Fake Trademark Specimens: An Empirical Study

In previous work, we studied "trademark depletion," the decreasing supply of unclaimed, competitively effective trademarks available to new market entrants. 1 The empirical results we reported confirm that it is becoming increasingly difficult for new commercial entities, and [Vol. 120:217 particularly small businesses, 2 to find a trademark that is competitively effective (in that the mark is reasonably marketable) but that has not yet been claimed by another commercial entity. 3 The problem of trademark depletion has been exacerbated in recent years by a surge of fraudulent applications originating in China. 4 In this Piece, we report the results of new empirical work showing that a substantial proportion of applications originating in China include fraudulent specimens of use. In particular, with respect to use-based applications originating in China that were filed at the U.S. Patent and Trademark Office (PTO) in 2017 solely in Class 25 (apparel goods), we estimate that 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.

In 2017, 42,728 use-based U.S. trademark applications originated in China. If Class 25 applications are representative of the overall population of Chinese-origin applications in that year, then approximately 28,585 of such applications were fraudulent. This number represents about 14.0% of total use-based applications filed with the PTO in 2017.

When fraudulent applications are allowed to register, they contribute to a problem closely related to the problem of trademark depletion, which is the problem of "clutter" on the Principal Register, the primary register of trademarks maintained by the PTO. 5 "Clutter" is the term that 2. Leah Chan Grinvald, Shaming Trademark Bullies, 2011 Wis. L. Rev. 625, 694 & n.139 (defining small businesses as having "few or no financial resources to defend against trademark bullying [by large corporations]").

3. Beebe & Fromer, supra note 1, at 951-53 ("The supply of word marks that are at least reasonably competitively effective as trademarks is finite and exhaustible . . . . [N]ew trademark applicants are increasingly being forced to resort to second-best, less competitive marks, and the trademark system is growing increasingly-perhaps inordinately-crowded, noisy, and complex.").

4. See, e.g., Jacob Gershman, Flood of Trademark Applications from China Alarms U.S. Officials, Wall St. J. (May 5, 2018), https://www.wsj.com/articles/flood-of-trademark-applications-fromchinaalarms-u-s-officials-1525521600 (on file with the Columbia Law Review) ("The Patent and Trademark Office has found numerous instances of Chinese applicants asserting that a proposed trademark is used in commerce, while submitting multiple nearly-identical images of the same consumer product with a different word on the brand tag."); Nikkei Staff Writers, Chinese Trademark Filings Flood Foreign Markets, Nikkei Asian Rev. (June 14, 2018), https://asia.nikkei.com/Business/Business-trends/Chinese-trademark-filings-flood-foreign-markets [https://perma.cc/3AK2-3V7M] ("Many of the applications in the U.S. come from small, obscure online retailers . . . using images of nearly identical products differing only in the brand on the tag.").

5. See PTO, U.S. Dep't of Com., TMEP § 801.02 (Oct. 2018) (describing the PTO's procedure to direct applications to the Principal Register if not specified otherwise); Beebe & Fromer, supra note 1, at 956 n.45 ("Marks that meet all requirements for registration are registered on the Principal Register.").

trademark law uses to describe marks that are registered but that are not actually used in commerce. 6 Both the U.S. House of Representatives and Senate recently introduced legislation designed to reduce filings with fraudulent trademark specimens and more easily remove registrations based on them from the register, 7 following on the heels of congressional hearings held to learn more about fraudulent Chinese trademark specimens. 8 For its part, the PTO has insisted it has already and will continue to undertake reforms to address the problem. 9 Citing our findings, one court has recently ruled that a U.S. trademark registration granted to a Chinese individual is likely to be canceled on the ground that it contains a fraudulent specimen. 10 Also relying on our findings, the media has reported that fraudulent trademark filings are "key to the trade dispute between the US and China." 11 In the recently concluded Phase One trade deal between the United States and China, China explicitly agreed to "ensure adequate and [Vol. 120:217 effective protection and enforcement of trademark rights, particularly against bad faith trademark registrations." 12 The extent of the problem of fraudulent specimens presented in this Piece suggests that legislative and regulatory reforms are necessary to preserve the proper functioning of the trademark system and advance its core purposes of promoting competition and enhancing consumer welfare.

In Part I of what follows, this Piece provides a brief overview of American trademark law, reviews the main findings of our empirical research on trademark depletion, and demonstrates the effect of trademark depletion on small businesses. Part II focuses on the recent influx of fraudulent trademark applications originating from China. Part III discusses the costs of trademark depletion and clutter on the Principal Register. Part IV addresses reforms that should help to improve the efficiency, integrity, and fairness of the trademark registration system.

Trademark law protects designations of commercial source, which are typically words or images. 13 For example, the word APPLE is a registered trademark of Apple Inc., 14 as is the bitten-apple symbol that the company uses. 15 The canary-yellow color of Post-it Notes is a registered trademark of 3M Company. 16 Trademark law protects such designations to promote fair competition and lower consumers' search costs in locating products of interest. 17 Section I.A provides an overview of applicable requirements for trademark registration, and section I.B examines challenges that new businesses face in the current registration system.

To qualify for registration at the PTO, a trademark must be used in commerce. 18 The use-in-commerce requirement ensures that only those (1) (providing for the registration of a mark "used in commerce"); id. § 1051(d) (providing for the registration of a mark filed on an intent-to-making actual use of a trademark can claim property rights in it. This requirement is of the utmost importance in preventing the "warehousing" or "squatting" of registered but unused marks. 19 In this respect, it ensures that the Principal Register is not cluttered with trademark registrations by businesses that are not actually using their marks in commerce. To satisfy the use-in-commerce requirement, the applicant must submit one or more specimens of use of its mark in commerce. 20 This typically takes the form of a digital image of the mark affixed to the applicant's goods or appearing in connection with the applicant's services. As we discuss below in Part II, fraudulent applications typically include fraudulent specimens of use that have been digitally altered or otherwise manipulated to falsely demonstrate use in commerce.

use basis upon filing of a statement that the mark is "used in commerce"). Certain foreign applications need not meet the use-in-commerce requirement to receive registration. See id. § 1126(e) (providing for the registration of a mark already registered in certain foreign jurisdictions provided that the applicant has a bona fide intention to use the mark in commerce in the United States, and specifying that "use in commerce shall not be required prior to registration"); id. § 1141f(a) (providing for the registration of a mark under the Madrid Protocol system provided that the applicant has a bona fide intention to use the mark in commerce in the United States). Applications filed under § 1126(e) (so-called "Section 44(e) applications") are rare. The PTO data indicate that, of the 385,249 Principal Register trademark applications filed in 2016, 5,585 were filed on this basis. Beebe & Fromer, supra note 1, at 955 n.32. Applications filed under § 1141f(a) (so-called "Section 66(a) applications") are also rare. There were 15,374 such applications filed in 2016. Id Applications are evaluated under the Lanham Act, the federal statute that governs the trademark system. 21 A mark must also not violate any of the Lanham Act's various statutory bars to protection, including Section 2(d)'s bar on registering a mark that is confusingly similar to an alreadyregistered mark. 22 Among other things, to register a mark, an applicant must set forth a written description of "the goods [and services] in connection with which the mark is used." 23 The applicant must also indicate in which of the forty-five classes in the so-called "Nice Classification" scheme the goods or services are classified. 24 Among the Nice classes that boast the highest number of registrations at the PTO is Class 25, for apparel goods. 25 The PTO publishes marks in the Official Gazette if it determines that they qualify for registration. 26 Within thirty days of the date of publication, third parties may oppose the registration. 27 If no opposition is brought or succeeds, use-based applications will then proceed to registration. 28 A trademark need not be registered to be protected by the Lanham Act. 29 Nonetheless, the law provides a variety of incentives to promote registration. Registration confers on the mark a prima facie presumption of validity and of the registrant's ownership of the mark. 30 It also confers on the registrant nationwide priority in the mark as of the date of application. 31 Furthermore, only registered marks may qualify for incontestable status, and only the owners of registered marks may qualify for statutory damages against counterfeiters of those marks. 32

Although trademark law has operated on the assumption that there exists an inexhaustible supply of unclaimed trademarks, the results we report in empirical work published in 2018 confirm the opposite: It is becoming increasingly difficult for new businesses to find a trademark that is competitively effective but that has not yet been claimed by another business. 33 To give a sense of the state of depletion, when people living in the United States speak English, three-quarters of the time they are using a word that someone has already claimed as a trademark. 34 Also, a majority of American residents carry a surname that is already claimed as a mark, meaning they may have been born too late to claim their own surname as a mark. 35 Even using a conservative similarity-matching protocol, nearly all the words Americans use on a daily basis and the surnames of a high proportion of American people are already registered or are confusingly similar to an already-registered mark. 36 Meanwhile, new applicants are increasingly shifting toward coined words and longer, more complexand thus less effective-marks. 37 Yet doing so does not appear to be succeeding. PTO refusal rates for confusing similarity to an alreadyregistered mark continue to rise. 38 The data suggest that an increasing proportion of trademark registration applications at the PTO are coming from small businesses. Applications from filers who have filed only a single application at the PTO from 1985 through 2016 may function as a proxy for small-business applicants. Figure 1 shows that, since the mid-1990s, the proportion of single-filer applications has been steadily increasing, to nearly one in three in 2016. The data indicate that such applicants are less successful in registering their trademarks. From 1985 through 2016, single-filers achieved a publication rate of 67.6%, while non-single-filers (that is, entities filing more than one trademark application during the period) achieved a significantly higher publication rate of 78.9%. Single-filers also suffered higher Section 2(d) refusal rates. Over the period 2003 through 2017, 14.9% of single-filer applications received a Section 2(d) refusal, while only 12.8% of non-single-filer applications did so. Many factors may account for these differences, but the depletion of available, competitively effective trademarks from which new businesses can choose has likely played a significant role in making it more difficult for firms to register new marks. This is particularly true for small businesses, which typically lack the sophistication and resources to find unclaimed, competitively effective marks and to prosecute those marks to registration.

These findings belie many of the basic assumptions underlying trademark law and policy. Most significantly, they show that the granting of exclusive rights in words is not costless. The supply of commerciallyviable words is exhaustible, and we are reaching a level of economic development that has begun to test the limits of our resource of words. Small businesses have proven to be especially affected by this development. As Part III discusses, trademark depletion undermines trademark law's goal of promoting fair competition by raising entry barriers for new businesses with regard to finding and claiming unclaimed, competitively effective marks. 39

A surge in fraudulent trademark registrations originating from China is making the problem of trademark depletion significantly worse. Since 1985, when trademark applications from China represented just over 0.07% of PTO trademark applications (or 42 overall applications), 39. Id. at 1021-26. Some explain this change in filing patterns as nothing other than more Chinese-branded goods and services being sold in the American market, including on platforms like Amazon. 40 Even so, there is broad suspicion that a significant part of this rise is attributable to fraudulent applications with false specimens of use that do not comply with U.S. trademark law's requirement that a mark be used in commerce in connection with goods or services to be registered. 41 This suspicion stems from evidence that some regional Chinese governments are offering their citizens a subsidy of approximately $800 for each U.S. trademark

Filing Year registration secured. 42 With a filing fee of under $300 for use-based applications, an applicant taking advantage of this subsidy would likely prefer to pocket the remainder of the subsidy rather than incur the additional cost of making actual use of the mark in U.S. commerce, which may lead applicants to submit false specimens of use. 43 The incentive to take advantage of these subsidy programs is great. As trademark attorney Josh Gerben explains, "Given that the median monthly income for a Chinese citizen is around $1,000, the government payments make it possible for someone in China to have a full time income by registering just two US trademarks per month." 44 This Part seeks to move beyond speculation as to the extent of fraudulent trademark applications from China by studying the issue empirically. It focuses on applications that include fraudulent specimens of use. As explained above, to qualify for registration, trademark applicants must submit images showing their use of the applied-for mark in commerce. 45 In an effort to satisfy this requirement, fraudulent applications typically include digitally altered or otherwise improperly manipulated images.

Section II.A below reports the results of our empirical study of fraudulent specimens of use in applications originating from China. In the study, we focus on use-based applications originating in China that were filed at the PTO in 2017 solely in Class 25 (apparel goods). 46 We estimate that 66.9% of such applications included fraudulent specimens. The PTO issued refusals on the basis that the applications included fraudulent specimens of use to only 13.4% of these applications. Moreover, of the applications in our sample that appear to have included fraudulent specimens, an extraordinary 59.8% proceeded to publication and then 38.9% proceeded to registration. We conclude that, at least in Class 25, a substantial proportion of fraudulent applications originating from China are surviving PTO review and proceeding to registration. Section II.B sets forth general data on the increasing number of refusals that the PTO has issued to trademark registration applications on the basis that the applications included fraudulent specimens of use. These general data present a disturbing picture, but given our study reported in section II.A, it appears that they significantly underestimate the scale of the problem.

To estimate the prevalence of fraudulent specimens, we randomly sampled 365 applications from the 6,752 use-based applications that originated in China and were filed at the PTO in 2017 solely in Class 25. 47 We reviewed all submitted specimens for each application in our sample and coded them for multiple factors that indicate to varying degrees the probability that the application's specimen is fraudulent. These factors consist of whether a specimen image:

1. Showed discontinuities that indicated digital alteration; 2.

Appeared in a Google reverse image search as matching an image of a product with a different mark of a different company, 48 or the specimen image matched a specimen image previously submitted to the PTO in connection with a different mark of a different company; 3.

Depicted a mark consisting of a nonsense word that is unpronounceable in English and that the applicant indicated has no meaning in any other language; 49 4.

Depicted a tag with irregularities, such as exceptionally poor print quality, the tag being placed on top of another tag, or tags across multiple specimen images having different appearances;

47. This sample allows us to estimate the proportion of fraudulent specimens for the entire population of Chinese use-based applications filed in 2017 solely in Class 25 with a confidence level of 95% and a margin of error of 5%.

48. Google operates a free online search service that allows users to reverse image search by uploading images and searching the internet for identical or similar images. Google Images, https://www.google.com/imghp [https://perma.cc/P2EL-L8Z6] (last visited Oct. 3, 2020). Google states that "[t]he pictures you upload in your search may be stored by Google for 7 days. They won't be a part of your search history, and we'll only use them during that time to make our products and services better." Find Related Images with Reverse Image Search, Google Search Help, https://support.google.com/websearch/ answer/1325808 [https://perma.cc/A57Y-ZXD7] (last visited Aug. 13, 2020).

49. Nonsense words can be consistent with legitimate use in commerce, particularly for business conducted on online platforms such as Amazon where traditional branding may be less relevant and so-called "pseudobrands" may suffice. See Herrman, supra note 40. For this reason, we report one set of findings based on inclusion of nonsense words as an indicator of probable fraudulence and another set of findings that does not rely on nonsense words. Although we are unable to measure it, we suspect that an appreciable portion of nonsense marks are fraudulent because they are easier to register without encountering a refusal for confusing similarity with an already-registered mark. At the same time, we appreciate that some portion of nonsense marks are legitimately being used in commerce, as suggested by searches on Amazon for various everyday items. See id. In any event, many of these marks may not merit trademark registration because consumers are likely to perceive them as barcodes or extraneous matter rather than source indicators. They would therefore fail to function as trademarks. See Alexandra J. Roberts, Trademark Failure to Function, 104 Iowa L. Rev. 1977, 1981 n.11 (2019) ("To be protectable, matter must be used in such a way that consumers will understand it as a trademark."). As such, it is critical that trademark examiners scrutinize all nonsense marks, to assess both fraudulence and capacity to function as a trademark.

Depicted pricing in a foreign rather than U.S. currency or commerce with delivery to a foreign rather than a U.S. address; 6.

Depicted a product that carried the mark of another company (such as CHANEL), indicating that the applicant was affixing its applied-for mark to a different company's product; 7.

Depicted a branding card not attached to the good; 9.

Depicted a screenshot of an ecommerce website (such as Amazon) to show use in commerce; 10.

Depicted a collar label to display the mark. Based on our sample, it is reasonable to conclude that a very high proportion of Chinese use-based applications filed in 2017 solely in Class 25 were fraudulent. The PTO itself refused 13.4% (or 49) of the 365 applications included in our sample on the basis that the submitted specimen image was digitally altered or otherwise improperly manipulated. 50 But this statistic significantly understates the degree of the problem. Taking into account a subset of the factors listed above that are especially indicative of fraud, we estimate that 66.9% of Chinese use-based applications filed in 2017 solely in Class 25 included fraudulent specimens. Specifically, 66.9% (or 244) of the applications in our sample included a specimen with a discontinuous image, a matching reverse-image-search image, a price displayed in foreign currency or delivery to a foreign address, the mark of another company, a nonsense word for a mark, tag irregularities, a misspelled mark, and/or a branding card not attached to the good. Many of these 244 applications raised multiple such red flags. Nonetheless, 59.8% proceeded to publication and then 38.9% proceeded to registration. 51 What follows provides more detail on certain of these factors. 1. Discontinuities in Specimen Images Indicating Digital Alterations. -We found that 26.0% (or 95) of the applications in our sample included specimen images showing discontinuities that indicated digital alteration. For example, the specimen shown below for Application Serial No.

50. With respect to the entire population of 6,752 Chinese use-based applications filed in 2017 solely in Class 25 from which we sampled, 899 (or 13.3%) of these applications received a digital-alteration refusal.

51. If one excludes nonsense words as an indicator of probable fraudulence, we estimate that 39.7% of Chinese use-based applications filed in 2017 solely in Class 25 include fraudulent specimens. Specifically, 39.7% (or 145) of the applications in the sample included a specimen with a discontinuous image, a matching reverse-image-search image, a price displayed in foreign currency or delivery to a foreign address, the mark of another company, tag irregularities, a misspelled mark, and/or a branding card not attached to the good. Nonetheless, 44.8% of the applications proceeded to publication and 24.1% then proceeded to registration. 87,300,262 for INMUINS and accompanying logo has discontinuities indicative of alteration, yet the PTO did not refuse registration on that basis, and the application eventually registered as Registration No. 5,275,912.

Submitted Specimens for Different Marks. -We further found that 12.6% (or 46) of the applications in our sample included specimen images that either matched a Google reverse-image-search image associated with a different company or mark or matched another specimen previously submitted to the PTO for another mark. 52 For example, shown below on Two other examples may be instructive. The set of images below depicts, on the left, the specimen submitted in connection with Application Serial No. 87,362,577 for BEAL, and on the right, an image yielded by a Google reverse image search of the specimen image. The reverse-image-search image shows the same image but with Burberry branding. 54 The PTO refused this application on the basis that the specimen was digitally altered but did not detect the reverse-image match. Similarly, Application Serial No. 87,309,926 for FIDIKO included the specimen image shown below on the left. On the right appears the reverseimage-search match as it appears with other branding online. 55 The PTO refused this application on the basis that the specimen was digitally altered, but it did not detect the reverse-image match. specimen-related refusals of this application but eventually registered the mark on April 16, 2019, as Registration No. 5,728,428. 56 3. Marks Consisting of Unpronounceable Nonsense Words. -We found that 44.4% (or 162) of the applications in our sample use a mark that consists of a nonsense word that is unpronounceable in English and that the applicant indicated has no meaning in any other language. Some examples of these nonsense marks in our sample are ALSYIQI, KELUOSIBODE, KXCFCYS, KIEDVLI, and KJAHSLK. 57 Although some businesses legitimately use nonsense words as marks, particularly in selling their goods on platforms like Amazon, choosing a mark made up of complex gobbledygook tends to indicate fraudulence because its owner would not be able to use it as an effective trademark in the marketplace. 58 4. Hang Tags, Stickers, Collar Labels, Ecommerce Websites, and Foreign Pricing. -Approximately 55.3% (or 202) of the applications in our sample submitted a specimen with a hanging tag showing the mark; 4.7% (or 17) submitted a specimen with a sticker showing the mark; and 32.6% (or 119) submitted a specimen with a collar label showing the mark. Although these three factors can be consistent with legitimate use in commerce, the presence of any one of these factors raises a red flag suggesting the need for further inquiry. Some of these tags had irregularities that we thought raised yet more significant red flags. We found tag irregularities in 3.0% (or 11) of the applications in our sample. These irregularities include the mark appearing crooked on the tag, the print quality of the mark on the tag 56. Along similar lines, in granting a preliminary injunction, a district court recently ruled 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 in light of the fact that the defendant fraudulently used a photograph of the plaintiff's product as its specimen of use. being exceptionally poor, the tag being placed atop a preexisting tag, tags across multiple specimens for the same mark having different appearances, and a tag being ripped off of the specimen. Shown below is an example of collar labels of different appearances across the specimens (U.S. Trademark Application Serial No. 87,544,646 for MAXSOFT, with two different styles of collar labels across two specimens). The examiner did not note this inconsistency, and the mark was registered as Registration No. 5,658,920.

We further found that 15.1% (or 55) of the applications in our sample submitted a screenshot of a good for sale on an ecommerce site, such as Amazon, to show use in commerce. Here, too, specimen images in the form of ecommerce websites can be consistent with legitimate use, but also raise concerns that merit further inquiry.

Additionally, 1.4% (or 5) of the applications in our sample submitted a specimen depicting pricing in a foreign currency or commerce with delivery to a foreign rather than a U.S. address. This may indicate use in commerce, but not use in U.S. commerce. An example (Application Serial No. 87,359,150 for COTTON COMING) is below. The tag depicted in the specimen shows a price of twenty-eight yuan. The PTO refused to register this application on the basis that the specimen image was digitally altered. The applicant then abandoned the application.

As another example, consider Application Serial No. 87,706,203 for ROOZOOE. The application included the specimen shown below, which displayed a hangtag for ROOZOOE but also displayed the mark of CHANEL repeatedly along the sleeves, something noted by the examining attorney in their refusal of registration due to the deficiency of the specimen.

6. Additional Indicia of Fraudulence. -In addition to those already discussed, other factors suggested that certain specimens in our sample were fraudulent. In two applications (0.6% of our sample of 365), the specimens displayed a misspelling of the mark. For example, shown below is a specimen image from Application Serial No. 87,534,972 in which the applied-for mark CYCLING STARS was misspelled on the specimen product as "Cyling Stars." The application was registered as Registration No. 5,411,978. Also shown below is the specimen image from another application, which used a fake store as a specimen (Application Serial No. 87,339,443, which registered as Registration No. 5,349,124).

With respect to all trademark applications (not just our sample, and not just those originating in China), our dataset of all trademark office actions issued by the PTO from 2003 through 2016 enables us to estimate the number of office actions that the PTO has issued on the basis that the application included a digitally-altered specimen. Our data indicate that the PTO first issued an office action on this basis in 2012 in connection with Application Serial No. 85,549,660 for the mark ALTER EGO, which originated in the United States. Applications originating in China account for a very large proportion of applications receiving digital-alteration refusals. Specifically, 9,229 (or 71.1%) of the 12,973 applications filed from 2012 through 2017 that received digital-alteration refusals were Chinese in origin. Figure 3 below shows the significant increase since 2005 in the number of Chinese-origin applications (in any class) and in the number of digital-alteration refusals issued to those applications. As for the distribution by class of goods or services of applications receiving a digital-alteration refusal, Figure 4 below indicates the proportion and number of applications in each class filed at the PTO (from all countries, not just from China) from 2014 through 2017 that received a digital-alteration refusal. Class 9 (electronics goods) had the highest number of digital-alteration refusals with 3,062, and these constituted 1.3% of all applications filed in that class for the time period. Class 25 (apparel goods) also had a very high number of digital-alteration refusals, with 2,180, and these constituted 1.3% of all applications filed in that class for the time period. Figure 5 shows comparable data for applications originating in China. 60 Certain classes show high proportions of applications originating from China receiving digital-alteration refusals. For example, 14.1% of Chinese applications in Class 21 (including household utensils, glassware, and porcelain) received digital-alteration refusals, and 13.0% of applications in Class 8 (including cutlery) received digital-alteration refusals.

60. Note that the scale of the x-axis is different from that in the previous figure. Applications may also be refused for including fake specimens for reasons other than the specimen image being digitally altered. For example, an unaltered specimen image may show the applied-for mark being used along with some other company's mark (as in the CHANEL example above), and on that basis the PTO may refuse registration. Our study reported in section II.A suggests how common fraudulent trademark specimens are in use-based applications originating from China, at least those filed solely in Class 25 (apparel goods). In 2017, 42,728 use-based U.S. trademark applications originated in China. If Class 25 applications are representative of the overall population of Chineseorigin applications in that year, then approximately 28,585 such applications were fraudulent. This number represents about 14.0% of total use-based applications filed in 2017. 61 Applicants submitting fraudulent specimens are not demonstrating the requisite use in commerce. 62 Each such application that proceeds to registration contributes to trademark clutter by adding a mark not being used in commerce on the Principal Register. Trademark clutter in turn worsens trademark depletion by making fewer marks available for businesses that legitimately want to use one of these registered marks. Although the PTO is detecting some fraudulent specimens at increasingly elevated levels and refusing applications on that basis, as shown in section II.B, our study in section II.A suggests that there is very possibly much more fraudulence in specimens than the PTO is currently detecting.

There are a series of practical implications that flow from increasing rates of trademark depletion and clutter on the Principal Register. Section III.A discusses the anticompetitive costs of trademark depletion and the increased barriers that new firms face upon entering the market. Section III.B examines the impact of clutter on the use, accuracy, and reliability of the Principal Register.

Consider first trademark depletion. As we discuss in previous work, 63 trademark depletion is a problem because it undermines trademark law's goals of promoting efficient and fair competition and minimizing consumer search costs. In particular, with increasing depletion, new market entrants face higher costs than incumbents did previously in developing a mark that is not confusingly similar with an already-registered mark and that is competitively effective. 64 constrained to settle for less effective marks, such as longer and more complex marks, that minimize the advertising power of these marks. 65 Although new firms are still finding trademarks to register and are managing to compete, the negative effects of trademark depletion continue to worsen:

[A]n insidious quality of depletion is that it proceeds gradually, and even though its pace has quickened in recent years, it remains a chronic rather than acute condition. We should expect no tipping point or moment of crisis in which there are suddenly no trademarks left at all and competition grinds to a halt. Instead, we should expect what the data report: a continuous process in which individual applicants are still able to find usable marks, but at ever-greater cost in pursuit of ever-less benefit. 66 Trademark depletion also increases consumer search costs-and in a similarly gradual way-because consumers must cope with less efficient marks and a more crowded field of marks in the marketplace.

The effects of trademark depletion on small businesses in particular are especially troubling. As section I.B shows, small businesses tend to have more difficulty registering their trademarks on the Principal Register. 67 As the supply of unclaimed, competitively effective marks continues to decline, these difficulties will only increase.

Trademark depletion causes harmful effects even when all registered trademarks are in compliance with trademark law's requirements, including use in commerce. Compounding these problems, however, it has become clear that some sizeable portion of registered marks are not in such compliance. Beginning in 2012, the PTO conducted a two-year pilot program that audited a random sample of 500 trademark registrations whose owners had filed a six-year declaration of continuing use to determine if the registrations satisfied the use-in-commerce requirement for the goods and services referenced in the registration. 68 In explaining the purpose of the program, the PTO expressed concern that a substantial proportion of registered marks were not being used in commerce for all or even any of the goods specified in the registration, with the result that such registrations were cluttering the register and preventing new market entrants from adopting the marks. 69 clutter that the PTO identified align with those associated with trademark depletion as well:

The accuracy of the trademark register as a reflection of marks that are actually in use in the United States for the goods/services identified in the registration serves an important purpose for the public. The public relies on the register to clear trademarks that they may wish to adopt or are already using. Where a party searching the register uncovers a similar mark, registered for goods or services that may result in confusion of consumers, that party may incur a variety of resulting costs and burdens, such as changing plans to avoid use of the mark, investigative costs to determine how the similar mark is actually used and assess the nature of any conflict, or cancellation proceedings or other litigation to resolve a dispute over the mark. If a registered mark is not actually in use in the United States, or is not in use on all the goods/services recited in the registration, these types of costs and burdens may be incurred unnecessarily. Thus, accuracy and reliability of the trademark register help avoid such needless costs and burdens, and thereby benefit the public. 70 The results of the pilot program were concerning. 71 Of the audited registrations, 50% did not meet the use-in-commerce requirement for all of the goods and services specified in the registration. 72 Indeed, 16% of the registrants failed to respond and their registrations were cancelled outright. 73 An additional 34% of the registrations were amended and narrowed to the goods and services in connection with which the marks were actually being used. 74 The problem of "deadwood" registrations prompted the PTO to make its auditing program permanent. 75 Under a recent rule change, each year the office will randomly audit up to 10% of continuing-use affidavits for registrations in which multiple goods or services are claimed in particular classes. 76 70. Id [Vol. 120:217 In sum, it is important to declutter the trademark register both to decrease-or at least not increase-rates of trademark depletion and to maintain the register's integrity. 77 Furthermore, decluttering the register will improve competition in the marketplace. 78 In particular, it will help to lower depletion-related barriers to entry that benefit incumbent businesses at the expense of new entrants and benefit large businesses at the expense of smaller ones. 79 A decluttered register also ensures that the Principal Register can be trusted and can be used as a reliable resource by businesses searching it to see which marks have already been claimed. 80 IV. REFORMS TO TRADEMARK LAW Because the worsening extent of trademark depletion and clutter is impairing the integrity of the register, damaging competition, and harming small businesses, we support a number of reforms to trademark law that build upon proposals for which we advocated in our previous work on depletion. 81 These reforms will help to clear the Principal Register of unused marks and thereby mitigate disturbing historical trends we identify in our Piece's empirical study that reduce competition and that hurt consumers.

Although we are sensitive to an argument that trademark registrants have settled expectations in the continued registration of their marks, those arguments hold little to no weight in the face of the challenges of trademark depletion and clutter on the Principal Register. For one thing, any settled expectations that trademark law has fostered in registrations occurred long ago based on very different background assumptions: that the supply of competitively effective trademarks was inexhaustible and that there were no resource constraints in granting new trademark rights. The increasing and alarming rate of trademark depletion shows that these assumptions are no longer justified, if they ever were. Furthermore, there should be no settled expectations in trademark rights that were never warranted or are no longer warranted based on trademark law's clear substantive requirements, such as that a mark be used in commerce. 83 If a registrant is not using a mark in commerce, the mark does not deserve to remain on the Principal Register.

Moreover, recent reforms by the PTO are not sufficient to stem the tide of fraudulent specimens being filed. Recently, the PTO created a "streamlined protest procedure for reporting improper specimens." 84 Third parties can submit by email objective evidence showing that a submitted specimen identically matches an image used by others (without the applicant's trademark), but they must do so "no later than the 30th day after publication for opposition." 85 Though it is a move in the right direction, this pilot program gives very little time to third parties to detect and report fraudulence. The program will likely be of minimal help in reducing fraudulent trademark registrations.

In response to the high proportion of fraudulent applications filed by applicants based in China, the PTO implemented in August 2019 a requirement that all foreign-domiciled parties appearing before the Office be represented by a U.S.-licensed attorney. 86 Media reports suggest that the rule has reduced the number of fraudulent applications, but as a recent World Trademark Review article reports, the "scourge of suspicious specimens continues." 87 The article records numerous examples of fraudulent specimens filed after the effective date of the U.S.-licensed attorney rule. 88 Whether by improperly borrowing the credentials of U.S. attorneys or by finding U.S. attorneys willing to file these specimens, foreign fraudulent filers are likely to find ways to circumvent this new rule. While it will no doubt reduce fraudulent trademark filings to some degree, the U.S. attorney rule alone will not likely eliminate, or at least significantly diminish, fraudulent filings. 89

Two reforms we support are new processes for ex parte reexamination of registered marks and expungement of registered marks that were not used in commerce before registration and should not have been registered in the first instance. Recently introduced federal legislation in the Trademark Modernization Act of 2020 would add both of these processes. 90 Trademark law currently provides third parties with two opportunities to ensure that a mark is not improperly on the Principal Register: opposition and cancellation. 91 Both are deficient in important respects. Opposition, discussed above, provides an opportunity for third parties to prevent a mark from being registered in the first instance. 92 Any party that believes it would be harmed by the registration may file an opposition to the registration of the mark within thirty days of the mark's publication. 93 If the registration is successfully opposed, the mark does not proceed to registration. 94 For all marks that proceed to registration, for a five-year period following the date of registration, a third party may petition to cancel the registration on any basis. 95 After five years have passed from the date of registration, a third party may petition to cancel a registration for only a limited number of reasons. 96 At this juncture, a third party cannot petition to cancel the registration on the ground that the mark is merely descriptive (and consumers have not learned that it is a designation of source) or on the ground that the registered mark is confusingly similar with a previously used mark.

Opposition and cancellation are insufficient in crucial ways, including the short time period in which an opposition can be filed and the limited grounds for cancellation after five years of registration. But their biggest limitation is their high cost, which can be burdensome to smaller businesses. One study reports that the average cost of a trademark opposition in the United States ranges from $100,000 to $325,000 97 established by trademark registration. Unused marks will continue to clutter the Principal Register. This clutter will continue to worsen the problem of trademark depletion discussed above and the difficulty small and new businesses face in finding good trademarks that whereas another estimates the median cost of a U.S. opposition or cancellation proceeding to be $95,000. 98 According to one trademark scholar, the cost of opposition or cancellation as they stand "remains above the level that some small businesses will be able to afford." 99 A study also finds that the cost of opposition proceedings in the United States is much higher than in other jurisdictions, including Australia, Brazil, Canada, the European Union, and Japan. 100 To help diminish rates of trademark depletion and bolster the integrity of the register, we therefore support the provision of less expensive, comprehensive ex parte proceedings to allow third parties to remove improperly registered trademarks from the register. It is crucial that trademark law make it easier to cull marks from the Principal Register that are not in use, that have been abandoned, or that otherwise do not warrant registration, freeing them for use by others. Businesses could then have a broader pool of competitively effective marks from which to choose new marks, both benefiting competition and lowering consumer search costs. Moreover, such reforms would boost the integrity of the register. In these respects, two laudable additions are the processes provided for in the Trademark Modernization Act of 2020 of ex parte reexamination of registered marks and expungement of registered marks that were not used in commerce before registration and should not have been registered in the first instance.

After the PTO has issued a trademark registration, its ability to cancel the registration on its own initiative is currently severely limited, even when the PTO has itself discovered new facts that show that the mark should not have been registered. During the sixth, tenth, and each successive tenth year following the date of registration, the registrant must file an affidavit verifying that it continues to use the mark in commerce. 101 This affidavit must include specimens of use. 102 The PTO's review of these affidavits and specimens provides the only means by which it can cancel a registration on its own initiative. 103 In essence, if the PTO becomes aware of new facts that show that a mark should not have been registered, it must either rely on third-party challenges to the registration or wait until the registrant is required to file an affidavit of continuing use.

By contrast, the PTO currently has significant authority to reexamine patents on its own initiative at any time. Section 303(a) of the Patent Act provides the Director with the authority "[o]n his own initiative, and any time" to determine whether prior art "discovered by him" raises a "substantial new question of patentability." 104 Based on our study of fraudulent trademark applications originating in China that Part II discusses, it is likely that trademark examining attorneys within the PTO will periodically become aware through their own research of facts that will lead them to believe that previous registrations should not have been issued. For example, in comparing pending applications to previous registrations, examining attorneys may recognize the use by multiple applicants or registrants of the same specimen-of-use images, each slightly digitally altered to show a different trademark. Under current law, the PTO has no authority after a registration has issued to demand further information from any registrant who submitted suspect specimen-of-use images. 105 We therefore support reforms that would provide the PTO with this authority, as the Trademark Modernization Act of 2020 does. 106 Furthermore, we anticipate that this authority may become more useful as the PTO develops enhanced technological means to compare specimen-of-use images and to expose other modes of fraudulent conduct.

Our study shows that an appreciable number of trademark applications originating in China for apparel goods in 2017 are likely to be fraudulent. These applications, particularly the large proportion that succeed to registration, exacerbate the problems of trademark depletion action against defunct marks which appear in registrations."). The PTO has relied on 15 U.S.C. § 1058 to cancel registrations, in whole or in part, based on audits to determine if the registrant's use actually meets the statutory requirement of use with respect to all or even any of the goods or services specified in the registration. See Changes in Requirements for Affidavits or Declarations of Use, Continued Use, or Excusable Nonuse in Trademark Cases, 82 Fed. Reg. 6,259, 6,262 (Jan. 19, 2017) (codified at 37 C.F.R. § § 2.161, 7.37 (2019)).

104. 35 U.S.C. § 303(a) (2018); see also 37 C.F.R. § 1.520 (2018) ("The Director, at any time during the period of enforceability of a patent, may determine whether or not a substantial new question of patentability is raised by patents or printed publications . . . even though no request for reexamination has been filed . . . ." (emphasis added)); PTO, U.S. Dep't of Com., MPEP § 2239 (9th ed., Rev. June 2020) (discussing the process for reexamination ordered at the Director's initiative, including when "an Office employee becomes aware of an unusual fact situation in a patent which he or she considers to clearly warrant reexamination").