# The Scope of Strong Marks: Should Trademark Law Protect the Strong More Than the Weak?

**Authors:** Barton Beebe, C. Scott Hemphill
**Citation:** "The Scope of Strong Marks: Should Trademark Law Protect the Strong More Than the Weak?," 92 *New York University Law Review* 1339 (2017) (with C. Scott Hemphill)
**Source:** https://www.nyulawreview.org/wp-content/uploads/2018/08/NYULawReview-92-5-BebbeHemphill.pdf

## INTRODUCTION

*p. 2*
It makes strong intuitive sense that as a trademark grows increasingly famous, the likelihood that consumers will confuse similar marks with it declines. Consider the example of COCA-COLA. Having been exposed to the COCA-COLA logo countless times, the average consumer is likely sensitized to even the slightest differences between the distinctive appearance of the logo and closely similar logos. The soundness of this intuition seems especially apparent today, as the strongest and most successful brands pursue ever more sophisticated ways to increase brand awareness through social media, product placement, and data-driven ad targeting, all to educate consumers about what makes them different. Recent studies show that now even three-year-olds can recognize famous trademarks. 1 In light of the extraordinary degree of consumer awareness of the very strongest marks, it makes sense that trademark law would adopt the principle that an exceptionally well-known mark is less susceptible to confusion precisely because consumers know the mark exceptionally well.

*p. 2*
Yet U.S. trademark law follows the opposite rule. The central question in most trademark litigation concerns the likelihood of con-sumer confusion: Is a significant proportion of relevant consumers likely to be confused as to the true source of the defendant's goods due to the similarity between the defendant's and plaintiff's marks? To answer this question, one of the most important factors that courts consider-in addition to assessing the similarity of the marks-is the strength of the plaintiff's mark. 2 Strength refers to the degree to which the mark, due to its notoriety and to some extent its intrinsic characteristics, identifies and distinguishes the product's source. It is currently taken for granted as blackletter doctrine, rehearsed in nearly every significant trademark infringement opinion, that the stronger a mark is, the more likely consumers are to confuse similar marks with it and thus the wider the scope of protection it should receive. This understanding of the role that trademark strength plays in increasing the likelihood of consumer confusion significantly influences courts' ultimate determination of liability. 3 Greatly to the benefit of the strongest brands, it always does so in one direction: Stronger marks invariably merit broader scope. The monotonic relation between strength and scope extends to the very strongest marks, which enjoy the widest scope of protection.

*p. 3*
This Article challenges this conventional wisdom, which we argue results in erroneous and anticompetitive findings of liability. Stronger marks should not always receive a broader scope of protection. To the contrary, the relation between strength and scope sometimes takes the form of an inverted U, as shown in Figure 1. Exceptionally wellknown "superstrong" marks often merit a narrower scope of protection than marks of lesser strength because their extraordinary strength reduces rather than increases the likelihood of consumer confusion. The Article examines the underpinnings of American courts' mistaken analysis, explains the errors, and urges essential reforms in this core area of trademark doctrine in order to advance the overarching policy goal of trademark law, which is not to enable the strongest to grow even stronger, but rather to promote effective competition.

*p. 3*
We call the conventional wisdom that stronger marks always merit broader protection the Kenner doctrine, after the case that is most responsible for its perpetuation. 4 Kenner overruled a line of cases embodying a competing alternative model of trademark strength, in which superstrong marks receive narrower protection. 5 The alternative model is premised on the insight that for the strongest marks, "the better known it is, the more readily the public becomes aware of even a small difference." 6 In essence, by virtue of repeated exposure to a famous mark and advertising emphasizing its distinctiveness, consumers grow more sophisticated in their ability to distinguish the mark from other marks. 7 We find support for this alternative model in numerous foreign trademark cases whose influence has apparently not yet been felt in the United States. 8

## Scope of Protection

*p. 4*
Today the cases following the alternative model-and Kenner itself-are largely forgotten. Reconsideration of the Kenner doctrine, however, is newly urgent, given both the steady growth in the strength of our most famous marks 9 and the significant and controversial expansion over several decades in the types of confusion that are actionable, including initial interest confusion, post-sale confusion, and sponsorship confusion. 10 A careful likelihood-of-confusion analysis has become increasingly important to ensure that desirable, lowconfusion conduct-particularly by competitors and other market 6 B.V.D. Licensing Corp. v. Body Action Design, Inc., 846 F.2d 727, 729 (Fed. Cir. 1988) (Rich, J.). 7 See infra Section II.A.3-4 (discussing social science research and case law supporting this view). 8 See infra notes 60-67 and accompanying text. 9 See Best Global Brands 2016 Rankings, INTERBRAND, http://interbrand.com/best-brands/best-global-brands/2016/ranking/ (last visited Aug. 1, 2017) (showing growth in brand value of a large proportion of the globally most valuable brands). 10 See, e.g., Mark A. Lemley & Mark P. McKenna, Owning Mark(et)s, 109 MICH. L. REV. 137, 146-56 (2010) (describing expanded protection against presale "initial-interest" confusion that is quickly dispelled on closer inspection, "post-sale" confusion by nonpurchasers, and unauthorized merchandise that assertedly causes confusion as to sponsorship).

*p. 5*
entrants-can flourish. Unfortunately, the Kenner doctrine curtails such conduct and substantially amplifies the advantages of incumbency for the strongest marks.

*p. 5*
The Kenner doctrine is undertheorized, its basis usually left unstated by courts. The idea that stronger marks are always associated with greater confusion is based on a jumble of unexamined empirical assumptions about consumer perception. These assumptions may support a positive relation between strength and scope at lower levels of trademark strength, 11 but they do not apply to superstrong marks, such as marks that qualify for federal antidilution protection because they are "widely recognized by the general consuming public of the United States." 12 Indeed, current doctrine recognizes this point in a limited way already by excusing trademark parodies on the ground that when extremely strong marks are parodied, confusion is particularly unlikely. 13 A second rationale, heavily emphasized in Kenner itself, sidesteps the analysis of confusion. The idea is that defendants with similar marks must not reap where they have not sown. 14 But neither can free-riding arguments justify expanded protection for the strongest marks. 15 As we explain, U.S. trademark law contains no clear antifree-riding principle, and even if it did, it does not follow that stronger marks necessarily merit stronger protection. Nor does it make sense for stronger protection (if it were warranted) to take the particular form of broader scope.

*p. 5*
Beyond our challenge to the conventional wisdom and the flawed rationales supporting it, our Article makes several further contributions. We recover a lost debate between competing accounts of trademark strength, illustrated by a simple, novel formal model of strength and scope. We furnish the first full account of the various rationales that might support a positive relation between mark strength and scope, including some that are new. We also distinguish two types of consumer confusion-signifier confusion and affiliation confusionthat are often lumped together. Courts often inappropriately apply doctrine developed with respect to one form of confusion to situations involving the other. 11 See infra Section II.A.1 (discussing "top of mind," "careless purchaser," and "consumer inference" rationales).

*p. 5*
12 15 U.S.C. § 1125(c)(2)(A) (2012). 13 See infra Section II.A.2 (reviewing case law addressing parodic uses of superstrong marks). 14 See infra Section III.A (presenting anti-free-riding rationale). 15 See infra Section III.B (critiquing the rationale).

*p. 6*
Finally, we distinguish two dimensions of mark scope-mark similarity and product relatedness-that are often conflated. As we explain, the Kenner doctrine's expansion of scope along the dimension of mark similarity, particularly when asserted against the producer of a competing product, is undesirable. This expansion often works to prevent competitors from using signifiers of product attributes. The recycling of such signifiers serves the important trademark goals of minimizing consumer search costs and promoting competitive entry. 16 Expansion along the product dimension has its own problems, as it improperly smuggles antidilution protection-a distinct body of trademark law with its own demanding limitations-into the ordinary analysis of confusion. 17 The Article proceeds in four Parts. In Part I, we explicate the Kenner doctrine and the alternative model of how mark strength affects scope. In Part II, we assess the empirical claim that the stronger a mark is, the more likely consumers are to confuse it with similar marks or otherwise assume some commercial affiliation between the two. We examine the flawed assumptions that underpin this claim and conclude that the alternative intuition has considerable force. We also present and critique several other confusion-based rationales for the Kenner doctrine. Part III explains and criticizes the free-riding rationale. Part IV recommends an abandonment of the Kenner doctrine with respect to superstrong marks and spells out additional implications of our analysis for trademark doctrine.

## I TWO MODELS OF TRADEMARK STRENGTH

*p. 6*
This Part sets out two competing models of trademark strength and scope. Part I.A describes the Kenner doctrine, the conventional wisdom that stronger marks get broader scope. Part I.B proposes an alternative model and situates it within Federal Circuit doctrine and case law from foreign jurisdictions. We illustrate both approaches using a simple formal model. Part I.C distinguishes the two dimensions of mark scope-mark similarity and product relatedness-along which heightened strength might have an effect.

## A. The Kenner Doctrine

*p. 6*
The Kenner doctrine takes its name from a 1992 Federal Circuit case about PLAY-DOH, one of the most famous and successful brands 16 See infra Section III.C (discussing the rationale in the context of competing goods). 17 See infra Section III.D (discussing the rationale in the context of noncompeting goods).

*p. 7*
in the history of children's toys. 18 In the 1980s, a majority of American youngsters owned PLAY-DOH modeling compound, and sixty percent of mothers-fathers apparently having not been asked-could name the product without prompting. 19 A competitor sold its own modeling compound under the FUNDOUGH mark. 20 Kenner, the producer of PLAY-DOH, opposed registration of FUNDOUGH. 21 Kenner argued that consumers would mistakenly believe that FUNDOUGH came from the same producer as PLAY-DOH. 22 The Trademark Trial and Appeal Board (Board) rejected this argument. 23 The Federal Circuit sided with Kenner and reversed the Board in an opinion that heavily emphasized the strength of the PLAY-DOH mark. 24 Judge Randall Rader, writing for the panel, declared: "The driving designs and origins of the Lanham Act demand the standard consistently applied by this court-namely, more protection against confusion for famous marks." 25 And even more pointedly: "In consonance with the purposes and origins of trademark protection, the Lanham Act provides a broader range of protection as a mark's fame grows." 26 This is the Kenner doctrine: that stronger marks always receive broader protection. In due course, we will scrutinize the "designs" and "purposes" of trademark law-to avoid confusion and (assertedly) to avoid free riding-advanced by courts in support of this doctrine. 27 But first, we need to understand what it means for a mark to be "strong" (or "famous," as Kenner synonymously put it 28 ), and what it means for a stronger mark to receive "broader" protection.

*p. 7*
Trademark strength is a measure of-and largely synonymous with-trademark distinctiveness. A trademark is strong or distinctive 18 See Allie Townsend, Play-Doh, TIME (Feb. 16, 2011), http://content.time.com/time/ specials/packages/article/0,28804,2049243_2048654_2049125,00.html (listing Play-Doh among "100 most influential toys" since 1923). 19 Kenner Parker Toys Inc. v. Rose Art Indus., Inc., 963 F.2d 350, 351 (Fed. Cir. 1992) (reporting that at least half of two-to seven-year-olds owned the product and that sixty percent of mothers could name the product without prompting). 20 Id. at 351-52. 21 Id. at 351. 22 Kenner Parker Toys, Inc. v. Rose Art Indus., Inc., Opposition No. 75,237, 1992 TTAB LEXIS 39, at *1-2 (T.T.A.B. Apr. 29, 1992). 23 Id. at *8-14. The Board's reasoning is discussed infra Section I.B. 24 Kenner, 963 F.2d 350. 25 Id. at 353. 26 Id. at 353-54 (footnote and citation omitted). 27 See infra Parts II and III.

*p. 7*
28 "Fame" in the context of likelihood of confusion means a high degree of strength. to the extent that it identifies and distinguishes the source of the goods to which it is affixed. Marks such as GOOGLE, AMAZON, or BUDWEISER are exceptionally strong marks in the sense that they are extremely well known to consumers as designations of source and very effectively differentiate their firms' offerings from those of other firms.

*p. 8*
Trademark law distinguishes between two forms of trademark strength. The more important form is "commercial strength," which is a measure of a mark's "acquired distinctiveness." A trademark acquires distinctiveness-that is, it gains commercial strength-to the degree that its sales, advertising, and other means of dissemination contribute to consumer awareness of and familiarity with the mark. 29 For example, when the APPLE mark first appeared in the marketplace, it was commercially very weak, minimally familiar to only a limited number of consumers as a designation of source. However, after decades of extensive use, advertising, and media coverage, it has emerged as one of the commercially strongest marks in the world.

*p. 8*
The second, less important form of trademark strength is "conceptual strength," which is a measure of a mark's "inherent distinctiveness." In general, a mark is inherently distinctive in inverse proportion to the degree to which it describes the product with which it is used. 30 Empirical evidence shows that when courts assess the overall strength of a mark, they weigh far more heavily a mark's commercial strength than its inherent strength. 31 In what follows, we will 29 See MCCARTHY ON TRADEMARKS AND UNFAIR COMPETITION § 11:83 (4th ed. 2016) (defining commercial strength as "the marketplace recognition value of the mark"). 30 See Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4, 9-11 (2d Cir. 1976). The so-called Abercrombie spectrum categorizes trademarks into one of five categories of distinctiveness: generic, descriptive, suggestive, arbitrary, and fanciful marks. Marks in the latter three categories qualify as inherently distinctive. See id. Suggestive marks do not directly describe the characteristics of the products to which they are affixed but are rather indirectly suggestive or evocative of those characteristics, such as GOURMET GARAGE for groceries or ROACH MOTEL for insect traps. See id. Arbitrary marks are common words that have no semantic connection with the products to which they are affixed, such as APPLE for mobile phones and other high-technology goods and services. focus on commercial strength and consider inherent strength only where appropriate. 32 "Broad" trademark protection is synonymous with a wide scope of protection-that is, the range of marks that if employed by a defendant (on a certain range of products) 33 would be held to be infringing. So when Kenner says that a stronger mark gets a broader scope of protection, that means that there is a larger range of marks that would be found infringing if used on another product. Broadened scope is not the only way in which a stronger mark could enjoy "better" protection. As we discuss below, a stronger mark could receive heightened protection within its scope of protection, without any expansion of that scope. 34 But here, as in Kenner, we are focused on the question of broader protection.

*p. 9*
We can formalize the "stronger mark, broader protection" idea in a simple model. Suppose two parties produce competing goods, as with PLAY-DOH and FUNDOUGH. We can denote the similarity of the parties' marks by x, which ranges from 0 (highly dissimilar) to 1 (identical). A second important variable is the strength of the plaintiff's mark. 35 The outcome of interest is the proportion z of the relevant consumer population confused as to source. This is the outcome of interest because when trademark law estimates the "likelihood of confusion," it is referring to the likelihood that a particular proportion of consumers in the relevant population are confused. 36 The proportion of confused consumers varies depending upon the degree of similarity. The more similar the parties' marks, the greater will be the proportion confused. This positive relationship is depicted in Figure 2, which plots z (the proportion confused) against x (the degree of similarity) for both a weak mark (the thin black curve) and a strong mark (the thick blue curve). An increasing proportion of consumers is confused as to source as the defendant's mark becomes more similar to the plaintiff's. At or near x = 0, there is no risk of confusion. At or near x = 1, confusion is at its peak. The height of this peak at x = 1 is an expression of the proportion of the relevant consumer population that upon perceiving the plaintiff's mark (or a 32 Trademark "strength" is different from patent "strength." Patent law typically speaks of strength to describe a patentee's overall probability of success in infringement litigation. Trademark strength refers only to the commercial and conceptual strength of the mark. Courts consider a variety of factors, including strength, in assessing infringement. 33 As explained infra Section I.C, scope has not only a mark dimension but also a product dimension. 34 See infra Section I.B and III.B. 35 Here we take the extreme case in which there is no significant difference between the plaintiff's and the defendant's goods. We relax this assumption below. 36 See MCCARTHY, supra note 29, at § 23:2.

*p. 11*
resulting level of confusion is greater than the threshold. Thus, the scope of plaintiff's weak mark extends from x 1 to 1. At a higher level of mark strength, the proportion of consumers confused as to source changes. As Kenner understands the relationship, this proportion is higher at all levels of mark similarity. 38 Compared to the weak mark, the strong mark crosses the 20% threshold at a lower degree of similarity (x 2 < x 1 ). The scope of this stronger mark is broader, extending from x 2 to 1. The difference between x 2 and x 1 is an expression of the Kenner doctrine.

*p. 11*
The Kenner doctrine is widely accepted. It is incorporated into each of the multifactor tests that the Courts of Appeals have adopted to assess the overall likelihood of confusion between two marks. The details of the multifactor test vary by circuit, but the strength of plaintiff's mark features in all of them. 39 In many, it is the first factor to be considered. 40 The Kenner doctrine also routinely appears in jury instructions. 41 Its inclusion is important because, as one of us has reported in previous work, this factor has an often significant and sometimes decisive impact on the overall outcome of the multifactor test. 42 In at least one circuit, the presence of a very strong mark plays a "dominant" role in balancing the various factors of the test. 43 Moreover, as in Kenner itself, the factor only points one way. Greater strength is uniformly understood without reflection as counting in favor of a likelihood of confusion. 44 Courts equate this point with greater breadth, routinely stating that for stronger marks, a wider range of marks is confusingly similar, 45 and that the strongest marks merit the widest range of protection. 46 To be clear, the proposition that we have called the Kenner doctrine predates Kenner and is invoked and applied in many cases that never cite Kenner. We emphasize the Kenner case because, as discussed later, it provides an unusually clear and frank expression of the true basis for the conventional wisdom. Moreover, Kenner is uniquely responsible for suppressing a competing model of trademark strength, which we consider next.

## B. An Alternative Model

*p. 12*
The alternative model was set out by Judge Giles Rich, a founding member of the Federal Circuit and "towering figure in patent and intellectual-property law for a half-century." 47 Judge Rich 43 See Bose Corp. v. QSC Audio Prods., Inc., 293 F.3d 1367, 1371 (Fed. Cir. 2002) ("Fame of an opposer's mark or marks, if it exists, plays a 'dominant role in the process of balancing the DuPont factors.'" (quoting Recot, Inc. v. Becton, 214 F.3d 1322, 1327 (Fed. Cir. 2000))). 44 See, e.g., Welding Servs., Inc. v. Forman, 509 F.3d 1351, 1361 (11th Cir. 2007) ("The stronger or more distinctive a trademark or service mark, the greater the likelihood of confusion . . . ."). The same point applies to conceptual strength. Trademark law conventionally holds that due to their higher level of conceptual strength, fanciful marks should enjoy a wider scope of protection than arbitrary marks, and arbitrary marks, a wider scope than suggestive marks. See Nutri/System, Inc. v. Con-Stan Indus., Inc., 809 F.2d 601, 605 (9th Cir. 1987) (considering precedent in which court "established a continuum of marks" on which "suggestive" marks are "given moderate protection" and "arbitrary" or "fanciful" marks are "awarded maximum protection" (citing Surgicenters of Am., Inc. v. Med. Dental Surgeries, Co., 601 F.2d 1011, 1014-15 (9th Cir. 1979))). 45 See, e.g., Welding Servs., 509 F.3d at 1361 ("The stronger . . . [the] mark, the greater the likelihood of confusion and the greater the scope of protection afforded it and conversely, the weaker the mark, the less protection it receives."). Often this is stated in the negative-that weak protection is defeated by only "minor alterations." See, e.g., First Sav. Bank, F.S.B. v. First Bank Sys., Inc., 101 F.3d 645, 655 (10th Cir. 1996) ("When the primary term is weakly protected to begin with, minor alterations may effectively negate any confusing similarity between the two marks."). 46 See, e.g. wrote landmark opinions expanding patent protection to then-new areas, such as genetically engineered organisms 48 and software-implemented inventions. 49 Writing four years before Kenner, Judge Rich took a different and, in its way, equally groundbreaking approach to trademark scope in B.V.D. Licensing Corp. v. Body Action Design, Inc. 50 The Federal Circuit had been asked to evaluate the likelihood of confusion between BVD, used as a mark on underwear, and a competing underwear manufacturer using the B•A•D mark. 51 BVD, like Kenner, complained that consumers would mistakenly believe that apparel bearing the B•A•D mark actually originated from BVD. 52 Judge Rich rejected BVD's argument, zeroing in on the strength of the BVD mark. 53 At the time (1988), BVD had "one of the three most recognized trademarks in the world," mentioned in the same breath as Coca-Cola. 54 Judge Rich observed that BVD was "practically a household word synonymous, primarily, with underwear for men." 55 But it did not follow that the scope of protection must increase. To the contrary, Judge Rich explained, consumers were so intensely familiar with the mark that they would be able to detect even slight differences between it and other marks:

*p. 13*
The fame of a mark cuts both ways with respect to likelihood of confusion. The better known it is, the more readily the public becomes aware of even a small difference. BVD has that wellknown quality which would trigger the observer to notice at once that B A D, with or without the periods in either mark, is a different symbol. 56 This reasoning echoes the position Judge Rich had taken as a judge on the Federal Circuit's predecessor court. 57 The court found no likeli-hood of confusion and affirmed the decision below to allow the registration of B•A•D. 58 The upshot of the B.V.D. case is that as a mark increases in strength beyond a certain point, the range of confusing marks might begin to contract. The result is a narrower scope of protection than at lower levels of strength. Figure 3 describes this effect. Kenner assumes that the relation between strength and scope is always positive. The alternative model recognizes that this relation may turn negative at very high levels of trademark strength. Importantly, the alternative model also recognizes that at the same time that a superstrong mark's scope of protection covers a reduced range of conduct, some conduct occurring within that scope of protection triggers higher levels of consumer confusion. In this sense, superstrong marks still get strong protection. Due to the heightened levels of consumer confusion, the likelihood that a court will issue an injunction is greater. In terms of our formal model, increasing strength results in higher levels of confusion at high levels of x, and less confusion at lower levels of x, with a sharper transition (an increased slope) in between.

*p. 14*
This effect is depicted in Figure 3, which extends Figure 2 by adding a superstrong mark (depicted as a red dotted curve). For a superstrong mark, the level of confusion exceeds the threshold at x 3 , which is to the right of x 2 . The scope of this superstrong mark, extending from x 3 to 1, is narrower than for the merely strong mark. At the same time, for certain unauthorized conduct that occurs within this scope-in particular, for marks close to 1-a higher proportion of consumers is confused. 59 the recent case of Delfi Chocolate Manufacturing S.A. v Mars Australia Pty Ltd, the Federal Court of Australia found no likelihood of confusion between the senior mark MALTESERS and the junior mark MALTITOS, both for chocolate confectionary, where the former mark had a "very widespread, solid, reputation in the area of packaged confectionary" 63 and "is strongly embedded in the consciousness of many of those of average intelligence." 64 The court concluded that given the enormous strength of the senior mark, the average consumer of confectionary "would be immediately struck by the differences between the two marks." 65 British, Canadian, and Japanese courts have reasoned similarly in cases involving extremely strong marks. 66 relation to particular goods or services" is a "relevant consideration" in "assessing the nature of a consumer's imperfect recollection of a mark"); see also Beecham 70 Id. at 354 (holding that B.V.D., "to the extent it treats fame as a liability, is confined to the facts of that case"). 71 The main exception is William E. Gallagher & Ronald C. Goodstein, Inference Versus Speculation in Trademark Infringement Litigation: Abandoning the Fiction of the Vulcan Mind Meld, 94 TRADEMARK REP. 1229, 1250-53 (2004). In a wide-ranging critique of the likelihood of confusion analysis in trademark law, Gallagher and Goodstein briefly propose, citing B.V.D., that "the possibility exists that the likelihood of confusion can decrease with increasing strength of the plaintiff's mark." Id. at 1250; see also Bone, supra note 42, at 1343-46 (raising doubts about the assumed positive relationship without reference to B.V.D.).

*p. 18*
reject the argument as contrary to the received doctrine of trademark law. 72

## C. Dimensions of Scope

*p. 18*
Our discussion of mark scope so far has been limited to the dimension of mark similarity, particularly in the context of competing goods. However, there is a second dimension of scope to consider: the range of products beyond those that are directly competing with the trademark holder's products that the holder can exclude from bearing an identical or similar mark. (Dissimilar marks are not actionable. 73 ) Thus trademark litigation may be classified along the two dimensions of mark similarity and product relatedness. With respect to the parties' marks at issue, the marks may be either identical or similar. With respect to the goods at issue, the goods may be competing, related, or unrelated. Table 1 depicts the set of possible disputes along these two dimensions. 73 See Beebe, supra note 31, at 1608 (reporting that of 192 preliminary injunction or bench trial opinions studied, defendant prevailed in all of the sixty-five opinions in which the court found that the similarity factor in the multifactor test for the likelihood of confusion did not favor the plaintiff).

*p. 19*
Mapping trademark disputes in this manner quickly shows where the easy cases are found. Consider, for example, the top-left corner, the case of identical marks on competing goods. It is likely that a significant proportion of the relevant consumer population will be confused as to source no matter how weak the plaintiff's mark. Indeed, European trademark law goes so far as to establish a presumption of infringement on the basis of this "double identity" of the parties' marks and products. 74 At the other extreme, the bottom-right corner, the case where the parties' marks are merely similar but their goods are unrelated, it is unlikely that consumers will be confused as to source.

*p. 19*
The hard cases-and the litigation-tend to arise in the shaded areas in the table, where the facts present similar marks on competing (or related) goods or else identical marks on unrelated (or related) goods. It is in these four scenarios that the strength of the plaintiff's mark can play a decisive role in a court's infringement determination.

*p. 19*
The application of the Kenner doctrine along the dimension of product scope can be understood by reference to our formal model. Instead of assuming that the parties produce competing (i.e. more or less identical) products, we assume that they use an identical mark (such as TIFFANY). And instead of considering a wide range of mark (dis)similarity, we consider a wide range of product relatedness, from competing (TIFFANY jewelry) to related (TIFFANY clothing) to unrelated (TIFFANY bicycles). Let us use y to denote the degree of relatedness of the parties' products, which ranges from 0 (unrelated products) to 1 (competing products). That assumption allows us to consider how the level of consumer confusion varies depending on the degree of relatedness of the products.

*p. 19*
This relationship is depicted in Figure 4, which plots the proportion of consumers confused z against the degree of product relatedness y. Once again the threshold level of confusion is 20%. The three curves-depicting, as before, a weak mark, strong mark, and superstrong mark-represent the proportion of confused consumers, which varies over a wide range of different defendant products using the 74 See Supreme Petfoods Ltd. v. Henry Bell & Co. (Grantham) [2015] EWHC 256 (Ch), ¶ 147 (Arnold, J.) (discussing the double identity presumption in European trademark law). Article 16(1) of the TRIPS Agreement obligates the United States to also apply this double identity principle. See Agreement on Trade-Related Aspects of Intellectual Property Rights, Apr. 15, 1994, Art. 16(1), Marrakesh Agreement Establishing the World Trade Organization, Annex 1C, Legal Instruments-Results of the Uruguay Round, 33 I.L.M. 1125 (1994) ("In case of the use of an identical sign for identical goods or services, a likelihood of confusion shall be presumed."). U.S. trademark doctrine does not have a clear double identity rule, though in practice the United States is very likely in compliance with this aspect of TRIPS.

*p. 22*
narrower scope of the superstrong mark along the mark dimension, combined with broader scope along the product dimension.

## II CONFUSION-BASED RATIONALES

*p. 22*
As noted in the Introduction, courts rarely explain the basis for the Kenner doctrine. Instead, they typically offer a perfunctory recitation of the basic blackletter rule-stronger marks enjoy wider scopeand move on, as if the underlying reasoning is too obvious to require explanation. Those courts that do pause to explain the rule tend to base their explanations on a jumble of untested empirical assertions about consumer perception.

*p. 22*
This Part considers a cluster of confusion-based rationales for the Kenner doctrine. Part II.A considers and critiques the core assertion that increased mark strength always increases the likelihood of consumer confusion, even at extremely high levels of strength, by altering consumer perceptions. While we find the rationales that support this assertion persuasive in explaining the positive relation between strength and scope at lower levels of trademark strength, they do not persuasively explain why extremely strong marks should enjoy the broadest scope of protection. We then turn to other confusion-based rationales for the positive relation. These fail to support the positive relation at any level of strength. Part II.B considers the rationale that strong marks, including very strong marks, are associated with (as opposed to causing) higher levels of confusion because they are disproportionately targeted for infringement. Finally, Part II.C considers an additional tack, that even holding constant the likelihood of confusion, the stronger a mark is, the greater the damage it suffers from confusion.

## A. Greater Consumer Confusion

*p. 22*
When courts conclude that stronger marks produce greater confusion, they are relying on a set of assumptions about consumer perception, which we unpack in Part II.A.1. There we introduce an important distinction between signifier confusion and affiliation confusion. As we explain in Part II.A.2, an important doctrinal exception runs counter to these assertions-the parody exception, in which consumers are rightly understood to be sensitized to small differences. In Parts II.A.3 and II.A.4, we show that the concerns that animate the parody exception are general, because superstrong marks tend to dispel both signifier confusion and affiliation confusion.

## Assumptions About Consumer Perception

*p. 23*
Courts have proposed three reasons why consumers will be more confused when the plaintiff's mark is stronger. They are simple and easily stated, which may account in part for their continuing appeal. We call them the "top of mind," "careless purchaser," and "consumer inference" rationales.

*p. 23*
To understand them, it is helpful to distinguish between two fundamentally different and mutually exclusive forms of consumer confusion, which we term signifier confusion and affiliation confusion. 76 Signifier confusion denotes those situations in which a consumer fails to detect the difference between two different marks and perceives each mark to be identical to the other. For example, a consumer may be exposed to the mark STARLUCKS and simply mistakenly read or hear the mark as STARBUCKS.

*p. 23*
By contrast, affiliation confusion denotes those situations in which a consumer detects the difference between two different marks (so there is no signifier confusion), but the consumer nevertheless concludes that due to the similarity of the marks, there must be some commercial connection between the users of the marks. For example, a consumer thinks STARLUCKS represents a brand extension, sponsorship or endorsement relationship, or some other form of commercial affiliation. The consumer perceives the plaintiff as the source of or somehow responsible for the defendant's goods.

*p. 23*
[1] Top of mind. The most common rationale asserts that the stronger a mark is, the more likely a consumer is to think of it when exposed to a similar or identical mark and thus the more likely it is that the consumer will conclude that the two marks originate in the same source. If the consumer fails even to think of the plaintiff's mark when exposed to the defendant's mark, there is no chance that the consumer will be confused. The greater the likelihood that the consumer will call to mind or "activate" 77 the plaintiff's mark in her consciousness, the greater the chance that the consumer will be in a position to confuse the source of the marks. Top of mind confusion 76 For a similar conceptualization, see Jacob Jacoby, The Psychological Foundations of Trademark Law: Secondary Meaning, Genericism, Fame, Confusion, and Dilution, 91 TRADEMARK REP. 1013, 1042-46 (2001) (conceptualizing the types of confusion as single source, which includes confusion as to origin and item, and multiple source, which includes confusion as to affiliation and independence). 77 Id. at 1018-21 (describing the associative network theory of memory, according to which information is stored in inter-connected nodes, which, upon activation, bring that information into consciousness).

*p. 24*
can take the form of either signifier or affiliation confusion. Courts invoke this "top of mind" rationale frequently. 78 [2] Careless purchaser. The careless purchaser rationale is an assertion about signifier confusion and superstrong marks. Courts reason that consumers are so familiar with superstrong marks that they devote less care to the purchase of goods bearing such marks. 79 Consumers are therefore more likely to overlook differences between superstrong marks and similar marks. For example, in B.V.D., a dissenting judge surmised:

*p. 24*
[A] purchaser is less likely to perceive differences from a famous mark. All that is needed is a suggestion of such mark to trigger a mental perception that it is the famous mark. Purchasers simply do not take the time to study the marks and see the differences. Nor are they expected to. 80 The dissent cited no evidence in support of this proposition. 81 [3] Consumer inference about inherently distinctive marks. A third rationale centers on affiliation confusion. Applicable only to inherently distinctive marks, this rationale asserts that the more inherently distinctive the plaintiff's mark, the more likely it is that consumers will 78 See, e.g., Virgin Enters. Ltd. v. Nawab, 335 F.3d 141, 148 (2d Cir. 2003) ("Widespread consumer recognition of a mark previously used in commerce increases the likelihood that consumers will assume it identifies the previously familiar user, and therefore increases the likelihood of consumer confusion if the new user is in fact not related to the first."); Brookfield Commc'ns, Inc. v. W. Coast Entm't Corp., 174 F.3d 1036, 1058 (9th Cir. 1999) ("The stronger a mark-meaning the more likely it is to be remembered and associated in the public mind with the mark's owner-the greater the protection it is accorded by the trademark laws."); Hormel Foods Corp. v. Jim Henson Prods., Inc., 73 F.3d 497, 503 (2d Cir. 1996) ("The more deeply a plaintiff's mark is embedded in the consumer's mind, the more likely it is that the defendant's mark will conjure up the image of the plaintiff's product instead of that of the junior user."); James Burrough, Ltd. v. Sign of the Beefeater, Inc., 540 F.2d 266, 276 (7th Cir. 1976) ("A mark that is strong because of its fame or its uniqueness, is more likely to be remembered and . . . associated in the public mind with a greater breadth of products . . . than is a mark that is weak because relatively unknown or very like similar marks . . . ."); see also RESTATEMENT (FIRST) OF TORTS § 731 cmt. e (AM. LAW INST. 1938) ("The more distinctive the trade-mark is, the greater its influence in stimulating sales, its hold on the memory of purchasers and the likelihood of associating similar designations on other goods with the same source."); TRADEMARK MANUAL OF EXAMINING PROCEDURE § 1207.01(d)(ix) (2017) ("Famous marks are afforded a broad scope of legal protection because they are more likely to be remembered and associated in the public mind than a weaker mark."). 79 See, e.g., Specialty Brands, Inc. v. Coffee Bean Distribs., Inc., 748 F.2d 669, 675 (Fed. Cir. 1984) ("The fame of a trademark may affect the likelihood purchasers will be confused inasmuch as less care may be taken in purchasing a product under a famous name.").

*p. 24*
80 B.V.D. Licensing Corp. v. Body Action Design, Inc., 846 F.2d 727, 730 (Fed. Cir. 1988) (Nies, J., dissenting). 81 See id. The dissent did cite a single case, Specialty Brands, Inc., 748 F.2d at 675. It also cited the McCarthy and Gilson treatises, but the single explanatory parenthetical provided-from the Gilson treatise-consisted of a quotation that did not speak to the issue. Id.

*p. 25*
assume any similar mark comes from the plaintiff. 82 For example, in TCPIP Holding Co. v. Haar Communications Inc., 83 the court focused on fanciful marks (that is, neologisms like EXXON) and arbitrary marks (common words, like APPLE, bearing no relation to the product). 84 The court asserted that "the consumer is likely to assume that a similar mark designates the owner of the first as the source of the goods. The arbitrariness of the mark in relation to those goods makes it unlikely that an unrelated merchant would select a similar mark for closely related goods." 85 Similarly, the Restatement of Torts asserts that for KODAK or other coined words, "it is more probable that all goods on which a similar designation is used will be regarded as emanating from the same source than when the trademark is one in common use on a variety of goods, such as 'Gold Seal' or 'Excelsior.' " 86 This inference is most frequently employed as to identical marks used on competing or related goods. 87 In explaining why strength expands scope, courts sometimes veer from one argument to another. For example, in TCPIP Holding Co. the court began with the consumer inference argument. 88 But then, to exemplify its point about affiliation confusion, the court suddenly switched to a careless purchaser example based on signifier confusion: "To illustrate the point, if a car is marked 'Buicke,' or film in a yellow box is labeled 'Kodok,' the unsophisticated observer may well fail to notice the slight differences and assume that they come from the makers of Buick and Kodak." 89 82 See Bone, supra note 42, at 1344 (acknowledging this argument but suggesting that its effect is probably small). 83 244 F.3d 88 (2d Cir. 2001). 84 See id. at 100-01. For a discussion of fanciful and arbitrary marks, see supra note 30 and accompanying text. 85 Id. at 100. The court adds a layer of complexity by identifying varying degrees of arbitrariness or fancifulness: "the more arbitrary or fanciful" the mark, "the more likely" the consumer will be confused as described in text. Id. (emphasis added). The court goes on to explain: "Conversely, the more descriptive the mark . . . the less likely a consumer is to assume that a similar mark used on related goods came from the same source," noting that "[t]he similarity between the two marks is as likely explained by the fact that each seeks to describe the goods, as by their coming from the same merchant." Id. at 100-01.

*p. 25*
86 RESTATEMENT (FIRST) OF TORTS § 731 cmt. e (AM. LAW INST. 1938). 87 See, e.g., Virgin Enters. Ltd. v. Nawab, 335 F.3d 141 (2d Cir. 2003) (competing goods); TCPIP Holding Co., 244 F.3d at 88 (related goods). 88 See TCPIP Holding Co., 244 F.3d at 100. 89 Id. at 101. In the court's analysis, BUICK is meant as an example of an inherently distinctive mark, but as the surname of its founder David Dunbar Buick, BUICK would qualify as a non-inherently-distinctive descriptive mark. See Abraham Zion Corp. v. Lebow, 761 F.2d 93, 104 (2d Cir. 1985) ("Personal names used as trademarks are generally regarded as descriptive terms, not arbitrary or fanciful terms.").

*p. 26*
Finally, the court shifted back to consumer inference and affiliation confusion:

*p. 26*
Conversely, if a consumer knows crunchy candy bars coming from Confectioner A under the mark "CRUNCHIES" and later encounters cookies branded "CRUNCHY delights," the consumer is as likely to attribute the similarity of name to two unrelated confectioners independently selecting names that describe and vaunt the crunchy texture of their products. 90 A sympathetic interpretation of this reasoning is that the court believed that multiple rationales apply-that inherently strong marks have both higher affiliation confusion (because of the consumer inference rationale) and higher signifier confusion (because consumers are careless). In other words, some consumers see STARLUCKS and misread it as STARBUCKS, while others assume it is a brand extension of STARBUCKS. As we argue below, however, neither argument is persuasive across all levels of mark strength.

## The Parody Exception

*p. 26*
Courts recognize an exception to the empirical proposition that increased strength results in an increased likelihood of confusion. When the defendant is engaging in parodic reference to the plaintiff's mark, courts routinely reason that the stronger the plaintiff's mark, the less likely it is that consumers will experience signifier confusion between the plaintiff's mark and the defendant's parodic version. Furthermore, consumers will not likely experience affiliation confusion because they will not likely believe that the plaintiff has developed a similar mark deliberately to criticize or parody itself.

*p. 26*
For example, in Louis Vuitton Malletier, S.A. v. Haute Diggity Dog, LLC, 91 the defendant produced various dog toys bearing trademarks cleverly altered to evoke both the targeted fashion brands and doghood. LOUIS VUITTON became CHEWY VUITON. JIMMY CHOO became JIMMY CHEW. DIOR became DOGIOR. 92 Louis Vuitton sued. The Fourth Circuit reasoned:

*p. 26*
It is a matter of common sense that the strength of a famous mark allows consumers immediately to perceive the target of the parody, while simultaneously allowing them to recognize the changes to the mark that make the parody funny or biting. In this case, precisely because LOUIS VUITTON is so strong a mark and so well recog-90 TCPIP Holding Co., 244 F.3d at 101. The court concluded: "Thus, similarity in descriptive marks is less likely to cause confusion than similarity in arbitrary or fanciful marks." Id.

*p. 26*
91 507 F.3d 252 (4th Cir. 2007). 92 See id. at 258.

*p. 27*
nized as a luxury handbag brand from LVM, consumers readily recognize that when they see a "Chewy Vuiton" pet toy, they see a parody. 93 Numerous opinions have reasoned similarly. 94 Nevertheless, courts have explicitly rejected attempts to apply this reasoning outside of the parody context. In the long-running case Starbucks Corp. v. Wolfe's Borough Coffee, Inc., the defendant had adopted the mark CHARBUCKS to identify an extremely dark-roasted blend of coffee it produced. 95 As the defendant's owner testified, "[t]he inspiration for the [mark] comes directly from Starbucks' tendency to roast its products more darkly than that of other major roasters." 96 The Second Circuit was somewhat amused, but not enough:

*p. 27*
Although we recognize some humor in "Char"bucks as a reference to the dark roast of the Starbucks coffees . . . the Charbucks parody is promoted not as a satire or irreverent commentary of Starbucks but, rather, as a beacon to identify Charbucks as a coffee that competes at the same level and quality as Starbucks in producing darkroasted coffees. 97 Because CHARBUCKS did not qualify as a "clear parody" with which there was "widespread familiarity," the Second Circuit refused to 93 Id. at 261-62 (citation omitted). 94 See, e.g., Radiance Found., Inc. v. NAACP, 786 F.3d 316, 328 (4th Cir. 2015) ("The implications for the likelihood of confusion factors are thus obvious: [P]arody or satire or critical opinion generally may be more effective if the mark is strong and the satirical or critical version is similar to the original."); Lyons P'ship v. Giannoulas, 179 F.3d 384, 389 (5th Cir. 1999) ("[W]hen a consumer encounters the use of a trademark in a setting that is clearly a parody, the strength of the mark may actually make it easier for the consumer to realize that the use is a parody."); Hormel Foods Corp. v. Jim Henson Prods., Inc., 73 F.3d 497, 502-03 (2d Cir. 1996) ("[W]here the plaintiff's mark is being used as part of a jest or commentary . . . [and] both plaintiff['s] and defendant's marks are strong, well recognized, and clearly associated in the consumers' mind with a particular distinct ethic . . . confusion is avoided . . . ." (second alteration in original) (quoting Yankee Publ'g Inc. v. News Am. Publ'g Inc., 809 F. Supp. 267, 273 (S.D.N.Y. 1992)); Louis Vuitton Malletier, S.A. v. My Other Bag, Inc., 156 F. Supp. 3d 425, 441 (S.D.N.Y. 2016) ("Louis Vuitton's marks are so well known that consumers are likely both immediately to recognize the target of the joke and to appreciate the obvious changes to the marks that constitute the joke." (internal quotation marks omitted)); Tommy Hilfiger Licensing, Inc. v. Nature Labs, LLC, 221 F. Supp. 2d 410, 416 (S.D.N.Y. 2002) (finding that strength did not favor the plaintiff's mark, TOMMY HILFIGER, in an action against TIMMY HOLEDIGGER pet perfume); Eveready Battery Co. v. Adolph Coors Co., 765 F. Supp. 440, 450 (N.D. Ill. 1991) ("However, the very strength of the Energizer Bunny mark in this case seems to weigh against a likelihood of confusion, particularly in light of the obvious parody depicted in defendant's use."). 95 Starbucks Corp. v. Wolfe's Borough Coffee, Inc., 588 F.3d 97, 103 (2d Cir. 2009). 96 Id. at 113 (first alteration in original) (internal quotation marks omitted). 97 Id.

*p. 28*
apply the parody exception. 98 As a result, the court rejected the argument that the enormous strength of the STARBUCKS mark would help to mitigate any likelihood of signifier or affiliation confusion. 99

## Strength and Signifier Confusion

*p. 28*
It is certainly plausible that in some circumstances increased strength may result in an increased likelihood of confusion. But the confusion-based rationales for the positive relation between strength and scope cannot justify the blanket empirical claim that increased strength always results, at all levels of strength, in an increased likelihood of either signifier confusion or affiliation confusion.

*p. 28*
The rationales are especially unpersuasive with respect to signifier confusion. As Judge Rich recognized in B.V.D., as a trademark's strength increases, at some point it may become so strong that the proportion of consumers experiencing signifier confusion with a similar mark begins to decline. 100 Such consumers may become so familiar with the mark and their sophistication with respect to it may increase to such an extent that they can quickly detect even the slightest differences between it and other similar marks, even when, to address the consumer inference hypothesis, the mark is inherently distinctive. In the language of the Australian cases on the issue, there is a lower likelihood of consumers' "imperfect recollection" 101 of extremely strong marks as compared to their recollection of marks of lesser strength.

*p. 28*
Social science research strongly supports this intuition. Extensive experimental research has long shown that familiarity results in stronger memory structures, and stronger memory structures increase one's ability to make fine distinctions between a familiar stimulus and other stimuli. 102 Research has also shown that the appearance within 98 Id. at 115-16. The court additionally found that "Charbucks is not a 'clear parody' as [defendant] Black Bear urges because the purported parody designates a product that is in direct competition with the products identified with the Starbucks Marks." Id. at 116. 99 Id. at 115 (rejecting this argument). Ultimately, after considering the degree of similarity and other factors, the court concluded that there was no likelihood of confusion. Id. at 119. 100 See supra note 56 and accompanying text. 101 an otherwise familiar schema of an unfamiliar stimulus results in the "covert, rapid, and automatic orientation of attention" towards that unfamiliar stimulus. 103 These findings comport with the intuition that a consumer's attention will immediately be drawn to any slight variation in the appearance of a superstrong mark with which the consumer is highly familiar. The implication of this research is that the relation between strength and scope may sometimes take the form of an inverted U, as represented in Figure 1. This is essentially the relation that the parody exception posits. At least with respect to signifier confusion, there is no reason to believe that the reasoning underlying the parody exception is limited only to that exception.

*p. 29*
The intuitive appeal of the inverted U helps to explain why soon after Judge Rich handed down his opinion in B.V.D., courts quickly adopted and improved on his reasoning when analyzing the possibility of signifier confusion. For example, one court cited B.V.D. to find that "the fame of the Girls Scouts and Boy Scouts is such that confusion with the name 'Pee Wee Scouts' in this context is highly unlikely." 104 In another case following B.V.D., the plaintiff produced a pinescented automobile air freshener in the shape of a thin, stylized profile of a pine tree. The defendant produced an air freshener also in a stylized pine tree shape. Citing B.V.D., the court found that very strong marks essentially work a recalibration of the scale of similarity, making it more sensitive to differences: "[T]he very strength of plaintiff's mark, including the distinctive design of its flat, pine tree-shaped Experience in Recognition Memory, 105 PSYCHOL. REV. 724, 724 (1998) (studying the "familiarity effect" by which "[w]hen we become familiar with something-an object, a person, or an item in a memory experiment-we come to recognize it more reliably"); see also George M. Zinkhan & Aydin Muderrisoglu, Involvement, Familiarity, Cognitive Differentiation, and Advertising Recall: A Test of Convergent and Discriminant Validity, in 12 ADVANCES IN CONSUMER RESEARCH 356, 357 (1985) ("The more well-developed a memory structure is (the more familiar a person is with a stimulus) the higher the probability of being able to make fine distinctions between that stimulus object and others."). product, leads this court to conclude that there is only a slight degree of similarity between the products." 105 In a registration dispute before the Board involving the Marshall Field's department store and Mrs. Field's cookies, it was the fame not just of the plaintiff's mark but also of the defendant's mark that minimized the likelihood of signifier confusion: "It is because both marks are famous that we believe the public will easily recognize the differences in the marks and distinguish between them." 106 As a justification for the Kenner doctrine, heightened signifier confusion faces an additional limitation. Where the parties' marks are identical, the mechanism by which increased strength results in increased source confusion cannot be an increase in signifier confusion because the signifiers are the same. There is no difference between the marks that increased strength causes consumers to overlook. Thus, increased scope on account of heightened signifier confusion is only meaningful in scenarios involving nonidentical marks.

## Strength and Affiliation Confusion

*p. 30*
Defenders of the Kenner doctrine may respond that even if increased strength does not increase signifier confusion, it nevertheless increases affiliation confusion. That is, even if strength may aid consumers in distinguishing between two similar marks, it will encourage consumers to assume some commercial connection between those marks. In its retreat to affiliation confusion, the argument thus assumes that the strong mark is strong enough to disabuse consumers of signifier confusion but apparently not strong enough to disabuse them of affiliation confusion. In other words, the argument assumes that there is some special segment of consumers who, because of the strength of the plaintiff's mark, are able to differentiate it from the defendant's similar mark, but then, despite that strength, conclude that the two different marks originate in the same source.

*p. 30*
Yet common sense and marketing research suggest that the opposite is often true, particularly for superstrong, highly distinctive marks. Due to the very strength of a strong mark, consumers of the products to which the mark is affixed may have a high degree of sophistication not simply with respect to the appearance of the mark (thus minimizing signifier confusion in nonidentical marks scenarios), but also with respect to the kinds of products to which the mark is affixed (thus minimizing affiliation confusion in all scenarios other than counterfeiting or "double identity"). Marketing research supports this intuition; as a brand grows more familiar, consumers develop increased "brand knowledge." 107 This knowledge enables them to differentiate the brand and make accurate inferences about the brand's characterisand conduct. 108 This effect is especially pronounced in certain mark-product scenarios. Consider, for example, the scenario involving the defendant's use of an identical mark on unrelated goods. If the mark TIFFANY appeared on a product completely unrelated in characteristics and quality to the high-end jewelry products that conventionally bear the TIFFANY mark, such as bicycles, it is highly likely that relevant consumers would use their knowledge of such a well-known brand to conclude that there is no affiliation between the producers of the goods bearing the mark. The strength of the TIFFANY mark would mitigate affiliation confusion. In a related vein, although for many years the CBS television network was referred to as "The Tiffany Network," 109 a term sometimes still used today, this did not to our knowledge generate significant affiliation confusion.

*p. 31*
The consumer's ability to disambiguate extends to similar marks on related goods. For example, in Jim Beam Brands Co. v. Beamish & Crawford, Ltd., 110 the court considered the likelihood of confusion between plaintiff's JIM BEAM whisky mark and defendant's BEAMISH mark for Irish stout. The court observed that JIM BEAM "may be one of the more recognized marks in this country, at least by those over the age of twenty-one"; 111 it "is a strong mark, long known for its Kentucky-style straight bourbon whisky and related products." 174 (1990) ("The more familiar consumers are with the various brand offerings within a product class, the more likely they will be able to make distinctions among brands, thereby reducing the likelihood of confusion."); see also Gallagher & Goodstein, supra note 71, at 1252 (noting that if defendant's brand image is inconsistent with plaintiff's brand image, "confusion may be less likely because the defendant's brand does not fit the established 'template' of the plaintiff's brand as it exists in the minds of relevant consumers."). 109 See ED SHANE, SELLING ELECTRONIC MEDIA 14 (1999) (discussing CBS's reputation in the 1960s and '70s as the "Tiffany Network" due to its high quality programming).

*p. 31*
110 852 F. Supp. 196 (S.D.N.Y. 1994). 111 Id. at 199. 112 Id.

*p. 32*
Given consumers' extensive familiarity with the JIM BEAM brand, the district court cited B.V.D. and found no likelihood of confusion: "it is precisely this strength that makes it unlikely to be confused with an Irish stout product." 113 Notably, the reasoning of the court, involving similar marks on related goods, applies a fortiori to a defendant's use of a similar mark on unrelated goods.

## * * *

*p. 32*
To summarize, the relation between mark strength and scope may be positive at lower levels of mark strength, but at extremely high levels of strength, the overall effect of increased strength may be less confusion and narrower scope. 114 This conclusion applies as to both signifier and affiliation confusion. To emphasize again, we are not arguing that as trademark strength increases, the positive relation always turns negative at some point. We do not wish to make the same mistake as current doctrine by making such a sweeping claim. The inverted U is most likely where increased strength brings with it greater consumer knowledge of the mark and its products rather than simply more widespread awareness of the mark. Furthermore, the inverted U more often implicates signifier confusion and is relevant to affiliation confusion only in certain circumstances.

*p. 32*
Instead, our more limited claim is that the traditional confusionbased rationales for the positive relation between trademark strength and scope cannot support the oft-rehearsed all-purpose empirical claim that increased strength should always result in an increased scope of protection. Nor can the traditional confusion-based rationales explain why trademark law has adopted the positive relation doctrine so unreservedly. For such an explanation, we must look elsewhere, specifically to the law's overriding concern with free riding-to which we turn in Part III.

## B. Greater Infringer Targeting

*p. 32*
The top of mind, careless purchaser, and consumer inference rationales for the Kenner doctrine have an important point of commonality. They all assert that strong marks provoke different behavior by consumers. A second strategy is to argue that strong marks induce different infringer behavior. For example, some courts have asserted 113 Id. 114 Because we accept that the top of mind and consumer inference rationales have some force with respect to marks at lower levels of strength, we do not believe that very weak marks should enjoy a wider scope of protection than marks that are not as weak. Instead, as Figure 1 suggests, the relation between strength and scope is positive for weak marks.

*p. 33*
that because strong marks are "more attractive as targets for would-be copyists," 115 they are associated with a higher likelihood of confusion. Kenner puts the point even more sharply, asserting that the opportunity to exploit a strong mark's goodwill "encourage[s] competitors to snuggle as close as possible to a famous mark." 116 Similarly, a line of doctrine first established by the Court of Customs and Patent Appeals holds that "there is . . . no excuse for even approaching the wellknown trademark of a competitor" and that "to do so raises but one inference-that of gaining advantage from the wide reputation established by appellant in the goods bearing its mark." 117 The logic of infringer targeting differs from the consumer behavior rationales in a fundamental way. This argument entails no "shift in the curve," in the language of the formal model. Only differences in consumer behavior shift the curve; differences in infringer behavior do not. Without a shift in the curve, there is no change to the scope of the mark. To this extent, Kenner is quite wrong to suggest that "snuggling" would imply broader scope.

*p. 33*
Instead of arguing that a change in strength causes a shift in z, the proportion of consumers confused, the targeting idea is to directly infer something about z (and indirectly about x and/or y) from the predicted behavior of infringers. The starting point is that copyists find it more profitable to target strong marks. Moreover, the copyist's profits are premised on a high degree of confusion caused by the defendant's choice of mark and product. If strong marks are frequently targeted by high-confusion defendants, and weak marks are rarely targeted, then if we observe a strong mark it might seem to suggest (all else equal) that the defendant's mark is highly confusing. In other words, the proposed inference is that the mark must be at the right-hand side of the curve, where x (and/or y) and z are high.

*p. 33*
To the extent firms are rational, this approach has a fallacy at its core. In general, rational firms do not choose high-confusion marks. In 115 Bose Corp. v. QSC Audio Prods., Inc., 293 F.3d 1367, 1371 (Fed. Cir. 2002) (stating that a mark's "fame" (for likelihood of confusion purposes) "plays a dominant role" in multifactor balancing because "famous marks are more likely to be remembered and associated in the public mind than a weaker mark, and are thus more attractive as targets for would-be copyists" (internal quotation marks omitted)). 116 2007) (noting that in the late nineteenth century, owners of technical trademarks did not have to show that an infringer possessed an intent to infringe because "[u]se of another's technical trademark was unlikely to have a legitimate explanation and could be condemned categorically"). fact, most firms choose marks with much lower levels of confusion. Rational firms seek to follow the law, whether out of a sense of duty or a fear of the costly consequences of infringement. Unlike an illmotivated counterfeiter with little to lose, most firms will not "snuggle" to the extent that maximizes closeness. 118 Thus, we cannot infer whether a defendant has chosen an actionably high degree of confusion simply by assessing the strength of the mark. To be sure, some firms deliberately "target" another's marks for one reason or another. But these firms do not ignore trademark law. To the contrary, they are likely to choose marks that minimize or manage their potential liability. For example, producers of "store brands" typically pattern their packaging (or in the terminology of trademark law, their "trade dress") after the well-known brands against which they are competing. 119 But as numerous courts have recognized, store brands are also typically careful to ensure that their packaging does not confuse consumers as to source, often by prominently featuring the store brand trademark, disclaimers, and "compare to" statements. 120 (Later, we consider one particular form of targeting by a competitor that actually serves important trademark interests. 121 )

*p. 34*
A smaller set of firms not only targets a mark but also goes too far, choosing a confusing mark. The intentional choice of a confusing mark, as by counterfeiters that have nothing to lose, is plausibly concentrated on strong marks. Meanwhile, other firms might deliberately target a mark for legitimate reasons, such as in comparative advertising, but accidentally go too far. About all we can say is that not only infringers, but also many non-infringers, target strong marks. The existence of a small set of firms that target and infringe does not support a special inference of infringement for strong marks, or even for strong marks where we have additional evidence of targeting. By way of analogy, we would not suspect someone of robbery simply because 118 If by snuggling "as close as possible" Kenner means merely to suggest optimal closeness without infringing, as opposed to maximum closeness, such an unobjectionable choice would not support the Kenner doctrine. 119 See McKeon Prods. Inc. v. Flents Prods. Co., 69 U.S.P.Q.2d 1032, 1035 (E.D. Mich. 2003) (discussing the "standard industry practice" of store brands using packaging similar to targeted name brands). 120 See, e.g., Versa Prods. Co. v. Bifold Co. (Mfg.), 50 F.3d 189, 213 (3d Cir. 1995) (discussing how defendant's extensive labeling helped to dispel any significant likelihood of confusion); Conopco, Inc. v. May Dep't Stores Co., 46 F.3d 1556, 1570-71 (Fed. Cir. 1994) (discussing the "prominent placement" of the defendant's logo on the defendant's store brand product); id. at 1565 (noting that defendant "expressly invites the consumer to compare its product with that of the national brand, by name"). 121 See infra Section III.C.

*p. 35*
they loitered in front of a bank rather than a fire station, without checking whether the bank was actually robbed. Moreover, often it will be unclear whether the firm in fact targeted the mark, as opposed to independently used the same descriptive or suggestive term ("fish fry"), or stumbled across a similar mark by accident. In those instances, the probative value is entirely absent.

## C. Higher Cost of Confusion

*p. 35*
A final tack is to argue that even if the likelihood of confusion is no higher for strong marks, the consequences at a particular level of confusion are greater. For example, suppose the per-consumer cost of mistaken purchases is higher for strong marks. An example might be a mark used on a high-quality luxury handbag, faced with a counterfeiter that is a low-quality imitation. There, the quality gap between the true source and the defendant product might be particularly high. The cost of confusion, for each consumer making a mistaken purchase, would also be high. This greater consequence might furnish an argument for insisting upon a lower threshold of confusion. In terms of the formal model, this would entail a downward shift of the threshold triggering a finding of a likelihood of confusion. Instead of a threshold of (say) 20% of confused consumers, 10% would be enough. 122 This argument does not justify the Kenner doctrine either. Although it is possible to identify examples that fit the theory, the full range of cases is much more varied. In Kenner and B.V.D., for example, the defendant's quality was similar, so far as it appears. Overall we see no reason to expect that the gap for strong marks is usually or systematically larger. The same is true if we instead consider the error costs of nonpurchasers due to post-sale confusion.

*p. 35*
A second theory about consequences is that, all else equal, an infringing defendant will cause a higher proportion of the relevant population to be confused when the plaintiff's mark is stronger. 123 In other words, the infringing defendant may exceed the threshold at 122 Comparable reasoning may partly motivate the Second Circuit's analysis of "the quality of defendant's product" as a factor in the circuit's Polaroid multifactor test for the likelihood of confusion. Polaroid Corp. v. Polarad Elecs. Corp., 287 F.2d 492, 495 (2d Cir. 1961); cf. Virgin Enters. Ltd. v. Nawab, 335 F.3d 141, 152 (2d Cir. 2003) ("The issue of the quality of the secondary user's product goes more to the harm that confusion can cause the plaintiff's mark and reputation than to the likelihood of confusion."). It appears, however, that courts give this reasoning little weight. Empirical work suggests that the quality factor correlates weakly, if at all, with courts' overall determination of liability. See Beebe, supra note 31, at 1612. 123 See, e.g., Bone, supra note 42, at 1346 (making an argument along these lines).

*p. 36*
which liability is found by a greater extent. Where infringement on a weak mark might confuse 30% of consumers, the same infringing conduct with respect to the same but stronger mark might confuse 50% of consumers, and this difference in confusion levels represents a cost that trademark law should take into account when it sets the scope of protection. This theory, however, amounts to a restatement of the basic theory of consumer perception examined in Part II.A. It is already reflected in-and, indeed, is the fundamental basis for-the upward shift in the curve. A third argument is that more expansive protection is justified when a mark affects the lives of, say, tens of millions of people rather than merely thousands. A cluster of cases suggests a relaxed threshold of liability in such instances, essentially shifting from the standard percentage of consumers confused to an absolute number. 124 A problem with this argument, however, is that it leads to the conclusion that if the plaintiff only has thousands rather than tens of millions of customers, it would apparently not matter if all of these customers were confused because the costs to society are so low. More fundamentally, the shift to absolute numbers when the proportion of confusion is low fails to recognize an implicit balance that courts strike when they engage in the confusion analysis. If 10% of consumers are confused in a relevant population of 100 million consumers, that leaves 90 million consumers who are not confused, and who may benefit from the information provided by the defendant's mark. 125 The point of basing the liability determination on the proportion rather than the absolute number of confused consumers is to incorporate into the cost-benefit analysis that underlies that determination not just the costs of bad information but also the benefits of good information. Rejecting this proposition would imply not only broader protection for superstrong 124 (concluding that a survey, in which 7.7% of respondents "perceived a business connection between the two companies" and 8.5% "confused the names," was "strong evidence" of a likelihood of confusion). 125 See Michael Grynberg, Trademark Litigation as Consumer Conflict, 83 N.Y.U. L. REV. 60, 61-62 (2008) (advocating that trademark doctrine take into account the interests of nonconfused consumers).

*p. 37*
marks, but also a wholesale rethinking of major elements of trademark doctrine. 126 A final theory about consequences is that strong marks get different treatment not because of a higher cost of consumer confusion but because the consequences for producers are different. This is the free-rider argument, considered in the next Part.

## III THE FREE-RIDING RATIONALE

*p. 37*
We saw in Part II that the traditional confusion-based rationales for the conventional wisdom have never been persuasive. Soon after B.V.D. declared, in essence, that the emperor has no clothes, courts began to act on the insight that increased trademark strength may sometimes justify narrower scope. Plaintiffs began to lose where formerly they would have won. Though B.V.D. has been forgotten, it had all the makings of an extraordinary turning point in the development of U.S. trademark doctrine.

*p. 37*
It also invited a strong reaction from those who believed that it simply cannot be that as a mark grows stronger, its scope of protection should ever narrow. That reaction entailed an abandonment of what was ostensibly the subject of analysis-the likelihood of confusion. Instead, courts sought refuge in a concern about free riding.

*p. 37*
This Part unpacks the free-riding rationale and exposes its shaky analytical underpinnings. While we recognized in Part II that some confusion-based arguments in support of the positive relation between strength and scope are plausible at lower levels of strength, we argue here that the free-riding rationale cannot justify the positive relation between strength and scope at any level of trademark strength. Part III.A describes reliance upon the free-riding rationale by Kenner and other courts. Part III.B identifies three sets of problems with the freeriding rationale: that trademark law lacks an anti-free-riding principle; that free-riding concerns do not necessarily imply extra protection only for strong marks; and that extra protection (even if warranted) ought not take the form of expanded scope. Part III.C discusses addi-126 Stated differently, if we assume that the defendant's mark has no informational value, then a cost-benefit analysis would recommend that we enjoin any conduct that causes any level of confusion. But the great majority of trademark courts take a different approach. As discussed above, supra note 37, courts insist that the plaintiff show that a significant proportion of the relevant population is confused for liability to be found. Implicit in this approach is a recognition that defendant's conduct, even if it causes confusion among some proportion of consumers, may nevertheless have informational value and should on that basis be allowed. Cf. KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 543 U.S. 111, 121 (2004) (holding that "some possibility of consumer confusion must be compatible with fair use" in trademark law).

*p. 38*
tional problems with the Kenner doctrine for cases (like Kenner itself) in which a mark owner asserts broad mark scope against a competing product. In such cases, there is a strong trademark interest in permitting free riding on signifiers of product attributes and thereby promoting competition and entry. Finally, Part III.D focuses on the operation of the Kenner doctrine in cases involving noncompeting goods. It explains why the extension of the Kenner doctrine to such cases improperly imports into the likelihood-of-confusion analysis concerns better addressed under antidilution law.

## A. Rationalizing Kenner

*p. 38*
As discussed above, Kenner ultimately convinced the court that FUNDOUGH posed an unacceptable likelihood of confusion with PLAY-DOH. The Kenner court did not reach this conclusion by refuting B.V.D.'s alternative model of consumer perception. Instead, Kenner simply dismissed the alternative model as irrelevant:

*p. 38*
While scholars might debate as a factual proposition whether fame heightens or dulls the public's awareness of variances in marks, the legal proposition is beyond debate. The driving designs and origins of the Lanham Act demand the standard consistently applied by this court-namely, more protection against confusion for famous marks. 127 In place of an analysis of heightened confusion, the court interposed a different justification for broadened scope-a concern about free riding. The argument proceeded in two simple steps. First, famous, superstrong marks attract free riders. 128 Indeed, the court deemed free riders likely to "snuggle as close as possible" to the mark 129 -an assertion, refuted above in our analysis of infringer targeting, that is not essential to the free-riding concern. Second, such free riding is problematic because it weakens investment in superstrong marks: Even in their earliest common law origins, trademarks functioned to benefit both producers who invest their good will and capital in a trademark and consumers who rely on those symbols. . . . [T]rademarks protected investments of property owners and 127 Kenner Parker Toys Inc. v. Rose Art Indus., Inc., 963 F.2d 350, 353 (Fed. Cir. 1992). The statement that the legal proposition was "beyond debate" is curious, given that opinions such as B.V.D. had reached a contrary conclusion. The court criticized the Board for taking B.V.D. "out of context," apparently a reference to other cases that had followed the conventional wisdom. Id. at 354. 128 Id. at 353 ("Both the mark's fame and the consumer's trust in that symbol, however, are subject to exploitation by free riders."). 129 Id. ("A competitor can quickly calculate the economic advantages of selling a similar product in an established market without advertising costs. These incentives encourage competitors to snuggle as close as possible to a famous mark.").

*p. 39*
ensured proper return to those who invested work and capital. These manifold purposes and benefits . . . only operate, however, if investments to secure a strong, recognizable mark bring the reward of certain legal protection. If investors forfeit legal protection by increasing a mark's fame, the law would then countenance a disincentive for investments in trademarks. The law is not so schizophrenic. In consonance with the purposes and origins of trademark protection, the Lanham Act provides a broader range of protection as a mark's fame grows. 130 This is not to say that Kenner altogether ignored the question of the likelihood of consumer confusion. Nor could it have. The statutory basis for opposition to registration, section 2(d) of the Lanham Act, provides that an opposition may be granted if a likelihood of confusion is shown. That section makes no mention of free riding, 131 nor does any other section of the Lanham Act. This required the court to return to the ostensibly empirical question of the likelihood of confusion. Rather than evaluate that likelihood, however, the court merely made brief reference to the semantic similarity of "play" and "fun" and the "graphic" similarity of "doh" and "dough," 132 and then implemented its concerns about free riding by imposing a heavy procedural burden on defendants. The court stated that when a market incumbent has established a famous mark, the new market entrant bears the burden of persuasion as to the likelihood of confusion: "In the event of doubts about the likelihood of confusion, the Board and this court should resolve those doubts against the newcomer, especially when the established mark is famous." 133 This special procedural burden effectively widens the scope of strong marks. 134 Other courts have also embraced the free-riding rationale to find infringement of very strong marks. 135 The adoption of a free-riding rationale for the Kenner doctrine can be understood as one compo-130 Kenner, 963 F.2d at 353-54 (footnote and citation omitted). 131 15 U.S.C. § 1052(d) (2012). 132 Kenner, 963 F.2d at 354-55. 133 Id. at 355 (citation omitted). 134 Fortunately, it appears that Kenner's burden-shifting framework has been adopted in only one other court opinion, an unpublished Federal Circuit opinion in 1995. See Money Station, Inc. v. Cash Station, Inc., No. 95-1240, 1995 WL 697313, at *3 (Fed. Cir. Nov. 27, 1995) (unpublished opinion). 135 See, e.g., Au-Tomotive Gold Inc. v. Volkswagen of Am., Inc., 603 F.3d 1133, 1138 (9th Cir. 2010) (finding infringement of Volkswagen's VW marks and asserting that "[p]ost-purchase confusion creates a free-rider problem"); Bd. of Supervisors for La. State Univ. Agric. & Mech. Coll. v. Smack Apparel Co., 550 F.3d 465, 488 (5th Cir. 2008) (finding that use by apparel maker of universities' well-known color schemes "is not an advantage to which it is entitled under the rubric of legitimate competition"); see also Brookfield Commc'ns, Inc. v. W. Coast Entm't Corp., 174 F.3d 1036, 1062 (9th Cir. 1999) (finding that by using Brookfield's MOVIEBUFF mark as a metatag in West Coast's website, nent of a broader acceptance of free-riding arguments in cases normally governed by a likelihood-of-confusion analysis. As Lemley and McKenna report, free-riding arguments have been adopted at the expense of an analysis of confusion in a variety of cases, including merchandising, trademarked characters and objects, initial interest confusion, and post-purchase confusion. 136 Its use here is subtle because it is disguised, a sort of Trojan horse. Put another way, the free-riding justification for the Kenner doctrine is an attack on the likelihood of confusion doctrine from within (in that courts use free riding to justify a finding of likelihood of confusion), rather than an overt substitute for it.

## B. Weaknesses and Limitations

*p. 40*
The free-riding rationale, though undertheorized, at first glance seems plausible. After all, concerns about free riding are pervasive in IP. The standard economic argument in favor of IP is that temporary monopoly internalizes a benefit and thereby provides an inducement to innovate, an incentive that free riding undercuts. This argument conventionally applies to patent and copyright, but an extension to trademark might seem natural-particularly to Federal Circuit judges such as Judge Rader whose expertise is primarily in patent law.

*p. 40*
On deeper inspection, however, the rationale suffers from several serious analytical problems. Most of our critique is directed to Kenner's incentive account-the unexamined idea that broader protection confers an important economic incentive that ought to be protected particularly for strong marks but not weak marks. Some of our discussion also responds to the belief that on moral grounds, or as a matter of unjust enrichment, trademark law does not permit a firm to reap where it has not sown, whatever its incentive effects. 137 Protection against free riding. The first problem is that U.S. trademark lacks a clear anti-free-riding principle that would support broader protection. 138 The Supreme Court has emphasized that trademark law focuses primarily on consumer search costs rather than pro-West Coast "improperly benefits from the goodwill that Brookfield developed in its mark"). 136 ducer incentives. It is designed to reduce consumers' "costs of shopping" 139 by assuring them that a product with a given mark has the characteristics they expect given previous purchases or advertising. These costs include the time and effort to identify a product with particular characteristics and the cost of error from choosing a product with inferior or otherwise different characteristics.

*p. 41*
This goal is quite different from patent and copyright's focus on producer incentives. In accordance with this goal, U.S. law prevents only uses that confuse consumers as to source or affiliation (or are diluting of the senior mark). In this respect, U.S. law differs from European trademark law, which contains an explicit provision that prohibits taking "unfair advantage" of the "distinctive character or the repute of [an] earlier trade mark." 140 To be sure, the Supreme Court has given some attention to producer incentives. For example, in Qualitex Co. v. Jacobson Products Co., the Court noted that trademark protection permits a producer to internalize the rewards for product quality. 141 But this proposition only applies to free riding by producers of confusing goods-those producers who "capitaliz[e]" 142 on confusion. Free riding of other types is excluded from the Court's rationale. That exclusion is consistent with widespread free riding undertaken by parodists, search engines, comparative advertising, and products that make use of functional aspects of an incumbent's trade dress. Free riding is not just permitted, but is an essential part of trademark's proper functioning. 143 The Court's limitation makes sense because stronger protection-and especially broader protection-imposes a heavy burden on the core trademark goal of minimizing consumer search costs. Broader protection actually increases consumer search costs by making it more expensive for other firms to invest in their own differentiated marks. To take one example of increased breadth, search costs would increase if a mark owner were able to control the generic 139 Qualitex Co. v. Jacobson Prods. Co., 514 U.S. 159, 163 (1995) (internal quotation marks omitted). 140 Council Directive 2015/2436, art. 10(2)(c), 2015 O.J. (L 336) 11. 141 Qualitex, 514 U.S. at 164 ("[T]he law helps assure a producer that it (and not an imitating competitor) will reap the financial, reputation-related rewards associated with a desirable product."). 142 Id. ("The law thereby encourage[s] the production of quality products, and simultaneously discourages those who hope to sell inferior products by capitalizing on a consumer's inability quickly to evaluate the quality of an item offered for sale.") (alteration in original) (citation and internal quotation marks omitted). 143 See David W. Barnes, Free-Riders and Trademark Law's First Sale Rule, 27 SANTA CLARA COMPUTER & HIGH TECH. L.J. 457, 475-80 (2011) (discussing ways in which trademark law permits free riding).

*p. 42*
term for a product rather than merely a particular brand. 144 Broader protection leads to lower investment by rivals, higher consumer search costs as to the rivals' products, decreased industry supply, and deadweight loss. 145 It is thus too strong to say, as Kenner did, that "trademarks protect[ ] investments of property owners and ensure[ ] proper return to those who invest[ ] work and capital." 146 Rejecting the free-riding rationale does not require embracing any "schizophrenia" in our approach to trademark law.

*p. 42*
Stronger protection for stronger marks. Even if free-riding concerns supported strong trademark protection, it does not follow that stronger marks merit stronger protection. It is not enough that stronger marks are associated with a higher level of investment. The relevant question, from an economic standpoint, is the effects from a change in the mark owner's investment induced by the extra opportunity to internalize benefits from the mark.

*p. 42*
For example, suppose that a superstrong mark sees a certain amount of extra investment thanks to extra protection from free riders, while a (merely) strong mark receives an even bigger boost. In that case, although both are benefitted, the Kenner doctrine has it backwards, and it is weaker marks that merit the extra protection, rather than superstrong marks.

*p. 42*
Whether stronger marks merit greater protection depends on the size of reduced profits due to free riding and the sensitivity of investment to that reduction, among other factors, for each type of mark. Stronger marks are plausibly subject to more free riding, with greater effect on profits. Whether this affects investment is another matter. For very strong marks, the level of investment might well be at a maximum level even with free riding. To oversimplify a complex question, even if the NFL has to contend with free riders making unauthorized (and often perfectly legal) uses of the term SUPER BOWL, the NFL's level of investment in promoting and producing the spectacle is likely unaffected. We see no firm basis for concluding that the strongest marks merit more, less, or the same amount of protection on freeriding grounds.

*p. 42*
Increased scope for stronger marks. Finally, even if free riding supported extra protection for stronger marks, it does not follow that the extra protection should take the form of broader scope. After all, stronger marks enjoy heightened protection already, within the "ordi-nary" scope of the mark. In particular, their protection against very similar marks is heightened because, as noted in Part I, stronger marks imply higher confusion at high levels of similarity, whatever the effect at lower levels of similarity. This higher level of confusion unlocks multiple forms of relief by tending to establish a high likelihood of confusion, actual confusion, and bad faith. A high likelihood of confusion clears the likelihood-ofconfusion threshold by a greater margin, which in turn makes it more likely in practice that a mark owner can enjoin the defendant's confusing use. Showing actual confusion is typically a prerequisite for damages. 147 Establishing bad faith is an important prerequisite for receiving enhanced damages. 148 This sort of extra protection might be what some courts implicitly have in mind in referring to a strong mark's "greater protection" or "latitude" rather than greater scope. 149 A final reason not to confer extra protection through extra scope is that the end has little connection to the means. The free-riding rationale implies no difference in consumer behavior, such as heightened confusion for strong marks. 150 Nor does it posit a difference in infringer behavior, such as infringer targeting of strong marks. 151 Instead, the free-riding rationale merely addresses the claimed consequence of consumer confusion (whatever its level) for mark owner behavior.

*p. 43*
The attenuated connection is particularly glaring because the extra trademark scope makes a difference only at low levels of confusion. The practical effect of granting broader scope, even though (on this theory) consumer behavior is no different for strong marks, is to lower the threshold of confusion for the strongest marks. In terms of the formal model, this is once again a downward shift of the threshold that would trigger a finding of a likelihood of confusion. Instead of (say) 20% of confused consumers, 10% is enough. The lower threshold makes a difference only when the defendant's use produces confusion between 10% to 20%. Above 20%, the plaintiff would win 147 See, e.g., Res. Developers, Inc. v. Statue of Liberty-Ellis Island Found., Inc., 926 F.2d 134, 139 (2d Cir. 1991); Web Printing Controls Co. v. Oxy-Dry Corp., 906 F.2d 1202, 1204-05 (7th Cir. 1990). See generally MCCARTHY, supra note 29, at § 30:74. 148 See MCCARTHY, supra note 29, at § 30:92 (collecting cases where damages were increased because infringement was intentional).

*p. 43*
149 Xtreme Lashes, LLC v. Xtended Beauty, Inc., 576 F.3d 221, 227 (5th Cir. 2009) ("Stronger marks are entitled to greater protection."); Recot, Inc. v. Becton, 214 F.3d 1322, 1327 (Fed. Cir. 2002) ("Famous marks thus enjoy a wide latitude of legal protection."). 150 See supra Section II.A (discussing rationales for the positive relation between mark strength and scope that proceed from the assumption that mark strength affects consumer perception). 151 See supra Section II.B (discussing the infringer-targeting rationale for the Kenner doctrine).

*p. 44*
either way; below 10%, it would lose either way. In other words, the extra benefit only accrues with defendant uses that raise little risk of confusion.

## C. Competing Products and Broad Mark Scope

*p. 44*
The previous section explains why free-riding concerns generally fail to justify the Kenner doctrine. This section considers the special case of competing products. If the defendant sells a competing product, the Kenner doctrine implies an expansion along the dimension of mark scope. In this setting, additional problems arise.

*p. 44*
Free riding by competitors is a neglected subject. For example, Lemley and McKenna's otherwise thorough analysis of free-riding arguments in trademark law largely avoids competing products. 152 Addressing this question is important, given that cases such as Kenner rely explicitly and heavily on a free-riding concern, rather than treating free riding as a makeweight.

*p. 44*
As Part III.C.1 explains, not all free riding is created equal. The free riding at issue in competitor cases often pertains to a signifier of product attributes, rather than source. Permitting a competitor to reuse the signifier lowers consumer search costs and thereby furthers a core trademark value. Doing so is desirable for a further reason spelled out in Part III.C.2, to promote competition and entry in a particular product category.

## Reducing Consumer Search Costs

*p. 44*
There are several forms of free riding that must be distinguished. The most obvious form is misappropriation of a trademark by a counterfeiter. Such conduct is irrelevant to the examination of free riding as an independent rationale for the Kenner doctrine, separate from the prospect of confusion, because such conduct is already easily condemned in any event due to the high probability of confusion. A second form of free riding is affiliation, paradigmatically engaged in by a maker of noncompeting goods seeking to invoke and profit from the resonance associated with (say) TIFFANY. This form of free riding is considered below in Part IV.D. 152 See Lemley & McKenna, supra note 10, at 174 (conceding that the argument that trademark protection incentivizes market entry and investment in product quality "makes much more sense in the context of competing, or at least closely related, goods"). At the same time, Lemley and McKenna clearly recognize that free riding by competitors is entirely proper. When they demonstrate the instability of lay intuitions about free ridingand the acceptability of free riding in a wide range of circumstances-they are drawn to examples featuring competing goods. See id. at 183 (discussing, inter alia, the permissibility of a firm's decision to locate a gas station directly across the street from a competitor).

*p. 45*
A third form of free riding is the recycling of a term that denotes a particular product attribute. For example, the DOUGH in FUNDOUGH denotes (as does the -DOH in PLAY-DOH) a water-based toy modeling compound. The -STER in GROKSTER denotes (echoing NAPSTER) an easy-to-use filesharing service. 153 CHARBUCKS (echoing STARBUCKS) is a "beacon" identifying high-quality "dark-roasted coffees." 154 HER- BROZAC denotes an herbal alternative to the antidepressant PROZAC. 155 And in a recent controversy, DIET RITE PURE ZERO (echoing COKE ZERO) denotes a zero calorie soft drink. 156 All five examples feature competing goods, where a second firm-call this firm the "entrant," though it need not be a new entrant-recycles a term already in use by an incumbent. This characteristic and distinctive feature of the competing products case requires separate analysis.

*p. 45*
An important benefit of the recycling is to communicate to consumers what it is, exactly, that the entrant is selling. The recycling thereby serves the consumer interest in minimizing search costs. This consumer interest is greatest as to the strongest marks. Strong marks are associated with successful incumbents. These marks contain signifiers worth recycling. This argument suggests that a strong mark ought to face a higher threshold level of confusion, all else equal, rather than a lower one. Id. at 32-35 (determining priority issue). But COKE ZERO is by far the stronger mark, so that most consumers will likely have heard of COKE ZERO first or in any case assume that COKE ZERO came first. 157 In addition, consumer familiarity already gives the senior mark owner a competitive advantage against competing products as a result of the mere exposure effect, in which consumers like familiar stimuli more than unfamiliar ones. Robert B. Zajonc, Attitudinal Effects of Mere Exposure, 9 J. PERSONALITY & SOC. PSYCHOL. MONOGRAPH SUPPLEMENT 1, 23 (1968) (concluding that "mere repeated exposure of an individual to a stimulus object enhances his attitude toward it"). Risk is inherent in the unknown, and many people are risk averse. Robert F. Bornstein, Exposure and Affect: Overview and Meta-Analysis of Research, 1968-1987, 106 PSYCHOL. BULL. 265, 282 (1989). Familiarity is a strong predictor of preference, particularly in purchase contexts in which consumers do not have the time or motivation to carefully evaluate competitors. See Paul Hekkert et al., Most Advanced, Yet Acceptable: Typicality and Novelty as Joint Predictors of Aesthetic Preference in Industrial Design, 94 BRIT. J. PSYCHOL. 111, 122 (2003) ("Since it has been argued that a preference for typicality reflects an automatic or immediate reaction towards easy-toclassify stimuli, it can be predicted that typicality will be a stronger predictor of aesthetic preference when there is less time to process the stimulus."); Wayne D. Hoyer, An Examination of Consumer Decision Making for a Common Repeat Purchase Product, 11 J. CONSUMER RES. 822, 828-29 (1984) (finding that when choosing repeatedly-purchased Recycling is a form of free riding, inasmuch as the entrant is making a deliberate choice to free ride on an incumbent's prior investments in educating consumers about a product attribute. For example, Grokster's choice of a -STER ending was likely deliberate. 158 The key distinction is that, though deliberate, the purpose and effect is to communicate the product attribute, not to confuse as to source or affiliation.

*p. 46*
Not all such uses are well characterized as free riding. Where the term is already well known to the public, there may be no relevant investment by the incumbent or else no targeting by the entrant. Moreover, the relevant investment may be a consumer investment. Consumers undertake costly efforts to understand the characteristics of a good and how to use it, and to understand how a particular label maps to those characteristics. A substantial literature details how user investments contribute to the success of trademarks. 159 In these instances, shared use may occur without free riding.

*p. 46*
Trademark doctrine explicitly protects this sort of recycling. For example, marks and parts of marks are recognized as generic for a product name and hence unprotectable. "Dough" and its phonetic equivalent "doh" are generic for water-based modeling compound, a point recognized by the Board in Kenner. 160 Moreover, the Board reasoned, where (as there) "the only common element . . . is a generic term," confusion is unlikely because the generic matter is not likely to be understood by the public as an indication of common affiliation or trade identity, but rather is likely to be viewed as indicative of the nature or characteristic of the product. Differences in the remaining parts of the marks are likely to avoid confusion in such cases. 161 low-importance products, consumers engage in minimal in-store deliberation and instead employ simple choice tactics that may rely on, among other things, familiarity with a brand and exposure to advertisements for it). The boost that strong, familiar products already have psychologically is another reason that counsels against expansion for the strongest marks.

*p. 46*
158 Metro-Goldwyn-Mayer Studios Inc., 545 U.S. at 939 ("Grokster's name is apparently derived from Napster."). 159 See, e.g., Deborah R. Gerhardt, Consumer Investment in Trademarks, 88 N.C. L. REV. 428, 450-58 (2010) (discussing how consumers participate in creating trademark value); Steven Wilf, Who Authors Trademarks?, 17 CARDOZO ARTS & ENT. L.J. 1, 1 (1999) (discussing the "public's significance as trademark creator"). 160 Kenner Parker Toys, Inc. v. Rose Art Indus., Inc., Opposition No. 75,237, 1992 TTAB Lexis 39, at *8 (T.T.A.B. Mar. 20, 1986); see also id. at *7 ("[I]t appears that virtually all manufacturers of water-based compounds use 'DOUGH' or its phonetic equivalent as part of their brand names."). 161 Id. at *9.

*p. 47*
The common reference to "dough" serves the public by indicating what the product is. Other examples include the prefixes "e" for "electronic" 162 and "i" for "internet." 163 This doctrine protects recycling of generic terms without regard to the incumbent's investment in teaching the consumer, which may be large, or the entrant's choice to free ride on that investment. Even clearer is the doctrine of genericide. When a mark comes to signify not a single source but the product itself-as with aspirin, elevator, and yo-yo-the mark passes into the public domain. 165 A finding of genericide allows competing producers to free ride on the mark-building investments of the incumbent. Genericide is particularly relevant because the type of free riding at stake is the same. The entrant wishes to take advantage of the work already done to educate the public about what the word "yo-yo" or "elevator" signifies. The entrant prefers to avoid advancing a new alternative, which would be costly-likely even more costly than the original effort, given consumers' familiarity with the standard term-and wasteful. These examples from trademark law suggest that free riding on product attributes is compatible with trademark policy. This is hardly surprising, given the central trademark goal of minimizing consumer search costs.

*p. 47*
Moreover, this reasoning fits well with the more general principle that trademark functions best when a mark leaves enough and as good for others. 166 Beyond the treatment of generic marks, including genericide, the principle appears in the Supreme Court's concern to avoid a depletion of colors as signifiers, 167 in the exclusion of functional features of trade dress from protectability, 168 and in the granted to firms to engage in the descriptive fair use of terms trademarked by others. 169 It would abrogate that principle to give an incumbent a monopoly on a signifier, granting extra protection where the incumbent finds, or through investment creates, a particularly apt or catchy way to describe the product.

*p. 48*
On the other side of the ledger, the cost of permitting such free riding, insofar as an assessment of the Kenner doctrine is concerned, is low. By construction, we are only considering recycling in contexts where the probability of confusion as to source or affiliation is low. If it were high, ordinary likelihood-of-confusion analysis would apply, and there would be no need for the Kenner doctrine to expand the boundaries.

*p. 48*
These low costs contrast favorably with the costs of genericide. With genericide, the incumbent is obliged to develop a new designator of source, 170 and some consumers may continue to believe that the generic term still refers uniquely to the incumbent. 171 Moreover, permitting rivals to use informatively similar marks acts as a safety valve against genericide. For example, if an entrant is free to adopt a variant with the same ending (GROKSTER; FUNDOUGH), there is less need to use the whole word as a product signifier, and less need for genericism doctrine to step in. This approach also avoids the need for a legal determination that the whole mark has become generic. Indeed, after losing in the Federal Circuit, the maker of FUNDOUGH argued that PLAY-DOH had become generic. 172 Another possibility is that the scope of genericide is narrowed, as just the attribute-signifying part of the mark-such as "cola" for drinks or "shuttle" for airline servicesis dedicated to the public. 167 Qualitex Co. v. Jacobson Prods. Co., 514 U.S. 159, 168-69 (1995) (discussing the possibility of "color depletion" and suggesting that trademark functionality doctrine would minimize its anticompetitive consequences). 168 171 See MCCARTHY, supra note 29, at § 12:6 (discussing scenarios in which a minority of the relevant consumer population still perceives a generic term as a designation of source). 172 The firm changed the product's name to FUN DOUGH, attracting a new suit, to which it responded with a defense of genericide, apparently not having made this argument the first time around. Tonka Corp. v. Rose Art Indus., Inc., 836 F. Supp. 200, 207, 209 (D.N.J. 1993). Following a denial of summary judgment on this argument, id. at 217, the parties subsequently settled. In the end, FUN DOUGH remained on the market.

## Promoting Competition

*p. 49*
This trademark interest-in preventing an incumbent from capturing single "best" way to describe a product and its attributescan also be understood as a form of competition policy internal to intellectual property doctrine. By preventing the incumbent from placing entrants at a disadvantage, we promote entry and competition.

*p. 49*
It is well recognized that trademark law seeks to promote competition. 173 It does so through several features already discussed, including the doctrinal treatment of generic marks (and parts of marks), genericide, and the nonprotectability of functional elements. Competition is further promoted by the trademark doctrines of comparative advertising and nominative use. 174 Limitations on the scope of a mark reduce consumer search costs, freeing up rivals to use similar marks and thereby increasing industry supply and consumer welfare. 175 The point also finds expression in the general idea that trademark enforcement should be held in abeyance where it does not serve this goal. 176 This perspective sharply conflicts with Kenner, which openly embraces the idea that trademark rights might serve as a barrier to entry. The Kenner doctrine grants the strongest incumbents extra freedom of action to prohibit entry. From the standpoint of competition policy, this has things backwards. Antitrust law is premised on the view that firms with a measure of market power are subject to a smaller freedom of action in inhibiting competitors. The most successful products, the ones with market power, get less leeway rather 173 Peaceable Planet, Inc. v. Ty, Inc., 362 F.3d 986, 990 (7th Cir. 2004) (Posner, J.) (noting a purpose to "promote competition and consumer welfare"). But see Deven R. Desai, The Chicago School Trap in Trademark: The Co-Evolution of Corporate, Antitrust, and Trademark Law, 37 CARDOZO L. REV. 551, 606-07 (2015) (drawing different implications from quoted language on the ground that "consumer welfare" is actually a reference to total surplus). 174 See, e.g., Hypertherm, Inc. v. Precision Prods., Inc., 832 F.2d 697, 701 (1st Cir. 1987) ("Prohibiting imitators from telling others the very purpose [of their products] would severely restrain competition without serving the slightest countervailing (valid) purpose."). 175 LANDES & POSNER, supra note 144, at 187, 191 (discussing economic costs if producer permitted to appropriate a nondistinctive mark or appropriate the name of a product). 176 This conclusion echoes recent academic calls to recognize "IP injury" and enforce IP law only where there is a violation of a core IP policy. than more. For most antitrust claims, market power is a for liability. 177 To be clear, antitrust law does not have much direct application here. A trademark, in itself, seldom confers substantial market power. A trademark is one among multiple instruments for preserving status as a differentiated product and attaining some degree of power over price. A trademark usually confers merely "slight monopoly power," and hence ordinarily is not worth the "heavy artillery of federal antitrust law." 178 At the same time, the antitrust analogy does suggest the function of a competition norm internal to trademark. 179 Denying heightened protection to marks that sue over competing products is a natural corollary of that norm.

*p. 50*
We observe a similar dynamic in copyright law, which, like trademark, has faced the question of user investments which serve to make a particular interface into a standard and thereby create a kind of lock-in. 180 In response, one appellate court denied protectability to a hierarchy of software menu commands that a competing, improved program was employing in order to enable compatibility for users. 181 In an influential concurring opinion, one judge explained that copyright should not force users to remain a "captive" of the incumbent, instead allowing users "to exploit their own prior investment in learning" the interface. 182 As the copyright example demonstrates, our point about trademark fits within a larger IP policy concern: that user investments ought to be protected in order to promote competitive entry.

## D. Noncompeting Products and Broad Product Scope

*p. 51*
The previous section explained why free-riding concerns cannot justify expansive mark scope in the particular context of competing goods. This section considers the alternative special case in which the defendant sells a noncompeting (either unrelated or related) product, and the Kenner doctrine is deployed to justify expansive product scope. 183 Consider, for example, a defendant's use of the TIFFANY mark on bicycles. Typically, when a defendant uses an identical or closely similar mark on unrelated goods, there is little consumer confusion as to source. Under an ordinary infringement analysis, a bicycle maker would be free to use the TIFFANY mark in this way. Despite the lack of confusion, free-riding concerns might be invoked to justify a broad scope of protection. The most important argument is that defendant's unlicensed use, in the course of taking advantage of plaintiff's investment, thereby damages plaintiff's mark. A second argument is that even if there is no damage, defendant's unlicensed use improperly denies compensation to the mark owner. Let us take these arguments in turn.

*p. 51*
The claim of damage is that defendant's use "blurs" the link between the plaintiff's mark and the plaintiff and its products. For example, even if TIFFANY on bicycles does not confuse consumers as to source, it will still blur the link in consumers' minds between the mark TIFFANY and the New York City jeweler. One justification for preventing blurring is that blurring increases consumer search costs. As Judge Richard Posner puts it, diluting uses require consumers to "think for a moment"foot_10 to determine to which company the blurred mark is referring. This may well be true-though the search costs explanation for antidilution protection has been roundly criticized. 185 Regardless, as Frank Schechter recognized nearly a century ago when he first introduced the concept of dilution to American law, the more fundamental harm from dilution is the harm to the mark itself. 186 Diluting uses blur the mark's associations and render the mark less unique, which damages the mark's "selling power." 187 This damage may be particularly severe for status brands like TIFFANY in that much of their value derives from their reputation for uniqueness and exclusivity.

*p. 52*
In U.S. trademark law, concerns about blurring are handled not by an ordinary claim of infringement, but by a special body of trademark antidilution law. A dilution claim is the legal channel through which mark owners vindicate their concerns about harm from nonconfusing uses of noncompeting goods. For example, Tiffany & Company has repeatedly brought antidilution causes of action against uses of the mark on noncompeting goods. 188 Antidilution protection remains controversial in American law because it essentially grants rights in gross in the mark as to all uses, be they competing or noncompeting. 189 But importantly, Section 43(c) of the Lanham Act, which is the basis of the federal antidilution cause of action, contains several significant limitations on the availability of relief. One is that in order to qualify for protection, the mark must be "famous," 190 which the section defines as "widely recognized by the general consuming public of the United States." 191 A second is that the plaintiff must do more than merely show that consumers associate the defendant's mark with the plaintiff's. The plaintiff must also show that this association "impairs the distinctiveness of the [plaintiff's] famous mark." 192 In other words, the plaintiff must show a likelihood of material harm to its mark. In practice, this requirement has proven to be extremely demanding, if not insurmountable. 193 To the extent that the Kenner doctrine is explained as a response to harm from free riders even when there is no likelihood of confusion, it treads nearly the same ground as antidilution law. The differ-ence is that antidilution law offers a reasonably clear theory of how noncompeting uses harm a mark and its owner's incentives to develop the mark. This theory of harm can support limitations on antidilution rights. 194 In contrast, Kenner simply asserts that any narrowing of protection harms incentives and any broadening of protection improves incentives, but it contains no limitations on this logic, which in its unlimited form is nothing more than an argument for in gross rights for all strong marks across all competing or noncompeting uses. Such a theory cannot support the proposition that the scope of famous marks should extend beyond that justified by the likelihood of consumer confusion. Thus, if defendant's (non-confusing) use arguably harms the mark through blurring, that contention is properly handled using specialized antidilution law, rather than by watering down ordinary trademark doctrine. 195 That leaves the residual free-riding argument that, even if defendant's use causes no harm to plaintiff's mark, plaintiff is still being denied licensing fees that properly accrue from defendant's use. After all, if a license is required, one might readily imagine a bicycle maker being willing to pay the owner of TIFFANY or ROLLS-ROYCE for use of an important mark. Moreover, if there is little or no harm from blur- 194 For example, the blurring theory of harm supports the demanding statutory requirement that the plaintiff show that consumers' association of the plaintiff's and defendant's marks "impairs the distinctiveness of the [plaintiff's] famous mark." 15 U.S.C.

*p. 53*
§ 1125(c)(2)(B). See supra notes 192-93 and accompanying text. 195 An alternative and far less common form of dilution is "dilution by tarnishment," which Section 43(c) of the Lanham Act defines as "association arising from the similarity between a mark or trade name and a famous mark that harms the reputation of the famous mark." 15 U.S.C. § 1125(c)(2)(C). Tarnishment causes of action have generally been limited to conduct that links the plaintiff's mark to low-quality products or degrades the positive associations of the mark. See L.L. Bean, Inc. v. Drake Publishers, Inc., 811 F.2d 26, 31 (1st Cir. 1987) ("The threat of tarnishment arises when the goodwill and reputation of a plaintiff's trademark is linked to products which are of shoddy quality or which conjure associations that clash with the associations generated by the owner's lawful use of the mark."). Tarnishment implicates concerns separate from those we address here. In addition to requiring the plaintiff to show that its mark is famous, the tarnishment cause of action also requires the plaintiff to show that the defendant's products are of significantly lesser quality or that the defendant's conduct links the plaintiff with salacious or degrading associations. Courts have generally been reluctant to find tarnishment. See, e.g., Louis Vuitton Malletier S.A. v. Haute Diggity Dog, LLC, 507 F.3d 252, 269 (4th Cir. 2007) (rejecting plaintiff's tarnishment claim). But see V Secret Catalogue, Inc. v. Moseley, 605 F.3d 382, 388-89 (6th Cir. 2010) (holding that defendant's use of the mark VICTOR'S LITTLE SECRET tarnished the VICTORIA'S SECRET mark by associating it with "sex related products"). However, the reasoning and holding in V Secret Catalogue has been very strongly criticized. See, e.g., MCCARTHY, supra note 29, at § 24:89 (characterizing the majority opinion in V Secret Catalogue as "wildly misguided").

*p. 54*
ring, one might imagine a mark owner willing to grant a license in exchange for payment. 196 The prospect of licensing fees cannot be easily dismissed as a matter of theory. In principle, such fees, if available, might motivate higher levels of investment in a trademark. 197 This licensing question arises distinctively for noncompeting products. For a mark owner facing a competitor, the minimum license price is higher than the competitor is willing to pay. Kenner will not offer the maker of Fundough a license at a price the latter is willing to accept because competing entry lowers the incumbent firm's profits by too much.

*p. 54*
However, it is unlikely that a right to control use of the mark on unrelated products would produce enough licensing fees to change the behavior of mark owners. In such an instance, the licensee would have a large set of marks to choose from-not only TIFFANY but also ROLLS-ROYCE and many others. Competition among licensors might drive down the license price to a low level. 198 Where the license price is low compared to the transaction costs, it makes sense simply to assign the right to the would-be licensee. 199 Moreover, a low fee 196 An additional argument in support of the proposition that a mark owner may be harmed by an unauthorized non-confusing use of its mark on non-competing products is that the unauthorized use will impair the ability of the mark owner to extend its mark into the product category in which the unauthorized use is being made. For example, if a party is allowed to make a non-confusing use of TIFFANY on bicycles, then this will severely limit the ability of the jewelry maker to extend its brand into bicycles. See Mark P. McKenna, Testing Modern Trademark Law's Theory of Harm, 95 IOWA L. REV. 63, 115-17 (2009) (discussing this "market preemption" theory of harm). It is, however, a fundamental principle of U.S. trademark law that a trademark owner's rights are not rights in gross in the mark but rather extend only so far as the owner's actual use of the mark. See, e.g., Standard Brands, Inc. v. Smidler, 151 F.2d 34, 36 (2d Cir. 1945) ("A trade-mark is not one in gross like a patent right but is a right of user in connection with a trade or business to designate the product to which the mark is applied in that trade or business."). The market preemption argument applies to non-confusing uses of a mark in any product category in which the mark owner is not currently using its mark; it is in this sense essentially an argument in favor of unlimited rights in gross. 197 For example, Lemley and McKenna note the "possib[ility] that refusing to protect a mark against some uses outside [a particular range] will somewhat reduce the incentive to invest in this brand 'personality,'" while concluding that this incentive is small. See Lemley & McKenna, supra note 10, at 177. Such an effect on incentives would escape the circularity of a "harm" defined solely by reference to the asserted presence of a legal entitlement. See, e.g., id. at 141 ("The claim that trademark owners are injured by not being able to control use in a remote market is ultimately a circular claim-mark owners are injured if, but only if, we define their trademark rights ex ante to include control over that remote market."). 198 See LANDES & POSNER, supra note 144, at 208 (making this argument). 199 An analogous argument has been advanced in copyright law as a reason to permit uncompensated use by others, particularly "productive" uses. See, e.g., Timothy J. Brennan, Harper & Row v. The Nation, Inc., Copyrightability and Fair Use, 33 J. COPYRIGHT SOC'Y U.S.A. 368, 385 (1986) (arguing that where "expected market price . . . would be close to zero," a "zero-price compulsory license" minimizes transaction costs).

*p. 55*
would have little incremental effect on the investments of a strong mark owner that already possesses a strong incentive to invest in the mark. In any event, this concern with lost licensing revenue is largely theoretical. In cases like TIFFANY bicycles, the real practical issue is typically an alleged harm from blurring-a harm that makes voluntary licensing a fanciful prospect-rather than lost licensing revenue that the mark owner would otherwise be happy to collect.

## IV IMPLICATIONS

*p. 55*
We have argued that the conventional wisdom that stronger marks always merit a wider scope of protection is incorrect. The strongest trademarks should sometimes receive narrower protection than marks that are of lesser strength. While the confusion-based rationales for the positive relation between strength and scope may make some sense at lower levels of strength, they break down at very high levels of strength. Meanwhile, Kenner's free-riding rationale fails to support the positive relation between strength and scope at any level of strength, let alone at very high levels of strength. If we are correct that the relation between strength and scope takes the form of an inverted U in some cases, then this supports several important reforms of trademark doctrine.

*p. 55*
The first is that courts should not apply the Kenner doctrine without some showing that the doctrine is appropriate to the facts of the cases before them. When conducting the multifactor test for the likelihood of consumer confusion, courts currently apply the rule that stronger marks merit wider scope as if it were true in all instances, even when it clearly is not. Particularly when the plaintiff's mark is a superstrong mark, the plaintiff should be required to explain why this strength supports a wider rather than a narrower scope of protection. This may not be difficult for plaintiffs in some contexts. For example, the plaintiff's case may rely on affiliation confusion rather than signifier confusion, or the plaintiff's consumers may be exceptionally unsophisticated and thus easily confused as to source. 200 In other cases, however, we expect that plaintiffs will not be able to persuade a court to apply the Kenner doctrine, resulting in a finding of no confusion. For example, in cases presenting typical dilution facts-the plaintiff's mark is extremely strong and the parties' goods are noncompeting-the Kenner doctrine will almost never be appropriate. The owner of a mark on the order of APPLE, DISNEY, or NIKE cannot reasonably claim that the strength of its mark increases the likelihood of confusion as to source if the defendant has applied a similar mark to, for example, laboratory equipment. 201 Consumers know such marks too well to assume any kind of affiliation. Such a case would present exactly the kinds of facts that show why the Kenner doctrine is not a one-size-fitsall doctrine-and why the plaintiff should look to antidilution law and be obligated to meet its requirements for relief.

*p. 56*
Second, and closely related, in determining whether the Kenner doctrine should apply to a particular set of facts, courts should recognize that the logic supporting application of the Kenner doctrine does not naturally carry over from one category of case to another. For example, if the parties' marks are identical, it would be inappropriate for a court to cite cases reasoning that the Kenner doctrine should apply because of signifier confusion. Similarly, an application of the Kenner doctrine in an unrelated products case does not necessarily apply to a competing products case. The unrelated products case might pertain to expansive product scope, whereas the competing products case would concern expansive mark scope. The unrelated products case might rely on a particular top-of-mind argument or a consumer inference argument about affiliation confusion, inapplicable to a competing products case focused on signifier confusion. The competing goods case also poses a unique challenge to the Kenner doctrine because of the consumer interest in recycling signifiers of product attributes. Nor, for similar reasons, can an application of the positive relationship in a competing products case (which might have focused on signifier confusion) necessarily support its use in an unrelated products case (which might focus on affiliation confusion). All of this suggests that the relation between strength and scope is far more complex than courts and commentators have generally appreciated, and that courts should not apply cases and reasoning that support only one version of the Kenner doctrine to facts calling for a different analysis. Perhaps unsurprisingly, Kenner itself violated this principle, indiscriminately citing both cases involving competing goods and others that involved entirely unrelated goods. 202 For example, in arguing that courts before and after B.V.D. held that stronger marks accede to broader protection, Kenner cited a number of cases about competing products. Kenner Parker Toys Inc. v. Rose Art Indus., Inc., 963 F.2d 350, 354 (Fed. Cir. 1992) (citing cases). But in the same paragraph of analysis, Kenner then also cited R.J. Reynolds Tobacco Co. v. R. Seeling & Hille, 201 U.S.P.Q. 856 (T.T.A.B. 1978), which involved Third, in deciding whether to apply the Kenner doctrine, courts should recognize that consumers' likelihood of confusion is largely endogenous to trademark law and to the Kenner doctrine itself. This is because consumers develop expectations about how different two marks should be based on the general degree of difference they find in the marketplace. Yet this general degree of difference is largely based on what the law allows. 203 If the law broadens the average scope of protection afforded to marks, consumer expectations will adjust over time, so that marks that were once perceived as non-confusing will come to be perceived as confusingly similar. The reverse process will occur if the law narrows the scope of protection. To the extent that the Kenner doctrine broadens protection for superstrong marks, consumers will adjust their expectations accordingly. But if courts decline to apply the Kenner doctrine and allow competitors to "snuggle" nearer to superstrong marks, consumers will learn to tell the difference between the superstrong mark and its competitors. Some confusion may result in the near term, but in the long term, the system can be expected to reach a new equilibrium, along with long-term benefits in enhanced competition and lower barriers to entry.

*p. 57*
Fourth and finally, the Federal Circuit should abandon the rule that a defendant, if it is a new entrant, bears the burden of persuasion on the question of likelihood of confusion with a superstrong mark. As we have seen, this is the very situation in which concerns about entry are highest. 203 This has long been a theme in trademark case law and commentary. See, e.g., Vornado Air Circulation Sys., Inc. v. Duracraft Corp., 58 F.3d 1498, 1509 (10th Cir. 1995) (describing the likelihood of consumer confusion with respect to product configuration trade dress as a "self-fulfilling prophecy" in that if "product configurations are protected as identifiers, consumers will come to rely on them for that purpose, but if copying is allowed, they will depend less on product shapes and more on labels and packaging"); Duraco Prods., Inc. v. Joy Plastic Enters., Ltd., 40 F.3d 1431, 1451 (3d Cir. 1994) (speculating that the protection of product configurations as trademarks could cause a "snowballing effect" in which consumers come to expect product configurations to designate source); Robert C. Denicola, Freedom to Copy, 108 YALE L.J. 1661, 1668 (1999) ("There is more than a little circularity in basing a legal right to control unauthorized ornamental use on the assumptions that consumers make about the official sponsorship of the ornamented items-assumptions that rest in turn on consumers' views about whether trademark owners have the legal right to control such use."). See also Barton Beebe, Search and Persuasion in Trademark Law, 103 MICH. L. REV. 2020, 2066-67 (2005) ("The law . . . sets the consumer expectations that are purportedly the gauge of the law's grant.").

*p. 59*
McDonald's. But given how well known the brand was, it is far more likely that consumers recognized that the fast food giant had not extended its business into the production of expensive Scottishthemed hand-sewn teddy bears. And as a normative matter, even if some consumers were confused, they should not have been. 213 Yet the Board in McKinley did not require McDonald's to show that the extreme strength of its family of MC/MAC marks increased the likelihood of confusion. Instead, the Board simply reiterated, without significant analysis, the conventional wisdom that very strong marks are more likely to be confused with similar marks. As in Kenner, the Board's adoption of this doctrine in McKinley was essentially dispositive in the case.

*p. 59*
Our analysis demonstrates why the reasoning and result in McKinley was incorrect and why, more generally, the Kenner doctrine should not apply to superstrong marks. Broader protection for the strongest marks is not supported by concerns that strength causes confusion or is otherwise associated with more (or more costly) confusion. Nor do free-riding concerns support the doctrine-even, or perhaps especially, for assertions of broad mark scope against the makers of competing products. This is not to say that there are no other arguments that might conceivably be advanced in favor of the Kenner doctrine. Our account has not been exhaustive. Confusion and free riding are the two most important, most frequently advanced, and, we believe, most plausible arguments. Of course, any argument in favor of trademark protection or a particular doctrine-as a response to concerns about "overuse," 214 say, or a second-best solution given the costs of trademark enforcement 215 -might be tried out as a defense of the doctrine. However, we do not believe that these plausibly justify stronger protection for strong marks that takes the form of more expansive scope. 216 213 See generally Graeme B. Dinwoodie, Trademarks and Territory: Detaching Trademark Law from the Nation-State, 41 HOUS. L. REV. 885, 889-90 (2004) (considering whether trademark law should "reactively" protect existing consumer perception or "proactively" seek to shape consumer perception). 214 The overuse concern is that widespread use of a signifier degrades its value, particularly in conveying status, partly because consumption is rivalrous and subject to congestion (a sort of common pool resource) and partly because overuse can cause permanent depletion (akin to overfishing that permanently damages the stock). See generally Mark F. Grady, A Positive Economic Theory of the Right of Publicity, 1 UCLA ENT. L. REV. 97 (1994) (discussing rent dissipation). 215 See generally Bone, supra note 137 (analyzing enforcement costs as an explanation for various trademark doctrines). 216 For example, though overuse is plausibly concentrated in strong marks, it only affects a small subset of strong marks, and in any event an expansion of scope that makes a difference only at low levels of confusion is poorly aimed at the goal of limiting overuse.

*p. 60*
Ultimately, the Kenner doctrine has eluded sustained critique and survived as long as it has not so much because of the force of the arguments that might be adduced in its support, many of which are wholly unpersuasive, but because it fits so well with the overall trend in trademark law toward enhanced protection for the strong as against the weak. This trend has quickened in recent decades with the development of antidilution protection for famous marks and the rise of initial interest and post-sale confusion doctrine, which arguably only apply to very strong marks. The rise and expansion of free-riding concerns as a basis for finding infringement have both reflected and furthered this trend. These developments run contrary to the overriding purpose of trademark law to enhance competition. Instead, they have worked to make the rich in the trademark system even richer. Our hope is that abandoning the Kenner doctrine may help to reverse this trend and promote trademark law's overarching procompetitive purpose.

## Footnotes

> 39-40 and accompanying text.3 See infra Section I.A (discussing role of mark strength within multifactor analysis of likelihood of confusion and empirical evidence of strength factor's importance).4 Kenner Parker Toys Inc. v. Rose Art Indus., 963 F.2d 350, 353 (Fed. Cir. 1992) (Rader, J.).5 See infra Section I.B (discussing the alternative model).

> B.V.D. Licensing Corp., 846 F.2d at 729.

> The scope is narrower provided that the range of reduced confusion includes the point where the level of confusion crosses the threshold. If the reduction occurs only at low levels of confusion, scope is not narrower.

> Car-Freshner Corp. v. Scentex, Inc., No. 88 Civ. 1247, 1989 WL 47373, at *5 (N.D.N.Y. May 4, 1989), aff'd, 927 F.2d 594 (2d Cir. 1991).

> Marshall Field & Co. v. Mrs. Fields Cookies, 25 U.S.P.Q. 2d 1321, 1332 (T.T.A.B. 1992).

> WILLIAM M. LANDES & RICHARD A. POSNER, THE ECONOMIC STRUCTURE OF INTELLECTUAL PROPERTY LAW 187-91 (2003).

> Kenner Parker Toys, Inc. v. Rose Art Indus., Inc., 963 F.2d 350, 354 (Fed. Cir. 1992).

> See, e.g., R.J. Reynolds Tobacco Co. v. R. Seelig & Hille, 201 U.S.P.Q. 856 (T.T.A.B. 1978) (cigarettes and teas). For competing products, by contrast, there is no need to lean on the Kenner doctrine to justify expansive product scope.

> Richard A. Posner, When Is Parody Fair Use?, 21 J. LEGAL STUD. 67, 75 (1992) ("A trademark seeks to economize on information costs by providing a compact, memorable, and unambiguous identifier of a product or service. The economy is less when, because the trademark has other associations, a person seeing it must think for a moment before recognizing it as the mark of the product or service.").

> See Rebecca Tushnet, Gone in Sixty Milliseconds: Trademark Law and Cognitive Science, 86 TEX. L. REV. 507, 527-42 (2008) (criticizing the search costs theory of antidilution protection).

> See BURRELL & HANDLER, supra note 62, at 252 (noting that "reputation can cut in two directions" in Australian law such that strength can minimize confusion in instances where "consumers are not going to misremember famous marks" but increase confusion in instances where "consumers might react to something that could be taken to be a brand extension").

> 31 Barton Beebe, An Empirical Study of the Multifactor Tests for Trademark Infringement, 94 CALIF. L. REV. 1581, 1635-36 (2006). But see Timothy Denny Greene & Jeff Wilkerson, Understanding Trademark Strength, 16 STAN. TECH. L. REV. 535, 558-61 (2013) (interpreting other data from Beebe as suggesting that inherent strength plays an important role in courts' strength determination and their overall determination of likelihood of confusion).

> "therefore entitled to the widest scope of protection"); L'Oreal S.A. v. Marcon, 102 U.S.P.Q.2d 1434, 1437 (T.T.A.B. 2012) ("[E]xtreme deference [is] accorded to a famous mark in terms of the wide latitude of legal protection it receives . . . ."). 47 Richard A. Oppel Jr., Giles S. Rich, Oldest Active Federal Judge, Dies at 95, N.Y. TIMES, June 12, 1999, at A13.

> Time Warner Entm't Co. v. Jones, 65 U.S.P.Q.2d 1650, 1660 (T.T.A.B. 2002) (stating that B.V.D. is "not controlling precedent").

> 104 Girl Scouts of U.S. v. Bantam Doubleday Dell Publ'g Grp., Inc., 808 F. Supp. 1112, 1124 (S.D.N.Y. 1992), aff'd, 996 F.2d 1477 (2d Cir. 1993).