Amazon's Quiet Overhaul of the Trademark System

Amazon's policies also create incentives for businesses to adopt different kinds of marks. Specifically, businesses are more likely to claim descriptive or generic terms, advantageously in stylized form or with accompanying images, and to game the scope limitations that would ordinarily attend registration of those marks. And the same Amazon policies have given rise to the phenomenon of "nonsense marks," which are strings of letters and numbers unrecognizable as words or symbols. In the midst of these systemic changes, Amazon has consolidated its own branding practices, focusing on a few core brands and expanding its use of those marks across a wide range of products. In combination, Amazon's business model and Brand Registry have overhauled the American trademark system, and they have done so with very little public recognition of the consequences of Amazon's business approach.

Amazon's impact raises profound questions for trademark law and for law more generally. There have been powerful players before, and other situations in which private dispute resolution procedures have affected parties' behavior. But Amazon's effect on the legal system is unprecedented in scale and scope. What does-or shouldit mean that one private party can so significantly affect a legal system? Do we want the trademark system to have to continually adapt to Amazon's rules? If not, how can the law disable Amazon from having such a profound impact? In this regard, we explore the ways in which Amazon's practices affect competition, harm the trademark system, and reshape how we think about trademark law at its foundation.

Amazon's dominance as a platform is widely documented. 1 Congress held hearings focused on Amazon for more than a year, 2 the Federal Trade Commission (FTC) recently sued Amazon for unfair competition, 3 and scholars and commentators (including Amazon critic Lina Khan, Chair of the FTC during the Biden Administration) debate the effects of this dominance on consumer welfare. 4 But one aspect of that dominance has not received sufficient attention-the Amazon Brand Registry's sweeping influence on firm behavior, particularly in relation to the formal trademark system. Amazon's Brand Registry serves as a shadow trademark system 5 that dramatically affects [Vol. 113:1169 businesses' incentives to seek registration of their marks and to choose certain types of marks to designate their goods.

The Brand Registry is, most basically, a private dispute resolution system that allows mark owners to object to unauthorized uses of their marks on Amazon without having to file formal legal proceedings. 6 What makes Amazon's system different than other private dispute resolution systems is the extent to which it influences parties' behavior within the legal system itself. 7 The Brand Registry not only gives parties a cheaper and more efficient way to resolve disputes, but it also creates incentives for parties to use the trademark system differently than they otherwise would, and in ways that were not anticipated in the design of that system.

Most directly, the Brand Registry has changed the incentives to register trademarks. The American trademark system is use-based: Trademark rights arise out of use, not registration, and registration in the U.S. Patent and Trademark Office (PTO) simply records those rights and provides certain enforcement benefits. 8 But because unregistered marks have long been enforceable under federal law on largely the same terms as registered marks, 9 unregistered rights have often been perfectly adequate for many smaller businesses. Indeed, the availability of those unregistered rights has traditionally been seen as a benefit of the American system for small and medium-sized businesses. 10 Amazon's policies have shifted that calculus because parties can participate in the Brand Registry in the United States only if their mark is registered with the PTO-or, as of very recently, if their application to register is pending. 11 Given Amazon's dominance as an online shopping platform, small and mediumsized businesses feel compelled to sell on the platform, and registration in the PTO is the ticket to the meaningful enforcement and search benefits provided to participants in the Brand Registry. 12 Businesses therefore have strong incentives to register marks when they previously would have relied on unregistered rights. 13 The result has been a dramatic increase in the number of applications to register, which has swamped the PTO and created delays for all applicants, including those that would have previously registered their marks. 14 Our data show that annual PTO applications estimated to originate with small businesses have approximately doubled since Amazon's Brand Registry began, from about one hundred thousand to two hundred thousand, increasing the proportion of filings from these entities from about 30 percent to about 40 percent annually. 15 In response to the delays that Amazon's policies helped create, Amazon has recently started qualifying parties for the Brand Registry based only on a pending application. 16 This change enables parties to privately enforce marks that might ultimately be rejected by the PTO and is likely to increase applications even more, creating further PTO delays.

By conditioning participation in the Brand Registry on PTO registration or at least an application to register, Amazon has created opportunities for pirates to extort sellers that operate on the platform using unregistered marks. 17 Like old-fashioned cybersquatters, these bad actors apply to register the unregistered marks in their own names and then seek to extort the true owners of those marks by threatening the owners with exclusion from Amazon's platform.

Not only does the Brand Registry increase incentives to register marks for which the owners would previously have relied on unregistered rights, but it also creates incentives for businesses to adopt different kinds of marks. Descriptive terms (like NATIONAL CAR RENTAL for nationwide car rental services) and generic terms (such as APPLE for apples) are especially more valuable in light of Amazon's policies. 18 Descriptive words are normally not protectable or registrable without evidence that consumers actually associate the term with a single source (what trademark law calls "secondary meaning"), and this additional proof requirement is supposed to be a deterrent to claiming descriptive terms. 19 Generic terms are categorically excluded from protection, even if they have secondary meaning. 20 Despite these rules, there has always been some incentive to claim descriptive and generic terms as trademarks because control over those terms can provide meaningful competitive benefits. 21 But with Amazon, there is an overwhelming incentive to control descriptive or generic terms because consumers often use those terms to search on Amazon.

14. See infra Part III.A. 15. See infra Part III.A. We do not suggest that all this increase is attributable to Amazon alone. As we discuss in Part The Brand Registry's structure also enables parties to game the trademark registration system and effectively get the full benefits of exclusive rights in descriptive and maybe even generic terms. 22 Of particular relevance, parties can avoid descriptiveness and genericness rejections in the PTO by applying to register those terms in a stylized format (such as a particular font) or with an accompanying image, sometimes disclaiming rights in the descriptive or generic word(s) themselves. In those cases, trademark law supposedly limits the scope of rights accorded to the registered mark to reflect the stylization or accompanying image. 23 But while Amazon is capable of considering the format in which a mark is presented, it does not require visual matching: A simple text match is sufficient to trigger exclusion. 24 That means a highly descriptive term, and maybe even a generic one, might be deemed registrable in the PTO because of its stylization but then enforced on Amazon as if it were a registration of the unprotectable word itself. That gaming disrupts the balance that the formal legal system tries to strike between recognition of the source-identifying capacity of design features on the one hand and the need for competitive access to descriptive and generic terms on the other, ignoring the reasons why those marks get any protection at all. 25 The Brand Registry also creates greater incentive to claim so-called "nonsense marks," which are strings of letters or numbers that are not comprehensible as words. For example, ELXXROONM, SUJIOWJNP, XUFFBV, and LXCJZY are all nonsense marks parties have recently applied to register in the PTO. 26 Indeed, the PTO data suggest a tremendous increase in filings for nonsense marks in the past few years, from almost none to over twenty thousand applications annually (0.5 percent of annual filings to approximately 4.5 percent). 27 Nonsense marks are currently easy to register as trademarks because they appear not to provide any information about the goods or services with which they are used, making them inherently distinctive and thus immediately protectable. 28 But those "marks" pose significant conceptual problems for trademark law, which presumes that parties are claiming terms that will have some meaning to consumers. 29 That is a sensible assumption in a world where consumers search and buy by brand. But when algorithms do the searching, businesses just need something that the algorithm can use to preference them. Nonsense will do. 30 Critics of excessive branding might rejoice about that de-centering and potential democratization of the online marketplace. But there is irony here: Amazon's de-centering third-party branding will likely amplify Amazon's own power by making its search function and algorithm even more important in finding products. And it certainly enhances the value of Amazon's own branding strategies, as reflected in the massive expansion of products sold under the Amazon Basics and Amazon Essentials brands. 31 Amazon controls the platform and can preference its own products in search results based on product descriptor keywords, making its house brand more important than product line brands. To take just one example, a search for "Hanes T-shirt" returns an Amazon Essentials T-shirt as the first result, followed by several Hanes results. 32 In all of these ways, actors affected by Amazon's business model and Brand Registry have overhauled central aspects of the trademark system in ways that are potentially troublesome. The Brand Registry has increased incentives to register and to register different types of marks, putting pressure on several substantive validity doctrines and forcing the PTO to deal with a huge influx of applications that examiners cannot manage in a timely way. Moreover, this legal overhaul has happened with very little public recognition of the consequences of Amazon's business approach. 33 We do not claim that Amazon specifically intended to affect the trademark system in any of the ways we describe or even that Amazon has fully understood the extent of its impact. It seems very likely that Amazon adopted policies that it thought were sensible for its business and that it created the Brand Registry in part to address actual concerns about counterfeit goods and product liabilityand, not incidentally, to avoid regulation like it would face under Congress's proposed SHOP SAFE Act. 34 In that regard, it may be that Amazon's policies and practices have simply had the unintended, though not necessarily unforeseeable, consequences that we describe.

Regardless of Amazon's intent, its business model and Brand Registry raise profound questions for trademark law and law more generally. Amazon is not the first commercial powerhouse, nor is it the first to create a private dispute 33. Even when there has been any recognition of the consequences, it has generally been of a single aspect, typically the phenomenon of nonsense marks, rather than a comprehensive sense of Amazon's impact on the trademark system. See John Herrman, All Your Favorite Brands, From BSTOEM to ZGGCD, N.Y. TIMES (Feb. 11, 2020), https://www.nytimes.com/2020/02/11/style/amazon-trademark-copyright.html [http://perma.cc/MLU2-6DJS] (highlighting the nonsense marks on Amazon); Note, Fanciful Failures: Keeping Nonsense Marks off the Trademark Register, 134 HARV. L. REV. 1804 (2021) [hereinafter Fanciful Failures] (thinking through how the trademark system should handle nonsense marks).

resolution system. But Amazon's effect on the trademark system is unprecedented. One set of questions is institutional and structural. What does, or should, it mean that one private party can so significantly affect a legal system? Do we want the legal system to have to continually adapt to Amazon's rules? If not, how can the law disable Amazon from having such a profound impact? Another set of questions focuses more specifically on the overall effects of Amazon's policies on competition and on trademark law's normative commitments. As we describe, Amazon's model and its policies likely increase its own power vis-à-vis third-party brands, de-centering branding more generally. The net value of that shift may be in the eye of the beholder: It depends on how one weighs the potential benefits of search simplification and lower prices for consumers, as well as the ease of marketplace entry for third-party sellers, against the costs of Amazon's increased power over third-party sellers. Likewise, the benefits of a reduction in the power of brands depend on whether alternative search tools, particularly algorithms that focus on product information and consumer reviews, convey relevant information to consumers as effectively as trademarks. In the end, whether and how we should respond to Amazon's effects on the trademark system depends on whether we want the trademark system to demand that trademarks play their traditional role or whether instead the facts on the ground have changed so much that the premises of that system no longer hold.

Many legal scholars have considered whether and how law might have to adjust to new platforms, asking, among other things, whether Uber drivers should be treated as employees and how to assess trademark liability for platforms. 35 These scholars, as exemplified by Julie Cohen, recognize that the platform is "the core organizational form of the emerging informational economy" and has replaced the more traditional marketplace as the locus for barter and exchange. 36 We extend this literature by demonstrating how a dominant platform like Amazon can spearhead an overhaul of a legal system, at least with the assistance of its third-party sellers. Khan seeks to reorient antitrust law for the platform era, suggesting that reorientation is actually a return to antitrust law's founding principles, by proposing that antitrust analysis focus on "the underlying Part I describes the conventional trademark system's aim and design. Part II turns to Amazon's business model and Brand Registry. Part III builds on these two parts by investigating how Amazon's practices have provoked the businesses selling wares there to change how they think about registering marks and the marks they choose, plus the trademark extortionists that have arisen in response to this ecosystem. Part IV considers whether potential precursors like Sears and Network Solutions provide useful analogies and concludes that Amazon's impact on the trademark system is unprecedented. That Part then discusses possible changes to PTO practice and Amazon policies that would address Amazon's impact. Finally, it considers the broader impact of Amazon's practices on trademark law and competition more broadly.

Scholars have long debated whose interests trademark law primarily serves. On one account, trademark law aims to protect businesses against illegitimate uses of their marks that would divert customers or confuse consumers about those businesses' relationship to the goods bearing their marks. 38 Other accounts, particularly in modern commentary, make consumer interests primary. Trademark law makes misleading uses of trademarks actionable so that consumers are not defrauded and can rely on marks to select goods from the sellers they wish to patronize. 39 structure and dynamics of markets" rather than consumer welfare measured through "short-term effects on price and output." Khan, Amazon's Antitrust Paradox, supra note 1, at 716-17. In particular, she argues that even though consumers generally love Amazon, its low prices, and broad availability of products, Amazon's "willingness to sustain losses and invest aggressively at the expense of profits and integration across multiple business lines" is problematic as a matter of antitrust. Id. at 746-47. Khan suggests that these features have caused numerous problems for competition, including in the e-book market, the delivery sector, and Amazon's direct competition with its third-party sellers. n all cases where rights to the exclusive use of a trade-mark are invaded, it is invariably held that the essence of the wrong consists in the sale of the goods of one manufacturer or vendor as those of another; and that it is only when this false representation is directly or indirectly made that the party who appeals to the court of equity can have relief. This is the doctrine of all the authorities.").

Here is a classic account from William Landes and Richard Posner:

Rather than reading the fine print on the package to determine whether the description matches his understanding of brand X, or investigating the attributes of all the different versions of the product (of which X is one brand) to determine which one is brand X, the consumer will find it much less costly to search by identifying the relevant trademark and purchasing the corresponding brand . . . . A trademark conveys information that allows the consumer to say to himself, "I need not investigate the attributes of the brand I am about to purchase because the trademark is a shorthand way of telling me that the attributes are the same as that of the brand I enjoyed earlier. Courts often present business and consumer interests as harmonious, so that trademark protection simultaneously advances both. As the Supreme Court has stated,

[T]rademark law, by preventing others from copying a sourceidentifying mark, "reduce[s] the customer's costs of shopping and making purchasing decisions" . . . for it quickly and easily assures a potential customer that this item-the item with this mark-is made by the same producer as other similarly marked items that he or she liked (or disliked) in the past. At the same time, the law helps assure a producer that it (and not an imitating competitor) will reap the financial, reputation-related rewards associated with a desirable product. 40 But regardless of how one prioritizes the respective interests, source indication is the conceptual center of every serious theoretical justification of trademark law. All of the benefits of trademark protection depend on a mark's capacity to identify goods or services as emanating from a single source, and all of the harms trademark law targets result from interference with that sourceindicating function.

Given that focus on source indication, it is no surprise that modern doctrine defines trademark subject matter fundamentally in terms of capacity to indicate source. According to the Lanham Act, a trademark is "any word, name, symbol, or device, or any combination thereof" that is used "to identify and distinguish his or her goods . . . from those manufactured or sold by others and to indicate the source of the goods." 41 Because "human beings might use as a 'symbol' or 'device' almost anything at all that is capable of carrying meaning," as the Supreme Court noted, trademark subject matter is broad and capacious, including colors, product packaging, and even the design of products, to the extent they identify source. 42 The following sections elaborate, respectively, on trademark's distinctiveness requirement, the requirement of trademark use, and the registration process of the trademark system.

Trademark law calls source designation "distinctiveness," and that concept is the foundation of protectability. 43 Consumers are unlikely to be confused about the source of a product or service if they do not recognize a particular designation as source-indicating in the first place. 44 Relatedly, consumers are only able to use marks to reduce their search costs-the costs of identifying goods or services with the characteristics they want-if those marks reliably indicate who is responsible for the goods or services they are used with. 45 Moreover, from a business's perspective, if consumers know that a term or symbol identifies the business as the source of goods or services, the business will be encouraged to invest in the consistent quality of its goods or services, an important goal of trademark law. 46 In addition to the benefits of requiring distinctiveness, there is a cost to granting trademark rights to words or symbols that do not identify source. The principal worry is that protecting those words or symbols would inefficiently prevent other businesses from using them in competitively useful ways. 47 The framework to assess distinctiveness and thus protectability in trademark law is set out most famously in Abercrombie & Fitch Co. v. Hunting World, Inc., a 1976 Second Circuit decision authored by Judge Friendly 48 that was widely relied upon by other courts, including the Supreme Court. 49 Abercrombie identified five different categories of terms with respect to trademark protection: "Arrayed in an ascending order which roughly reflects their eligibility to trademark status and the degree of protection accorded, these classes are (1) generic, (2) descriptive, (3) suggestive, and (4) arbitrary or [(5)] fanciful." 50 As explained by the Second Circuit, a term is categorized based on the amount of information it supplies about the goods or services with which the term is being used. 51 As per Abercrombie, a "generic" term "refers, or has come to be understood as referring, to the genus of which the particular product is a species." 52 For example, IVORY would be generic for products made from elephant tusks, but not for soap products. 53 Generic terms are never protectable as trademarks, even if they develop secondary meaning. 54 As courts routinely say, competitors should have the absolute right to call their goods or services by their category name; if they could not, they would be unable to compete effectively and consumers would ultimately be hurt by confusion and illegitimate restriction of competition. 55 "Descriptive" terms are presumptively unprotectable, but they can earn their way into trademark status. As the Second Circuit has explained, a descriptive term "describe[s] a product or its attributes." 56 These are terms like HOLIDAY INN for inns in which people stay while on holiday, ALL BRAN for all-bran cereal, and AMERICAN GIRL for American girl dolls. 57 To the Seventh Circuit, "[a] descriptive mark is not a complete description, . . . but it picks out a product characteristic that figures prominently in the consumer's decision whether to buy the product or service in question." 58 Similar to generic terms, the fear with protecting descriptive marks is that competitors might want to use a term because it describes their product too, and those competitors would be unfairly disadvantaged if one business in the competitive landscape has exclusive rights in such a term. 59 Descriptive terms are not inherently distinctive because they give direct information about the nature or characteristics of the goods or services they are used with and therefore do not automatically signify source. 60 Only when consumers have to come to understand a descriptive term primarily as a source indicator, and not just as a descriptor of the products or services, can it be protected as a trademark. 61 The upshot of that rule is that descriptive terms cannot be protected immediately upon use; it takes time (and usually substantial advertising) for them to become trademarks. 62 As Judge Friendly explained in Abercrombie, in allowing protection for descriptive marks that have acquired secondary meaning, trademark law "strikes the balance . . . between the hardships to a competitor in hampering the use of an appropriate word and those to the owner who, having invested money and energy to endow a word with the good will adhering to his enterprise, would be deprived of the fruits of his efforts." 63 Even when descriptive terms develop secondary meaning, trademark law recognizes that competitors might still need to use the terms in their primary, descriptive sense. The descriptive fair use defense permits a business to use a competitor's protected descriptive mark, not as a mark, but "fairly and in good faith only to describe the [business's] goods or services." 64 For example, if Delta Airlines described itself patriotically as "an American airline," that might be permissible as a descriptive fair use of the AMERICAN AIRLINES mark, as long as Delta was using "American" in a non-trademark way. The defense is an important one, 65 but it is relatively narrow because it does not allow others to use the term as a mark-including a domain name or slogan-even if that term also describes the defendant's goods or services. 66 Lisa Ramsey documents that the defense is murky, and "[r]elevant factors for determining whether a use is a trademark or descriptive use include the size, style, location, and prominence of the descriptive term in comparison to the defendant's use of its own trademark or other descriptive matter in advertising or product packaging." 67 "Suggestive," "arbitrary," and "fanciful" terms are considered inherently distinctive and therefore protectable without proof of secondary meaning. 68 As explained by the Fifth Circuit, a "suggestive" term "suggests, rather than describes, some particular characteristic of the goods or services to which it applies and requires the consumer to exercise the imagination in order to draw a 63. Abercrombie, 537 F.2d at 10. 64. 15 U.S.C. § 1115(b) (4); see also KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 543 U.S. 111, 124 (2004) (holding that a defendant raising such a defense need not demonstrate that consumers will not be confused by the use).

65. See Deborah R. Gerhardt, A Masterclass in Trademark's Descriptive Fair Use Defense, 52 AKRON L. REV. 739, 740 (2019) (explaining how this defense "involves careful balancing of a mark owner's right to protect its reputation against the rights of competitors to compete fairly and effectively so that consumers will have access to accurate, informative speech that is not bounded by monopolies on descriptive words").

66 conclusion as to the nature of the goods and services." 69 For example, courts have found SWAP for a watch with interchangeable parts, 70 5 HR ENERGY for an energy drink, 71 and GLASS DOCTOR for glass installation and repair services all to be suggestive. 72 "Arbitrary" terms are preexisting words that are used in "an unfamiliar way" that is conceptually unrelated to the category of goods or services at hand. 73 Examples of marks courts have classified as arbitrary are STARBUCKS for coffee, 74 VEUVE (meaning "widow" in French) for champagne, 75 and KIRBY for vacuum cleaners. 76 "Fanciful" terms are "invented solely for their use as trademarks." 77 Courts have deemed fanciful marks to include CARSONITE for highway markers 78 and LUMAR for fabric softener. 79 Suggestive, arbitrary, and fanciful terms are immediately protectable because they do not provide any direct information about the goods or services and are therefore assumed automatically to convey source information. Protection of those terms also does not "depriv[e] others of a means of describing their products to the market." 80 Thus, a claimed mark's protectability depends on its distinctiveness, which is determined according to Abercrombie's classification scheme. As discussed in the next Section, a claimed mark's protectability also depends on the way the claimant uses the purported mark.

Eligibility for trademark status depends not only on a sign's capacity to identify source, but also on evidence that the claimant has actually used that sign in a particular trademark way. 81 "Use" in this sense has special meaning: It has a quantitative dimension (some, rather than none), and it also connotes particular functional characteristics, namely that the sign as used has the effect of identifying source. 82 As the Supreme Court recently explained, a trademark "identifies a product's source (this is a Nike) and distinguishes that source from others (not any other sneaker brand)." 83 Indeed, to the Court, "whatever else it may do, a trademark is not a trademark unless it" performs this sourceidentifying function, "tell[ing] the public who is responsible for a product." 84 It is easy enough to state the principle that a sign must be used as a trademark to qualify for protection; it is much harder in practice to say what that means. In theory, the ultimate issue is whether consumers regard the mark as used as identifying source, which suggests that trademark use is an empirical question. 85 But because it is often impractical to assess consumer understanding directly, courts have long used proxies to make that judgment, focusing on the prominence and location of a use; the consistency of presentation of the sign, stylization, and coloration; and the presence or absence of other identifying signs. 86 In one recent case, for example, the court concluded that Jaymo's use of "S'Awesome" for food sauces, as shown in Figure 1, was not trademark use. 87 The court explained that despite the consistency of Jaymo's use, "S'Awesome blend[ed] into a smattering of text on the label style used on most of Jaymo's sauces," and that "placement and emphasis on other terms coupled with the comparatively small, plain font of the term fail[ed] to adequately demonstrate it [was] being used as a source indicator on the bottle labels." 88 The PTO considers whether a sign functions as a mark when considering a claimed mark's registrability. In that context, the rule operates in the negative. A claimed mark is not registrable when it fails to function as a mark. The PTO's Trademark Manual of Examining Procedure describes a purported mark's failure to function primarily by reference to other functions performed by the claimed matter. A claimed mark might fail to function as a mark, for example, "because it is merely ornamentation," 89 it constitutes "informational matter," 90 or is a "model or grade designation." 91 Especially when evaluating whether a claimed mark is merely ornamentation, the Trademark Office puts significant weight on the location of the claimed mark, as shown in the specimen filed with the PTO showing the mark's use, presuming that signs in "trademark spaces" function as marks. 92 To illustrate, the PTO rejected an application to register the "Shorebilly" word and design mark shown in Figure 2 In sum, while the principle that a sign must be used as a trademark to qualify for protection is straightforward in theory, its application often hinges on nuanced and context-specific factors, such as presentation, placement, and consumer perception. This complexity underscores the pivotal role of both judicial proxies and PTO guidelines in evaluating whether a claimed mark functions as a source identifier, setting the stage for further exploration of the trademark registration process.

Signs that satisfy trademark law's legal requirements, including distinctiveness and use as a mark, are eligible to be registered with the PTO. 95 But registration is not central to American trademark law, or at least historically it has not been. 96 That is not to say that registration is irrelevant; indeed, registration has several important legal benefits. For one thing, registration confers nationwide priority as of the date of application, subject to uses that predate the registrant's use or application to register. 99 Unregistered marks, by contrast, are protected only in the geographic areas in which they were in use prior to the allegedly infringing use. 100 Registered marks are also presumed valid and owned by the registrant, 101 and certain of those presumptions become irrebuttable if the registration becomes incontestable. 102 Moreover, registered marks are subject to customs enforcement, 103 and some enhanced remedies are only available for registered marks. 104 Under the Lanham Act, then, registration serves as a carrot rather than a stick. Some of the legal incentives to register are more significant for certain types of businesses. There are meaningful incentives to register for businesses that anticipate significant geographic expansion, as nationwide constructive use effectively secures rights across the country, even in advance of actual use in many places. 105 Incontestability is particularly valuable for marks-like descriptive terms, geographic terms, surnames, and product design-that owe their existence to secondary meaning, 106 because marks that are incontestably registered cannot be challenged for lack of secondary meaning. 107 Customs enforcement is valuable for parties that sell on a scale that is likely to attract widespread copying. And registration can be extremely helpful for parties with international aspirations because various treaty provisions make it easier to secure foreign trademark rights when a party has registered in the United States. 108. Section 44 of the Lanham Act, for example, allows a foreign trademark owner to file a U.S. trademark application and claim priority based on a foreign trademark registration that was filed within six months prior to the U.S. application. 15 U.S.C. § 1126. That treatment is available only to applicants whose country of origin is a party to a convention or treaty that provides reciprocal rights to U.S. nationals. Id. § 1126(b). U.S. nationals seeking registration in any country granting such reciprocity therefore benefit from the U.S. filing date.

Beyond these legal encouragements, practical considerations also might counsel in favor of registration. Registered marks are more easily findable, particularly in the PTO's publicly-available database, and public notice of claims to those registered marks helps avoid conflict to the extent it prompts others to avoid using similar marks. 109 Registration also enables the PTO to refuse thirdparty trademark applications for confusingly-similar marks. 110 At the same time, there are also advantages to unregistered rights. Most obviously, unregistered rights are free: The rights attach naturally as soon as the party starts using the mark to indicate the source of its goods. That means that less-resourced parties can develop rights without having to spend time or money filing trademark applications. 111 Registration favors large, sophisticated companies, which generally are familiar with the registration system and have the resources to seek registration of each new potential trademark. 112 Given the lack of expense associated with unregistered rights and the ability to enforce those rights under federal law on substantially the same terms as applicable to registered marks, unregistered rights are perfectly adequate, and may be superior, for many trademark owners. 113 Even sophisticated parties might have reason to prefer unregistered rights for marks that are less likely to be durable branding elements over time. A party that goes to the expense of registering a mark has some incentive to stick with that mark over longer periods of time, whereas unregistered rights are better suited to marks that might be adapted or used in connection with different goods or services in the future. 114 Trademark registration, while historically secondary to the use-based foundation of American trademark law, offers meaningful legal and practical 109. Gerhardt & Lee, supra note 105, at 874-75. 110. Id. The statute empowers the PTO to deny registration to any mark that is likely to cause confusion with any mark previously registered or in use, 15 U.S.C. § 1052(d), but the PTO proactively searches only registered marks and pending applications during ex parte examination. TMEP, supra note 89, § 1207.01. Owners of marks previously in use must establish their priority in an opposition proceeding. See id. § 1503.01.

111. PTO trademark application filing fees range from $250-$750 per class of goods and services, with additional fees in the three figures for special circumstances per class. U.S. PAT. & TRADEMARK OFF., USPTO FEE SCHEDULE (2025), https://www.uspto.gov/learning-and-resources/fees-and-payment/uspto-fee-schedule [https://perma.cc/AS6U-9MUQ]. In a run-of-the-mill case, legal representation for trademark registration runs in the three figures on the low end but often costs four figures or more. s] businesses to think through their designs and how they intend to commercialize and market those designs," and acts as "an incentive to articulate claims that correspond to their market intentions, which they might not otherwise have thought through as thoroughly at that stage").

advantages that make it a valuable tool for many businesses. However, the cost and complexity of registration often favor larger, resource-rich entities, leaving unregistered rights as a viable and often preferred option for smaller or more flexible businesses. With this understanding of trademark law's framework, we now shift focus to Amazon's business model and its Brand Registry.

Since Amazon launched, it has not only grown what is perhaps the most vibrant online commerce platform, with 9.7 million third-party businesses selling goods on Amazon, 115 but it has also created a brand that is valued at over $576.6 billion. 116 To better understand how Amazon's business practices have upended central aspects of the conventional trademark system, we must first examine how Amazon has achieved such market dominance. Part II.A provides background on Amazon's business model and its evolution, and Part II.B turns to Amazon's Brand Registry and its role in Amazon's business model.

Amazon was founded in 1995 as an online bookseller, and it has since evolved into a pervasive e-commerce platform and then some. 117 Indeed, a recent in-depth cultural study of Amazon describes it as the most ubiquitous company in history: "the 'everything' brand for 'everyone.'" 118 It is the biggest online retailer in the United States, controlling an estimated half of online retail sales. 119 And its reach is global: Amazon serves customers in nearly two hundred countries. 120 From the start, Amazon has had grand ambitions, as evidenced by early marketing materials drawing on its trademark: 121 "Amazon.com's name pays homage to the Amazon River. Just as the Amazon River is more than six times the size of the next largest river in the world, Amazon.com's catalog is more than six times the size of the largest conventional bookstore." 122 Even with these aspirations, founder Jeff Bezos always intended for Amazon to be much more than the largest online bookstore. He began selling books only after considering twenty product categories, with the books as the entry point to, as communications scholar Emily West puts it, ultimately "build[ing] a mammoth e[-]commerce website." 123 According to Bezos himself, "we're not trying to be a book company or trying to be a music company-we're trying to be a customer company." 124 Amazon's business strategies have generally been in service of this overarching goal of attracting loyal customers and distributing to them, rather than just selling books. In particular, Amazon has sold books and other products at very low prices as loss leaders to attract customers, which has also led to accusations of predatory pricing. 125 In doing so, Amazon has demonstrated its willingness to delay profits to build up its customer base, all the while drawing consumers away from its competitors. 126 Indeed, Amazon only became consistently profitable in 2015. 127 More generally, especially in its early years, Amazon did not spend much on traditional advertising and marketing but instead spent its money improving the platform's customer experience, including its unprecedented fast, free shipping that ultimately became the central feature of its popular Prime membership service. 128 To broaden its product base and attract yet more customers, since 1999 Amazon has allowed third parties to sell their products on the Amazon platform. 129 Amazon now has almost ten million third-party sellers on its 121. The company considered several other names, including Cadabra, Awake.com, Bookmall.com, Aard.com, and Relentless.com, the last of which still redirects to Amazon's website. BRAD STONE, THE EVERYTHING STORE: JEFF BEZOS AND THE AGE OF AMAZON 31 (2013).

122. WEST, supra note 118, at 172. 123. Id. at 87. Though Amazon has expanded well beyond books and now sells just about everything, books are still Amazon's largest product category (16 percent of items sold). Id. at 83. Moreover, Amazon is by far the largest retailer of physical books and e-books in the United States. Id.

124. Id. at 6; cf. STONE, supra note 121, at 24 (referring to Jeff Bezos's 1994 plans with David Shaw for an "everything store").

125. WEST, supra note 118, at 87-88; Khan, Amazon's Antitrust Paradox, supra note 1, at 753 ("The fact that Amazon has been willing to forego profits for growth undercuts a central premise of contemporary predatory pricing doctrine, which assumes that predation is irrational precisely because firms prioritize profits over growth. In this way, Amazon's strategy has enabled it to use predatory pricing tactics without triggering the scrutiny of predatory pricing laws.").

126 platform, 130 which has created network effects to lure consumers, which in turn attracts more sellers, ad infinitum. 131 Amazon's offering of one-click ordering (and the resulting patent it obtained on it) typifies how the platform has sought to provide extreme convenience for consumers. 132 Since its launch, Amazon has sought to gain consumer trust by collecting and sharing consumer reviews of the products it sells. Initially, competitors and experts scoffed at that practice, believing it would be counterproductive because consumers would at least sometimes leave bad reviews. 133 But Amazon seems to have won that bet: Its collection of reviews, one of the world's largest, has fostered a "reputation economy" and propelled Amazon's market dominance. 134 Amazon also collects reams of data about consumer behavior in service of developing its predictive models. 135 It uses these models to continually adapt its platform and product offerings, which encourages consumers to keep using Amazon. 136 All of these strategies are reflected in the logo that Amazon redesigned in 2000. 137 The logo has an arrow going from the 'a' to 'z' in AMAZON, to suggest that all products from A-Z can be found and bought on the platform. 138 And the 130. See supra text accompanying note 115. 131. WEST, supra note 118, at 9, 34. Relatedly, Amazon began to extract higher margins from the third-party transactions completed on its platform than from its own sales. arrow also suggests that Amazon brings products from all locations and sellers to consumers' homes. 139 As a result of these unique strategies and practices, Amazon's business model has been a smashing success. 140 After Walmart, Amazon is the second largest retailer in the United States, with $355.1 billion in sales in 2023. 141 And it is by far the largest online retailer in the United States, controlling about half of the online retail market. 142 Third-party sales have become a critical part of Amazon's business model. Indeed, the money Amazon makes from charging third parties to use its platform, such as listing fees, represented 23 percent of Amazon's revenues in 2022 ($117.7 billion), second only to the 43 percent of Amazon's revenues generated in first-party sales that year ($220 billion). 143 As important as the third-party sellers are to Amazon, Amazon is even more essential to the third-party sellers. As one seller pointed out, "If you say no to Amazon, you're closing the door on tons of sales." 144 Amazon's market dominance is also reflected in the great degree of public confidence in Amazon. Amazon is at or near the top of the list of most-loved brands in the United States. 145 In fact, one recent poll done by Georgetown University found that Americans trust Amazon more than any institution except the military, ranking the company above all other parts of the U.S. government and above universities, non-profit institutions, and major businesses. 146 And Amazon is also one of the world's most market-capitalized companies. 147 With this exploration of Amazon's business model generally, we now turn to how Amazon has deployed its Brand Registry to advance its business goals.

As Amazon sought to advance its overall business model, it encountered concerns from the third-party sellers and consumers (both of which were essential to attract to and keep on its platform), as well as from the government.

Third-party sellers-both potential and actual-wanted Amazon to do more to prevent counterfeit versions of their goods from appearing on its platform. 148 Those sellers expressed unwillingness to sell their genuine goods on Amazon unless Amazon took further action to exclude counterfeits. 149 And consumers were unhappy when they accidentally purchased knockoffs instead of the genuine goods they were trying to buy. 150 As complaints mounted, Congress held hearings on counterfeit goods being sold on online platforms like Amazon. 151 Several legislators introduced the SHOP SAFE Act, which would make online platforms "liable for infringement of a registered trademark by a third-party seller of goods that implicate health and safety unless the platform takes certain actions." 152 Sellers' and consumers' anxiety about Amazon's platform, and the looming threat of regulation, posed real threats to Amazon's market dominance. 153 In response, Amazon launched the first version of its Brand Registry in 2015. 154 That fairly limited program allowed businesses to better control their own listings and to contest other listings on copyright grounds. 155 Yet the filing process under that program was time-consuming and cumbersome, 156 and the program did little to address counterfeit goods, which are principally targeted through trademark claims, not copyright claims. 157 In 2017, Amazon launched the second version of its Brand Registry. In addition to providing enhanced branding capabilities for businesses' own listings, the new Registry made it easier to remove listings of counterfeit goods. 158 Businesses could qualify for the Brand Registry in the United States if they had registered their trademark on the Principal Register of the PTO and were using that mark on their products or packaging. 159 Only trademarks that contain alphanumeric characters can be listed in the Brand Registry, though the marks can also be stylized or include an image. 160 The word(s) in the trademark must identically match the spelling, spacing, and punctuation found in the U.S. trademark registration. 161 With the launch of the second version of the Brand Registry, Amazon put in place a three-hundred-person customer service team dedicated to addressing reports of trademark and copyright infringement from Brand Registry members. 162 Amazon now promises round-the-clock service to address these reports, twenty-four hours a day, seven days a week. 163 Rather than taking days to address such seller claims, the team would resolve these claims within a few hours and without a court order. 164 The Brand Registry made it easier for registered businesses to identify potential infringements by providing search tools, including reverse-image search technology, to help locate other products using the same name or packaging as the registrant. 165 And the Registry enables mark owners to benefit from predictive protections that block improper listings from third parties in the first place. 166 The Brand Registry has also been attractive to third-party sellers because it offers them higher visibility in consumer search results on Amazon, brand analytic tools, and the ability to give one's products to credible buyers for Amazon reviews. 167 In 2019, Amazon made it even easier for sellers to protect their brands and qualify for the Brand Registry by launching the Intellectual Property Accelerator, which is a curated network of intellectual property law firms providing trademark registration services at pre-negotiated rates. 168 it "created [the] Accelerator specifically with small and medium businesses in mind," so as to help them "more quickly obtain intellectual property . . . rights and brand protection in Amazon's stores." 169 Businesses participating in the Accelerator get charged only by the law firm they are using, not Amazon. 170 Businesses that use the Accelerator program get "accelerated access to brand protection" on Amazon. 171 Rather than having to wait for their trademark registration to issue, Amazon provides Accelerator participants access to the Brand Registry as soon as they have filed a trademark application with the PTO. 172 Amazon says that it provides that early access because "the participating law firms have been thoroughly vetted," and the marks Accelerator participants apply to register will therefore "be strong candidates for registration." 173 Though Amazon does not make public all of the specific Brand Registry benefits that it provides on this accelerated basis, it has indicated that Accelerator participants get "automated brand protections, which proactively block bad listings from Amazon's stores, increased authority over product data in our store, and access to our Report a Violation tool, a powerful tool to search for and report bad listings that have made it past our automated protections." 174 More recently, as of approximately 2023, Amazon made all sellers, not just those using the Accelerator program, eligible for its Brand Registry as soon as they have a pending application to register a trademark with the PTO. 175 Amazon has not publicly explained its reasons for that expanded eligibility, but it certainly calls into question the previous claim that Accelerator participants warranted early access because the marks they applied to register were particularly likely to be registered. 176 The new and improved Brand Registry has been a hit among third-party sellers. In 2021, there were more than seven hundred thousand active marks enrolled in the Brand Registry worldwide, a 40 percent increase over the previous year. 177 In 2022, more than sixteen thousand trademarks were the [Vol. 113:1169 subject of the Accelerator program. 178 Amazon advertises the successes of the Brand Registry and the Accelerator program in promoting its participants and removing infringing listings. 179 For example, Amazon boasts that it has blocked or removed 99 percent of listings suspected of offering counterfeit goods. 180 The growth and success of the Brand Registry have been noticed by businesses and business writers, who have written about the obvious advantages to being part of the Brand Registry. 181 Despite the general success of the Brand Registry from the perspective of many, some larger companies like Nike have not been as impressed with Amazon's anti-counterfeiting measures. 182 Nike began selling on Amazon only after Amazon created the Brand Registry, believing the Registry would help stop counterfeiting. But Nike later reversed course and stopped selling directly on the platform because it thought Amazon was still not sufficiently controlling counterfeit sales. 183 Not many companies can afford to take that position. According to Emily West, "Nike had confidence in the power of its brand to leave Amazon, but as one industry analyst put it, 'I don't think as many brands can be as selective as Nike.'" 184 Because most sellers do not have Nike's power and need to sell on Amazon, the Brand Registry is essential for them.

All in all, Amazon's Brand Registry undergirds the platform's business model by helping to keep the vast majority of third-party businesses comfortable and motivated to sell their wares on Amazon, which in turn keeps customers hooked on the platform. The Registry does so by making it easier for registrants to have infringing sellers removed from the site expeditiously, and by giving registrants superior search optimization tools. The resulting seller and consumer satisfaction removes some of the ongoing pressures for the government to regulate Amazon in this regard, such as through the SHOP SAFE Act, which would expose Amazon to significantly greater liability for selling counterfeit goods. 185

Third-party sellers' widespread participation in Amazon's Brand Registry has not only promoted Amazon's business model. As this Part addresses, it has also put significant hydraulic pressure on the U.S. trademark system, in effect overhauling the system and calling into question many of trademark law's foundational assumptions. 186 Because Amazon's Brand Registry is built on the U.S. trademark registration system-as opposed to being a system entirely of Amazon's creation-businesses have developed very different practices with regard to the selection and registration of trademarks. In this Part, we detail some of the most significant of these changed practices: small businesses' increased use of the trademark register, trademark extortion, registration of descriptive and generic marks, and registration of nonsense marks. We also discuss how Amazon's use of its own internal house brands fits into this story. These changes have happened relatively quietly without much public attention, but they have materially overhauled the trademark system.

Even though Amazon's Brand Registry might be seen as a shadow trademark system, the story here is not one about a community that relies primarily on norms rather than formal legal rules, such as those described by other legal scholars focusing on the fashion industry, cuisine, stand-up comedy, roller derby, tattoos, or magic. 187 Those situations are often described as involving "intellectual production without intellectual property" (or "IP without IP") or a "negative space." 188 Amazon's Brand Registry is different because it influences parties' use of the formal trademark system. Indeed, the story here is more like "IP plus IP" or an "exponential space."

As noted above, American trademark law has long protected unregistered marks on largely the same terms as registered marks, making registration a set of advantages rather than a requirement. 189 For many small and medium-sized businesses, those advantages were not significant enough to justify the time and expense of registration, which means that registration has traditionally been more the province of larger, established companies.

Amazon is changing that dynamic. Because the Brand Registry requires PTO registration, or now at least a pending application to register, 190 the incentives for small and medium-sized businesses selling on Amazon are very different. Because most smaller businesses want to be in the Brand Registry, they are much more likely to register than they once were. 191 PTO data suggest that the incentives to register are growing. 192 Specifically, as shown in Figure 5, the proportion of new applications filed by single filers (entities that have not previously filed other trademark applications) has risen substantially since about 2015, from approximately 30 percent to 40 percent annually. This rate increased most sharply after 2019. 193 193. This data includes applications filed based on an intent to use the mark rather than on current use of the mark. Applications based on an intent to use the mark require that a statement of actual use be filed before the registration is issued. 15 U.S.C. § 1051(d). Approximately 10 percent of annual filings by single filers are based on an intent to use the mark rather than on current use of the mark. applied to register ten or more marks (likely bigger companies) has correspondingly declined during this time. These data are supported by Figure 6, which shows that the gross number of applications by single filers has increased slowly from the 1980s through the early 2010s. The number has doubled since 2015, from about one hundred thousand to two hundred thousand.

it is hard to imagine that Amazon's policies were not substantial drivers. 194 There are now just under ten million third-party businesses selling on Amazon, 195 while there were over seven hundred thousand brands in the Brand Registry just in 2021. 196 In other words, at least 7 percent of third-party businesses selling on Amazon either have PTO trademark registrations or have pending trademark applications. Many of these trademarks probably were not registered in the PTO before their entry into the Brand Registry, either because the brands were new or because the businesses using them had made the rational, pre-Amazon Brand Registry choice not to register. 197 Approximately sixteen thousand new brands enrolled in Amazon's accelerator program in 2022-the trademark applications filed by those brands represent the ones least likely to have been filed but for the desire to be part of Amazon's Brand Registry. 198 It is true, of course, that Amazon has enabled many more small and medium-sized businesses to engage in interstate commerce, and that alone might explain some increase in applications to register. 199 But it seems clear that Amazon's Brand Registry is an extra push toward registration. In Figures 567, we can see a small increase in trademark registrations after Amazon enabled third-party selling on its platform in 1999. However, that increase pales in comparison to the jump in small-business trademark applications after the launch of Amazon's second version of the Brand Registry in 2017.

The increase in applications in the PTO has significantly increased examiner workload and lengthened the pendency of applications. The PTO now reports an average total application pendency of 14.4 months, compared to 9.6 months in the first quarter of 2021 (which was similar to prior years). 200 The average time to receive a first office action is 8.5 months, up from less than 5 months in the first quarter of 2021 (also similar to prior years). 201 As one expert recently said about the costs of the PTO's delays:

Until recently, the average time until a first office action . . . allowed business owners to file trademark applications for new products or ventures, and to obtain feedback and a 'read' on the position of the []PTO before the trademark was placed in commercial use. Now, the longer wait time before examination has been 'too long to wait', [sic] and has forced many businesses to move forward with commercial introductions without this initial feedback, and with more uncertainty about their trademark rights. 202 Likely as a response to the PTO delays and business complaints, Amazon has now made the Brand Registry available to its sellers based only on an application to register, a move that will presumably further increase the number of applications and delay registration even more. 203

Not only has Amazon's Brand Registry created incentives for legitimate small businesses to register marks they might not have felt the need to register in the past, but it has also created incentives for other parties to seek registration of unregistered trademarks that are used by others on Amazon. Why? Because if that registration is successful (and now perhaps as long as the application is pending), the registrant can threaten to invoke the Brand Registry against the prior user, the legitimate owner of the mark. This situation could lead to trademark extortion, with fraudulent registrants extracting payments from legitimate businesses that are trying to avoid having their businesses taken down by Amazon.

A trademark extortion scheme of this nature was recently at issue in a case in the Eastern District of New York. In that case, the district court granted a preliminary injunction, ruling that a New York-based plaintiff was likely to succeed in its claim seeking the cancellation of the China-based defendant's U.S. trademark registration because the defendant fraudulently used a photograph of the plaintiff's product as its specimen of use. 204 The plaintiff had been selling home furniture and organizers on Amazon for many years using the mark SAGANIZER, but it had never attempted to register that mark. 205 The defendant seized the opportunity, filing an application to register SAGANIZER and submitting a photo of one of the plaintiff's products as its specimen of use. 206 The PTO registered the mark in the name of the defendant on this basis. 207 The defendant then relied on its registration to complain to Amazon about the plaintiff's use of the SAGANIZER mark, and Amazon delisted some of the plaintiff's products. 208 Given the number of complaints by the defendant against the plaintiff's products on Amazon, the plaintiff was at imminent risk of being suspended on Amazon altogether. 209 The plaintiff alleged that this would destroy its business, given how focused its model was on Amazon sales, 210 as many small businesses are. This is not an isolated example of this form of trademark extortion. Presumably with the goal of maintaining the legitimacy of its Brand Registry, Amazon has recently filed multiple lawsuits, including some seeking cancellations of PTO registrations, against entities operating under the same fraudulent model as the SAGANIZER registrant. 211 According to Amazon, these entities fraudulently obtained PTO registrations of marks owned and used by businesses operating on Amazon and then used those registrations to join the Brand Registry. 212 The fraudulent registrants "then created fake, disposable websites, with product images scraped from the Amazon store, to use as false evidence when making thousands of claims that selling partners were violating their [intellectual property]." 213 Amazon alleged that one of the three entities filed almost four thousand takedown requests over a few months. 214 Amazon ultimately detected these entities' behavior, shut down their accounts, and sued them. 215 In response to incidents like these, Amazon has announced it is working with the PTO to prevent trademark fraud. 216 It claims to "directly receive[] and act[] upon information from the []PTO regarding registration status and parties that have been subject to []PTO sanctions" for fraudulent filings, which it uses to remove the fraudulent registrants from its Brand Registry. 217 Amazon claims to have removed five thousand false brands from its platform in this way. 218 In recent years, as Barton Beebe and one of us have demonstrated empirically, fraudulent trademark filings using fake specimens of use have become a significant problem at the PTO, even outside of the Amazon context. 219 This work estimates that "with respect to use-based applications originating in China that were filed at the . . . PTO[] in 2017 solely in Class 25 (apparel goods), . . . 66.9% of such applications included fraudulent specimens. Yet 59.8% of these fraudulent applications proceeded to publication, and 38.9% then proceeded to registration." 220 The extent of this fraud led Congress to pass the Trademark Modernization Act in 2020, which provided for new reexamination and expungement procedures to remove fraudulent marks from the trademark register. 221 There are many reasons for the substantial and increasing number of fraudulent trademark filings. Until now, those investigating the issue have suspected that a major reason for this fraud-often coming from applications originating in China-is that some regional Chinese governments have been offering their citizens a financial subsidy for each U.S. trademark registration secured. 222 This subsidy encourages Chinese citizens to file fraudulent PTO trademark applications, while avoiding the costs of operating an actual business. 223 But Amazon's business model and Brand Registry also have contributed to the rise in fraudulent PTO trademark applications. Shrewd operators can and have extorted legitimate businesses operating on Amazon by fraudulently registering their unregistered marks, sometimes even using these businesses' real specimens of use.

For the reasons described in Part III.A, Amazon's Brand Registry increases businesses' incentives to seek PTO registration. It also affects the types of marks for which parties seek registration, significantly increasing business incentives to register generic and descriptive terms. Recall that trademark law categorically refuses to protect or register generic terms like "apple" for a company selling apples; it does so to prevent businesses from monopolizing those terms to the detriment of competition. 224 And descriptive terms like AMERICAN AIRLINES for an American airline are protectable only if they have developed secondary meaning. 225 As discussed by one of us in prior work, businesses have long had an incentive to choose a descriptive or generic term as a mark when they think that "consumers will rely on [the term] to seek out their products even though consumers do not associate that term with them as a source." 226 For example, once upon a time, a business might have chosen a generic term in the hope that a telephone operator would direct business to them when a consumer requested a particular category of goods or services. 227 But the value of that strategy has likely increased sharply in the search engine era. Now, businesses can capitalize on consumers using generic or descriptive terms as search terms, even when those consumers are not looking for any particular provider of goods or services. 228 Indeed, courts have sometimes recognized the competitive advantage that highly descriptive marks have in these search listings-marks like 24 HOUR FITNESS for a gym that is always open, 229 1-800 CONTACTS for contact lenses, 230 HOME-MARKET.COM for homeowner referral services, 231 and BOOKING.COM for travel booking services. 232 But there have also been historical disadvantages to choosing a generic or descriptive term to identify the source of particular goods or services. 233 Businesses that do so understand that they cannot obtain exclusive rights in a generic term, so they cannot prevent other businesses from using that term in the course of competition. 234 Likewise, businesses have incentives not to choose descriptive terms as their marks because they have to deal with the cost and uncertainty of developing secondary meaning; they also cannot use their exclusive rights to prevent competitors from using the term in its descriptive sense. 235 Amazon's business model, combined with its Brand Registry, disrupts this traditional calculus by increasing the benefits of using and even seeking to register descriptive or generic terms. A business might reasonably conclude that it wants to use a descriptive or generic term as a mark for its Amazon-sold goods to improve the odds of prominent placement in search results. Companies have already been adopting names like "Thai Food Near Me" or "Plumber Near Me" to promote themselves in Google search results for those exact terms. 236 The incentive to use generic or descriptive terms is likely even greater for goods sold on Amazon because Amazon's search results put consumers directly into a position to buy the listed goods they see, regardless of brand. As one recent academic analysis of Amazon puts it, "Although in theory Amazon's digital shelf space is limitless, in practice the first few results-especially those on the first page of the smartphone screen-are tremendously important. According to industry research, more than two-thirds of product clicks happen on the first page of Amazon's search results, with half of those focused on the first two rows of products that appear." 237 Importantly, the search-related benefits are amplified for any mark in the Brand Registry because of the search result preference entailed in that program. 238 That means there is extra incentive not just to use those marks, but to try to register them. The PTO might erroneously register the mark, and even if it does not, the applicant can now get the benefits of the Brand Registry for at least the time during which the application is pending. While pending applications have always had a notice function, specifically by making a party's claim of ownership visible, 239 applications have never before had this kind of enforceable "legal" significance. 240 Additionally, participation in the Brand Registry makes it more likely that the business can prevent others from selling competing goods with the same generic or descriptive term. 241 For example, one Amazon seller that has filed a PTO application to register "tactical hanger" as a trademark has purportedly gotten Amazon to deactivate another seller's listings, even though the reported seller claims to be using the term descriptively. 242 While descriptive marks risk invalidation when enforced in court, the owners of those marks enjoy significant competitive advantages with relatively little risk when they enforce their rights primarily within Amazon's system.

Of course, none of those benefits would be available if trademark law effectively disincentivized registration of generic or descriptive terms. Amazon relies on the PTO's trademark registry, after all. And, of course, there are rules that attempt to do just that. As we noted, generic marks are not at all protectable or registrable, and descriptive marks are not protectable or registrable without secondary meaning. 243 But as both of us have separately argued, those rules are not sufficient. Businesses have often found loopholes that allow them to claim and even register with the PTO seemingly generic or descriptive terms, thus gaining access to the Brand Registry. For one thing, the threshold for establishing secondary meaning is regarded by many (including us) as often being too low, so a business might be able to easily clear that bar to obtain a registration. 244 For another thing, a business can often get a registration for marks that contain generic or descriptive 239. Cf. Greg Reilly, Misleading Patent Signals, 38 HARV. J.L. & TECH. 107, 167-68 (2024) (arguing that marking a product as "patent pending" is a misleading signal because it does not have legal significance even though many see it as an important market signal).

240. A pending application can give a trademark applicant provisional priority over a later-filed application in an administrative proceeding, preventing the later-filed application from proceeding to registration. But that effect is only provisional: The PTO will only tentatively reject the later-filed application and then stay further consideration during the pendency of the earlier-filed application. If the earlier-filed application does not mature into registration, the PTO will reverse course and allow the later-filed application to go forward. Larami Corp. v 457-58 (explaining how courts often fail to consider threshold questions about the number of consumers that must treat the term as a trademark for a term to acquire secondary meaning).

words as long as it disclaims its rights to the unprotectable word components. 245 According to the Lanham Act, the PTO "may require the applicant to disclaim an unregistrable component of a mark otherwise registrable. An applicant may voluntarily disclaim a component of a mark sought to be registered." 246 Indeed, 26.9 percent of applications filed from 1985 through 2016 contain disclaimed matter. 247 On Amazon, those disclaimers mean nothing. 248 Consider the following example, depicted in Figure 8. In 2017, a business applied to register a mark that contained a single word, MUSTACHES, for fake mustaches. 249 As is selfevident, this term is generic for mustaches. Yet the business was able to obtain a registration for this mark because the word was part of an image, and the applicant disclaimed "the exclusive right to use 246. 15 U.S.C. § 1056(a). Nonetheless, a disclaimer does not prejudice an applicant's common law rights or any future rights that might arise as to disclaimed words. Id. § 1056(b). Furthermore, disclaimed language still might be protectable because courts evaluate it together with nondisclaimed language in assessing trademark infringement. See, e.g., Juice Generation, Inc. v. GS Enters. LLC, 794 F.3d 1334, 1341 (Fed. Cir. 2015) (evaluating the nondisclaimed words "PEACE LOVE" in conjunction with the disclaimed word "JUICE"); Shen Mfg. Co. v. Ritz Hotel, Ltd., 393 F.3d 1238, 1243 (Fed. Cir. 2004) (emphasizing the importance of evaluating the nondisclaimed word "RITZ" in conjunction with the disclaimed words "PARIS" and "HOTEL" when assessing similarity).

247. Beebe & Fromer, supra note 23, at 985 n.162. 248. There is another way the Brand Registry potentially gives trademark registrations greater scope than trademark law would. Marks are registered for particular goods or services, in recognition of the fact that trademark rights are not rights in gross. Fromer & McKenna, supra note 10, at 147. It is unclear to us whether the Brand Registry enforces the goods and services restriction. There is nothing in Amazon's publicly available information that clearly limits enforcement to the goods or services for which a mark is registered. If enforcement is not limited that way, then parties might be able to game registration by registering for one set of goods or services but enforce that mark against other goods or services for which they could not have gotten a registration (because, among other things, the registered mark would have been generic or descriptive as applied to those goods or services).

249. U.S. Armed with this registration and despite having disclaimed rights to the word MUSTACHES for mustaches, the business can now take part in Amazon's Brand Registry claiming, as per Amazon's rules, the exact wording in its trademark registration: MUSTACHES. 251 The business is now prioritized in search results-depicted in Figure 9-and can call upon Amazon to prevent other fake-mustache sellers from using the term MUSTACHES. This trademark registrant has effectively bootstrapped its trademark registration, which disclaims protection for a generic term, into protection-at least on Amazonof exactly that generic term. And although Amazon claims to consider descriptive fair use when considering infringement, there is no available public information about how it does that, 252 and at least some sellers have alleged that Amazon does not, in fact, insulate those uses. 253 251. Supra Part II.B (describing this Brand Registry requirement). This feature of Amazon's Brand Registry is an active choice: Amazon allows mark owners to file complaints based on claimed rights that include design elements, so it is capable of matching more than text. Amazon's choice to focus on text matching even when the qualifying registration is in stylized form or has accompanying images asymmetrically benefits registrants.

252. Supra note 243. Descriptive fair use is notoriously fact-specific, requiring consideration of whether the use is "otherwise than as a mark," "fair," and in "good faith"-all questions that are illsuited to rapid and consistent implementation in private dispute resolution. On As another more technologically-focused example, consider German business MXP Prime (operating as SellerX), which buys up small Amazon businesses and has received the rare unicorn valuation. 254 It recently sought to register over thirty marks for different electronics parts using the parts' generic identifiers, such as IRF520, ATMEGA328, and DHT11 (the letters represent their maker and the number identifies the part). 255 The PTO trademark examiner, likely wondering if these identifiers are generic, and therefore unprotectable, responded with office actions asking SellerX to explain the significance of the symbols in the industry. 256 While SellerX subsequently abandoned these applications in the face of the office actions, it could have used the MUSTACHE trick and resubmitted an application to register the same alphanumeric combinations but with a drawing of anything-a sun, a clown, a hataccompanying the word while disclaiming the alphanumeric combinations themselves. This would have allowed SellerX to register trademarks with generic text that would then hold muster in the Amazon Brand Registry.

In all, Amazon's business policies have bulked up the incentive for businesses to seek registration of generic and descriptive terms, often using loopholes to get the benefits of both trademark registration with the PTO and admission to the Amazon Brand Registry.

Another way Amazon's policies have influenced the marks parties seek to register is reflected in the new phenomenon of nonsense marks. These are marks that are comprised of random strings of letters or numbers that are not comprehensible as words or as symbols with any meaning. 257 As Grace McLaughlin has noted, these marks pose serious conceptual problems for trademark law. Most obviously, they confound distinctiveness determinations because the marks seem to be fanciful (and therefore inherently distinctive). Nonsense marks do not provide any information about the goods or services and do not have any other ordinary meaning. 258 But fanciful terms are generally considered especially strong trademarks because they are assumed to be understandable only as trademarks. 259 Nonsense marks flout that assumption because they are not comprehensible as trademarks or, for that matter, as anything at all.

It is also extremely difficult to determine whether these marks are being used as trademarks. 260 As discussed above, the PTO refuses to register claimed marks that do not function as marks because those features do not indicate the source of the goods or services with which they are used. 261 Those refusals for what the PTO calls "failure to function" are typically based on contextual determinations: They focus on whether a particular sign functions as a mark as it is shown in a particular specimen of use. 262 That is why the failure-to-function doctrine has primarily focused on the location of a claimed mark and not its intrinsic nature. Consumers often recognize that a sign is a trademark when it is located in a prototypical "trademark space," even if they have not previously encountered that mark, but signs used in other places may not be understood by consumers as trademarks at all. 263 Nonsense marks likely do not function as marks, but the reason is their intrinsic nature, not that they do not indicate source when used in a particular manner. They are not vehicles for any meaning, let alone trademark meaning. 264 Even likelihood of confusion, the standard used to assess trademark infringement, is complicated in the context of nonsense marks. Trademark law does not have a good way of assessing similarity when one of the things being compared is not comprehensible as a word or understandable as a symbol. Similarity is usually assessed in terms of sight, sound, and meaning, and only sight is even possibly relevant for nonsense marks. 265 Is NXLYP confusingly similar to NYLPX, or for that matter to PTXWA? On the one hand, these marks might not be confusing because confusion depends on the ability to attach external meaning to the terms-in that sense, because people do not attach any meaning to nonsense marks, they might rarely be similar enough to cause confusion. On the other hand, it might be that all nonsense marks are potentially confused with other nonsense marks because none of them are distinct. Either way, trademark law has no good framework to evaluate such confusion. 266 Perhaps just as troubling is that a nonsense mark might be seen as confusingly similar to a more traditional mark, such as McLaughlin's examples of MAJCF being confused with MAJI, or JANRSTIC with JANSTICK, preventing the more traditional mark applicant from being able to register their mark in the face of the already-registered nonsense mark. 267 261. Supra Part I.B. 262. See Roberts, supra note 81 (arguing for this understanding, all the while noting that courts and the PTO tend to emphasize the distinctiveness of a mark over use of a mark when assessing protectability).

263. See generally Lemley & McKenna, supra note 92 (discussing the development of "trademark spaces" as a factor in the PTO's determination of trademark status).

264. Cf. Fanciful Failures, supra note 33, at 1821 (positing that while failure to function as a mark means something different in other contexts, "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").

265. Precisely because nonsense marks are not comprehensible, they are extremely unlikely to be memorable as marks. For that reason, there was never previously much incentive to use nonsense marks. 268 Regardless of the availability of legal protection, a mark is first and foremost a marketing tool that is supposed to indicate the source of goods. 269 If the mark a business chooses is not memorable, it will not provide real commercial benefits because consumers are not likely to attach any meaning to it.

Amazon's policies significantly change those incentives. The usual disincentive against a nonsense mark disappears or is greatly diminished for a third-party business selling on Amazon. For one thing, participating in Amazon's Brand Registry does not just help a business enforce its mark, a benefit that probably does not matter much to a nonsense mark user, but it gives the participating business valuable preference in Amazon's search algorithm. 270 Moreover, Amazon's business model diminishes the incentive to choose memorable marks in the traditional sense because businesses can rely on consumers being attracted to the AMAZON mark and the Amazon platform. Many consumers also focus more heavily on consumer reviews and search results listing products based on searches for the type of good rather than for the branded good than they would in other shopping contexts. 271 When searching and purchasing are not necessarily done by people who are looking for particular brand names, businesses just need something to make the algorithm prefer them. 272 Indeed, the forgettability of nonsense marks might be precisely their point. Owners of nonsense marks can collect product reviews on their listings. If they are positive, they can rely on the search algorithm to deliver them more customers. If the reviews are negative, they can easily relaunch under another forgettable nonsense mark and avoid the reputational consequences of those 268. The possible exceptions are the few now-archaic names (like A AAAAA BCALVY for fire and water carpet-damage specialists) that were chosen so that the owner would appear first in alphabetically-ordered Yellow Pages business listings. 'A Alphabetical Advantage' Helps Businesses Listed in Yellow Pages, CHI. TRIB. (Nov. 8, 1992), https://www.chicagotribune.com/1992/11/08/a-alphabetical-advantage-helps-businesses-listed-in-yellow-pages [https://perma.cc/A3Z3-UC9N].

269. See supra text accompanying notes 38-42. 270. Supra Part II.B. 271. As Emily West elaborates, "Since 2015, products that perform well on price, customer reviews, return rate, and availability to ship immediately via Prime have the chance to be highlighted as 'Amazon's Choice' (although the exact formula and decision process remains a mystery, even to companies whose products are chosen)." WEST, supra note 118, at 37. Additionally, since 2012, businesses can pay to be in sponsored search results, appearing near the top, or at the top, of search results for certain types of goods, further reducing the dominance of trademarks and brands in this context. Id. 272. WEST, supra note 118, at 37 ("[S]trong retail brands like Sears, Walmart, or Amazon can erode the value of product brands. Retailers [otherwise] generally carry particular product brands to bring consumers into their stores."); cf. id. ("Amazon's platform characteristics and relationship with consumers position it particularly well to erode the value of well-known product brands. The ability to sort product searches by price, the availability of customer reviews, and advantageous placement in searches (at a cost) can all level the playing field across brands.").

reviews. 273 In this way, nonsense marks undermine the very function of trademarks: an easy way for consumers to attach reputation to the right party. 274 Despite their conceptual difficulty, the current substantive requirements for PTO registration make nonsense marks attractive to businesses selling on Amazon that simply want a registration to participate in the Brand Registry. Nonsense marks are likely to be treated as fanciful and therefore inherently distinctive, they are likely to be seen as functioning as marks despite being gibberish, and they are unlikely to be confusingly similar to other marks given their composition. For these reasons, nonsense marks are relatively easy to register and to bootstrap into the benefits of the Amazon Brand Registry.

The PTO data bear this out in our analysis. To approximate the rate of nonsense marks in trademark applications and publications, 275 we counted the number of applications and publications with a word mark of more than four characters, comprising only one word, that was not of the one hundred thousand most frequently used words in American English, 276 and that contained either four consonants in a row or three vowels in a row. 277 This approach properly counts ELXXROONM, SUJIOWJNP, XUFFBV, and LXCJZY as nonsense marks. But the approach is both somewhat overinclusive and underinclusive. It counts OLDSMOBILE and SHIRTCRAFT as nonsense marks when it should 274. Supra Part I. 275. If the PTO determines that the trademark application satisfies all requirements for registration, it will approve the mark for publication in the Official Gazette. 15 U.S.C. § 1062(a). Third parties then have thirty days from the date of publication to oppose the registration. Id. § 1063(a). Unless there is a successful opposition, a use-based application (an application based on the applicant's current use of the mark) will automatically proceed to registration. Id. § 1063(b). Applications based on an intent to use the mark require that a statement of actual use be filed before the registration will issue. Id. § 1051(d). We focus on publication rates rather than registration rates because many intent-to-use applications succeed to publication but then are not registered because the applicant fails to file a statement of use. See Barton 277. We count three vowels in a row when a "y" is the second or third vowel in the sequence, but not the first.

not, but it does not include EARKOHA as a nonsense mark when it likely should. Even with these mistakes, we think counting marks using a metric like this one can reveal trends in nonsense marks over time, especially when there is no reason to think it undercounts or overcounts marks at different rates out of proportion to true nonsense marks over time.

We find that the number of applications to register nonsense marks has increased markedly in recent years. As Figure 10 shows, the proportion of applications with one-word nonsense marks has risen sharply in just the past few years. For decades, applications for nonsense marks accounted for only about 0.5 percent of applications. That proportion is now approximately 4.5 percent. As the number of applications has risen steadily over time, we also find, as depicted in Figures 11 and12, that the absolute number of applications comprised of a nonsense mark has risen from almost none for decades, to over twenty thousand annually in the past few years. marks are published at roughly the same rate as all other non-nonsense marks. Thus, the rising number of nonsense mark applications proceeding to publication in recent years reflects the substantial increase in the number of newly filed nonsense mark applications, not any increased propensity of the PTO to publish nonsense mark applications. As Figure 14 shows, over twenty-five thousand nonsense marks proceeded to publication in 2021, the most recent year for which we are likely to have nearly complete rates of publication, as compared to nearly zero such marks annually going back decades, except in recent years.

A final important way that Amazon's business model and Brand Registry have shifted the trademark system, and competition more broadly, is Amazon's focus on first-party sales, rather than the third-party sales explored thus far. Recall that when Amazon launched, it engaged exclusively in first-party sales. 278 While its business model has shifted toward substantial numbers of third-party sales, Amazon still engages heavily in first-party sales, and those sales generate the largest share of its revenue. 279 In recent years, Amazon's first-party sales practices have garnered substantial attention, particularly its practice of launching products under Amazon brands in product categories where third-party sellers are doing well. As Eric Johnson describes it, "[t]hanks to the massive amounts of data that platform giants [such as Amazon] are able to collect about what is sold through their site, they can cherry-pick the bestselling items from third-party retailers and then enter as a retailer, grabbing much of the sales volume." 280 Indeed, there are many stories of third-party sellers that lose half of their sales or more when Amazon enters the space, undercuts the price of the third-party sellers, and gives itself prominent search-result placement. 281 For example, a ProPublica study found that Amazon gave the products it sells under its own brands better searchresult placement approximately 75 percent of the time, even when other sellers' prices for these products are lower. 282 Feng Zhu and Qihong Liu provide more systematic evidence of Amazon's approach, finding that over the course of ten months, Amazon began directly competing against third-party sellers on 3 percent of over 160,000 products offered across twenty-two product spaces. 283 The spaces Amazon entered already had higher sales and better product reviews, whereas the spaces they did not enter tended to be those requiring greater seller effort for growth. 284 Amazon is in a unique position when it decides to sell in a product category because it is not only a platform participant like third-party sellers, it is also the platform provider. 285 By leveraging the massive amount of sales data it collects in its platform provider role, Amazon can outcompete the third-party sellers it relies on for its platform's success. 286 Scholars and regulators have expressed a range of views on Amazon's firstparty sales practices. Some view these practices as worrisome. Lina Khan says it is "anticompetitive" for Amazon to use its dominance to exploit its customers, the third-party sellers, as competitors. 287 Johnson calls Amazon's practice "salesjacking" and often suppressive of third-party innovation. 288 Indeed, a House of Representatives subcommittee led a sixteen-month investigation into Amazon on these and other practices, 289 and the FTC, under Khan's leadership, recently sued Amazon for unfair competition for related practices. 290 By contrast, Daniel Francis views Amazon's self-preferencing more benignly as part and parcel of competition in a vertically integrated entity. 291 Others have even suggested it might be procompetitive because consumers get more competitors offering products, often at better prices. 292 Whether or not Amazon's self-preferencing is problematic from an antitrust perspective, it has important consequences within the trademark system. When Amazon sells its own goods, it does so under its own mark. That mark might be an Amazon sub-brand that contains the term AMAZON, most commonly AMAZON BASICS or AMAZON ESSENTIALS, or a separate house brand like GOODTHREADS apparel or RIVET furniture. 293 In 2020, Amazon's privatelabel business had 243,000 products across forty-five different house brands. 294 For cost-cutting reasons, and perhaps out of fear of government regulation, Amazon recently eliminated a number of its house brands, winnowing them down to fewer than twenty. 295 For example, Amazon dropped twenty-seven of its thirty clothing brands, leaving just AMAZON ESSENTIALS, AMAZON COLLECTION, and AMAZON AWARE. 296 Amazon also opted to focus on its AMAZON BASICS brands for a range of home goods and technology accessories. 297 Amazon's vice president of private brands explained to the Wall Street Journal that "[w]e always make decisions based on what our customers want, and we've learned that customers seek out our biggest brands-like Amazon Basics and Amazon Essentials-for great value with high quality products at great price points." 298 In all, it seems that Amazon has both consolidated most of its house brands to those containing the AMAZON mark and cut tens of thousands of products. 299 PTO data give a sense of the range of goods (and services) for which Amazon uses its AMAZON BASICS and AMAZON marks. Between 2009 and 2022, Amazon filed thirty-eight applications to register AMAZON BASICS. Twenty-four of those applications proceeded to publication; the remaining applications, which were all filed since the beginning of 2022, have not yet proceeded to publication or have been abandoned. Like the fictional ACME mark used in the Looney Toons cartoons for just about every possible item, including rubber bands, anvils, cars, nitroglycerin, artificial rocks, and superhero suits, 300 Amazon's applications collectively are for goods in twenty-four of the thirty-four classes of goods in the "Nice Classification" system and for services in one of the eleven classes of services. 301 Between 2007 and 2022, Amazon filed fifty-five applications to register AMAZON. Fifty-two of those applications proceeded to publication, while the one application filed in 2022 has neither proceeded yet to publication nor been abandoned. These applications are for goods and services in fifteen of the thirty-four Nice classes of goods and nine of the eleven Nice classes of services. 302 Amazon has two mutually reinforcing trademark-related advantages in selling its private goods with a house mark. First, it controls its search algorithm and can preference its own goods in product search results. 303 This aspect is trademark-related because it gives increased prominence and emphasis to the 301. A trademark applicant must specify the goods and services in connection with which the applicant claims the exclusive right to use the mark. 15 U.S.C. § 1051(a) (2). The applicant must do so in the form of a written description of the goods and services and by reference to one or more of the forty-five categories of goods and services contained in the International Classification of Goods and Services for the Purposes of the Registration of Marks, otherwise known as the "Nice Classification" after the French city where it was established in 1957. [https://perma.cc/RY3F-WJDM]. The Nice classes in which Amazon has applied for the AMAZON BASICS mark are 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 15, 16, 17, 18, 19, 20, 21, 22, 24, 25, 26, 27, 28, and 35. 302. The Nice classes in which Amazon has applied for the AMAZON mark are 2, 3, 5, 7, 8, 9, 10, 11, 14, 16, 18, 20, 21, 25, 28, 35, 36, 37, 38, 39, 40, 41, 42, and 45. 303. Angwin & Mattu, supra note 282; Khan, Amazon's Antitrust Paradox, supra note 1, at 781; Leon Yin & Adrianne Jeffries, How We Analyzed Amazon's Treatment of Its "Brands" in Search Results, MARKUP (Oct. 14, 2021), https://themarkup.org/amazons-advantage/2021/10/14/how-we-analyzed-amazons-treatment-of-its-brands-in-search-results [https://perma.cc/VCT5-UNGL].

Amazon house brands over third-party brands, thereby further advancing the Amazon marks' strength. Even when consumers search on Amazon for thirdparty branded products, Amazon can give prominent search-result placement to its own Amazon-branded products over the third-party products for which the consumers were explicitly searching. Consider this striking example from February 2023, shown in Figure 15 The second trademark-related advantage for Amazon in selling under its own brands, particularly AMAZON BASICS or AMAZON ESSENTIALS, is that Amazon is a beloved, trusted, and widely recognized brand. 306 Indeed, communications scholar West emphasizes that "Amazon has normalized its own ubiquity[, which] should be viewed as an accomplishment borne of branding, public relations, and relationship marketing." 307 Consumers who shop on Amazon are very likely to be drawn to Amazon-branded products. In that vein, West observes that "Amazon has . . . leveraged the trust that consumers have in its brand to launch sub-brands" like AMAZON BASICS and AMAZON ESSENTIALS. 308 Furthermore, Amazon's de-centering of third-party branding, which is the same de-centering that leads many businesses to opt for nonsense marks, 309 likely amplifies Amazon's own branding strategies. The fact that Amazon is consolidating its house brands under the AMAZON BASICS, AMAZON ESSENTIALS, and related house brands containing the AMAZON mark, while discarding unrelated house marks, suggests that Amazon recognizes the power of its AMAZON-centered marks.

Unlike the incentives for small and medium-sized business registration and for adoption of nonsense marks, Amazon's branding practices have not profoundly affected the total number of trademark applications filed with the PTO. But Amazon's practices are part of a larger story of the increasing importance of Amazon's marks in the Amazon marketplace and the corresponding diminishment of third-party sellers' marks. We consider the implications for the trademark system and competition writ large in the next Part.

This Part addresses how, if at all, trademark law, the PTO, and Amazon should address the overhaul of the trademark system wrought by Amazon's business model and Brand Registry. After discussing some precursors to Amazon's pervasive effect on third-party business practices in Part IV.A, we consider what it means for a single dominant company to have such an impact on trademark law in Part IV.B. In Part IV.C, we discuss how the law should be adjusted. Part IV.D considers what Amazon's practices might mean for the future of trademark law and competition more broadly.

306. Supra Part II.A. 307. WEST, supra note 118, at 13. She elaborates: "Amazon's brand can stretch like taffy because its focus is not the things it sells so much as the service relationship it cultivates with customers." Id. at 37. 308. Id. at 37. 309. See supra Part III.D.

clothing, harnesses, saddles, firearms, wagons, buggies, bicycles, shoes, baby carriages, and musical instruments. 315 A few years into the business, Sears' catalogs were hundreds of pages long. 316 Sears would generally choose which categories of goods he wanted to sell, such as sewing machines and bicycles in earlier years, and then he would create a market for these products. 317 As the business grew, it offered an alternative retail option not just to rural consumers-who typically had no significant retail stores nearby-but to all consumers, including those located near independent retail merchants. 318 Feeling hostility from these merchants, Sears decided to target a rural audience by placing advertising for his catalogs in magazines catering to that audience. 319 Sears faced several challenges in getting small manufacturers to make and supply goods for Sears to sell by mail order. For one thing, the small manufacturers feared boycotts from the independent retailers threatened by the Sears model. 320 Sears also had to find manufacturers that would provide goods at the low prices Sears would pay. 321 Finally, Sears had to overcome manufacturers' reluctance to commit all of their product to Sears because of their fear of the chokehold the distributors would have on them. 322 Sears solved these problems in two ways: first, by locating manufacturers that wanted to work with Sears, providing capital to those companies if necessary, and second, by making goods in house if outside manufacturers could not be located. 323 Sears generally sold all of these goods without branding other than the Sears name or house brands that Sears chose. 324 Starting around 1925, Sears launched retail stores and married them with its mail-order business. 325 With this combination of urban retail and rural mail order, Sears became the largest U.S. retailer of general merchandise in the midtwentieth century. 326 Julie Cohen has observed that Sears' mail-order catalog business can be understood as a proto-platform. 327 As she explains, "[i]nclusion of a product in the Sears, Roebuck catalog gave its manufacturer access to a marketing juggernaut with the ability to reach consumers nationwide, the range to offer concert gran[d] pianos and engraved shotguns, and the power to undercut the prices charged by local 'five-and-ten-cent stores' for everyday essentials." 328 In that sense, Sears' business model bears more than a passing resemblance to Amazon's. 329 But Amazon's model differs from Sears's in two key ways. First, just about any third-party business can partner with Amazon to sell any of its wares on Amazon's platform. 330 By contrast, Sears would internally determine which products it wanted to sell and then solicit particular third-party businesses to manufacture those items. Second, Amazon has considerably more power vis-àvis small businesses than Sears did. Businesses operating in Sears' heyday could still thrive without Sears by engaging in local commerce, which was then more prominent. 331 Sears therefore did not have the grip that Amazon has on thirdparty business practices. One recent report has described this phenomenon as Amazon's "trickle-down monopoly." 332 These two differences in combination make Amazon's operation unlike Sears'. As Emily West puts it, even though distributors like Sears have network effects and economies of scale in common with Amazon, "the logics of digital capitalism have launched Amazon into a sphere of market dominance and expansion into horizontal and vertical integrations that are unprecedented relative to" other distributors. 333 Because of that dominance, Amazon's practices have a much more substantial impact on the U.S. trademark system. One important difference is that Sears used its own brand or internally branded the goods it sold via its mailorder catalogue, whereas Amazon has many third-party branded goods. Sears therefore was much less likely to impact third parties' trademark practices. The second precursor, to which we now turn, is more like Amazon in terms of its effect on the trademark system.

As the commercial internet took off in the 1990s, everyone and their dog 334 rushed to claim internet domain names. Whereas in 1992 there were fifteen thousand registered domain names, 335 by 2000, there were over thirty million. 336 By 2022, there were 350.4 million. 337 Many businesses wanted to, and still want to, register their brand name as a domain name in the .com top-level domain. 338 Some businesses readily claimed a .com domain that matched their trademark, as Cisco did with cisco.com and Apple did with apple.com. 339 But one problem quickly became apparent: Domain names are, by their very nature, exclusive, but it is common for multiple different companies to use the same or very similar marks for different goods or services. So, for example, Delta Airlines coexists with Delta Financial and Delta Faucets because the uses are different enough that consumers are unlikely to be confused by the concurrent uses. 340 But as the domain system is set up, there can only be one owner of delta.com. 341 Sometimes, businesses in this situation amicably worked out allocation. DeltaComm Internet Services originally claimed delta.com; it then transferred the domain name to Delta Financial, which subsequently transferred to Delta Airlines, the current registrant of that domain name. 342 But multiple potentially legitimate claims to a domain name often led to conflict, such as when Nissan Motor Company sued Nissan Computer Corporation, alleging that the latter's registration and use of nissan.com constituted trademark infringement. 343 Those conflicts had a different character when they involved domain name registrants with no prior interest in marks corresponding to the domain names they registered. Because domain name registration was initially a gold rush, many opportunists rushed to claim domain names corresponding to existing marks whose owners had not yet claimed the domains. They then turned around and offered to sell them to the mark owners. One individual registered over one hundred domain names by 1995, including domain names corresponding to the names of apparel company Eddie Bauer and the airline Lufthansa, and then attempted to sell the domain names to the companies operating under those names. 344 Others registered domain names corresponding to generic category names, such as cars.com and insurance.com, because they thought those domain names would give them a unique advantage among consumers looking for goods or in those categories. 345 This domain name activity spilled over into the trademark system to a large extent, with businesses racing to the PTO to apply to register domain names as trademarks. 346 By 1995, the PTO announced a policy that it would register domain names so long as they were used as trademarks. 347 Indeed, empirical data on trends in trademark applications and registrations over time tend to have spikes in the data from 1999 to 2000, which are primarily attributable to this domain name activity. 348 Policies outside the trademark system regarding domain name registration and dispute resolution shaped trademark registration practice during this time period. In the 1990s, Network Solutions, a private company, held a U.S. government-sanctioned monopoly on registration of domain names. 349 Beginning in 1995, Network Solutions adopted a series of policies under which it could suspend a challenged domain name on the complaint of the owner of a registered trademark that exactly matched the domain name, minus the top-level domain, like .com. 350 It would suspend the domain name even though the domain name registrant might have its own relevant trademark, if the complainant could show that it registered its trademark before the domain name holder activated its domain or registered its trademark. 351 Trademark registrations in any country qualified under the Network Solutions policies. 352 Network Solutions' policy did not align with domestic trademark law in important ways. In particular, Network Solutions would suspend a domain name based on the complaint of the owner of a foreign trademark registration that predated the domain name registrant's use, even though American trademark law would give priority to the first user in the United States. 353 Network Solutions also gave priority to the owner of a trademark registration even if the domain name registrant had use in the United States that predated the registration or application to register. 354 Relatedly, the first to register a domain name with Network Solutions would win any challenge even if there were two parties holding concurrent trademark registrations for unrelated geographic areas. 355 According to scholars writing at the time of the domain name rush, trademark owners sometimes felt compelled to challenge domain name registrations because of their concern that acquiescence in a competing use would weaken their mark's strength and limit the scope of their rights. 356 Perhaps most significantly for current purposes, Network Solutions' policy encouraged businesses to apply to register their trademarks when they might otherwise not have done so because those businesses needed trademark registrations to be able to challenge domain names with Network Solutions. 357 And because Network Solutions gave priority to any registration that issued before a domain name was registered, those businesses had a strong incentive to seek registration wherever it was fastest. That turned out to be Tunisia, which would register an applied-for mark in a matter of days, rather than the year or so it took at the time in the U.S. PTO. 358 Indeed, so many domain name registrants and challengers registered in Tunisia that Network Solutions eventually amended its policies to erase the impact of the Tunisian registrations. 359 The effects of the Network Solutions policy turned out to be relatively short-lived because domain name dispute resolution came to be governed overwhelmingly by the Uniform Dispute Resolution Policy (UDRP) adopted by the Internet Corporation for Assigned Names and Numbers (ICANN) in 1999. 360 The UDRP established procedures that were grounded in recommendations by a United Nations World Intellectual Property Organization study. 361 Under the UDRP, which all domain name registrants must accept as a condition of registration, trademark owners can object to any identical or confusingly similar domain name on the ground that the domain name registrant has no rights or legitimate interest in the domain name and that the domain name has been registered and used in bad faith. 362 Complaints are resolved under the UDRP through ICANN-accredited dispute resolution service providers via arbitration. 363 The UDRP effectively nullified Network Solutions' policy and reduced the impact of domain name registration on trademark registration practice. In one of the most cited decisions under the UDRP, a case involving the domain name madonna.com, the panel ruled that a Tunisian trademark registration by a business not located in Tunisia did not reflect a legitimate interest in the disputed name and, in fact, might have reflected the registrant's bad faith. 364 Decisions like that significantly decreased interest in Tunisian trademark registration among non-Tunisian businesses and pushed domain name dispute resolution policy more in the direction of substantive trademark law.

Congress also enacted the Anti-Cybersquatting Consumer Protection Act (ACPA) in 1999. That statute prohibits the act of registering with the "bad faith intent to profit . . . a domain name that . . . [is] confusingly similar" to a registered or unregistered mark or personal name, or is dilutive of a famous mark. 365 The ACPA provided a more effective legal framework for resolution of trademark disputes relating to domain names. Together with the UDRP and the general cooling of the internet boom, the ACPA decreased the number of applications to register domain names as trademarks. 366 The effect of the Network Solutions policy and the domain name system paled in comparison to Amazon's effect in terms of its magnitude. Figure 16 shows the small and brief bump in applications filed during the internet boom of the early aughts as compared with the larger increase in filings in recent years. In scale and in duration, Amazon's effect on the formal trademark system is truly unprecedented. The Sears experience helps demonstrate why: Amazon's influence as a platform is orders of magnitude larger than any conceivably analogous predecessor. These two examples highlight the uniqueness of Amazon's effects. Amazon is a market-dominant platform, and as a result, it has affected the trademark system on an unprecedented scale. Those effects are also qualitatively different in that they are traceable to the policies of a single company rather than more general economic conditions or changing technology generally.

Amazon's recent and ongoing impact on the shape of the trademark system is staggering. Its Brand Registry and business model are largely responsible for a huge increase in trademark applications from small businesses that might never have otherwise filed for registration. Its policies have provoked trademark extortionists to file fraudulent applications, increased the incentive to apply to register descriptive and generic terms, and driven up applications for nonsense marks. All of those effects will no doubt be further amplified now that Amazon has begun allowing businesses with pending trademark registration applications to join its Brand Registry. As we have noted, there is irony here because the acceptance of pending applications seems to have been motivated by longer pendency of trademark applications in the PTO, to which Amazon surely contributed. Any other alterations in Amazon's business model or changes to the Brand Registry qualification rules and advantages are likely to shape-shift the trademark system in other unforeseen ways.

A recent Data and Society report describes and analyzes the "trickle-down monopoly" that Amazon has imposed on its third-party sellers. In particular, the report suggests that "[b]y platformizing such a huge swath of retail, Amazon has enrolled countless [third-party] sellers in expanding the company's influence. But it has also projected [its] own logics of monopoly onto these small-tomidsized scale sellers, who stockpile inventory in their own homes, sell at losses to try to corner niche markets, and diligently guard all information about their businesses." 367 For similar reasons, its capture of the market has also crept across to and seized the PTO.

Amazon's impact on the trademark system is perhaps not so surprising when considering the role that trademarks play for many third-party businesses selling on the Amazon platform. 368 As Sonia Katyal and Leah Grinvald explain, "the platform economy facilitates the emergence of . . . 'macrobrands'-the rise of platform economies whose sole source of capital inheres in the value of the brand itself-the Airbnbs, Ubers, and eBays of the world." 369 There is also the "parallel emergence of the 'microbrand'-the rise of discrete, small enterprises made up of individual businesses." 370 In many ways, consumers are attracted to Amazon because they trust it, because of the platform's network effects, and because of the company's consumer-focused model. 371 They are drawn to Amazon's macrobrand, much like consumers are drawn to a franchise brand regardless of its operators. 372 The third parties that sell their wares on Amazon are attracted to the platform for similar reasons. 373 Much like franchisees, third-party sellers realize that any microbrands they use might not matter as much to consumers as they would in other contexts. 374 Still, those sellers are likely to recognize the value of applying Amazon sellers resemble . . . franchisees in that they assume forms of risk and responsibility that entrepreneurship entails, while relinquishing much of the freedom it has historically provided. Insofar as sellers give Amazon, on average, 34% of each sale that they make through the platform, they also facilitate the transfer of loans and other resources that the state has designated to support small businesses-including minority-owned businesses-to one of the largest corporations in the world.").

373. to register marks that will benefit them in terms of search result placement and qualify them for the Brand Registry. That is why sellers are picking nonsense marks and descriptive or generic terms as marks when they would be much less likely to do so if they were not selling on Amazon's platform. Whereas Katyal and Grinvald suggest that microbrands "have a strong interest in utilizing the basic principles of branding and trademark protection," 375 at least some of Amazon's third-party sellers are throwing these basic principles out the window to maximize their impact on Amazon. The increased importance of Amazon's macrobrand and the decreased importance of third-party sellers' brands likely explain why Amazon is tripling down on its AMAZON-centered trademarks like Amazon Basics.

Without wading into the merits of Lina Khan's antitrust analysis 376 or the contrary positions taken by others, 377 we think it is clear that Amazon's dominant position in internet commerce substantially affects the shape of competition and dramatically influences the operation of areas of law that intersect with Amazon's business practices. Because of Amazon's dominance, its business model and Brand Registry have changed private parties' use of the trademark system so much that it has effectively overhauled that system.

To some, Amazon's singular impact on the trademark system will suggest that the platform is simply too powerful, whatever one thinks of the antitrust issues. At the very least, the effects we have described here should force us to reflect on whether we are comfortable with a single company setting internal rules for its own benefit when the effect is to reconfigure a legal system that was not developed with such a powerful actor in mind. Amazon is leveraging the existing trademark system, outsourcing decisions about trademark validity to resolve conflicts on its platform and protect its business model. Importantly, it is seeking to ward off prospective regulation that would expose the company to products liability claims and liability for counterfeits sold on its platform. 378 Amazon's leveraging of the trademark system is not incidental to that goal. By piggybacking on the U.S. trademark system rather than building an independent brand verification system from scratch, Amazon is able to claim the high ground of the government's own gold-standard system, making government regulation unwarranted. 379 But Amazon's practices have profoundly affected the trademark system in ways that impact consumers and every other trademark applicant. 380 Should trademark law be adapted to accommodate or respond to Amazon's practices? On the one hand, it seems absurd to suggest that trademark law should continually adapt to Amazon's internal business decisions. On the other hand, it seems absurd for it not to recognize commercial reality. Either way, Amazon's seismic impact on the trademark landscape forces us to consider where changes to the current system should be made to accommodate the platform's continued, unprecedented efforts.

After considering the various trademark and other legal harms at stake here, this Section considers how the trademark system might be adjusted in light of Amazon's influence on it. It also considers how Amazon's influence might be used to restore aspects of the trademark system. Indeed, the combination of the trademark system and Amazon's business model might be used to advance the goals of trademark and competition.

Amazon's considerable impact on the trademark system raises important rule-of-law questions. The practices of a single company have dramatically changed the functioning of the legal system, and those changes have not been the result of any public process with legislative or regulatory legitimacy. 381 Several scholars have explored the related, but analytically distinct, issue of 379. It is also likely significantly cheaper to incorporate the U.S. trademark system at the base of the Brand Registry rather than create a fully independent system. That is surely attractive to Amazon, a notoriously frugal company. See generally KRISTI COULTER, EXIT INTERVIEW: THE LIFE AND DEATH OF MY AMBITIOUS CAREER (2023) (emphasizing repeatedly, in a memoir by an ex-Amazon employee, the company's thrift).

380. In addition to the impact Amazon has on the trademark system because of its business model and Brand Registry, the company also has until recently had an arguably outsized role in further setting trademark policy by having one of its trademark lawyers occupy one of the nine seats on the PTO's Trademark Public Advisory Committee. Trademark Public Advisory Committee Members Biographical Information, U.S. PAT. & TRADEMARK OFF., https://www.uspto.gov/about-us/organizational-offices/public-advisory-committees/trademark-public-advisory-committee-0 [https://perma.cc/BV5G-ESNM]. In March 2025, Secretary of Commerce Howard Lutnick ended all current appointments to this advisory committee, thereby removing Amazon's lawyer from this committee, without yet appointing new members. See Gene Quinn & Steve Brachmann, Commerce Secretary Disbands PPAC and TPAC, IP WATCHDOG (Mar. 18, 2025) https://ipwatchdog.com/2025/03/18/secretary-lutnick-disbands-ppac-tpac [PERMA]. In this role, Amazon's lawyer might have provided helpful insight into its role in shaping the trademark system or other trademark issues it faces, including fraud and counterfeit goods. But it also further enlarged Amazon's outsized influence on the trademark system. platforms' role as private sovereigns 382 that engage in rulemaking or adjudicatory acts. 383 For example, Rory Van Loo has considered corporations' development of large-scale dispute resolution systems for customers. 384 As Van Loo notes, on the one hand, these forms of dispute resolution have some of the features of court adjudication, such as access to redress, accountability, truth, and justice. But these systems lack other judicial features, such as transparency, procedural equality, and aggregation mechanisms. 385 Van Loo has generally recommended both governmental oversight of these private systems and procedural rules similar to those used in courts, to promote due process and transparency. 386 Similarly, Hannah Bloch-Wehba analyzes how platforms act as regulators and "are performing quintessentially administrative functions." 387 She therefore proposes that "platform governance [be] accountable to the public." 388 Even though we could tell a similar story about Amazon creating a private trademark dispute resolution system parallel to the government's, 389 our focus is different and is on the extent to which Amazon's "system" influences parties' behavior within the legal system itself. 390 This is not a story of Amazon overtly seeking to capture the trademark system. But it is perhaps a story of indirect capture, with Amazon taking over the PTO, even without realizing as much, by shaping third parties' trademark registration behavior. 391 The worry here is that Amazon's model is materially affecting the trademark system as an indirect result of the company's pursuit of its own interests, which might diverge from the interests of the public, as reflected in trademark policy. 392 As Rachel Barkow points out in generally thinking through agency capture, "one person's political pressure is another person's democratic accountability. What policy makers who seek insulation want to avoid are particular pitfalls of politicization, such as pressures that prioritize narrow short-term interests at the expense of long-term public welfare." 393 This concern suggests taking a hard look at the substantive ways in which the trademark system's operation has changed and exploring whether any of those changes are out of line with the way the trademark system ought to operate. 394 We return to that constellation of issues shortly in discussing trademark-specific harms provoked by Amazon.

Moreover, the indirectness of this capture itself might be problematic because the effects materialized without the PTO or other actors in the trademark system being aware of the role of Amazon's policies or attempting to account for them. To the extent this hiddenness is worrisome, it can be addressed through sunlight provided by this and other scholarship and more self-reflection by the PTO on the changes it is experiencing, as well as an attempt to grapple with whether and how to respond to such changes.

Another general concern with capture is the instability it can foster when there are future political changes in elected offices like Congress or the presidency. 395 Amazon's de facto capture here brings potential instability in a different sense: To the extent Amazon decides further to adjust its practices in ways that affect third-party seller behavior vis-à-vis the trademark system, the trademark system could repeatedly experience massive and relatively abrupt legal shifts in whichever direction Amazon's winds blow. 396 To be sure, market and social conditions often shift, create demands, or subvert existing regulatory premises, and government actors routinely adapt to those changes. When cars became mainstream, we suddenly needed a Department of Motor Vehicles and driver's licenses. New voter identification requirements increased the demand for driver's licenses. The advent of the internet opened the floodgates of businesses rushing to register domain names as trademarks. 397 But Amazon's influence on the trademark system is different. For one thing, Amazon's effects are different in scale. For another, those effects are attributable to the policies of one company and its market dominance rather than being the result of more diffuse background conditions. The singularity of Amazon's influence creates challenges, given that Amazon can unilaterally provoke massive legal shifts. But it also creates opportunities because Amazon might be amenable to helpful changes; if it is not, it is a single locus for regulation.

Amazon's effect on the trademark system may be a species of a broader problem and dominant platforms might provoke similar shifts in other areas of the law, such as Uber with insurance and employment laws or Facebook with privacy law. In that way, these rule-of-law concerns are not Amazon-or trademark-specific. Yet Amazon's policies have also inflicted several trademark-specific harms. Trademark law assumes that businesses will use the trademark system in certain ways. Amazon has upended many of those assumptions, leaving the PTO and other legal actors ill-equipped to deal with the kinds of applications many businesses now file. For example, consider applications for descriptive or generic terms that businesses operating on Amazon might be tempted to seek. 398 For well-considered reasons, the trademark system makes it harder, if not impossible, to obtain protection for rights in these terms because of the impact that protection can have on fair competition. 399 It is much easier to register descriptive or generic terms if they are stylized or accompanied by an image. 400 Indeed, the PTO might very well require an applicant to disclaim rights in the descriptive or generic term itself. 401 If a business were to succeed in registering a descriptive or generic term because of its stylization or accompanying image and then seek enforcement of its rights in court, a court would be sensitive to the aspects of the mark that are either not protectable or weak by virtue of them being descriptive or generic. For example, the Second Circuit found no likelihood of confusion between the PARENTS magazine and PARENT'S DIGEST magazine trademarks in part because the PARENTS registration "protect[ed] not the name or the word 'parents,' but rather the stylized logo of that name including the unusual form and shape of the letters comprising the word," and thus did not prevent others' use of the generic term "parent." 402 Yet Amazon's brand registry rules are not similarly sensitive because Amazon ignores stylization and accompanying images, focusing only on the text of a registered mark. Functionally speaking, Amazon will allow a descriptive or generic term to be the basis for protection and preferential treatment, 403 which means that the as-applied trademark system has a different character on Amazon: Amazon lacks the ability to meter scope in the way that justifies registration of these terms in the first place. More broadly, Amazon's policies result in behavior that tends to undermine the trademark system's core assumptions. In particular, that happens when thirdparty sellers pick descriptive, generic, or nonsense terms to use on Amazon. Use of those marks contravenes trademark law's central premise that marks serve to identify the source of goods or services and as shorthand for their qualities and characteristics. 404 The reasons are assorted yet related. By their nature, those kinds of marks are unlikely to be memorable to consumers, let alone associated with a particular source. But those features are less important on Amazon, where consumers can rely on product searches, product reviews, and the pull of Amazon as a brand to find and buy products. 405 Some kind of mark is needed to be in the Brand Registry, but that mark does not need to carry all of the weight that trademarks traditionally have been expected to.

That kind of shift in the functions of trademarks opens an intriguing possibility. Modern trademark scholars have long lamented the extent to which trademarks enable artificial differentiation and the development of pure brand value, which is disconnected from real information about the nature or characteristics of a product or service. 406 405. For these reasons, the mark is not much different than a UPC code to uniquely identify the goods at hand. A related harm is that many of these businesses, particularly the ones registering nonsense marks, are primed to rebrand with a new (nonsense) mark if they get sufficiently negative product reviews on Amazon. Such rebranding would give them an opportunity to resurface on Amazon with the same product but with a clean slate. And it is not as if consumers remembered the mark anyhow. While such rebranding might also see to contravene trademark's core values, trademark law does not stand in the way of businesses rebranding, as can happen when a business wants to escape scandal, as with ValuJet Airlines becoming AirTran Airways after a plane crash; an association with a negative term, such as ISIS mobile banking app becoming Softcard; or a dated symbol, as with American Telephone & Telegraph becoming primarily known as AT&T. over how best to limit trademark law's reach, most have accepted that some enablement of brand exploitation is an inevitable, if unfortunate, cost of a system that protects trademarks' core functions. If, however, trademarks are less important for conveying product-related information because algorithmic tools can provide that information equally or better, then we should not worry if Amazon's practices lead more companies away from source-indicating marks. Indeed, we might celebrate that result and hope it undermines trademark law's incentives.

But it is not clear to us that we are yet at a point where the alternatives to trademarks are better. Empirical evidence suggests that Amazon consumer reviews are inaccurate in important ways (because consumers tend to write reviews only if they are extremely positive or negative), that consumers do not process reviews in cognitively accurate ways, and that as many as one in every three reviews are fake. 407 And of course the Amazon algorithm is not designed to neutrally provide the products consumers are really looking for. As we have noted, Amazon has routinely given preference to its own products in search results even when consumers search for other products by brand name. Furthermore, it gives participants in the Brand Registry access to various search optimization tools. 408 For those reasons, one might reasonably conclude that trademark law's emphasis on source designation remains important, and that Amazon's policies are undermining the incentives that trademark law properly creates.

Another trademark-related harm is the clutter of the PTO's trademark register. The increasing number of applications, particularly for marks that do not function as trademarks, imposes costs on the PTO, other businesses, and consumers. Increased registrations lead to a so-called "trademark thicket." 409 A thicket makes it harder for trademark examiners and businesses to search the 407. Gibson, supra note 273, at 24-29, 39-41. Amazon is making some effort to deal with fake reviews. For example, it created a team to address the issue and filing lawsuits against sellers or platform users that create and post those reviews. Amazon's Latest Actions Against Fake Review Brokers: Lawsuits Against Fraudsters Target the Source of Fake Reviews, AMAZON (May 1, 2024), https://www.aboutamazon.com/news/policy-news-views/amazons-latest-actions-against-fake-review-brokers [https://perma.cc/5S4M-F39J]. But even if Amazon could make a significant dent in the number of fake reviews, the other problems with relying on reviews would remain.

408. Supra Part III. 409. As one of us describes in a previous work, "[a] trademark thicket is analogous in some ways to the patent thicket, a crowded area of patent rights, in which rights to the many patents comprising the thicket must be secured for freedom of operation in the space, which can raise cost issues and anticompetitive concerns. register to ascertain whether there are existing registrations that are potentially confusingly similar. This is a difficult cost to impose, particularly when many of these registrations are undermining trademark's core assumptions. The clutter also makes it harder for businesses that want to choose a new mark, particularly one that is used in a traditional trademark sense, to settle on one that is not confusingly similar to previously-registered marks. 410 Moreover, even when a business finds a mark that it can clear through this thicket, that mark might be less useful because it is more difficult for its user to develop source distinction, compared to other options that might have been available. 411 Relatedly, a thicket can also harm consumers by making it harder for them to distinguish between marks in the crowd. 412 To be sure, this cluster of harms is less sharp and perhaps not significant for third-party sellers that are registering source-designating marks that they would have used but not registered in the absence of Amazon's policies. 413 On one side, notwithstanding their contribution to clutter, registration of these marks might be welcomed for making the register more comprehensive. This registration, in turn, makes it easier for third parties and the PTO to locate these marks and gives these businesses the benefits of registration. 414 On the flip side, registration of these marks might be seen as wasteful papering of rights and only very marginal improvement of notice, given the widespread availability of search engines.

But the increase in the number of applications itself causes a distinct trademark harm. Specifically, the delay imposed on everyone seeking registrations with the PTO is harmful even if the marks for which registration is sought would otherwise have been used. 415 The PTO's months-long backlog is harmful to businesses that want to use the trademark system for its core purposes because they must wait that much longer to get the benefits of registration and operate with some legal uncertainty. The volume of applications might also undermine the quality of trademark examinations because examiners are juggling more applications at a time and waiting longer periods before being able to return to an application they have already begun working on.

An additional harm to the trademark system is the fraud committed on the PTO by trademark extortion. The fraud harms legitimate businesses whose marks have been commandeered, as these businesses are the true source designation of those marks. All the while, it consumes scarce PTO resources.

Finally, third-party sellers' increased use of descriptive and generic terms as marks might be problematic for competing sellers on Amazon whose products get ranked lower in search results by virtue of the descriptive or generic term being claimed on the Brand Registry. Those competing sellers using Brand Registry-protected marks in their Amazon listings might also be accused of infringement. Even if Amazon ultimately clears them of any violation, the costs of investigation and possible suspension can be significant. By contrast, trademark law has developed tools to make it harder to protect these terms in the first place. 416 And when the law does allow protection, it has developed defenses of fair use to enable competitors to use such terms descriptively and in other ways that do not put them at a competitive disadvantage. 417 Now that we have cataloged some of the harms that might manifest from Amazon's capture of the trademark system, we turn to what the PTO and Amazon might do to ameliorate them.

With regard to the rule-of-law or capture concerns, the PTO ought to at least pay attention to the effect of Amazon's policies on trademark filings and consider whether it wants to change its approach to examining applications that are likely attributable to Amazon's influence. The PTO can also engage in discussions with Amazon and its Trademark Public Advisory Committee, as well as more publicly ask for feedback, about whether the new trends reflect the assumptions and approaches of the current trademark system, the PTO should adapt, or whether the PTO should encourage Amazon to make its own changes to ameliorate these trends. Proactivity and self-reflection will be critical, especially because small initial shifts in filing trends can quickly become massive given the number of third-party sellers on Amazon.

One of the biggest and most immediate problems the PTO faces is its backlog of applications and the accompanying delays in registration. The PTO has expressed sensitivity to this backlog. It recently announced plans to hire eighty-six more trademark examiners before the end of 2023 and up to sixty more in 2024, as well as considering incentives to encourage speedier examination. 418 The PTO was correspondingly targeting an average total application pendency of eight and a half months and an initial office action within five months, targets that would likely take two to three years to achieve. 419 the executive branch. 420 Even if the hiring freeze is eventually lifted, the PTO ought to be cautious in trading off speed for careful examination, especially because Amazon's policies have seemed to encourage fraudulent applications. PTO Trademark Commissioner David Gooder concedes as much, stating that in a world with fraudulent and different sorts of applications, trademark examiners will need to proceed slower and with more caution. 421 Dealing with this volume undoubtedly adds cost, and those costs must be paid somehow. The PTO should consider whether to adjust fees to better align the increased costs with their source. That will not be entirely straightforward. New applications will not, on their face, announce themselves to be Amazonrelated, even if the trends can be identified in broad categories. 422 But, for example, the PTO might consider different fees for descriptive terms and for any mark that is registrable only because of a disclaimer of words.

The PTO has also already begun to address the issues it faces with fraudulent filings, including from trademark extortionists capitalizing on Amazon's system and other fraudulent filings (many of which originate from China for other reasons). 423 In particular, the PTO has announced that it will seek to identify scams and other untoward filings and shuttle them to its newly-created Register Protection Unit. 424 Moreover, the PTO is working cooperatively with Amazon to identify fraudulent filings. 425 Beyond fraud wrought by trademark extortionists, the PTO ought to confront the other substantive issues raised by the shifts in filing for different types of marks. In particular, it ought to consider whether and how to apply different, more appropriate rules for determining the protectability of nonsense marks. It should first grapple with what qualifies as a nonsense mark. It should then assess whether to continue to classify such marks as fanciful and thus inherently distinctive. 426 While these terms are indeed coined, they are quite distinct from "conventional" fanciful marks like KODAK, EXXON, and PEPSI. Those words have no existing meaning, but they are pronounceable in English and more memorable than nonsense marks because they can be assimilated as words. 427 Indeed, nonsense marks do not sit easily within any of the Abercrombie distinctiveness categories. They are like generic terms or descriptive terms without secondary meaning, in that they do not identify source. But they are unlike generic or descriptive terms in that they do not have any alternate meaning. Nonsense marks should also frequently be rejected on failure-tofunction grounds. 428 Even if registration is allowed, the PTO should give more consideration to what makes a nonsense mark confusingly similar to another mark. Given that nonsense marks can only be compared in terms of sight and not sound or meaning, perhaps all nonsense marks are confusingly similar to one another as a mere jumble of letters.

The PTO should also reconsider the practice of registering generic or descriptive terms because of their stylization or accompanying images. As we noted, the justification for those registrations is that the scope can be limited to reflect the source-indicating value of the stylization or images. 429 But when those marks can effectively be enforced without the scope limitations, there is additional reason to be concerned about their registration even if the generic or descriptive terms are disclaimed. 430

The PTO cannot fully address Amazon's impact on the trademark system alone. Ideally, Amazon would play a role too. 431 Indeed, it might want to play such a role. Much of Amazon's effect on the trademark system is a byproduct of Amazon's legitimate interest in the concerns of its branded third-party sellers and the company's interest in avoiding government regulation. 432 In fact, Amazon has been piggybacking on the U.S. trademark system as a way to signal to regulators that it is taking the government's concerns with fraudulent goods in its marketplace seriously. 433 Instead of building its own fully functioning trademark system, Amazon is using the U.S. trademark system and its validation of trademarks by registration as the core of its Brand Registry. In this regard, Amazon should not be shocked that its sanctification of the U.S. trademark system has reverberated back into the U.S. trademark system itself. Amazon has a responsibility to support the U.S. trademark system in return for its piggybacking on it.

Beyond moral obligation, Amazon might be motivated by self-interest to cooperate more robustly with the PTO to smooth out the trademark system. In recent years, a number of Amazon's practices have been criticized as anticompetitive, most recently in an FTC lawsuit and congressional investigations. 434 If only for the public-relations benefits, Amazon might emphasize how it wants to help third-party sellers compete fairly on its platform. To that end, it could ramp up its cooperation with the PTO to improve its Brand Registry rules in ways that promote the operation of the trademark system.

Most fittingly, Amazon could help rein in the ways in which it is enabling the unfettered protection of descriptive and generic terms. For example, it could change its rules to allow for participation in the Brand Registry only when the corresponding mark is registered in plain text, without stylization or accompanying images. 435 In this way, third-party sellers would not be able to acquire rights to descriptive or generic terms in Amazon's Brand Registry without clearing them through the PTO. In most cases, given trademark's distinctiveness rules, they would not be able to get a PTO registration unless they have a descriptive term that has developed secondary meaning. 436 In this vein, Amazon also should clarify its rules on descriptive fair use 437 to ensure that, even when a third-party seller secures a PTO registration for a descriptive term, Amazon would not prioritize the third-party seller in its search results for that descriptive term. Moreover, in this instance, Amazon should allow other thirdparty sellers to use that term descriptively in its listings without fear of repercussion on the platform. By making these changes, Amazon might very well tamp down third-party sellers' incentive to seek registrations of descriptive and generic terms.

Amazon should also reconsider its policy of allowing businesses to participate in the Brand Registry with only a pending application for registration, as opposed to an issued registration. Amazon likely adopted this policy because of the long pendency of applications, a problem it is largely responsible for having created. But that policy exacerbates the "thicket" at the PTO. Additionally, it further encourages fraudulent applications and enables parties to enforce marks that they are unlikely to be able to register, enticing more businesses to file baseless applications.

More generally, Amazon should be more transparent about the ways in which it advantages Brand Registry participants and its rules of qualification, many aspects of which are not clear to the public. Transparency would better 434. Supra text accompanying notes 2-4. 435. Amazon might apply this rule to all marks seeking registration in its Brand Registry or only to those that are descriptive or generic. We thank Rebecca Tushnet for this suggestion.

436. Supra Part I.A. That is, the PTO is applying its rules in this regard properly and the distortion to the trademark system happens only because of Amazon's Brand Registry rules. Therefore, the focus should be on adjusting Amazon's rules.

enable the PTO and watchdogs to respond, when appropriate, to the effects of these rules on third-party sellers. 438 With regard to the delays in the PTO provoked by the Brand Registry's requirement that a seller obtain a PTO registration, and now merely have a pending application, perhaps Amazon ought to bear some of the cost of the delays it has caused in the PTO. 439 Such a tax could take many forms. For example, a tax could correspond to the number of applications filed in a year that are then registered in the Brand Registry. Alternatively, the PTO could consider changes to the fee structure that would specifically target the kinds of applications most likely attributable to Amazon policies.

As the driver of these changes, Amazon bears some responsibility for helping adjust the trademark system to its impact. However, doing so may require Amazon to take more of an enforcement role on its platform and to change its own successful business practices. Regardless, effectively responding to these changes and upholding the trademark system will require interaction and even collaboration between Amazon and the PTO.

The previous Parts focus on how Amazon's policies, and especially the Brand Registry, warp existing trademark law, and they offer suggestions about how to change either PTO rules or Amazon policies to better approximate the pre-Amazon balance in the trademark system. With this in mind, it is worth considering broader concerns about Amazon's impact on branding practice, and the implications for the role of trademark law in the future.

As we noted, Amazon's model is largely responsible for the rise of nonsense marks and has motivated sellers to claim descriptive and generic terms as marks. 440 Because consumers search for products on Amazon using either product categories, descriptive and generic terms, or consumer reviews, and because search results use that kind of information as much or more than brand information, Amazon sellers have diminished incentive to select a memorable brand name. Sellers need some mark that they can register in the PTO; nonsense, descriptive, or generic terms will often do. At the same time, Amazon's own branding practices, particularly its increasing emphasis on AMAZON-centered brands like AMAZON BASICS and AMAZON ESSENTIALS, and its preferencing of its own products in search results, have diminished the importance of third-party brands in favor of the Amazon brand.

On the one hand, these features have positive value for many smaller thirdparty sellers, whose brands typically cannot compete against large, better-known brands. Indeed, there is an important way in which Amazon's practices de-center brands altogether. Whereas search-costs theory has maintained that the value of a trademark is in its ability to decrease search costs by enabling consumers to use marks as a shorthand for product information, 441 Amazon's model makes that product information more directly available and the basis for algorithmic search. 442 The result is a more democratic marketplace, less dominated by big brands. For the many critics of expansive trademark protection and its contribution to an overly brand-focused economy, 443 that should sound like a win. 444 On the other hand, Amazon de-centers third-party brands in large part to center its own brand. In that respect, the effects are hardly democratic, as they further entrench the power of a dominant platform.

In terms of the net effects on competition, then, Amazon's practices are a mixed bag: Amazon to some extent enhances the visibility of third-party businesses, but it does so in a structure that primarily benefits Amazon and may ultimately promote Amazon's brand over all others. 445 In some ways, this is a long-term response to the rise of brands and their ascendancy vis-à-vis retailers, in which powerful brands wrested value away from retailers by reaching over the shoulders of the retailers and creating direct relationships with consumers who specifically demand those brands. 446 Amazon reverses those trends, reasserting the dominance of the platform over third-party brands. In that respect, one's views of Amazon's competitive effects are like a Rorschach test about Amazon: The combined effect of Amazon's policies is likely to be a more powerful Amazon, however one perceives the platform.

Perhaps more radically, we might see some of the de-centering of brands as evidence that Amazon is merely emblematic of a decreasing significance of the brand. In this sense, signs may be transforming into what Barton Beebe and one of us characterize in a different context as "indistinguishable ambient noise." 447 That would be welcome news for those persuaded by Naomi Klein's No Logo, which puts some of the blame for the costs of globalization on brands. 448 If the emergence of nonsense marks-not to mention the increased incentive to register descriptive and generic terms-indicates a broader decentering of brands, then Amazon's practices go to the very core of trademark law's justification. Trademarks have long been understood to have an important informational function: They are the shorthand for information about the qualities or characteristics of goods. 449 Critics have frequently observed that brands also, and maybe even primarily, create artificial product differentiation, allowing brand owners to extract value from consumers based on brand values that have little to do with underlying product quality. 450 To the extent those two functions are in conflict, we seem to accept the enablement of pure brand building as a necessary byproduct of protecting trademark law's central informational function. In Beebe's language, we have allowed mark owners to increase their ability to persuade so that they can assume consumers' search costs. 451 That balance may well need rethinking. If Amazon's algorithm and consumer reviews are as good or better at conveying information about products, then the informational function of brands is less important. In that respect, branding critics might celebrate the emergence of nonsense marks because they have the effect of de-centering brands, thereby reducing brands' ability to create artificial differentiation. In that sense, nonsense marks might seem like a partial antidote to trademark law's decades-long promotion of brand value, with everexpanding protections across a range of trademark doctrines. 452 Although largely the same formal legal rules apply to all marks, in practice the modern trademark system functions as two different systems: one for luxury brands like CHANEL, and one for the workaday brands. Many of trademark scope limitations those terms would ordinarily have. Amazon has increasedand really, created-incentives to claim nonsense marks. And businesses have more opportunities to extort legitimate Amazon sellers with fraudulent applications.

These effects may not have been intended by Amazon, but as our data show, they have had a profound impact on trademark registration practice. Some of the effects are problematic from a trademark perspective, and there are broader ruleof-law concerns about the fact that a single company's policies can so dramatically change the legal landscape. Though Amazon's market dominance has been unprecedented, the net effect of its policies on competition is potentially more mixed. On one hand, the collective effect of Amazon's policies, combined with Amazon's own concentration of its branding practices for its own goods, has been to deemphasize third-party brands on the platform. In some ways, this democratizing effect enables smaller parties to better compete with established brands. On the other hand, Amazon's practices de-emphasize third-party brands to the benefit of Amazon, whose algorithm chooses how to present the goods sold on its platform-regardless of brand strength, for example-and allows Amazon to prioritize its house-branded goods over those offered by third-party sellers. In that respect, the diminished significance of third-party brands is likely to further entrench Amazon and enhance its dominance.

Amazon does not seem to be slowing down anytime soon. As the platform keeps pushing its Brand Registry, sellers will continue to find reasons to participate in the program to protect their marks and remain competitive. While the full consequences of these practices are yet to be seen, as this Article shows, Amazon affects much more than the sellers on its platform: It is starting to reshape the normative and structural underpinnings of the U.S. trademark system. In the end, whether and how we respond to these profound effects on the trademark system depends on the extent to which we consider Amazon's overhaul troublesome for the goals that drive our trademark system.