recent SCOTUS Ruling Means for Trademark Law

My notes are messed up because I had technical difficulties! But the McCarthy Institute will apparently publish video on its YouTube channel.

David Bernstein, Debevoise & Plimpton: intro to case/overview. 

Hal Poret: Used a Teflon survey to educate consumers about difference between brand & common name, including .com terms to show they could be either (staples.com = brand name; officesupplies.com = generic term). Also used washingmachine.com to account for bias towards considering .coms to be brand names.  Pattern of responses for booking.com and wm.com: clear that booking.com & wm.com produced essentially opposite results—heavily in favor of brand, and heavily in favor of common name respectively. Booking.com’s results not just from bias in favor of considering a .com to be a brand. Dissent/PTO misunderstands significance of control: substantial noise doesn’t reflect a problem w/a survey but w/a real world bias that the survey needed to grapple with.  Even if you remove the people who misclassified wm.com, those who didn’t thought booking.com classified it as a brand by a 2:1 margin. 

So this is perhaps reflecting [de facto] secondary meaning and not lack of genericness. What are possibilities for accounting for this? Could do additional education in the survey training. Could include a .com in the mini-test respondents must pass. Could make the entire survey about .com terms, so all the educational section and examples would be about .com. Another possibility: change answer choices to instead of asking about brand/common, could focus on distinguishing b/t descriptiveness and genericness, which can be done. Could also look for secondary meaning to look for respondents who are responding to previous knowledge of the website versus those who weren’t previously familiar of the term—to weed out influence of actual secondary meaning. Could also ask those who were aware of the site questions to distinguish between whether they thought it was descriptive or generic: is the primary meaning a description of the products/services or is the primary meaning a type of product/service. 

RT: Brief note: the Poret report relied on the 74.8% total “brand name” response for booking.com as the relevant number rather than treating the net 44.8% as the key when the 30% who responded “brand name” for wm.com are taken into account as a control. This matters if (but only if) one holds the opinion that a threshold over 50% in the relevant consumer group is the correct threshold for showing that the primary meaning of a term is as a trademark. But realistically, measuring brand awareness is going to be hard in a Teflon survey, which is primarily directed at a different question. 

Given that the majority explicitly says that it’s not reaching the survey criticisms because of the PTO’s concessions, I think Mr. Poret’s suggestions for future surveys are useful. A few specific comments: asking if people have heard of generic.com suggests looking for secondary meaning; the interpretive question will be what to make of people who say, as 1/3 seem to have done with wm.com, “I’ve never heard of this particular generic.com but forced to choose between generic/brand/not sure, I’ll go with brand.” If domain names are not inherently distinctive by virtue of being domain names, it may be necessary to offer a fourth option to make a Teflon survey work, something like “neither a brand name nor a generic word.” Really interesting to see if people could be trained on the Abercrombie spectrum!

Caveat: classifications are legal matters and surveying on them may be a bad idea. Very few trademark lawyers can predict the suggestive/distinctive line in a given dispute by looking at prior cases. If we can’t get useful answers from five minutes of training an ordinary consumer, we will have to look elsewhere. 

Another possibility: use Likert scales where people give answers and then rate their degree of certainty—using Likert scales can reveal a lot about where respondents are basically guessing. Some empirical work by Sprigman et al on this forthcoming.

The larger problem is the still uncertain issue of de facto secondary meaning: if the SCt opinion means that there is no remaining doctrine of de facto secondary meaning, then the survey as conducted in Booking is fine. If there is still such a doctrine, then the survey shows that it may not be possible to ask consumers questions they can reasonably be expected to answer. Respondents were (a) trained on the brand/generic distinction, (b) classified screening examples correctly in order to have their responses counted, and (c) uniformly got “supermarket” correct and 398 out of 400 got “sporting goods” correct—if 30% of those people, who clearly understood the task, still considered wm.com to be a brand name, when everyone agrees that it lacks trademark function, then this type of survey may simply not work for the relevant question. 

To put it another way, post-decision, a Teflon survey will likely always produce the result “not generic” for a generic.com. But “not generic” is not necessarily the same thing as “a trademark,” and especially given the PTO’s increased and welcome attention to trademark function, we may see much more attention to whether that has been shown. Bold applicants may even argue that the Court majority at least implicitly endorsed the district court’s reasoning that domain names are usually inherently distinctive b/c everyone understands that they are exclusive—but trademark function is a requirement even for matter that could in theory be inherently distinctive, so we might see more attention to that point.  

Bernstein: footnote 6 says other evidence might be better than surveys. Voice of the consumer evidence.

Peter Golder, Prof. Marketing, Dartmouth: Hearing the Voice of Consumer through Archival Evidence. Archival evidence can avoid certain biases. Consumer perceptions can change over time: Howard Johnson’s was once the most well known fast food chain, and quotes you think are about McDonald’s were about it. Here: look at consumer perceptions from tweets, referring to booking.com as one among several members of a class like Travelocity.  Can also look at prevalence of consumer serach. Can look at historical “name of class”—what did competitor call itself? Example: Portable electronic camera was one early name; took a while to converge on “digital camera.” Carefully designed studies can be scientifically valid and persuasive in litigation. 

RT: Relevant, but normative issues remain about interpreting things like usage versus comprehension and whether de facto secondary meaning is a thing. Test suite: determine what would this methodology have done for some of the key cases already decided, “hog” for motorcycles, shredded wheat, etc. I don’t think that if there’s a divergence b/t the results and the case law it’s fatal, because it might be that the courts were wrong, but I do think it’s important to see what kind of results you’d have gotten in such cases to see what the match is. More overarchingly: still have the normative issue: is there such a thing as de facto secondary meaning? If not—if de facto secondary meaning is de jure secondary meaning—then this methodology is perfectly good in itself to answer the relevant questions. 

Q: competition issues?

Issues still to come: if wm.com might be inherently distinctive, what about washingmachines.com? Is that confusingly similar and how will we know? Consistent with past Supreme Court decisions, questions about the scope of the mark have all been deferred to the infringement inquiry: Freecycle example of how this can allow overclaiming against people who are using only the generic portion. The majority says that this risk attends any descriptive mark, but not all descriptive terms are comprised of generic components: while one need not use “soft” to sell one’s mattresses, one generally must use mattress to do so. Open question: will courts use plausibility to put an early end to such cases? 

Related issue: prosecution history estoppel, well known in patent but unknown in TM until recently: Booking will likely increase both the PTO and the courts’ interest in the concept. Two important examples pre-Booking: First, Bottega Veneta’s application to register a specific width and angle of basket weave for shoes and bags, which the PTO granted but with an explicit statement that the registered rights did not extend past the width and angle of straps specifieds in the registration; second, the Lettuce Turnip the Beet case in which a court said that where the PTO found that a phrase failed to function as a mark when used to decorate a T shirt but did function as a mark on a tag, the court would not find liability for the defendant’s decorative use. 

If other registrants say something like what booking did—that it wasn’t going to assert its rights against variations—courts and the PTO may see a competitive need to enforce those concessions. 

Golder: use of generic.com hasn’t been key component of competitive advantage—market leaders don’t have the generic term in their names in key industries. [As Christine Farley pointed out in comments, that was against a different background legal regime. Given the installed base, the real question is what happens in new industries or new forms of advertising. It's not surprising that freecycle, a newly invented generic, is one place that problems have developed.]

Bernstein: booking.com is looking to prevent confusing uses. Cars.com: a brick & mortar dealer used a sign cars.com and was unaffiliated. Similarly w/UDRP rights. But there will be recognized limits on the breadth b/c the TM is still composed of highly descriptive if not generic components. Booking is a generic term for booking things and .com is a generic tld. So the ability to prevent other uses will depend on the totality of the facts: there’s a lot of mechanisms to prevent against overenforcement: likely confusion; descriptive fair use. 

Q: can this opinion can be cabined as holding only that a generic.com domain name isn’t necessarily generic? 

RT: yes; the opinion doesn’t preclude the idea that .coms are inherently distinctive because unique, but nor does it embrace it. Depends on the cases that people bring whether the doctrine develops in a way that protects users of generic terms from having to go through full trials/at least develop a full summary judgment record for. Descriptive fair use won’t be any help b/c using a term in the domain name will at least create a fact issue on “use as a mark.” 

Bernstein: if you have a registration, UDRP panelists will usually defer to it and find confusing similarity. If you don’t have a registration, you really need to show acquired distinctiveness. UDRP doesn’t have access to discovery tools—would probably have to be pretty overwhelming to persuade a panelist that the domain name is serving as a mark. Booking.com was already registered in many countries around the world; brings US into conformity; key takeaway goes beyond domain names, to hashtags, addresses, combinations of marks, etc. Issue is not whether the term is composed of terms that are generic but what is the perception of the term as a whole.

Q: booking.com had so much secondary meaning—isn’t it difficult for a survey to separate out that from what people think of the term in the abstract? Wouldn’t the survey have had to be done years before to separate that out? 

Poret: it’s a recurring problem that often shows up. You can’t go back in time and measure it back then. May be able to help people to distinguish between describing characteristics and identifying a class. 

Golder: archival analysis can be a complement to these methods. Can estimate the life of a brand even before it dies. Can find how terms were being used in archived materials. Federal Circuit in Converse case raised issue of historical secondary meaning being a challenge, but archive can serve as a time machine.


No comments: