Harvard Law Review Harvard Law Review Harvard Law Review


Fanciful Failures: Keeping Nonsense Marks off the Trademark Register

A search of the United States Patent and Trademark Office’s (USPTO) trademark database reveals that in 2020, a single attorney named Elizabeth Yang filed over 8,000 proposed trademarks.1×1. This figure is based on a search of the Trademark Electronic Search System (TESS). See TESS, U.S. Pat. & Trademark Off., http://tmsearch.uspto.gov [https://perma.cc/U5EL-DSBK] (select Word and/or Design Mark Search (Free Form) and use the query “(Elizabeth Yang)[AT] and (2020????)[FD]”). Trademark attorney Erik Pelton has made public a striking visual representation of Yang’s many filings. See Erik Pelton (@tm4smallbiz), Twitter (May 16, 2020, 9:45 AM), https://twitter.com/tm4smallbiz/status/1261653986446368768 [https://perma.cc/3BH8-GZP5]. This number is surprising in and of itself, but the proposed marks are perhaps even more unusual. In general, companies developing new trademarks seek marks that are clear and memorable, that suggest something about the goods with which they’re associated.2×2. See Rebecca Tushnet, Registering Disagreement: Registration in Modern American Trademark Law, 130 Harv. L. Rev. 867, 922 n.242 (2017). Indeed, trademark lawyers sometimes clash with marketing experts, who want marks so descriptive of the goods they sell as to be unprotectable under trademark law.3×3. Jake Linford, The False Dichotomy Between Suggestive and Descriptive Trademarks, 76 Ohio St. L.J. 1367, 1419 (2015). But the marks filed by Yang include LJXOAIEU for hair clips, QIANDLEE for apparel, JANRSTIC for headphones, and AEZLHJYA for jewelry.4×4. U.S. Trademark Application Serial Nos. 90/183,371 (filed Sept. 15, 2020); 90/183,413 (filed Sept. 15, 2020); 90/180,782 (filed Sept. 15, 2020); and 90/180,940 (filed Sept. 15, 2020), respectively. These words are not close to any in English, and the applications make clear that they have no meaning in any other language either.

It’s possible that these submissions are fraudulent: that the companies involved have not used and do not intend to use the marks in commerce and are lying to the USPTO. Apparently driven in part by economic incentives from the Chinese government,5×5. Josh Gerben, Massive Wave of Fraudulent US Trademark Filings Likely Caused by Chinese Government Payments, Gerben, https://www.gerbenlaw.com/blog/chinese-business-subsidies-linked-to-fraudulent-trademark-filings [https://perma.cc/7J33-8DZQ]. fraudulent trademark filings have been a growing problem in the United States and are a subject of concern among academics, politicians, and practitioners.6×6. See Barton Beebe & Jeanne C. Fromer, Fake Trademark Specimens: An Empirical Analysis, 120 Colum. L. Rev. F. 217, 218–20 (2020); Gerben, supra note 5. Fraudulent filings can impose significant burdens on the U.S. trademark system.7×7. See Beebe & Fromer, supra note 6, at 241–44. However, if marks like LJXOAIEU and AEZLHJYA are indeed fraudulent, existing trademark law, which requires that marks be used in commerce to receive protection, provides a clear basis to deny their registration.8×8. See 15 U.S.C. § 1051(a)(3)(C) (requiring use of the mark in commerce); id. § 1051(b) (providing an exception to the use-in-commerce requirement when an applicant shows bona fide intent to use).

But there’s another possibility for these proposed marks, based not on the incentives offered by the Chinese government, but on those offered by Amazon. In 2015, Amazon created the Brand Registry, a program that gives brand owners control over listings that contain their products, protection from unauthorized sellers, and “more access to advertising solutions, which can help [owners] increase [their] brand presence on Amazon.”9×9. John Herrman, All Your Favorite Brands, From BSTOEM to ZGGCD: How Amazon Is Causing Us to Drown in Trademarks., N.Y. Times (Feb. 11, 2020), https://nyti.ms/38uhKvi [https://perma.cc/QFM7-R3M2] (quoting Announcing the New Amazon Brand Registry, Amazon Advert. (Nov. 17, 2017), https://advertising.amazon.com/blog/brand-registry [https://perma.cc/PY8G-MK86]). As journalist John Herrman describes it: “If you’re feeding a brand-new listing into the Amazon machine . . . and doing so without a pre-existing brand or customers, getting into Brand Registry is extremely important. To achieve real and lasting success on Amazon, it’s vital.”10×10. Id. Since 2017, Amazon has required a registered trademark for Brand Registry access.11×11. Id. Amazon sellers therefore have a powerful motivation to register their trademarks. However, many of these sellers, particularly those that sell inexpensive everyday objects for which consumers do not have strong brand loyalty, like nail clippers or phone chargers, are not driven by the desire to build and protect a recognizable brand. For these items, if a product is well-reviewed and prominently featured on Amazon, the seller “can succeed without a brand, or even despite one.”12×12. Id. It is loyalty to Amazon and its algorithms that drives consumers’ purchasing decisions, rather than loyalty to the seller’s specific product or brand. So sellers, many based outside the United States, just want to get a registered trademark as quickly as possible, so that they can make it onto the Brand Registry. As explained by Timothy Wang, an attorney with a number of clients in cross-border e-commerce, these sellers therefore “just come up with random letters,” because “[i]f [you] just have a bunch of random letters, [other sellers] can’t say this is too similar to their marks, and it’s easier for you to get the trademark registered.”13×13. Id. Thus, marks like ZGGCD and MAJCF,14×14. Registration Nos. 5,486,138 and 5,846,959, respectively. which appear in connection with goods for sale on Amazon,15×15. Herrman, supra note 9. make it onto the Trademark Register (the Register).

At least some of the applications filed by Yang, rather than being falsified, might then be for real products that their sellers are hoping to get on the Brand Registry. It is not clear how many applied-for and registered marks at the USPTO fall into this category: marks that are actually used in commerce but that consist of an apparently random string of letters unpronounceable in English and with no meaning in another language — what this Note refers to as “nonsense marks.” In their study of a random sample of applications for marks for apparel filed at the USPTO in 2017, Professors Barton Beebe and Jeanne Fromer found that 162 out of 365 applications from China, or 44.4%, involved a “nonsense word that is unpronounceable in English and that the applicant indicated has no meaning in any other language.”16×16. Beebe & Fromer, supra note 6, at 232; see also id. at 227. This is a significant number, though as Beebe and Fromer acknowledge, it is unclear whether these applications are fraudulent or for marks actually in use.17×17. Id. at 227 n.49. Mary Boney Denison, the former USPTO Commissioner for Trademarks, has suggested that at least some of the applications fall into the latter category, citing the Brand Registry as a reason behind the recent increase in unusual trademark filings.18×18. Making a Mark with Action, Inventors Dig. (Feb. 14, 2019), https://www.inventorsdigest.com/articles/making-a-mark-with-action [https://perma.cc/Q9AR-F42V]. Denison refers to “fraudulent filings.” Id. However, applicants motivated by the Brand Registry are unlikely to be committing conventional trademark fraud: if they want to be on the Brand Registry, presumably this is because they have actual products to sell. Denison therefore seems to mean “fraudulent” in a broader sense, beyond falsified applications. It is beyond the scope of this Note to attempt to quantify the number of applications for nonsense marks, but suffice it to say that a cognizable number of these marks are making their way through the USPTO and ultimately appearing on the Register.

Nonsense marks like ZGGCD and MAJCF seem to meet all of the requirements for trademark registration. However, because they consist of random, unpronounceable sets of letters nearly impossible to remember in the long term, they are at odds with what the trademark system is designed to do: protect source-distinguishing marks for the benefit of both consumers and producers. These marks are not just an odd fit with trademark law, though. Their presence on the Register causes concrete harm, creating obstacles for others trying to register and use trademarks, destabilizing the metrics trademark law uses to establish what is protectable, and draining the time and resources of the USPTO. Fortunately, existing trademark law already provides the tool to prevent the potential harm that these marks cause. Failure to function doctrine, which denies registration to marks that consumers will not perceive as source identifying, ensures that marks on the Register actually “work” as trademarks. By extending failure to function to encompass marks that do not work because they are linguistically incoherent and unmemorable, courts and the USPTO can keep these marks off the Register and prevent them from further muddying the trademark system.

Part I of this Note explains why nonsense marks do not fit within the established purposes of trademark law and why, nevertheless, existing barriers to registration do not filter them out. Part II then explores the problems with having these marks on the Register, outlining the harms to other brand owners, trademark examiners, and the system of meaning established by trademark law. Part III provides an overview of failure to function doctrine, while Part IV advocates for an extension of this doctrine, so that it captures marks that do not function because they have no staying power in consumers’ minds. By focusing on whether these marks work as trademarks are supposed to, those applying failure to function can offer a legal basis for the common sense conclusion that the only mark that matters to consumers in this context is Amazon’s.19×19. Cf. Michael Grynberg, AI and the “Death of Trademark,” 108 Ky. L.J. 199, 230 (2019) (“[T]here is trademark work going on” for many Amazon shoppers, “but it applies to platform competition (the selection of Amazon) and not necessarily purchasing decisions beyond that (what is bought on Amazon).”).

I. Nonsense Marks Within Existing Trademark Law

Foundational to trademark law is an understanding that trademarks offer social value and therefore should be legally protected. Though different scholars offer a number of theories of how trademarks produce value for consumers and businesses,20×20. See, e.g., Rochelle Cooper Dreyfuss, Expressive Genericity: Trademarks as Language in the Pepsi Generation, 65 Notre Dame L. Rev. 397, 400–03 (1990); Jessica Litman, Breakfast with Batman: The Public Interest in the Advertising Age, 108 Yale L.J. 1717, 1725–28 (1999); Jeremy N. Sheff, Marks, Morals, and Markets, 65 Stan. L. Rev. 761, 797–801 (2013). the dominant justification for trademark law within the United States is search-costs theory.21×21. See Barton Beebe, The Semiotic Analysis of Trademark Law, 51 UCLA L. Rev. 621, 623 (2004) (describing search costs as part of “a totalizing and, for many, quite definitive theory of American trademark law”); Mark P. McKenna, A Consumer Decision-Making Theory of Trademark Law, 98 Va. L. Rev. 67, 73 (2012) (referring to search costs as “the dominant theoretical account” of trademark law). This theory was first articulated by and is generally associated with Professor William Landes and Judge Richard Posner. See William M. Landes & Richard A. Posner, Trademark Law: An Economic Perspective, 30 J.L. & Econ. 265, 268–70 (1987). Search-costs theory maintains that trademarks are beneficial because consumers recognize them as identifiers of the source of goods they encounter. This recognition then allows those consumers to rely on trademarks to streamline their purchasing decisions, minimizing the burden of having to sort through and compare the characteristics of goods each time they want to buy something.22×22. Barton Beebe & Jeanne C. Fromer, Are We Running Out of Trademarks? An Empirical Study of Trademark Depletion and Congestion, 131 Harv. L. Rev. 945, 954–55 (2018). If a consumer has experienced the Coca-Cola brand in the past and knows that she likes its soda, the trademark allows her to identify efficiently the product she wants when she is next in the market for a soft drink. In short, consumers can rely on the trademark over time as an indication of the quality of the goods it’s attached to. Because trademarks allow consumers to reward the brands they like and deny patronage to the ones they don’t, producers are in turn incentivized to maintain consistent, high standards for their goods.23×23. See Landes & Posner, supra note 21, at 269. And because trademark law prevents others from using deceptively similar marks on their goods, those producers do not have to worry that their investment will be wasted.24×24. Stacey L. Dogan & Mark A. Lemley, Trademarks and Consumer Search Costs on the Internet, 41 Hous. L. Rev. 777, 787 (2004). Referring to these mutually reinforcing benefits, Professors Stacey Dogan and Mark Lemley describe trademarks as providing a “reliable vocabulary for communications between producers and consumers.”25×25. Id. Though it generally goes unspoken, a basic assumption underlying search-costs theory — and thus underlying trademark law — is that consumers will remember the brands they encounter.26×26. See Beebe & Fromer, supra note 22, at 1024 (“Consumers’ ability to quickly link a mark with the source and qualities it is intended to represent is directly related to how memorable the mark is to consumers.”). None of the consumer- or producer-based benefits of trademarks exist if marks do not form a durable, lasting impression in consumers’ minds.

Traditional producers know the importance of developing strong, memorable marks and invest considerable resources in doing so.27×27. See John Colapinto, Famous Names, New Yorker (Sept. 26, 2011), https://www.newyorker.com/magazine/2011/10/03/famous-names [https://perma.cc/P4V9-UD4W]. Scholars and branding experts draw on market research, experience, and linguistic theory to understand what makes a mark for a new product “work” — what makes it catchy, effective, and memorable to consumers.28×28. See Jake Linford, Are Trademarks Ever Fanciful?, 105 Geo. L.J. 731, 753–54 (2017); Colapinto, supra note 27; Neal Gabler, The Weird Science of Naming New Products, N.Y. Times Mag. (Jan. 15, 2015), https://nyti.ms/1E3fpSY [https://perma.cc/K2BU-R6BR]. Though coming up with a new mark can be a lengthy process,29×29. Gabler, supra note 28. some of the principles these players have developed are fairly intuitive. After interviewing a number of them, journalist John Colapinto reported:

Naming experts agree on several universals of great names. It’s probably best to keep them short. . . . Names that display a consonant-vowel-consonant pattern, like Gatorade, Lipitor, and Amazon, are often easiest to say, since these sorts of letter combinations are among the first that infants learn in any language.30×30. Colapinto, supra note 27.

Beebe and Fromer, rebutting conventional wisdom that “the supply of potential trademarks is inexhaustible,” likewise describe a “finite universe of ‘good’ trademarks” that traditional producers would actually want to use.31×31. Beebe & Fromer, supra note 22, at 962 (capitalization and italics omitted). As they explain, it is “obvious that given the limits of human cognition and communication, incumbent firms using shorter, less complex, more familiar, more easily pronounced, and more evocative marks will enjoy a significant competitive advantage over new firms that must resort to brand names that are less effective along these dimensions.”32×32. Id. at 963; see also id. at 965 (explaining that one feature of good marks is that they be “relatively easy to pronounce, hear, read, and remember”). Numerous blogs offering advice for those attempting to create memorable brand names similarly endorse words that are easy to pronounce.33×33. See, e.g., Nellie Akalp, Your New Business Name Should Be Memorable, Spell-able and Available, Entrepreneur (Feb. 29, 2016), https://www.entrepreneur.com/article/269669 [https://perma.cc/7L66-8PB7]. This common focus on short, pronounceable words makes sense: consumers encounter thousands of brands a day, and words that are relatively easy to understand and articulate seem more likely to “stick” in a person’s brain.

It is for this reason that the nonsense marks on Amazon’s Brand Registry are at odds with the goals of both trademark law and most trademark owners. Words like ZGGCD and MAJCF are difficult to break down into constituent syllables — recognizable and distinct units of sound that comprise written words.34×34. See Susan Bartlett, Grzegorz Kondrak & Colin Cherry, On the Syllabification of Phonemes, in Human Language Technologies 308, 308–09 (Ass’n for Computational Linguistics ed., 2009). This quality indicates that it would be difficult for an English speaker to understand how to pronounce them or recognize them as “possible words,”35×35. Abigail Cohn, English Syllable Structure, in Fundamentals of English Phonetics and Phonology (forthcoming 2021) (manuscript at 4) (on file with the Harvard Law School Library) (“[S]peakers can judge whether non-words are possible words. . . . Speakers of English know which sounds occur in their language and which don’t.”). The insight that some made-up words “work” in a given language and the study of what makes up those words are central to the field of phonology. Software engineers, operating from a similar assumption, have created algorithms to generate made-up words that sound “real.” See, e.g., Luke Dormehl, This A.I. Makes Up Gibberish Words and Definitions that Sound Astonishingly Real, Digit. Trends (May 15, 2020), https://www.digitaltrends.com/news/this-word-does-not-exist [https://perma.cc/83UD-W926]. rather than a string of random letters. As these words do not consist of recognizable linguistic units, people are accordingly less likely to process and store them mentally — or to recall them in the future.36×36. In cognitive psychology, this tendency is captured by the Pseudoword Superiority Effect (PSE). Donna Coch & Priya Mitra, Word and Pseudoword Superiority Effects Reflected in the ERP Waveform, 1329 Brain Rsch. 159, 160 (2010). Studies on the PSE indicate that subjects are more likely to recognize letters embedded in real words or real-sounding “pseudowords” than in “nonwords” made up of letter strings, see id., and that pseudowords are easier to remember than random nonwords, see, e.g., Stefan A. Frisch, Nathan R. Large & David B. Pisoni, Perception of Wordlikeness: Effects of Segment Probability and Length on the Processing of Nonwords, 42 J. Memory & Language 481, 489–91 (2000); Craig E.L. Stark & James L. McClelland, Repetition Priming of Words, Pseudowords, and Nonwords, 26 J. Experimental Psych. 945, 951–54 (2000).
The notion that an arbitrary string of letters is less likely to be remembered and distinguished is one that trademark law has already recognized. See, e.g., Tyr Sport, Inc. v. Datanation, LLC, 2004 WL 474682, at *3 (T.T.A.B. Mar. 9, 2004) (“[W]e also note that courts and this Board have often held that consumers have more difficulty recalling differences in what appear to be arbitrary letter strings.”); 4 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 23:33 (5th ed. 2020) (“[I]t is more difficult to remember a series of arbitrarily arranged letters than it is to remember words, figures, phrases, or syllables.”).
The problems compound when proposed nonsense marks get longer and when the market is crowded with them, so no one unusual mark stands out.37×37. Cf. Brian C. Stanton & Kristen K. Greene, Character Strings, Memory and Passwords: What a Recall Study Can Tell Us, in Human Aspects of Information Security, Privacy, and Trust 195, 203–04 (Theo Tryfonas & Ioannis Askoxylakis eds., 2014); Lorenzo Franceschi-Bicchierai, Your Brain Actually Doesn’t Suck at Passwords, Vice (June 13, 2015, 12:00 PM), https://www.vice.com/en_us/article/mgbnzq/your-brain-actually-doesnt-suck-at-passwords [https://perma.cc/Q2CA-UAL4] (“[R]emembering one [random password] is not the problem, remembering dozens of [them] is the problem.”). Try, for example, to recall one, much less all four, of the proposed marks in the first paragraph of this Note.

Because they are generally unpronounceable and difficult to remember or distinguish, nonsense marks fail to do what most companies want marks to do and what the theoretical foundations of trademark law assume that they will do: create a lasting impression on consumers. Indeed, they seem to be a manifestation of the “Borgesian infinite universe of theoretically possible trademarks,” which Beebe and Fromer consider but dismiss, since “much of that universe would consist of trademarks that are comparable to, if not worse than, no trademark at all.”38×38. Beebe & Fromer, supra note 22, at 963. At least some of the companies filing these marks are not looking for good marks, however: they seek trademark registration to gain access to Amazon’s Brand Registry, and “the name of the brand has become less and less important” to them, as they rely instead on Amazon’s marketing tools.39×39. Herrman, supra note 9; see also Grynberg, supra note 19, at 231.

But while these marks do not work in the way the trademark system assumes they will, paradoxically, it is for that precise reason that they face virtually no barriers to trademark registration. For most applied-for marks, “the primary protectability hurdle” is the requirement that they be distinctive of their source.40×40. Alexandra J. Roberts, Trademark Failure to Function, 104 Iowa L. Rev. 1977, 1979 (2019); see also Beebe & Fromer, supra note 22, at 957 (“The most important requirement by far that a mark must meet to qualify for registration is that it be perceived by consumers as distinctive of its source.”). To assess whether a mark is distinctive, courts and USPTO examiners apply the Abercrombie41×41. Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4 (2d Cir. 1976). spectrum, determining the distinctiveness of a mark by considering it in relation to the goods on which it is used and placing it into one of several categories.42×42. See id. at 9; see also 2 McCarthy, supra note 36, §§ 11:1–:2. A mark is generic when it refers to “the genus of which the particular product is a species”43×43. Abercrombie, 537 F.2d at 9. — when the mark tells consumers what a good is. IVORY, for example, is generic for a product made from elephant tusks.44×44. Id. at 9 n.6. Generic terms are never protectable.45×45. Id. at 9. But see U.S. Pat. & Trademark Off. v. Booking.com B.V., 140 S. Ct. 2298, 2301 (2020) (allowing registration of some “generic.com” marks). Next on the spectrum are descriptive marks, which describe a quality, function, or purpose of the relevant goods.46×46. U.S. Pat. & Trademark Off., Trademark Manual of Examining Procedure § 1209.01(b) (2018) [hereinafter TMEP]. SEALTIGHT for fasteners and FROSTY TREATS for ice cream are both descriptive.47×47. Cf. Frosty Treats, Inc. v. Sony Comput. Ent. Am., Inc., 426 F.3d 1001, 1005 (8th Cir. 2005); B & B Hardware, Inc. v. Hargis Indus., 252 F.3d 1010, 1012 (8th Cir. 2001). As such, they are protectable only if their owner can demonstrate that they have developed secondary meaning: that consumers understand them as source identifying.48×48. Abercrombie, 537 F.2d at 9. As opposed to generic and descriptive marks, those categorized as suggestive, arbitrary, or fanciful are considered inherently distinctive and therefore registrable, assuming they do not face other bases for opposition.49×49. TMEP, supra note 46, § 1209.01(a). Suggestive marks are those that hint at some feature of the associated good, but that “require imagination, thought, or perception” to get from the mark to the good50×50. Id. — like COPPERTONE for sunscreen.51×51. Sara Lee Corp. v. Kayser-Roth Corp., 81 F.3d 455, 464 (4th Cir. 1996). Arbitrary marks are made of existing words applied to goods unassociated with their traditional meaning, such as APPLE for computers.52×52. TMEP, supra note 46, § 1209.01(a). Finally, fanciful marks are made-up terms invented specifically to be trademarks.53×53. Id. PEPSI and EXXON are both fanciful marks.54×54. Id.

Though suggestive, arbitrary, and fanciful marks are all considered inherently distinctive, because fanciful marks do not draw on already-existing words, they are seen as especially strong and sometimes afforded even greater protection in trademark disputes.55×55. Linford, supra note 28, at 740–42. As made-up words with no meaning in any language, nonsense marks are fanciful, making them inherently distinctive — and even especially protected — on the Abercrombie spectrum and allowing them to overcome a major barrier to trademark registration.

Most nonsense marks are also likely to satisfy a second important requirement for trademark registration. The Lanham Act prevents registration of any mark that is so similar to an already-registered mark “as to be likely, when used on or in connection with the goods of the applicant, to cause confusion, or to cause mistake, or to deceive.”56×56. 15 U.S.C. § 1052(d). The Lanham Act also prevents registration of marks confusingly similar to nonabandoned unregistered marks. Id. Multiple factors go into an assessment of whether an applicant’s mark is confusingly similar to someone else’s, but two particularly significant ones are “[t]he similarity or dissimilarity of the marks in their entireties as to appearance, sound, connotation and commercial impression”57×57. TMEP, supra note 46, § 1207.01(b) (quoting In re E.I. du Pont de Nemours & Co., 476 F.2d 1357, 1361 (C.C.P.A. 1973)). and “[t]he relatedness of the goods or services as described in the application and registration(s).”58×58. Id. § 1207.01. Like distinctiveness, likely confusion is a major barrier to those seeking registration and can often be the basis for rejection. See Beebe & Fromer, supra note 22, at 955–56, 955 n.34. These inquiries always involve a case-by-case analysis, and the outcome depends on the two marks at issue. However, given the traditional incentives of trademark owners, it seems unlikely that a mark like LJXOAIEU or QIANDLEE will appear confusingly similar to others on the Register — except perhaps to other nonsense marks.59×59. Even where confusion is possible, likelihood of confusion is unlikely to screen out nonsense marks effectively for two reasons. First, it is unclear how a decisionmaker would even assess likelihood of confusion when confronted with an unpronounceable mark that leaves little lasting commercial impression in the first place. See infra pp. 1816. Second, because nonsense marks are not commercially viable in a traditional sense, existing mark owners with debatably similar marks are unlikely to have much incentive to spend the time and resources required to challenge their registration. Nonsense marks thus easily overcome another major “protectability hurdle.”

Nonsense marks do not look like trademarks typically look or work the way the trademark system assumes they are supposed to. But because of that quality, rather than in spite of it, nonsense marks easily meet all of the formal requirements for trademark registration and may even receive stronger protections than do many more typical marks.60×60. There are other barriers to trademark registration besides those discussed above, but these are highly specific and do not apply to most nonsense marks. See, e.g., 15 U.S.C. § 1052(c) (barring the use of names of living people without their consent); id. § 1052(e)(2) (barring most geographically descriptive terms). Put another way, within the existing registration system, everything but common sense indicates that these words should be valid trademarks.

II. The Harms of Nonsense Marks

Though nonsense marks are at odds with the purposes of the trademark system, the ease with which they qualify for registration is not necessarily a problem in and of itself. Unlike marks that might confuse or deceive about a product’s source of origin, it does not seem that the presence of nonsense marks in the market will directly harm consumers or lead them to buy products they do not intend to. Since consumer protection has become the guiding normative justification for trademark law,61×61. 3 McCarthy, supra note 36, § 19:1.75. the question for nonsense marks, as it is for other nontraditional proposed marks, is why not offer trademark protection?62×62. Deven Desai, Should Trademark Law Protect Non-traditional Trademarks?, in The Protection of Non-traditional Trademarks 125, 125–26 (Irene Calboli & Martin Senftleben eds., 2018) (“Why not? That question taunts trademark law and scholarship. . . . [W]hen enterprises . . . pursue new ways of developing and selling goods and services, trademark law has been more than happy to sanction such practices with the force of law. That has happened even when a given practice or outcome strays far from trademark law’s alleged search costs foundation. The question ‘Why not?’ reflects this deferential posture.” (footnotes omitted)). However, as scholars like Professors Rebecca Tushnet and Mark McKenna have noted, trademark law’s primary focus on consumer protection is both recent63×63. See Mark P. McKenna, The Normative Foundations of Trademark Law, 82 Notre Dame L. Rev. 1839, 1848–73 (2007) (describing trademark law in its historical origins as a way to prevent unfair competition among producers). and limited.64×64. See Tushnet, supra note 2, at 916; see also Park ’N Fly, Inc. v. Dollar Park & Fly, Inc., 469 U.S. 189, 213–15 (1985) (Stevens, J., dissenting); William McGeveran & Mark P. McKenna, Confusion Isn’t Everything, 89 Notre Dame L. Rev. 253, 274–95 (2013). Any choice to expand trademark protection in a new way can impose burdens beyond those for consumers, including burdens on the USPTO, on competitors, and on the trademark system in general. It is along these metrics that the existence of nonsense marks on the Register presents the potential for tangible harm.

First, and most basically, applications for nonsense marks place further stress on an already-strained USPTO. Trademark applications have increased continually in recent years. In 2019, the USPTO received 485,444 applications, the tenth year in a row in which the number of applications broke the previous record.65×65. In 2019, USPTO Received Record Number of Applications for the 10th Year in a Row, Erik M. Pelton & Assocs. (Jan. 7, 2020), https://www.erikpelton.com/in-2019-uspto-received-record-number-of-applications-for-the-10th-year-in-a-row [https://perma.cc/LY7K-TDJY]; see also Tim Lince, “Never Seen Anything Like This” – USPTO Experiences Massive Surge in Trademark Applications from China, World Trademark Rev. (Aug. 1, 2019), https://www.worldtrademarkreview.com/brand-management/never-seen-anything-uspto-experiences-massive-surge-trademark-applications [https://perma.cc/V9LD-YNF3]. This means that USPTO examiners are receiving thousands of applications a day and are tasked with assessing whether they should proceed to the Register. Even under normal circumstances, “maintaining the registration system requires the investment of substantial government and private resources.”66×66. Tushnet, supra note 2, at 872. But faced with the twin problems of fraudulent filings and unused marks that remain on the Register, the USPTO has instituted reforms aimed at ensuring that the Register consists only of marks that are currently, actually being used in commerce.67×67. See Counterfeits and Cluttering: Emerging Threats to the Integrity of the Trademark System and the Impact on American Consumers and Businesses: Hearing Before the H. Comm. on the Judiciary, 116th Cong. (2019) (statement of Mary Boney Denison, Comm’r for Trademarks, USPTO) (describing the USPTO’s efforts to prevent registration of fraudulent filings); Beebe & Fromer, supra note 22, at 1034–35 (describing the USPTO’s efforts to clear the Register of deadwood). While these are welcome changes, they consume additional time and resources of the USPTO. Against this backdrop, when the owners of nonsense marks, many of whom are not seeking the traditional benefits of trademark registration, file applications with the USPTO, they impose a further burden on the agency. Indeed, examiners, wary of fraudulent marks from China in general, may spend even more time than usual reviewing nonsense marks from Chinese companies.68×68. See Lince, supra note 65. By calling on the resources of the USPTO to achieve their owners’ Amazon-related objectives, nonsense-mark applications themselves impose costs on the USPTO, as well as on other applicants affected by examiners’ greater caseloads and slower review processes.69×69. Id.

Once nonsense marks make it onto the Register, they have the potential to inflict further harm, particularly on traditional mark owners. While trademark registration imparts many benefits to mark owners, its primary public benefit is “in the value of the trademark register as a source of information,” allowing businesses to “discover quickly and cheaply which signs third parties already have claimed.”70×70. Robert Burrell & Michael Handler, Dilution and Trademark Registration, 17 Transnat’l L. & Contemp. Probs. 713, 715–16 (2008); see also id. at 722 (“The . . . entry of [a] mark on a public register allows third parties to identify what marks are protected in any given sphere of trade.”). To avoid opposition from the USPTO or other mark owners, those seeking to register new marks must avoid words that might be seen as confusingly similar to existing marks.71×71. Beebe & Fromer, supra note 22, at 1021–22. This is where the anticompetitive potential of registered nonsense marks begins. While traditional mark owners probably will not be interested in nonsense marks, at least some of these marks are just a couple of letters off from “working” in English. For example, MAJCF is quite close to MAJI, while JANRSTIC is not far from the more viable JANSTICK. If JANSTICK were already on the Register, the mark owner might be compelled to invest time and resources arguing that the similar but nonsensical mark presents a likelihood of confusion. Worse, if nonsense marks make it onto the Register first, potential registrants, already facing a diminishing number of “good,” available trademarks,72×72. Id. at 977–1012. might be hesitant to apply with their marks out of fear of a likelihood-of-confusion rejection.73×73. Alexandra J. Roberts, How to Do Things with Word Marks: A Speech-Act Theory of Distinctiveness, 65 Ala. L. Rev. 1035, 1053 (2014). This chilling effect would be particularly acute for smaller businesses, which may not be able to afford a potentially lengthy and expensive registration battle.74×74. See Fraudulent Trademarks: How They Undermine the Trademark System and Harm American Consumers and Businesses: Hearing Before the S. Comm. on the Judiciary, 116th Cong. (2019) (statement of Megan K. Bannigan, Counsel, Debevoise & Plimpton LLP) (describing the costs of opposing a mark’s registration).

Compounding the problem, trademarks that appear on the Register are afforded a “robust presumption of rights that is incredibly difficult to unravel,”75×75. Roberts, supra note 40, at 1985. including a presumption of validity.76×76. Tushnet, supra note 2, at 876. These protections would make it more difficult for would-be users of similar marks to challenge the nonsense marks’ position on the Register or prevail in a lawsuit. And, most cynically, owners of nonsense marks might use their registrations in overtly anticompetitive ways, bringing lawsuits against or sending cease-and-desist letters to users of arguably similar marks in bad faith, in order to extort money from those users.77×77. For more on trademark trolls, see, for example, Stacey Dogan, Bullying and Opportunism in Trademark and Right-of-Publicity Law, 96 B.U. L. Rev. 1293, 1312–13, 1316–22 (2016); Kenneth L. Port, Trademark Extortion: The End of Trademark Law, 65 Wash. & Lee L. Rev. 585, 589–92 (2008). Tushnet has noted that “anecdotally, registrations play a big role in the effectiveness of cease-and-desist letters,” particularly for small companies without legal counsel or experience. Tushnet, supra note 2, at 920; see also Dogan, supra, at 1312–13. Thus, in spite of their relative uselessness as traditional source signifiers, nonsense marks on the Register have the potential to chill producers seeking to develop marks in an increasingly crowded field, to prompt costly litigation, and even to be weaponized by bad faith actors taking advantage of the trademark system. These anticompetitive possibilities can and should be the concern of trademark law.78×78. It is worth noting that the potential for anticompetitive effects might cut both ways. The Brand Registry can be very helpful to new market entrants. Because entry requires trademark registration, keeping nonsense marks off the Trademark Register has the potential to harm smaller companies trying to reach consumers on Amazon while benefiting incumbent sellers — often Amazon itself. See John Herrman, Everything on Amazon Is Amazon!, N.Y. Times (Nov. 15, 2018), https://nyti.ms/2DomF4z [https://perma.cc/3F8B-L8XY].
Though these concerns are well taken, they might be better understood as a critique of Amazon and the market conditions it creates. Cf. Grynberg, supra note 19, at 225–34 (discussing how Amazon creates secondary effects for trademark law and use). As such, they are beyond the scope of this Note, which seeks to explain the problems of having these marks on the Register, regardless of the motivations of nonsense mark applicants.

Finally, nonsense marks place considerable strain on the metrics that trademark law relies on to function. As an initial matter, and as suggested above, nonsense marks challenge the ability of the Abercrombie spectrum — thought of as “the most black-letter of all the black-letter trademark law”79×79. Mark P. McKenna, Property and Equity in Trademark Law, 23 Marq. Intell. Prop. L. Rev. 117, 124 (2019). — to identify the value of and consumer reaction to different categories of marks. Though Abercrombie may be an imperfect standard to begin with,80×80. For critiques of the Abercrombie spectrum, see, for example, Thomas R. Lee, Eric D. DeRosia & Glenn L. Christensen, An Empirical and Consumer Psychology Analysis of Trademark Distinctiveness, 41 Ariz. St. L.J. 1033, 1037–38 (2009); Rebecca Tushnet, Looking at the Lanham Act: Images in Trademark and Advertising Law, 48 Hous. L. Rev. 861, 871–74 (2011). See also Roberts, supra note 40, at 1987 (referring to distinctiveness doctrine as “incomplete” as “an account of what makes matter protectable”). it remains a foundation of trademark law, frequently relied on by courts and USPTO examiners.81×81. See Joseph Scott Miller, Abercrombie 2.0 — Can We Get There from Here? Thoughts on “Suggestive Fair Use,” 77 Ohio St. L.J. Furthermore 1, 2 (2016) (finding over 800 cases citing Abercrombie over forty years — though, as Professor Alexandra Roberts points out, this number does not include the thousands of cases that rely on Abercrombie without citing it or the “thousands or millions of USPTO filewrappers that call on the Abercrombie categories in assessing trademark registrability,” Roberts, supra note 40, at 1982 n.15). Nonsense marks on the Register deepen the cracks in the foundation of one of trademark law’s major sorting mechanisms, calling into question Abercrombie’s legitimacy as an indicator of what makes a “strong” mark and leaving it less stable for those who look to it for guidance.

Nonsense marks also defy trademark law’s ordinary heuristics for determining whether two marks are confusingly similar — both at the registration stage and in infringement suits. Presented with a likelihood-of-confusion case in which one or both marks consist of a random string of letters, it is not immediately clear how a decisionmaker would assess the “similarity or dissimilarity of the marks in their entireties as to appearance, sound, connotation and commercial impression,”82×82. TMEP, supra note 46, § 1207.01(b) (quoting In re E.I. du Pont de Nemours & Co., 476 F.2d 1357, 1361 (C.C.P.A. 1973)). a “key consideration[] in any likelihood of confusion determination.”83×83. Id. § 1207.01. While that decisionmaker could walk through the likelihood-of-confusion factors, the ultimate question these factors are meant to answer is whether the relevant consumers are likely to be confused about the source of the goods at issue.84×84. Id.; see also du Pont, 476 F.2d at 1360 (“[T]he question of confusion is related not to the nature of the mark but to its effect when applied to the goods of the applicant.” (internal quotation marks omitted)). Considering marks that, if similar, are arbitrarily so (for example, LJXOAIEU and LJXZXMY85×85. Registration No. 6,128,282. ), the decisionmaker applying the factors would probably find it difficult to determine if consumers are likely to be confused about whether the goods come from the same source — much less whether consumers will make any meaningful assessment of the marks at all.86×86. In cases involving acronym marks, courts and the USPTO often find in favor of plaintiffs “based on the premise that it is more difficult to remember a series of arbitrarily arranged letters than it is to remember words, figures, phrases, or syllables” — and that acronym marks are therefore more likely to be confused. 4 McCarthy, supra note 36, § 23:33; cf. Jellibeans, Inc. v. Skating Clubs of Ga., Inc., 716 F.2d 833, 841–42 (11th Cir. 1983) (finding the marks “Lollipops” and “Jellibeans” for roller rinks similar in part because of the “similar general impression” they had on consumers). Decisionmakers confronting nonsense marks might draw on this line of cases and find in favor of firstcomers claiming that a new mark is likely to cause confusion. However, this result is particularly unsatisfying in cases involving nonsense marks, as it seemingly arbitrarily favors the first party to register a random string of letters. Nonsense marks thus throw into disarray the established likelihood-of-confusion approach.

To be clear, the existing metrics trademark law relies on are not necessarily the best or the only options. But they are the ones available to and relied on by participants and new entrants in the trademark system. The presence on the Register of marks that erode trademark law’s existing standards renders those standards less reliable and less predictable, making it harder for all involved to understand what trademark doctrines are supposed to do.

Like other registration systems, the Trademark Register acts as a kind of commons, publicly available for all who seek to use it.87×87. See Tushnet, supra note 2, at 918. Such registries are subject to the tragedy of the commons: as an ever-expanding number of registrants joins, the Register becomes overloaded by entrants that should not be there, people stop relying on it, and “the very attractiveness of making clear one’s claims by recording them defeats the purpose of the system, that is, to clarify all claims against a given property.”88×88. Carol M. Rose, Crystals and Mud in Property Law, 40 Stan. L. Rev. 577, 596 (1988). On an already strained Register, nonsense marks are not just a subject of academic curiosity. Rather, they have the potential to cause real harm to the USPTO, other market entrants, and trademark law’s existing systems of meaning.

III. Failure to Function Doctrine

Despite the novelty of the issues nonsense marks present, the tool needed to address them already exists in trademark law, in the form of failure to function doctrine. Described by Professor Alexandra Roberts as the “red-headed stepchild of trademark law,”89×89. Roberts, supra note 40, at 1983. failure to function has received relatively little scholarly attention.90×90. Roberts’s article is a notable exception. See generally id. A number of trademark scholars have also considered similar themes in discussing “use as a mark” as a defense to trademark infringement. See generally, e.g., Stacey L. Dogan & Mark A. Lemley, Grounding Trademark Law Through Trademark Use, 92 Iowa L. Rev. 1669 (2007); Mark P. McKenna, Trademark Use and the Problem of Source, 2009 U. Ill. L. Rev. 773. However, the question of whether a mark actually functions as a trademark is critical to assessing whether it should receive protection, and any analysis focused just on distinctiveness, likelihood of confusion, or other more traditional grounds for refusal is necessarily incomplete.91×91. See Roberts, supra note 40, at 1987. The USPTO has taken note: in recent years, failure to function has played an increasing role in registration refusals.92×92. As one indication, a search of “failure to function” in the Trademark Trial and Appeal Board (TTAB) database yields thirty-one results from 2019. TTAB Reading Room, USPTO, https://ttab-reading-room.uspto.gov/efoia/efoia-ui/#/search/decisions [https://perma.cc/W9RM-9QEX] (search “failure to function” with date range “1/1/19” to “12/31/19”). Ten years earlier, that number was seven. Id. (search “failure to function” with date range “1/1/09” to “12/31/09”). As one prominent trademark scholar has put it, “failure to function is having a moment.”93×93. Rebecca Tushnet (@rtushnet), Twitter (July 24, 2020, 3:04 PM), https://twitter.com/rtushnet/status/1286739126948179968 [https://perma.cc/6F9M-5ZVJ].

The doctrine relies on the basic premise that, to be registrable, a proposed mark must be used as a trademark. “[N]ot every designation that is placed or used on or in connection with a product necessarily functions or is recognized as a trademark for said product.”94×94. Am. Velcro, Inc. v. Charles Mayer Studios, Inc., 177 U.S.P.Q. 149, 154 (T.T.A.B. 1973). To establish valid trademark use, an applicant must show that the mark is used on or in connection with goods in commerce,95×95. 15 U.S.C. §§ 1051, 1127. but also that it actually works the way a trademark is supposed to. In other words, that it accomplishes “the job of identification: to identify one source and distinguish it from other sources.”96×96. 1 McCarthy, supra note 36, § 3:4; see also, e.g., MicroStrategy Inc. v. Motorola, Inc., 245 F.3d 335, 341 (4th Cir. 2001).

Failure to function has developed as a way to assess whether a given mark will be understood as a source indicator and to weed out those marks that will not. It is not specifically listed as one of the statutory grounds for registration refusal but is based on the Lanham Act’s definition of “trademark” as “any word, name, symbol, or device, or any combination thereof — (1) used by a person . . . to identify and distinguish his or her goods . . . from those manufactured or sold by others and to indicate the source of the goods.”97×97. 15 U.S.C. § 1127. When issuing a failure to function refusal, the USPTO relies on this definition of trademark, in addition to the sections of the Lanham Act covering registration requirements. TMEP, supra note 46, § 1202. Though the USPTO has long relied on this threshold requirement to reject proposed marks that consumers will not understand to have trademark significance,98×98. See, e.g., In re Bose Corp., 192 U.S.P.Q. 213, 215–16 (C.C.P.A. 1976). in recent years it has organized such refusals under the heading “failure to function.” In failure to function cases, the USPTO considers whether the proposed mark does what a mark is meant to: whether it “performs the function of a trademark by signifying to purchasers the source of the goods sold or offered for sale.”99×99. In re Valvoline Licensing & Intell. Prop. LLC, 2020 WL 919225, at *6 (T.T.A.B. Jan. 27, 2020) (quoting In re Paramount Pictures Corp., 213 U.S.P.Q. 1111, 1115 (T.T.A.B. 1982)). As Roberts explains, failure to function doctrine ensures that a mark is “not only used in commerce, but used in a trademark way.”100×100. Roberts, supra note 40, at 1981 (footnote omitted). Assessing whether a mark properly fulfills this role involves a contextual analysis focused on how the mark appears on the goods, how potential consumers will perceive the mark, and whether those consumers will understand it as something that identifies and distinguishes the source of the goods it is attached to.101×101. See In re Eagle Crest, Inc., 96 U.S.P.Q.2d 1227, 1229 (T.T.A.B. 2010). This inquiry is necessarily consumer focused. In the words of the USPTO, “[t]he critical question in determining whether [a mark] functions as a trademark is the commercial impression it makes on the relevant public, i.e., whether the term sought to be registered would be perceived as a mark identifying the source of the goods.”102×102. In re Peace Love World Live, LLC, 127 U.S.P.Q.2d 1400, 1403 (T.T.A.B. 2018). When the relevant public would not attach trademark significance to a given mark, the USPTO will conclude that it fails to function and refuse registration.

Though united by the same objection — that the proposed trademark will not be received as such by the relevant consumers — failure to function refusals come in a variety of flavors. In some cases, the issue is whether the mark appears where and how consumers expect marks to appear. Under this line, a mark fails to function if, “when viewed in context, it is not immediately obvious that a certain word or design is being used as an indication of origin.”103×103. 1 McCarthy, supra note 36, § 3:4; see also In re Riviana Foods Inc., Serial No. 87414879, at 4–5 (T.T.A.B. July 20, 2020) (“To make [a failure to function] determination, we look to specimens and other evidence of record showing how the designation actually is used in the marketplace.”). For example, many consumers look to what they call a “trademark spot” on a given good or its packaging, expecting the designation of origin to show up there.104×104. Lee et al., supra note 80, at 1076; see also id. at 1076–78. Other contextual indications that a word is serving the role of a trademark include large font relative to small surrounding text, distinctive print style or color, and prominent position on the label.105×105. See 1 McCarthy, supra note 36, § 3:4; Roberts, supra note 40, at 2002–10. When a proposed mark appears in a context not classically associated with trademark placement or style, this can indicate that consumers are likely to understand it not as a source identifier, but rather as an additional informational, branding, or design feature. For example, in 2018 the Trademark Trial and Appeal Board (TTAB) rejected the mark MAX-FLOW for a sleep apnea device on failure to function grounds.106×106. In re Fallon, Serial No. 86915495, at 5, 12 (T.T.A.B. Dec. 6, 2018). Explaining its decision, the TTAB pointed to the proposed mark’s relatively small size and placement on the packaging near text describing how the product works, as well as the presence of other marks that were styled more like traditional trademarks.107×107. Id. at 9–11. These factors all suggested that “the MAX-FLOW mark would be perceived as identifying a feature of the oral appliance,” rather than its source and therefore would not function as a trademark.108×108. Id. at 11. In cases like MAX-FLOW, “[t]he manner in which the mark is actually used is the determinative factor,”109×109. Am. Council on Exercise v. Health Care Fitness Integrations, LLC, 2019 WL 7423403, at *4 (T.T.A.B. Dec. 19, 2019). and a mark that doesn’t look like a trademark will fail to function as one.

In other cases, marks fail to function because they have informational or social meaning that exceeds their tie to any given product or source. In this genre of failure to function, the USPTO is guided by the principle that the “more commonly a phrase is used, the less likely that the public will use it to identify only one source and the less likely that it will be recognized by purchasers as a trademark.”110×110. In re Eagle Crest Inc., 96 U.S.P.Q.2d 1227, 1229 (T.T.A.B. 2010). In In re Deporter,111×111. 2019 WL 460492 (T.T.A.B. Jan. 29, 2019). for example, the TTAB considered the proposed mark #MAGICNUMBER108 for apparel. Rejecting the application on failure to function grounds, the TTAB explained that, though #MAGICNUMBER108 may be arbitrary or fanciful and may have been used on the applicant’s clothing, third-party evidence revealed that it is also widely used online “in a non-trademark manner to consistently convey information about the Chicago Cubs’ World Series appearance and win after a 108-year drought.”112×112. Id. at *6; see also id. at *7–9. Because prospective purchasers would associate the mark with this general social use, they would be “unlikely to consider it to indicate the source of Applicant’s goods” specifically, and the TTAB concluded that it did not function as a mark.113×113. Id. at *6; see also id. at *9. In addition to #MAGICNUMBER108, the TTAB has rejected ONCE A MARINE, ALWAYS A MARINE for apparel,114×114. Eagle Crest, 96 U.S.P.Q.2d at 1232. INVESTING IN AMERICAN JOBS as a service mark,115×115. In re Wal-Mart Stores, Inc., 2019 WL 193990, at *14 (T.T.A.B. Jan. 11, 2019). FARM TO TABLE for wine,116×116. In re Fowles Wine Pty Ltd., 2017 WL 4460470, at *11 (T.T.A.B. Sept. 15, 2017). and I ♥ DC for apparel,117×117. D.C. One Wholesaler, Inc. v. Chien, 120 U.S.P.Q.2d 1710, 1716–17 (T.T.A.B. 2016). among others. In all of these cases, the TTAB concluded that the proposed marks were made up of terms widely used by third parties, with that use shaping “the environment in which the [marks were] perceived by the public.”118×118. In re Hulting, 107 U.S.P.Q.2d 1175, 1178 (T.T.A.B. 2013) (quoting In re Tilcon Warren, Inc., 221 U.S.P.Q. 86, 88 (T.T.A.B. 1984)). Consumers were likely to bring those social meanings with them and understand the marks as ornamental, informational, or expressive of allegiance, rather than as indicative of a particular source.119×119. Even use in the “trademark spot” and other contextual indicators that a mark might designate source cannot defeat the fundamentally informational quality of these marks. See, e.g., In re AC Webconnecting Holding B.V., Serial Nos. 85635277 & 85635287, at 29–30 (T.T.A.B. Sept. 15, 2020). Marks of this kind therefore fail to function because they are too common in a given vernacular, such that they cannot be tied to one origin or brand.

Across these cases, failure to function imposes the common sense requirement that applicants for registration demonstrate that their marks actually do what marks are supposed to do and weeds out those that do not. This inquiry is necessarily consumer focused, centered on a mark’s context, social use, and impression on the consuming public and on the baseline question of whether the proposed mark has trademark meaning for that public.

IV. Linguistic Failure to Function

The nonsense marks that this Note is concerned with do not fit into any of the traditional failure to function categories. They may appear in a “trademark spot” and be styled like trademarks; they do not have any informational or social meaning that prevents them from acting as a single source identifier; and they are not model numbers, names, or titles.120×120. Cf. TMEP, supra note 46, §§ 1202.08, .09(a), .16; Lee et al., supra note 80, at 1076. But the concerns animating failure to function doctrine apply to nonsense marks, and trademark examiners should extend the doctrine to them, asking whether they perform the source-indicating function trademarks are meant to and denying registration to those that do not.

As explained above, failure to function refusals are based in part on the Lanham Act’s definition of trademark as “any word, name, symbol, or device, or any combination thereof” that is used “to identify and distinguish [the owner’s] goods.”121×121. 15 U.S.C. § 1127. Failure to function ensures that marks on the Register actually identify the goods to which they are attached, with the “critical question” being “the commercial impression [a mark] makes on the relevant public.”122×122. In re Peace Love World Live, LLC, 127 U.S.P.Q.2d 1400, 1403 (T.T.A.B. 2018). In the case of nonsense marks, consumers might initially perceive these marks as trademarks: they appear in the expected place, and they aren’t merely describing or touting any feature of the product. The issue arises when the consumer steps away from the physical product or Amazon listing. Because these marks are essentially just a random string of letters, they are unlikely to stick in consumers’ minds.123×123. See supra pp. 1809–10. Consumers will have a difficult time remembering, much less articulating, that they purchased the headband they’re wearing from LJXOAIEU or that their shirt is QIANDLEE brand. While they may be able to tell their friends where they bought their phone cable — Amazon — they are much less likely to recall who made and sold the product. And if they are again in the market for the same good or a similar one, the mark’s nonstickiness means that they probably will not be able to remember it well enough to look for or stay away from it in their new search. The volume of nonsense marks in the marketplace compounds the problem. Facing pages of options online for gloves, how are consumers to remember whether they got their last pair from WDDBHSKN, CYJAGNY, or OUBBGLVS?124×124. U.S. Trademark Application Serial Nos. 90/181,274 (filed Sept. 15, 2020); 90/180,948 (filed Sept. 15, 2020); and 90/180,906 (filed Sept. 15, 2020), respectively.

Paying attention to how consumers are likely to receive, understand, and use nonsense marks, it becomes clear that, as these marks are composed of arbitrary, unpronounceable strings of letters, they do not work as trademarks should. Though they may be legible as source identifiers at the point of sale, most nonsense marks fail to create a lasting commercial impression, so that, in practice, consumers are unlikely to use them to describe where their products came from or rely on them in future commercial transactions. In other words, they are not source identifying in the long term because consumers will find it difficult if not impossible to use them in this way.125×125. Cf. Tushnet, supra note 80, at 870 (“A term a consumer can’t remember is by definition not doing a good job as an indicator of source and, thus, can’t serve the functions we attribute to trademarks of protecting consumers from confusion and incentivizing producers to maintain quality.”). And, because they are not organized around a linguistic system of meaning, as soon as more than one nonsense mark exists, it becomes difficult to use any of them to distinguish the sources of different goods.126×126. See Barton Beebe, Search and Persuasion in Trademark Law, 103 Mich. L. Rev. 2020, 2028–31 (2005) (explaining the importance of both source distinctiveness and differential distinctiveness in trademark law). Where a mark like #MAGICNUMBER108 cannot distinguish the source of goods because too many people use the same phrase, a mark like CYJAGNY fails to perform a distinguishing function when it exists in a marketplace of marks made up of equally random, functionally interchangeable strings of letters.127×127. See Tyr Sport, Inc. v. Datanation, LLC, 2004 WL 474682, at *3 (T.T.A.B. Mar. 9, 2004) (“[W]e also note that courts and this Board have often held that consumers have more difficulty recalling differences in what appear to be arbitrary letter strings.”).

Though failure to function does not currently cover this form of failure, the doctrine’s focus on how consumers perceive and use marks means both that it ought to be concerned with nonsense marks and that it provides the tool to prevent them from making it onto the Register. Failure to function should therefore extend to encompass a new category of marks: those that fail linguistically. In this context, linguistic failure to function refers to word marks that do not function at the level of language — that, because of the linguistic form they take, are unlikely to “stick” in consumers’ minds or be relied on to refer to the origin of goods. Key questions examiners might ask in assessing whether a mark fails to function linguistically include whether it is pronounceable in English, how long it is, and whether the applicant can explain why it chose that specific combination of letters.128×128. A company crafting a new, fanciful mark that it intends to have source-identifying significance is likely to have a reason to pick that mark as opposed to another. See supra pp. 1808–09. If a proposed mark seems like an unpronounceable, truly random string of letters, and if the examiner concludes that it is unlikely to be meaningful or memorable to consumers, it should be rejected on failure to function grounds — unless, as discussed further below, the applicant can provide evidence that the mark is not random and has a meaning likely to be registered by its intended audience. The introduction of a linguistic failure to function basis for refusal will capture many nonsense marks that currently clutter the Register. But it also might serve as the grounds for rejecting other marks that do not do what trademarks are supposed to do, including product numbers,129×129. See, e.g., Tenneco Auto. Operating Co. v. Kingdom Auto Parts, 410 F. App’x 841 (6th Cir. 2010); see also Roberts, supra note 40, at 2032–33 (discussing this case). Though the TMEP already includes model designators as a category of failure to function, see TMEP, supra note 46, § 1202.16, its definition of the term may not always capture other similar features, such as bar codes. some phone numbers,130×130. In re Michelin N. Am., Inc., 2015 WL 5675633 (T.T.A.B. Sept. 11, 2015); see also Roberts, supra note 40, at 1978–82. and phrases too long to be remembered or registered as source indicators, such as one proposed mark comprised of a 101-word-long inspirational phrase.131×131. “SO, WHAT DO YOU WANT TO DO WITH YOUR LIFE?” “YOU SHOULD BE A LAWYER, A DOCTOR, AN ACCOUNTANT A CONSULTANT…BLAH, BLAH, BLAH.” EVERYWHERE YOU TURN PEOPLE TRY TO TELL YOU WHO TO BE AND WHAT TO DO WITH YOUR LIFE.  WE CALL THAT THE NOISE. BLOCK IT. SHED IT. LEAVE IT FOR THE CONFORMISTS. AS A GENERATION, WE NEED TO GET BACK TO FOCUSING ON INDIVIDUALITY. SELF-CONSTRUCTION RATHER THAN MASS PRODUCTION. DEFINE YOUR OWN ROAD IN LIFE INSTEAD OF TRAVELING DOWN SOMEONE ELSE’S. LISTEN TO YOURSELF. YOUR ROAD IS THE OPEN ROAD. FIND IT. FIND THE OPEN ROAD. Registration No. 3,719,214. Expanding failure to function in this manner can thus capture a variety of marks that, in spite or perhaps because of their distinctiveness, consumers are unlikely to actually use in a trademark way.

Two main objections might be raised to the introduction of linguistic failure to function. First, some might respond that this new basis for refusal invites a “know it when you see it” approach,132×132. Mark P. McKenna, Trademark Use Rides Again, 104 Iowa L. Rev. Online 105, 106 (2020). which poses the risk of inconsistent or arbitrary registration decisions based on the examiners’ own instincts about the marks before them. As failure to function necessarily involves imagining how consumers will receive a mark, this concern may be especially acute. Though the affirmative defense discussed in the following paragraph will hopefully minimize some of these concerns, it is the case that linguistic failure to function gives examiners a fairly flexible standard with which to assess marks. But much of trademark registration — and, indeed, much of trademark law — involves decisionmakers applying standards in highly context-specific and case-by-case ways and imagining how consumers would perceive a given mark.133×133. See McKenna, supra note 90, at 814. Currently, our trademark system trusts examiners and courts to determine where a mark falls on the Abercrombie spectrum, whether a descriptive mark has acquired secondary meaning, whether a mark is confusingly similar to one already registered, and whether consumers will perceive a mark as informational or ornamental rather than source indicating. These assessments often involve close calls, with similar cases sometimes coming out differently based on the decisionmaker’s analysis of the precise facts before her. To the extent that linguistic failure to function relies on examiners to perform a fact-specific and discretionary analysis about likely consumer perception, it is no different from many other grounds for registration refusal. Indeed, linguistic failure to function may pose less of a risk of arbitrariness than other inquiries, in light of insights from linguistics and psychology indicating that English speakers have a shared reaction to which words are linguistically “possible” and which are not.134×134. See Cohn, supra note 35, at 4.

The second possible concern about linguistic failure to function is that it will block marks that look like meaningless gibberish but that consumers actually register and recognize as source indicators. KUWTK, for example, a mark applied for in association with a variety of goods,135×135. U.S. Trademark Application Serial Nos. 88/435,498 (filed May 17, 2019); 88/435,506 (filed May 17, 2019); and 88/435,514 (filed May 17, 2019). may seem to the uninitiated to be as arbitrary as ZGGCD. However, it is in fact the well-known acronym for the television show Keeping Up with the Kardashians, and, as such, it seems likely that potential consumers would be able to remember the mark and use it to identify and distinguish the source of the goods sold under it. Because of marks like KUWTK, linguistic failure to function should include a defense that applicants can invoke in response to allegations that they are seeking to register a nonsense mark. If an applicant can provide evidence that a meaningful number of consumers actually recognize and use the mark across time and understand it to mean something specific and consistent, this should serve as an indication that the mark does in fact work as a mark and should not be refused registration on this basis. Relevant indicia of trademark meaning might resemble those used for descriptive marks with claimed secondary meaning, including the nature and amount of advertising, the volume of sales, consumer statements and survey evidence, and unsolicited media coverage.136×136. See 2 McCarthy, supra note 36, §§ 15:30, :70. Applicants might also point to specific reasons for choosing the mark or to its meaning in another language. This sort of defense is not foreign to failure to function doctrine. In the context of telephone numbers and ornamental phrases, which fail to function as marks in most cases, the USPTO has indicated that registration can be granted when the applicant provides evidence indicating that consumers understand the proposed mark to be source identifying.137×137. See Disorderly Kids, LLC v. Atwood, 2018 WL 6120080, at *7 (T.T.A.B. Nov. 20, 2018); In re Michelin N. Am., Inc., 2015 WL 5675633, at *4 (T.T.A.B. Sept. 11, 2015). But see In re Wal-Mart Stores, Inc., 2019 WL 193990, at *12 (T.T.A.B. Jan. 11, 2019). Where the obstacle to registration is the proposed mark’s nonstickiness and the associated trademark problems that poses, applicants should similarly be allowed to offer evidence that the mark is memorable and therefore likely to be relied on as a source signifier.

As with all bases for refusal, linguistic failure to function will likely present close, difficult cases. However, if applied thoughtfully, it will also offer an effective, tailored tool for keeping off the Register most nonsense marks, which clearly do not work as trademarks should but which current trademark law is ill-equipped to handle.

V. Conclusion

Nonsense marks, the product of a changed landscape for consumer goods and of incentives created by Amazon, at once pose new challenges to the U.S. trademark system and contribute to the larger problem of an increasingly overburdened Register.138×138. For work on the many strains the Register is under and how trademark law might adapt in response, see Beebe & Fromer, supra note 6; Beebe & Fromer, supra note 22; and Tushnet, supra note 2, at 916–40. Fortunately, trademark’s failure to function doctrine, in its focus on whether seemingly valid marks actually perform the work of trademarks, provides a path to refuse registration to nonsense marks. By extending the principles behind failure to function and recognizing linguistic failure to function as a ground for refusal, trademark examiners can capture these marks and articulate why they should not be registered. Refusals on the grounds of linguistic failure to function may in turn demonstrate how a doctrinal focus on the impressions of the public receiving a mark might limit the costs trademark registration can impose on the USPTO, competitors, and the trademark system.139×139. In this regard, the application of failure to function that this Note advocates is in line with proposals from Tushnet, McKenna, and Lemley, all of whom argue that a more precise analysis of likely consumer behavior within trademark law might limit the anticompetitive effects of ever-expanding theories of infringement. See Mark A. Lemley & Mark McKenna, Irrelevant Confusion, 62 Stan. L. Rev. 413, 427–44 (2010); Rebecca Tushnet, What’s the Harm of Trademark Infringement?, 49 Akron L. Rev. 627, 643–44 (2016).