USPTO v. Booking.com – A Linguistic Justification for ‘Generic Trademarks’

Guest post by Jake Linford, Loula Fuller and Dan Myers Professor, Florida State University College of Law, whose trademark law scholarship I have highlighted on Jotwell

I want to thank Lisa Ouellette for inviting me to blog about United States Patent & Trademark Office v. Booking.com, a trademark case argued before the Supreme Court in May, with a decision expected soon. The Court selected that case for its first live telephonic oral argument. The night of the historic oral argument, Christine Farley of American University Washington College of Law hosted a discussion where I joined Rebecca Tushnet, Marty Schwimmer, and Cara Gagliano to recap the argument and discuss the case. Below, I summarize the oral argument in some detail (with page and line references to the transcript) and offer my prediction of what the forthcoming opinion might hold.

To recap, respondent Booking.com sought to register trademarks containing the term ‘Booking.com’ for online hotel reservation services. The PTO denied registration, concluding that ‘booking’ is generic for respondent’s reservation services, that the top-level domain ‘.com’ is generic for online services; and that the combination did not create a protectable mark. The Trademark Trial and Appeal Board affirmed the PTO. On review, the United States District Court for the Eastern District of Virginia held that Booking.com is not generic. The Fourth Circuit affirmed the District Court. Both courts credited as probative a survey demonstrating that nearly 75% of respondents identified Booking.com as a trademark. The PTO petitioned for a writ of certioriari, urging the Supreme Court to adopt a categorical rule that would prevent courts from considering evidence of source significance if a mark is deemed generic prior to its adoption.

The theory that trademark protection should be withheld from terms once deemed generic even if they have acquired distinctiveness is called the de facto secondary meaning doctrine or the trademark incapacity doctrine. This theory gives greater weight to dictionary definitions and the term’s etymology than to current consumer use, mistakenly treating the generic nature of the trademark as essentially unchangeable. I argue in an earlier article and in an amicus brief joined by trademark and internet law professors that the doctrine is misguided for reasons highlighted in the oral argument.

Like the recently decided Romag Fasteners, Inc. v. Fossil Group, Inc., this case exposes tension between an older, more restrictive judicial rule and the more recent and arguably less restrictive federal statute. The PTO’s support for its categorical rule is drawn primarily from a pre-Lanham Act Supreme Court opinion, Goodyear’s India Rubber Glove Manufacturing Co. v. Goodyear Rubber Co., 128 U.S. 598 (1888). The Court in Goodyear’s held that a mark combining a generic or descriptive term (‘Goodyear’s’ for vulcanized rubber products) and ‘Company’ or ‘Co.’ cannot qualify for trademark protection. Following that logic, the PTO argues courts and examiners are required to ignore evidence that consumers see a ‘generic.com’ mark as a source signifier, because ‘.com’ is the functional equivalent of ‘Co.’

Erica L. Ross, Assistant to the Solicitor General, made an outstanding argument for the PTO, repeatedly pointing the Court back to Goodyear’s as providing an easy disposition for these issues. Some of the Justices, however, seemed inclined to turn to the primary significance test Congress embedded in the cancellation provision of the Lanham Act, 15 U.S.C. § 1064(3), which states that “[t]he primary significance of the registered mark to the relevant public rather than purchaser motivation shall be the test for determining whether the registered mark has become the generic name of goods or services on or in connection with which it has been used.” Courts look for primary significance when considering whether a source signifying mark has become generic. In Kellogg Co. v. National Biscuit Co., 305 U.S. 111 (1938), another pre-Lanham Act decision, the Supreme Court held that secondary meaning required a showing that the “primary significance of the term in the minds of the consuming public is not the product but the producer.” Id. at 118. The search for primary significance typically includes survey evidence in ex post genericness cases. The Fourth and Federal Circuits have used the primary significance test to assess genericness in registration proceedings. As Justice Roberts observed, “it makes more sense to follow the language that Congress chose in the statute rather than a 130-year-old case of ours.” Tr. 6:6-8. Justices Alito (Tr. 18:18-20), Sotomayor (Tr. 22:3-11), Kagan (Tr. 29:2-11), and Gorsuch (Tr. 30:13-17) also pressed petitioner’s counsel on the government’s preferred alternative to the primary significance test. 

Goodyear’s is an artifact of its era. It does not distinguish between descriptive and generic terms, consistent with the practice of the time, when neither type of mark qualified for protection. It also builds on cases like Canal Co. v. Clark (1871) that are no longer good law. In Canal Co., the Court held that a firm could not acquire trademark rights in a geographic designation. Congress exercised its legislative prerogative and reached a different conclusion, extending registrability to geographic designations that have acquired distinctiveness. See 15 U.S.C. § 1052(e)(2), (f); Robert Brauneis & Roger E. Schechter, Geographic Trademarks and the Protection of Competitor Communication, 96 Trademark Rep. 782, 784 (2006). Justice Kavanaugh’s phrasing may hint at his opinion of the Goodyear’s case, asking counsel for Booking.com whether “Goodyear[’s] just has no value anymore?” Tr. at 70:10.

Pressed by Justice Gorsuch to provide a justification other than Goodyear's for its categorical rule, counsel for the PTO turned to the statutory language in § 1052(e)(2), (f) as “command[ing]… that generic terms generally are not susceptible to trademark registration based on secondary meaning.” Tr. 31:8-11. But the statutory text is not as clear as the government might prefer. Section 1052 does not clearly require rejecting an application or treating as unprotected a mark that was once generic for the goods in question. In fact, the term ‘generic’ is not used at all. The only bases, under 1052(f), for preventing the registration of a mark that has acquired distinctiveness, are geographic misdescriptiveness under 1052(e)(3), and functionality under 1052(e)(5), barring registration for “any matter that, as a whole, is functional.”

As we argued in our brief, the Court would err to conclude that functionality and genericness are equivalent to one another. Functional features are fixed aspects of the product that sellers must be able to use in order to enter the market with competing goods. But as linguists understand, language changes continually as it is used. This is true for trademarks as well as other words. Otherwise, descriptive terms could not acquire distinctiveness, distinctive terms could not become famous, and protected marks would never become generic.

While the government did not highlight this approach, some advocates for the de facto secondary meaning doctrine conclude that marks will so rarely narrow from a generic product designator to an indicator of source that error is best avoided by ignoring narrowing shift when it occurs. In fact, narrowing shift may well be the most common type of documented language change. I’ll share two examples here. Queen once meant “wife” or “woman,” but the meaning narrowed to “king’s wife” or “female sovereign.” Likewise, hound or hund once meant any kind of dog, but the meaning narrowed over time to long-eared hunting dogs.

The reporters are similarly replete with cases where consumer use narrows an ostensible generic term to a source signifier. In some cases, courts have properly extended protection to marks despite their generic origins. For example, kiss was once any sweetmeat or confection, but now brings to mind a particular brand of candy—the teardrop-shaped chocolates offered for sale by the Hershey Company. See, e.g., Hershey Co. v. Promotion in Motion, Inc., 2011 WL 5508481 (D.N.J. Nov. 7, 2011). Similarly, the Goodyear mark itself, once seen as generic or descriptive for Charles Goodyear’s patented method of vulcanizing rubber, has shed that meaning and narrowed to a source indicator. See, e.g., Goodyear Tire & Rubber Co. v. H. Rosenthal Co., 246 F. Supp. 724 (D. Minn. 1965).

One might discount this linguistic evidence that consumers vest ‘generic’ terms with trademark meaning by focusing on concerns about monopoly power and competitive harm. At oral argument, some Justices signaled that the monopoly question should not predominate. For example, in his conversation with counsel for respondent, Justice Gorsuch seemed disposed to conclude the harm in this case was limited because the breadth of generic.com marks would be narrow. Tr. at 65:12-25. Other Justices indicated that the competition question was a more pressing concern. Justice Breyer voiced the strongest apprehension that domain name owners allowed to register a generic.com mark would have an advantage over other advertisers. Tr. 52:19-53:4. Justices Alito (Tr. 56:12:16), Kagan (Tr. 61:19 - 62:10), and Sotomayor (Tr. 58:8-10) echoed Justice Breyer’s concerns.

The evidence collected by the parties indicates that ‘generic.com’ marks like Booking.com have been registered rarely, but not never. Although some commenters worry that mark owners might be able to force the narrowing of generic terms, history is full of failed attempts to inject new meaning into the public lexicon. Counsel for petitioner asserted that the primary significance test provided examiners proper tools to reject most generic marks, and to allow those few that have acquired source significance to be registered. Tr. 46:21-23; 60:2-6. Our brief likewise argues that potential competition harms can be managed using the primary significance test in ex ante genericness cases.

Moreover, ignoring the primary significance of a trademark also impairs competition. If a majority of consumers use Booking.com as a source signifier, then refusing registration imposes costs on those consumers. The etymology of the word in question should not dominate evidence of consumer use. Indeed, overreliance on dictionaries is problematic for many reasons, not merely because the practice fails to properly account for consumer perception.

Another way to understand the case is as a battle about the weight that should be given survey evidence of distinctiveness acquired in an ostensibly generic term. The survey in question was a standard Teflon survey, in which respondents are taught to distinguish between source signifying trademarks and generic terms, and then asked to categorize a series of marks including the mark in question. Nearly three quarters of respondents (74.8%) categorized Booking.com as a trademark. That’s typically a strong enough response to keep a mark from being cancelled or held generic. 

But the survey also raised a troubling implication. Just under one third (30%) of respondents also identified washingmachine.com as a trademark, and it’s not. (If you are curious, the URL redirects to wayfair.com.) That control question tripped up a lot of the respondents, even though they were trained on the difference between marks and generic terms. Justice Breyer suggested that result is evidence of “internet power” that could be multiplied by the power of a trademark registration. Tr. 52:14-53:6. If every .com mark gets a 30% head start because of the .com, then surveys might measure less source significance than we think they measure.

An excellent brief authored by Rebecca Tushnet argues that the proper way to net out source significance from underlying noise is to subtract the 30% of confused washingmachine.com respondents from the 74.8% who see Booking.com as a trademark. That would treat respondents confused about washingmachine.com as effective votes for genericness. Booking.com’s net distinctiveness under that formula would only be 41.8%, lower than the 50% threshold needed to show that the term in question has acquired trademark meaning.

I find an alternative approach discussed by counsel for respondent, Lisa Blatt of Williams & Connolly, more consistent with standard survey practice. Often, when survey respondents fail an internal control question in a Teflon survey, their responses are excluded. Here, if the hopelessly confused are removed from the survey count, we could consider how the remaining subset viewed the term in question. Of that subset, 64% categorized Booking.com as a source signifier. Tr. 53:14-54:1; Joint Appendix at 165-168.

Even if the Court were to adopt the higher threshold for .com marks proposed by the Tushnet brief, that doesn't require applying a categorical rule against survey evidence in ex ante genericness cases. Some marks might well net out over 50% of responses. Indeed, one of the other controls in this survey asked respondents to categorize eTrade.com, and nearly every respondent (96.8%) categorized it as a trademark. Subtracting out the washingmachine.com 30% would still leave 68.8% of respondents who categorized eTrade.com as a mark.

In my discussion of the case on the night of oral argument, I hazarded a guess that the majority of the Justices would vote to affirm the Fourth Circuit. I still think that’s likely, although I can’t concretely ground that impression in the oral argument, and it would be foolish to try. But that result would be consistent with the bulk of the Court’s work in trademark cases over the past 30 years, eschewing bright line rules that would deny protection to trademark owners, and instead favoring context sensitive inquiries. I look forward to the Court’s answer.