Rights and Remedies in Trademark Law: The Curious Distinction Between Trademark Infringement and Unfair Competition

Article - Volume 98 - Symposium Issue


Ever since marks became important selling devices in the late nineteenth century, the law has recognized two distinct theories of protection: trademark infringement and unfair competition. In the latter part of the nineteenth century, when modern trademark law began to take shape, unfair competition stood ready to reach cases that trademark infringement could not. Over the ensuing decades, courts used the unfair competition tort to provide relief in cases where the trademark infringement tort gave out, and they relied on unfair competition principles to justify broader trademark rules. In the 1930s, some commentators argued that trademark and unfair competition were one and the same and should be merged into a single legal theory protecting source-identifying marks. Nevertheless, the distinction persists today.

This duality of liability theories makes no sense. Trademark infringement and unfair competition share the same underlying policies and the same liability principles. The only meaningful difference has to do with remedy. Unfair competition tends to be more flexible than trademark law; it supports limited injunctions that take account of competing policies. However, even this remedial distinction is misguided. It leads courts to give excessively broad protection to source-identifying symbols just because they are classified as “trademarks” and support a cause of action for “trademark infringement.”

In this Article, I argue for eliminating the distinction and applying a unitary approach. Courts should craft injunctions in all trademark cases based on the competing interests and policies at stake. However, remedial flexibility has a price, and this is where the complexity and cost of trademark litigation come into play. The high litigation costs and risks of trademark cases can chill socially desirable uses, such as uses of product features to compete, uses of marks for expressive purposes, and innovative uses of marks on the internet.

This poses a problem for remedial flexibility. While greater flexibility allows courts to adjust remedy to mitigate adverse effects, it also increases the costs of litigating a trademark suit. Moreover, it can increase the likelihood of liability if judges know they can mitigate adverse effects by limiting the scope of injunctive relief. In view of the potential chilling effects created by these additional costs and risks, I propose excluding certain socially desirable uses from liability altogether, regardless of whether the claim sounds in trademark infringement or unfair competition.

The body of this Article is divided into three parts. Part I briefly traces the history of the distinction between trademark infringement and unfair competition. This distinction made sense in the late nineteenth century, but it lost whatever sense it had when legal realism debunked the late nineteenth-century theory of property ownership. Still, the distinction between trademark infringement and unfair competition somehow survived the realist challenge and influenced the expansion of trademark law during the first half of the twentieth century.

Part II explains why the distinction, which persists to this day, makes no sense, and it describes some of the problems that the distinction creates, including doctrinal inconsistency and poorly justified rules. It then focuses on the one significant difference between the two theories: the scope of injunctive relief. It argues that there is no justification for a remedial distinction and that unfair competition’s flexible approach should be available in all trademark cases.

Part III revisits the question of flexible injunctive relief in light of the high costs and significant risks of trademark litigation. These costs and risks do not justify different remedial treatment along the old trademark-infringement/unfair-competition divide, but they do support creating categorical exclusions from liability for certain types of socially valuable uses.

I. A Brief History of Trademark Infringement and Unfair Competition

This Article focuses on the legal protection given to source-identifying symbols used by others in ways that risk consumer confusion. It does not address other competitive activities that might be harmful. For example, it does not deal with false advertising, in which a firm makes misleading representations of fact about its products without using a trade symbol.[1] Nor does it address reverse passing off, in which a firm sells someone else’s product as its own.[2]

A source-identifier is a word, phrase, logo, product feature, or other symbol or combination of symbols that consumers use to identify a single source of products or associate with a unique individual or firm.[3] For example, the word CREST is a source-identifier for toothpaste. When a consumer sees CREST on a tube of toothpaste, she believes that the tube she observes comes from the same source as every other toothpaste tube bearing the same word. She doesn’t need to know that the source is Procter & Gamble to be confident that all tubes marked CREST have uniform quality and carry the seller’s reputation.

If consumers can be confident that all toothpaste with the word CREST has the same quality, they can use the mark to access information that they acquire through advertising, word of mouth, and experience when they make their purchasing decisions. Thus, source-identifying symbols reduce consumer search costs.[4] Moreover, sellers can use marks to communicate the high quality of hidden features of their products not apparent to the consumer at the time of purchase, and this gives the seller an incentive to maintain high quality as well as an incentive to advertise.[5]

For a symbol to confer these benefits, however, consumers must have confidence that it identifies a unique source or uniquely refers to a particular individual or firm. If other sellers could use CREST to sell toothpaste with a different quality, for example, consumers would not be able to count on the mark for reliable information about the toothpaste they wish to buy. Consumer search costs would rise, and seller incentives to maintain product quality would weaken. Moreover, if others could use a symbol after it acquires meaning as a unique referent, consumers could be misled, producing allocative inefficiency and possibly serious consumer harm, as well as reputational injury to the seller or other referent of the symbol. [6]

In the United States, the law of trademark and the law of unfair competition protect the uniqueness of source-identifiers by preventing others from using the same symbol in ways that are likely to cause confusion as to source or sponsorship.[7] These two bodies of law began to develop along modern lines after the Civil War, when local markets became national, railway transportation exploded, and population and per capita income increased sharply.[8] The following gives a brief historical account with specific attention to the relationship between trademark infringement and unfair competition. I have outlined some of this history elsewhere,[9] so I will only briefly summarize it here.

A. 1870–1920

The laws of trademark and unfair competition developed as common law in the late nineteenth and early twentieth centuries. The common law treated trademarks as property, and the existing rules were strongly influenced by the natural law conception of common law property rights that prevailed at the time.[10]

The common law distinguished between technical trademarks protected by the tort of trademark infringement and other source-identifying symbols protected by what became known as the tort of unfair competition. Technical trademarks in the late nineteenth century included what we today call “fanciful” and “arbitrary” word marks.[11] A fanciful word mark, such as KODAK, is a made-up word fabricated just for use as a source-identifying mark. An arbitrary word mark is an existing word in the vocabulary, such as IVORY for soap, that says nothing at all about the product with which it is associated.[12]

The key to technical trademark status in the late nineteenth century was exclusivity.[13] To qualify as a technical trademark, a symbol had to be capable of exclusive possession, and the owner had to have an exclusive right—that is, a right to exclude all others from using the symbol.[14] A fanciful term could qualify because it was capable of exclusive possession, and it was capable of exclusive possession because it was not part of the ordinary vocabulary that everyone was free to use. So too for an arbitrary word mark, like IVORY for soap. Everyone was free to use the word’s ordinary meaning, but since the trademark owner did not make use of that meaning, it could obtain exclusivity in the mark.

Descriptive, geographic, and personal name marks were treated differently. They were known as tradenames to distinguish them from technical trademarks, and they could be protected only with proof of secondary meaning—that is, with proof that a substantial portion of the consuming public in fact used the mark as a source-identifier.[15] For example, a descriptive word, such as TASTY for donuts, could not be a technical trademark because it was not capable of exclusive possession. Everyone was free to use the word tasty in its descriptive sense, so the trademark owner could not obtain exclusive rights in the word’s descriptive meaning. The same was true for geographic designations, such as BOSTON for lager, as well as personal names. Everyone was free to use these terms in their primary geographic and nominative senses. Courts applied this same reasoning beyond tradenames to product features, what we today call trade dress. A product feature could not qualify for protection as a technical trademark under the tort of trademark infringement because product features were free for everyone to copy and use unless they were protected by copyright, patent, or some other statutory right.[16]

It is important to understand that the exclusivity late nineteenth-century courts and commentators had in mind was not an exclusivity conferred by law. It existed prior to any legal rules creating or enforcing it.[17] Exclusivity was a natural law concept with roots in the natural law theory of property that prevailed at the time.[18] The idea was that some symbols, such as descriptive terms, geographic terms, personal names, and trade dress, were “common property”—publici juris—and everyone had an equal right to use them.[19] Because these symbols were not capable of exclusive possession, they could not qualify as technical trademarks. However, they could still be protected if they were given a “secondary meaning”—a meaning to consumers as symbols designating a single source of the goods or services with which they were associated.[20] In that case, the public was free to use the symbol in its primary sense or for its primary purpose—as a descriptor, geographic indicator, name, or product feature—but not in its secondary sense as a source-identifier. A firm that imbued one of these symbols with secondary meaning could rely on the tort of unfair competition to obtain relief against passing off.[21]

A plaintiff bringing an unfair competition claim to protect a symbol that did not qualify as a technical trademark had to prove that the symbol had acquired secondary meaning and that the defendant used the symbol with an intent to deceive or confuse consumers.[22] The first requirement followed from the fact that the symbol could not be exclusively possessed, and the second requirement ensured that the defendant was not using the symbol in its primary sense or for its primary purpose. Eventually, courts relaxed the intent requirement so that by the early twentieth century, a defendant could be held liable for unfair competition based only on consumer confusion without regard to the presence or absence of intent.[23]

By contrast, a plaintiff seeking to protect a technical trademark through an action for trademark infringement could obtain injunctive relief simply by proving that the defendant used the same (or very similar) mark on the same (or very similar) goods in competition with the plaintiff.[24] There was no need to prove secondary meaning for a technical trademark. Moreover, liability attached without any showing of an intent to deceive or even proof of likely confusion.[25] Because the plaintiff had an exclusive right to the symbol, no one could appropriate it for the same purpose without the plaintiff’s consent.[26]

Trademark infringement and unfair competition also differed at the remedy stage. In trademark infringement cases, courts were inclined to grant broad injunctions prohibiting any use of the technical trademark on competing goods, in keeping with the idea of an exclusive right.[27] In unfair competition cases, courts were more inclined to grant qualified injunctions that ordered precautions against confusion, such as adding terms to the mark or including a prominent disclaimer.[28] The idea in unfair competition cases was to craft the injunction to prohibit use of the symbol’s secondary meaning but not its primary meaning. Thus, a defendant was allowed to use a descriptive or geographic term, a personal name, or trade dress in the way that all were free to do, but it had to add qualifiers or disclaimers to avoid using the symbol’s source-identifying (secondary) meaning.[29]

Over time, unfair competition developed into an open-ended legal category capable of reaching a variety of practices that judges deemed sufficiently unfair to warrant redress.[30] At various points in its development, the tort encompassed passing off, tradename and trade dress infringement, false or deceptive advertising, trade libel, interference with contract or business relationships, misappropriation of trade secrets, and other conduct.[31] In 1909, one commentator described unfair competition as “a recognition by the courts of the duty to be honest and fair in all relations of business life,”[32] and others linked it to norms of “fair play” and “honesty and fair dealing” in the marketplace.[33]

In effect, the tort of unfair competition empowered courts of equity to police business practices for fairness on a case-by-case basis without much guidance as to what was unfair. Judges used the tort to provide remedies for harm from marketplace conduct that did not fit an established liability theory.[34] Indeed, it was quite common during this period for commentators openly to profess confusion about exactly what unfair competition entailed.[35] Some celebrated this vagueness and open-endedness. They pointed to, among other things, the need for judicial intervention to deal with clever business strategies aimed at circumventing legal restrictions.[36] Others worried about the lack of guidance and constraint. They sought an organizing principle that might cabin judicial intervention, and many found that principle in the prohibition against “passing off” one’s goods or services as those of another.[37] Yet unfair competition was not easily contained in this way, and courts applied it more broadly.[38]

B. 1920–1960

With the rise of sociological jurisprudence and then legal realism in the 1920s and 1930s, the sharp division between trademark infringement and unfair competition came under attack.[39] The realists argued that legal exclusivity was created by positive law and was, as a result, just a consequence of whatever legal rights the law chose to confer.[40] And they attacked the idea that the law should protect something simply because someone invested in making it valuable. It made no sense in their view to argue that the law should protect a thing because it has value when its value depended on the law protecting it.[41] In short, legal rights were not derived from natural rights or based simply on the existence of economic value; society created legal rights to serve policy goals.

According to this functional view, there was no fundamental difference between trademark infringement and unfair competition. Both protected source-identifying trade symbols against uses that involved passing off or consumer confusion. The kind of symbol didn’t matter as long as it served as a source-identifier. Moreover, because the two torts, as applied to source-identifying symbols, shared the same policies, it made sense that they should share similar rules.

In two celebrated articles published in 1930, Milton Handler and Charles Pickett argued that courts, while professing to apply different rules, were in fact using similar rules in the two areas.[42] Handler and Pickett cited many examples of doctrinal convergence and argued that any differences in legal treatment were better explained by the policies at stake—the social costs and benefits of recognizing legal rights—than by technical classifications.[43] Others agreed. In 1940, for example, Zechariah Chafee published an important article on unfair competition in which he identified policies that courts should consider when deciding whether to intervene in the market and grant relief.[44]

Despite this criticism, judges continued to distinguish between trademark infringement and unfair competition.[45] Moreover, they continued to invoke the natural law idea that a firm has property rights in its goodwill and to rely on intuitions about the unfairness of free riding. This is particularly apparent in the way courts approached the most pressing trademark issue of the day: whether and how far to extend trademark protection to noncompeting uses. Plaintiffs often relied on unfair competition and unfair-competition principles when seeking to extend protection more broadly.[46] Perhaps because of this, judges felt free to rely on moral intuition, liberally mixing concerns about consumer confusion and injury to seller reputation with more general intuitions about the unfairness of free riding and the undesirability of tolerating trademark dilution.[47]

As a result, judges gradually expanded protection into more and more remote product markets. In some cases, defendants were found liable and enjoined from using marks where the products were so distantly related that there could be little risk of consumer confusion or passing off. In these cases, judges put great weight on the fact that the defendant was free riding on the plaintiff’s reputation and appropriating the luster of the plaintiff’s mark.[48] Some judges worried about the potential anticompetitive effects of these broad expansions,[49] but others were willing to find unfair competition based on little more than an aversion to the defendant’s free riding on the plaintiff’s goodwill.[50]

II. The Trademark-Infringement/Unfair-Competition Divide Today

Today, courts and commentators continue to mark a distinction between trademark infringement and unfair competition. Trademark infringement is no longer limited to technical trademarks; it includes what used to be called tradenames—descriptive, geographic, and personal name marks—and even trade dress. But there are situations where trademark infringement gives out and unfair competition comes in to fill the gap. The way courts analyze these situations reveals a great deal about how trademark infringement and unfair competition are conceived today. What emerges is a view that resembles the technical-trademark/unfair-competition divide in the nineteenth and early twentieth centuries.

The following discussion is divided into four subparts. Subpart A briefly describes the modern distinction and the assumptions that underlie it. Subpart B explains why the distinction makes no sense conceptually, doctrinally, or normatively. Subpart C examines some of the costs of maintaining the dual system. And subpart D describes the benefits of adopting a unitary system, including a flexible remedial approach to accommodating competing interests and policies in all trademark cases.

A. The Modern Distinction and Its Supporting Assumptions

It is commonplace for judges, lawyers, and scholars today to distinguish between trademark infringement and unfair competition. Unfair competition is sometimes used to refer to the broad genus of all marketplace wrongs, of which trademark infringement is one species.[51] It is also sometimes used to refer to wrongful conduct that does not involve a source-identifying symbol, such as false advertising or passing off without a mark, although this usage is less common now that many of those torts have their own causes of action.[52]

However, there is a third type of usage that is quite puzzling. In these cases, the two terms—trademark infringement and unfair competition—are both applied to the same source-identifying symbol and confusion-generating use. Typically, these are cases in which one party seeks to protect what it claims to be a source-identifying symbol against another party’s use that allegedly risks consumer confusion. For some reason, the symbol cannot be protected by one of the two theories, usually trademark infringement, so the court turns to the other theory to fill the gap. The problem is that the two theories are doing precisely the same work: protecting against use of a source-identifying symbol that is likely to cause consumer confusion. Why do we need two theories to do the work of one?

Subpart B below discusses two examples of this pattern: first, the use of unfair competition to protect against confusion caused by a mark that cannot be protected with trademark infringement because it is generic, and second, the use of unfair competition to protect a source-identifying mark that cannot be protected with trademark infringement because of the territoriality principle. In both situations, as we shall see, the switch to unfair competition clouds the issues and distorts the analysis.

It is not just courts that make this switch; scholars do as well. For example, a scholar who favors limiting the scope of injunctive relief in some set of trademark cases is likely to recommend a double-barreled reform: exclude those cases from liability for trademark infringement, but subject them to liability for unfair competition when confusion is a risk.[53] The switch from trademark infringement to unfair competition makes sense under current law: unfair competition is the home of remedial flexibility and limited injunctions. What is puzzling, however, is the fact that current law makes the switch necessary. If the optimal approach is some form of limited injunction that balances competing interests, then there should be no need to switch to unfair competition. Limited injunctions should be available across the board, without regard to the liability theory the plaintiff invokes.

What is it about the distinction between trademark infringement and unfair competition that makes the dual system hang on so tenaciously? The answer is not completely clear. However, it is possible to piece together an answer from what courts and commentators say about the two theories, and that answer has to do with the idea of exclusivity.

Even today, trademark infringement is assumed to be about protecting a party’s exclusive rights in a mark.[54] Unfair competition, by contrast, is about preventing confusion-related harm. To be sure, the exclusivity that modern jurists have in mind is not a prelegal, natural law construct, as it was in the nineteenth century. It is a creature of positive law. Nevertheless, it drives the difference between the two theories. In trademark infringement, the mark is conceived as a thing, a discretely defined symbol, protected by an exclusive legal right—although an exclusive right conditioned on likely confusion.[55]

Put simply, trademark infringement focuses on the plaintiff’s rights in a source-identifying symbol; unfair competition focuses on the marketplace effects of defendant’s conduct. Trademark infringement aims to assure that a mark owner’s exclusive rights are protected; unfair competition aims to assure that consumers are not deceived or confused and that sellers are not placed at risk of harm from that deception or confusion. This difference in focus becomes apparent when the two theories are compared. As one court put it when referring to trademark infringement: “In trademark law . . . it is only the exclusive symbol characterized as a ‘trademark’ that is juxtaposed against another’s usage to determine whether or not the two uses by two sellers are likely to confuse customers.”[56] By contrast, another court noted when discussing unfair competition: “Every facet of the parties’ selling program is relevant—from the symbols, letters, pictures, colors, shapes, and sizes connected with the products to the advertising representations made.”[57] It follows that a defendant might be liable for unfair competition “without having technically infringed [a mark].”[58]

These assumptions are strongly reminiscent of the property conception of marks that supported the late nineteenth- and early twentieth-century distinction between technical trademark infringement and unfair competition.[59] The focus on the mark as a thing capable of exclusivity and the assumption that trademark infringement is about protecting exclusive rights in the mark call to mind the exclusivity and property rights concepts that informed the earlier model. And this is the problem. Whatever sense those ideas had in the nineteenth century, they make no sense today. The following subpart explains why.

B. The Distinction Critiqued

Trademark infringement and unfair competition are not significantly different doctrinally; any differences are mostly superficial.[60] For example, a plaintiff must prove use in trade to have rights in a mark enforceable through a trademark infringement suit.[61] But use is also necessary as a practical matter for unfair competition. A plaintiff cannot allege unfair competition without creating a public association with the symbol, and a public association requires public use.[62] It is true that in a trademark infringement suit, there is no need to prove secondary meaning if the mark is fanciful, arbitrary, or suggestive or if the trade dress is inherently distinctive product packaging. But the reason for excusing proof of secondary meaning is that consumers are very likely to adopt these symbols as source-identifiers spontaneously without any help from the trademark owner.[63] Indeed, it is difficult to see how there could be a likelihood of consumer confusion if a symbol had no secondary meaning.

It is possible to deceive or confuse consumers about source or sponsorship without using a source-identifying symbol. A competing firm might, for example, state that its product is the “original” when another firm is actually the original seller.[64] Used in this way, the word original is not a source-identifier; it directly describes a feature of the defendant’s product that implicitly associates it with the plaintiff. Misleading statements of this sort are a form of false advertising, and false advertising does not necessarily use source-identifying symbols. Traditionally, unfair competition was thought to include false advertising, though in an extremely limited way, but false advertising is now an independent claim under the Lanham Act and no longer needs to fit within the unfair competition rubric.[65] In any case, I am concerned here with the use of source-identifying symbols.[66]

The one notable area of difference has to do with remedy and, in particular, with the scope of injunctive relief.[67] Courts continue to favor broad prohibitory injunctions in cases of trademark infringement, just as they did in the past.[68] They might limit injunctive relief to geographic territories where the mark has goodwill or to particular product markets that are sufficiently proximate to the plaintiff’s market. But within those bounds, courts tend to enjoin use of the mark absolutely rather than allow use with qualifiers, disclaimers, or other modifications. There are exceptions, of course, but many of the exceptions involve personal names, geographic marks, and the like, which used to be handled by unfair competition and which courts are accustomed to protecting with flexible remedies.[69]

By contrast, when courts apply unfair competition today, they tend to be more open to flexibility, allowing use of the mark as long as the defendant attaches qualifiers or disclaimers, just as courts did in the past. I do not mean to suggest that courts always grant limited injunctions in unfair competition cases—that depends on the competing interests and policies at stake—but they seem more willing to do so.

It is not surprising that the two torts share so much in common. Their focus is the same: the effect of the defendant’s activities on consumers and sellers. It does not matter whether confusion is created by use of a single symbol, a composite of symbols, or some other mix of things that consumers associate with the plaintiff. Nor does it matter that the case involves a discrete symbol selected and developed by the plaintiff to serve as a source-identifier. The test is the same in all these cases: the likelihood of consumer confusion.[70] And this likelihood is what defines the rights the parties possess.

It might be tempting to treat registered marks differently. The registration focuses attention on the mark itself as something definite and well-defined, a thing capable of being the object of exclusive rights. But this is a mistake. Registration does not matter in any fundamental way.[71] It is not required for trademark protection, and it cannot turn a mark that is not valid into one that is.[72] What registration does is confer certain benefits on the registrant.[73] Liability for use of a registered mark still turns on the effects of the use, and the same likelihood-of-confusion test applies.

In addition, trademark infringement and unfair competition applied to source-identifying symbols share the same underlying policies. The reason to prevent consumer deception or confusion about source, sponsorship, or endorsement is to reduce search costs, support incentives to maintain the quality of goods and services, and avoid unfair harms to sellers and other individuals identified by the symbols being used.[74] To be sure, in trademark infringement, there is a discretely defined symbol that consumers rely on to identify a single source and in which the trademark owner has built up goodwill. But the same is true for unfair competition. It too presumes the existence of symbols that consumers rely on to identify the plaintiff, even though the symbols might not be as clearly defined. Otherwise, there could be no confusion stemming from the defendant’s use. Moreover, even if the plaintiff does not invest directly in developing goodwill, the symbols still have value for the plaintiff—or the plaintiff would not be harmed. Some unfair competition cases involve intentional deception, which might trigger moral concerns, but so too do some trademark infringement cases. And intentional deception is not required for unfair competition, just as it is not required for trademark infringement.[75]

Even the remedial distinction cannot be justified. There is no reason to limit relief to broad prohibitory injunctions just because a trademark is involved or the case sounds in trademark infringement. Perhaps the idea of exclusivity in trademarks has led courts to assume that injunctions should eliminate infringing uses completely. But exclusivity is a conclusory label without meaningful content. The exclusivity that trademark law confers is merely the product of the remedies it provides. And limited injunctive relief should be available whenever the competing interests and policies support it, whether the case sounds in trademark infringement or unfair competition.

One might object at this point that there are cases in which trademark law focuses on the mark—usually a popular brand—as a thing of value in itself, a symbol of a lifestyle or a way to signal status. In these cases, courts are less concerned with market effects or consumer confusion and more concerned with protecting a trademark owner against appropriation by others of the value of its mark.[76] Indeed, these cases approach trademark infringement in a way that resembles the property theory underlying the late nineteenth-century trademark infringement tort.

However, one should not overstate the importance of these cases. Property-type protection of powerful brands is a relatively new phenomenon and quite controversial.[77] Most trademark cases focus on preventing confusion, and consumer confusion has been at the core of trademark law since the early twentieth century. Moreover, a property theory of trademark law is difficult to justify, hard to fit within core trademark theory, and difficult to square with established trademark precedent.[78] In any event, if there is a place in trademark law for property-type protection of brands without consumer confusion, it might be one area where trademark infringement diverges from unfair competition. But the vast bulk of trademark law does not.

C. Costs of a Dual System

Thus, the existing bifurcation of trade-symbol protection is indefensible on normative grounds and mostly illusory on doctrinal grounds.[79] However, there might be no problem if trademark law did what it is supposed to do and did it just as well with a split between two legal theories as with everything packed into one. In fact, however, the current bifurcation of the field has had pernicious effects. The availability of unfair competition removes the pressure to think hard about trademark law. Courts are able to accept trademark infringement’s limitations without adequately considering the policy implications, knowing that they can make creative use of unfair competition to address any problems. Moreover, judges invoke the crude moral intuitions associated with unfair competition to justify trademark expansions rather than engage in a rigorous analysis of social costs and benefits that those expansions require. The result is doctrinal confusion, poorly justified rules, and a crabbed approach to injunctive relief.

We have seen one example of this already. During the first half of the twentieth century, courts expanded trademark protection to distantly related product markets on the strength of thinly justified intuitions about unfair competition. The rest of this discussion presents two more examples: the use of unfair competition to grant relief in cases of “creeping genericity” and its use to circumvent the territoriality principle.

1. Creeping Genericity.

Creeping genericity refers to a source-identifying mark gradually becoming generic over time. A generic term is one that consumers use to identify the general type of product. Thermos, for example, is a generic term for a vacuum-insulated bottle that keeps liquids warm and cold.[80] What makes this term generic is the fact that consumers use it to refer to vacuum-insulated bottles no matter who sells them. It is a well-established rule of trademark law that generic terms cannot be marks and do not receive protection.[81] There are good reasons for this rule. A generic term, by definition, does not identify a single source and therefore cannot further the policy goals of trademark law. Moreover, preventing competitors from using a generic term can impede competition and impoverish the public’s working vocabulary.

Sometimes, however, a word or phrase starts out as source-identifying but becomes generic over time. This often happens when a firm has a patent on a product and sells the product under a mark that consumers gradually appropriate to refer to the product type. When the patent monopoly expires and competitors enter the market, they use the mark to communicate that their products are the same type. This is what happened with the THERMOS mark. THERMOS was once a source-identifier for a vacuum-insulated bottle sold by King-Seeley, but it gradually lost its source-identifying meaning as more and more consumers appropriated the word to refer to any vacuum-insulated bottle regardless of who sold it.[82] When King-Seeley’s patent expired, Aladdin entered the market, selling its vacuum-insulated bottle as a Thermos. The same thing happened to CELLOPHANE, ASPIRIN, MURPHY BED, and many other marks.[83]

Problems arise during the period when a term’s meaning is changing, when some consumers still use the mark as a source-identifier while others use it generically.[84] When a majority of consumers use the mark primarily in a generic sense, the mark is classified as legally generic, and the owner cannot sue others for trademark infringement.[85] However, classifying the mark as generic based on its meaning to a majority of consumers can leave the minority at risk of harm from source confusion.

Courts deal with this problem by switching to the tort of unfair competition (or passing off). While the original firm is unable to sue for trademark infringement, it can still sue for unfair competition and obtain limited injunctive relief. The typical injunction in these cases allows a competitor to use the term as part of its own trademark provided it takes reasonable steps to reduce confusion, such as by adding a conspicuous disclaimer, changing the mark’s font or format, prefacing it with the firm name, or combining it with a house mark.[86]

This solution makes sense. What does not make sense is implementing it by switching to a different legal claim. Doing so produces an odd result: a mark is protected (by unfair competition) even when it is not protected (by trademark infringement). The reason to provide relief in these cases is the same as the reason to provide relief in an ordinary trademark infringement case: to prevent consumers from being confused by the defendant’s use of the same mark.[87] Unfair competition therefore serves the same purposes as trademark law—it protects consumers from source or sponsorship confusion.

There is a more sensible and more productive way to understand this doctrinal pattern. All source-identifying symbols receive protection in trademark law, even those that have generic meaning. It is just that the scope of protection varies with the source-identifying strength of the mark and the social costs of restricting others from using it. The switch to unfair competition makes it seem that remedial flexibility and case-specific balancing are appropriate only for unfair competition claims and that trademark infringement is about providing exclusive rights through broad injunctive relief. But that is misleading, and worse yet, it impedes a more careful analysis of the competing policies at stake.

It is possible that the current approach—full protection before a majority of consumers gives the mark generic meaning and only partial protection after that—is optimal as a policy matter. But it is also possible that the scope of injunctive relief ought to vary in a more continuous way with the mix of secondary and generic meaning over time. It is also possible that a majority cutoff point is not optimal for all cases and that the cutoff itself ought to vary with the facts.[88] However, the bifurcated system of trademark protection with its doctrinal duality makes it hard for courts to engage these alternatives.

2. Territoriality.

The second example involves the territoriality principle. According to this principle, a foreign trademark owner cannot assert priority in the United States based exclusively on foreign use and cannot enforce its mark against another firm that adopts and uses it in the United States first. This principle creates problems when a foreign mark is well known to a significant segment of consumers in the United States despite never having been used to sell products there. If a U.S. firm adopts and uses the same mark to sell the same product, consumers who know the foreign mark might be confused into believing that the U.S. firm is connected with the foreign owner. Nevertheless, the territoriality principle, strictly applied, bars the foreign owner from obtaining relief, a result that can lead to confusion for a significant number of U.S. consumers.

Some courts try to solve this problem by switching from trademark infringement to unfair competition. They reason that while the foreign owner cannot protect its mark with a claim for trademark infringement, it can sue for unfair competition. In Belmora LLC v. Bayer Consumer Care AG,[89] for example, the Fourth Circuit relied on this distinction for this purpose. It interpreted the Lanham Act to allow a foreign mark owner to sue a U.S. company that used the same mark to sell the same product despite the fact that the foreign mark owner had never used or registered the mark in the United States.[90] While the opinion’s legal analysis involved a close textualist interpretation of the relevant statutory provision—§ 43(a) of the Lanham Act—and standing law, the court also relied to a considerable extent on the fact that the suit was for unfair competition rather than trademark infringement and thus that the plaintiff did not need to have a protectable mark.[91]

Once again, the classification shell game gets in the way of a straightforward analysis. The policy question in cases involving the territoriality principle is whether territoriality makes sense in an increasingly global world.[92] The legal question is whether U.S. trademark law should adopt the famous marks doctrine that recognizes an exception when a foreign mark has substantial secondary meaning domestically.[93] The answers to these questions do not depend on whether the claim is denominated trademark infringement or unfair competition. What matters is that some group of U.S. consumers uses the symbol as a source-identifier for the foreign owner.

D. Benefits of Eliminating the Distinction and Universalizing Remedial Flexibility

Given that the distinction between trademark infringement and unfair competition cannot be justified normatively, makes no sense legally, and produces bad effects, it should be abolished. We should retire unfair competition as a separate body of law available when trademark infringement gives out. The field of trademark law is a unitary one, and its rules should be designed to serve its underlying policies, not to fit antiquated ideas about trademarks as property or trademark law as protecting exclusive rights. Making this change does not guarantee good results. But without unfair competition as an easy way out, there is a good chance that judges will be more inclined to consider the general policies informing trademark protection and to contain the broad discretion that unfair competition licensed historically. Without recourse to the separate tort of unfair competition, for example, judges and lawyers will have to consult trademark policies to come to terms with the idea that generic marks receive protection. Also, they will not be able to dodge easily the policy questions raised by territoriality and other issues. And while concerns about free riding are not unique to unfair competition, it might be harder to justify relying on antifree-riding norms without the unfair competition imprimatur.

One of the most important benefits of abolishing the distinction is removing impediments to extending remedial flexibility to all trademark cases. This is particularly significant because injunctive relief is the most common and most important trademark remedy.[94] Other forms of relief are extremely limited, both as a legal and as a practical matter.[95] A plaintiff cannot recover compensation for its own losses unless it can prove actual (not just likely) confusion and offers some way to measure the loss, both of which can be difficult as an evidentiary matter. Moreover, plaintiffs can have trouble recovering defendant’s profits without proving willfulness or some other mental state.[96] Thus, most of the policy work trademark law performs must be done through injunctive relief.

The Supreme Court’s eBay Inc. v. MercExchange, LLC[97] decision restored equitable discretion to patent injunctions, and many courts, most notably those in the Ninth Circuit, have applied the holding to trademark injunctions as well.[98] Although eBay focused on the decision whether to grant an injunction and not what kind of injunction to grant, the Supreme Court’s reminder that injunctions are subject to equitable balancing might push judges toward greater flexibility when they can no longer relegate the balancing to the special domain of unfair competition law.

Greater flexibility in designing trademark injunctions can be a useful way to balance the benefits and costs of trademark protection in individual cases.[99] One possible application is in the area of brand protection. Suppose the plaintiff has a particularly strong brand that gives it considerable power in the product market and creates a barrier to entry. The defendant, a new entrant, uses a mark similar to the brand in order to borrow some of the brand’s consumer appeal and surmount the entry barrier. Today, a judge is likely to find liability and order the defendant to cease using the mark, especially if she decides that the defendant intended to benefit from the brand’s popularity. However, intermediate solutions are available, and they might be superior in some cases. It might make sense, for example, to grant a limited injunction allowing the defendant to use the mark but requiring it to add a qualifier, position its logo near the mark, or change the font or format in other ways. This should reduce the likelihood of confusion and minimize the risk of dilution by distinguishing the two marks, while still permitting the defendant to gain a toehold in the market and compete effectively with a powerful incumbent.

Moreover, a limited injunction also makes sense when the evidence supports a risk of consumer confusion, but the mark is relatively weak and the magnitude of the confusion risk uncertain.[100] And perhaps the strongest case for a flexible approach is where the defendant’s use has special social value. For example, a firm might use a mark with descriptive meaning to describe its own products but in a way that risks confusing a substantial segment of potential consumers into believing that there is a connection to the mark’s owner.[101] Or someone might use the mark to refer to the mark’s owner or to create a parody or other form of expression that, in addition to providing expressive value, also happens to confuse the consuming public about source or sponsorship.[102] Or a firm might use a mark in connection with the internet to facilitate consumer shopping among different brands or copy the plaintiff’s trade dress in order to compete by selling a similar product.[103] In all these examples, the defendant’s use has substantial social value even if it also creates harmful confusion.[104] If trademark law prohibits use of the mark completely, confusion is avoided but at the cost of scuttling the defendant’s socially valuable use.

Currently, trademark law balances benefits and costs in cases like these mainly through special liability doctrines, such as additional requirements for obtaining rights or tailored affirmative defenses.[105] This solution to the problem has the disadvantage of being binary: either the defendant is liable and enjoined or it is not liable and free to conduct the use unimpeded. In theory at least, a remedy-focused approach can be more fine-tuned. A court can adjust the balance by adjusting the injunction’s scope. A narrower injunction eliminates some harmful confusion and the confusion that remains is tolerated to make room for the socially valuable use.[106]

Of course, there are limits to what judges can do to balance costs and benefits at the remedy stage. While parties have some incentives to supply information about an injunction’s future impact, that information can be difficult to acquire and process. Moreover, predicting how those affected will respond to a decree is a highly uncertain business. And the best a court can do, in any event, is make rough adjustments to scope. Still, a remedy-focused approach might be superior to current liability doctrines for some types of cases and a useful supplement for others.

This does not mean that courts should consider limited injunctions in all trademark cases. In particular, it makes sense to adopt a rule requiring fully prohibitory injunctions and rejecting limited relief in cases where the defendant uses a fanciful or arbitrary word mark or inherently distinctive product packaging to sell identical or nearly identical products in direct competition with the plaintiff. In these cases, the probability of harmful confusion is likely to be substantial and the benefits from allowing the defendant’s use minimal, given the availability of alternative symbols to use as marks. Thus, fact-specific balancing is almost certain to support a broad injunction in most of these cases. To be sure, a few cases might warrant limited relief, but the administrative benefits of a general rule are likely to exceed the costs of granting a broad injunction if the exceptional cases are relatively rare.[107]

III. Accommodating Remedial Flexibility to Costly Trademark Litigation

So far, the argument for remedial flexibility has ignored trademark litigation and the indirect effect of litigation cost and risk on willingness to use marks in permissible ways. Case-specific remedial balancing is likely to increase litigation costs and possibly litigation risk. And higher costs and risks can produce serious chilling effects.

This is a particular concern because trademark litigation is complex. It is not complex in the same way class actions and multidistrict litigation are complex. A trademark suit does not have a particularly complex party structure or demand particularly complex judicial management or innovative procedural techniques. What makes trademark litigation complex are its factual and evidentiary demands. For example, survey evidence and expert testimony feature prominently in many trademark cases. Surveys are critical for a number of trademark issues, including genericity, secondary meaning, and likelihood of confusion. Moreover, trademark suits often involve substantial discovery.

This complexity translates into high litigation costs. It compounds the costs of fact-finding and invites adversarial contestation and strategic maneuvering. A 2017 survey of practicing trademark lawyers, conducted by the American Intellectual Property Law Association, reported median total litigation costs in trademark cases ranging from a low of $300,000 for suits with less than $1 million at risk to a high of $1 million for suits with more than $25 million at risk.[108]

Moreover, the open-ended nature of factual determinations in trademark cases generates uncertainty about outcome. One of the main sources of this uncertainty is the vague multifactor test for likelihood of confusion.[109] Courts balance many different factors, including the strength of the plaintiff’s mark, the similarity of the marks in market context from an ordinary consumer’s point of view, the proximity of the products, the defendant’s motivation, and so on.[110] Different circuits use different sets of factors, and judges within the same circuit balance the factors differently.[111] All of this increases the risks of trademark litigation.

Litigation cost and risk can produce chilling effects. A party interested in making what society considers a socially valuable use of the plaintiff’s mark might be discouraged from doing so by the prospect of facing costly and risky litigation. In such a case, it is the litigation cost and risk—not the substantive law—that creates the obstacle. A prospective user might decide not to make a socially valuable use even when she is confident that the use is proper under the substantive law. The costs of litigating to a successful conclusion and the risk that a judge might not see the matter the same way she does can be enough to dissuade her, especially if she is risk averse.[112]

Remedial flexibility exacerbates this problem. Parties will present more evidence and litigate more issues when a judge is tasked with considering the costs and benefits of injunctions of varying scope. Moreover, flexibility can increase litigation uncertainty and risk. Judges might be more inclined to find likelihood of confusion if they know they can mitigate adverse effects by limiting the scope of injunctive relief. If so, a prospective user would have to worry not only about liability but also about what type of injunction a judge might choose to impose. Thus, remedial flexibility has two countervailing effects: one ex post and the other ex ante. Ex post, it moderates the adverse impact of injunctive relief and makes room for socially valuable uses. Ex ante, it strengthens a trademark owner’s litigation threat and thus its power to deter socially valuable uses.[113]

One way to handle the chilling-effect problem is to create categorical exclusions from trademark liability for some socially valuable uses. I have discussed this idea in other writing. There, I propose applying this approach to two types of uses: (1) uses of word marks, logos, trade dress, and other protected symbols in parodies, music, film, and other forms of expression and (2) uses of product-design trade dress to compete in the product market.[114] In both cases, the risk of harmful confusion is low, and the social value of the use high. In an ideal world of costless and perfectly accurate litigation, it might well be optimal to proceed on a case-by-case basis. In that way, we could catch the exceptional case of serious confusion and deal with it by granting a limited injunction that balances the competing policies. However, in the real world of costly and risky litigation with associated chilling effects, it makes sense to tolerate confusion in the exceptional case in order to avoid discouraging socially valuable uses.[115]

To illustrate, consider the expressive use of marks.[116] Trademark law today relies primarily on affirmative defenses to create breathing room for expressive use.[117] Affirmative defenses, however, must be litigated, and judges sometimes interpret a defense to incorporate case-specific factual determinations, all of which contribute to chilling effects.[118] In addition, as we have seen, relying on remedial balancing is likely to exacerbate the problem. A better way to handle expressive use is simply to exclude this type of use from trademark liability altogether. This solution makes sense because the likelihood of harmful confusion from expressive use is very low and the social value of the use is very high.[119]

Confusion is unlikely because the expressive context of the use should signal others that the trademark owner is not involved. Moreover, even when members of the audience are confused, neither they nor the trademark owner are likely to be harmed in any way trademark law cares about.[120] On the other side of the balance, the user’s interest in self-expression rises to a constitutional level.[121] Thus, balancing the expected social costs of confusion against the expected costs of discouraging the use—and taking account of the First Amendment values at stake—a strong case can be made for categorically excluding all expressive use from trademark law. Adding remedial flexibility to the mix only strengthens the case.[122]


The old distinction between trademark infringement and unfair competition continues to distort trademark law well after its original justification has disappeared. Even differences in injunctive relief make no sense. It is past time to jettison the separate tort of unfair competition and embrace flexibility in determining the scope of injunctive relief in all trademark cases. The result of doing so will be a system of trademark law where unfair competition is retired as a distinct theory, where judges are willing to limit injunctive relief as a way to balance the benefits and costs of protecting marks, and where some uses are categorically excluded from trademark liability. This system rids trademark law of the formalisms that still haunt it, addresses confusion risks when they are serious, and respects uses of marks that have substantial social value—and thus strikes a superior balance among the relevant policies.

  1. .See 5 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition §§ 27:24–25 (5th ed. 2019) (outlining the prima facie case for false advertising).
  2. .See 4 id. §§ 25:6, 25:8 (describing express and implied reverse passing off). Moreover, I focus on confusion theories and do not address dilution.
  3. .See, e.g., Qualitex Co. v. Jacobson Prods. Co., 514 U.S. 159, 162–63 (1995) (noting that “almost anything at all that is capable of carrying meaning” can serve as a mark and that “[i]t is the source-distinguishing ability of a mark . . . that permits it to serve” trademark’s purposes). I add “associate with a unique individual or firm” to make clear that the set of source-identifying symbols includes those that the plaintiff does not intentionally adopt or develop as source-identifying but that the public nevertheless associates with the plaintiff. These might include elements of the plaintiff’s persona or aspects of the plaintiff’s business. For example, a firm might sell clothes with a particular design never intending the design to be a source-identifier, yet consumers might come to associate the design uniquely with the firm, thereby making it possible for the firm to sue competitors who use a similar design. See Wal-Mart Stores, Inc. v. Samara Bros., 529 U.S. 205, 207, 216 (2000) (holding that designs on children’s clothing could acquire distinctiveness and become a protectable symbol if the public comes to believe that the designs indicate a single source). Or the public might learn to associate a singer’s voice style uniquely with the singer, enabling the singer to bring a false endorsement claim. See 5 McCarthy, supra note 1, § 28:15 (describing cases of performers alleging false endorsement). These cases all involve source-identifying symbols in just the same way that more conventional trademark cases do.
  4. .See Qualitex, 514 U.S. at 163–64 (“[P]reventing others from copying a source-identifying mark . . . quickly and easily assures a potential customer that this item—the item with this mark—is made by the same producer as other similarly marked items that he or she liked (or disliked) in the past.”); William M. Landes & Richard A. Posner, The Economic Structure of Intellectual Property Law 166–68 (2003) (“The value of a trademark to the firm that uses it . . . is the saving in consumers’ search costs made possible by the information that the trademark conveys . . . .”); 1 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 2:5 (5th ed. March 2020 Update) (“Another important economic function of trademarks is that they reduce the customer’s cost of collecting information about products.”).
  5. .See Qualitex, 514 U.S. at 164 (noting that “the law helps assure a producer that it (and not an imitating competitor) will reap the financial, reputation-related rewards associated with a desirable product” and thus supports incentives to maintain quality products); Landes & Posner, supra note 4, at 168 (“[A] firm with a valuable trademark will be reluctant to lower the quality of its brand because it would suffer a capital loss on its investment in the trademark.”); 1 McCarthy, supra note 4, § 2:4 (“[T]rademarks create an incentive to keep up a good reputation for a predictable quality of goods.”).
  6. .For an overview of trademark policies, see Robert G. Bone, Enforcement Costs and Trademark Puzzles, 90 Va. L. Rev. 2099, 2104–14 (2004) [hereinafter Bone, Enforcement Costs].
  7. .See 4 McCarthy, supra note 4, § 23:1–4 (describing the likelihood-of-confusion test).
  8. .Robert G. Bone, Hunting Goodwill: A History of the Concept of Goodwill in Trademark Law, 86 B.U. L. Rev. 547, 576 (2006) [hereinafter Bone, Hunting Goodwill].
  9. .For these previous historical accounts, see Bone, Hunting Goodwill, supra note 8 and Robert G. Bone, Trademark Functionality Reexamined, 7 J. Legal Analysis 183, 190–210 (2015) [hereinafter Bone, Trademark Functionality].
  10. .See Bone, Hunting Goodwill, supra note 8, at 560–69 (describing trademark and unfair competition law as they took shape in the late nineteenth century). As a leading commentator explained at the time, “[t]he right to a trade-mark is a right of property” and a trademark must be something that others do not have an “equal right to employ for the same purpose.” James L. Hopkins, The Law of Unfair Trade: Including Trade-Marks, Trade Secrets, and Good-Will §§ 7, 8, at 8, 10 (1900) (quoting Newman v. Alvord, 51 N.Y. 189–193 (1872)) (summarizing cases); see also William H. Browne, A Treatise on the Law of Trade-marks and Analogous Subjects § 86, at 100 (2d ed. 1885) (claiming that the property right in a trademark “has its foundation in immutable law”).
  11. .Bone, Hunting Goodwill, supra note 8, at 563–64.
  12. .There are three other categories: suggestive (a word that only suggests attributes of the product), descriptive (a word that actually describes attributes), and generic (a word that designates the product class not a particular brand). 2 McCarthy, supra note 1, § 11:1.
  13. .See James L. Hopkins, The Law of Trademarks, Tradenames, and Unfair Competition § 10, at 17 (2d ed. 1905) (“The trademark right must be exclusive; by this test it stands or falls.”); Amasa C. Paul, The Law of Trade-Marks, Including Trade-Names and Unfair Competition § 22, at 35 (1903) (stating that a trademark “must be of such nature that it can be rightfully appropriated by one person, to the exclusion of all others”); see also Dennison Mfg. Co. v. Thomas Mfg. Co., 94 F. 651, 657 (D. Del. 1899) (describing how trademarks are appropriate only when an exclusive right would be appropriate); Avery & Sons v. Meikle & Co., 81 Ky. 73, 86, 90 (1883) (stating that trademarks derive from exclusive ownership).
  14. .See Bone, Hunting Goodwill, supra note 8, at 563–64 (discussing “technical trademarks” and their historical significance).
  15. .See, e.g., Browne, supra note 10, § 91, at 60–61 (distinguishing between trademark and tradename); Hopkins, supra note 13, § 3, at 9–12 (defining tradename as a word or phrase by which a product is known to the public and which is either descriptive or generic). The term tradename means something different today. See, e.g., 15 U.S.C. § 1127 (2000) (defining a tradename as the name of a firm or business, as opposed to a trademark (used to denote goods) and a service mark (used to denote services)).
  16. .Bone, Trademark Functionality, supra note 9, at 191–95.
  17. .If exclusivity were created by law, one could not argue that a mark should be protected because it was capable of exclusive possession without creating a vicious circularity. In that case, the argument would be that the law should give exclusive rights in the mark because the law gives exclusive rights in the mark.
  18. .Several scholars have written accounts of the impact of natural law theory on nineteenth‑century thinking about property rights, intellectual property rights, and trademark rights. See Morton J. Horwitz, The Transformation of American Law 18701960: The Crisis of Legal Orthodoxy 155 (1992) (noting that “the orthodox idea of property was that it was a pre-political, Lockean natural right not created by law”); Bone, Hunting Goodwill, supra note 8, at 562 (observing that common law rights in property were conceived as natural rights existing independently of positive law); Mark P. McKenna, The Normative Foundations of Trademark Law, 82 Notre Dame L. Rev. 1839, 1873–75 (2007) (noting that traditional trademark law was based on a “natural property rights theory”).
  19. .E.g., Avery & Sons v. Meikle & Co., 81 Ky. 73, 86, 90 (1883) (concluding that descriptive terms “are common property which all may use, but which none may exclusively appropriate”); accord Dennison Mfg. Co. v. Thomas Mfg. Co., 94 F. 651, 657 (D. Del. 1899) (arguing that “no one can acquire an exclusive right to the use, as a trade-mark, of a generic name, or word . . . .”); Hopkins, supra note 10, § 15, at 27–28 (distinguishing unfair competition from trademark infringement and noting that unfair competition applies when someone adopts a “geographical name, a generic term, or words otherwise publici juris,” which are not methods of distinguishing goods “which the law recognizes as a right of property and denominates ‘trade-mark’”).
  20. .See, e.g., E. R. Coffin, Fraud as an Element of Unfair Competition, 16 Harv. L. Rev. 272, 274–78 (1903) (distinguishing technical trademarks “capable of exclusive appropriation” and other marks “to which a trader cannot acquire this exclusive right” but which can be protected once they acquire secondary meaning).
  21. .Milton Handler & Charles Pickett, Trade-Marks and Trade Names—An Analysis and Synthesis: I, 30 Colum. L. Rev. 168, 168 (1930) [hereinafter Handler & Pickett I] (“Trade-marks are protected in a suit for trade-mark infringement; trade names in an action to restrain passing-off or unfair competition.”). Also, it was not uncommon for courts to mix formalistic arguments with expressions of concern about monopoly, noting that broad protection to descriptive, geographic, and personal-name marks might give their owners the ability to restrain competition.
  22. .Milton Handler, Unfair Competition, 21 Iowa L. Rev. 175, 183–84 (1936).
  23. .See, e.g., Milton Handler & Charles Pickett, Trade-Marks and Trade Names—An Analysis and Synthesis: II, 30 Colum. L. Rev. 759, 779–81 (1930) [hereinafter Handler & Pickett II] (observing that for unfair competition, “[m]any courts” were “satisfied by proof of confusion of source”); Edward S. Rogers, Comments on the Modern Law of Unfair Trade, 3 Ill. L. Rev. 551, 554 (1909) (noting that unfair competition should require only that the defendant’s use is likely to confuse consumers, regardless of intent).
  24. .See Handler, supra note 22, at 183–84 (“Commercial usage of the identical trademark in the same business field is taboo, regardless of the element of confusion.”).
  25. .See Paul, supra note 13, § 19, at 31–32 (claiming that it is not necessary to prove fraud or that defendant’s product is of inferior quality because “a trade-mark, when in use, is property itself”); Handler & Pickett I, supra note 21, at 168–69 (listing all the doctrinal differences between trademark infringement and unfair competition, including the rule that unfair competition requires fraud while trademark infringement does not).
  26. .Moreover, to qualify as a technical trademark, the mark had to be physically attached to the goods with which it was associated (known as the affixation requirement). See, e.g., Hopkins, supra note 10, § 19, at 39 (noting that “the mark must be affixed to the subject it serves to identify”); 2 McCarthy, supra note 4, § 16:22 (“The old rule at common law up to about the first half of the 20th Century was that ownership and priority of trademark rights flowed only from the physical ‘affixation’ of the trademark to a label on the goods or their container.”).
  27. .See Harry D. Nims, The Law of Unfair Competition and Trade-Marks, § 367, at 648–49 (2d ed. 1917) (“Where the plaintiff possesses full rights to a trade use of a fanciful word or term the injunction may be absolute.”); Handler & Pickett I, supra note 21, at 169 (“If a trade-mark is substantially copied, its use will be enjoined notwithstanding that it is accompanied by such distinguishing features as render it unlikely that the public will mistake the goods bearing the simulated mark for those stamped with the original.”); id. at 184 (contrasting a court’s limited injunction in an unfair competition case involving a descriptive tradename with the “broad injunction which is normally issued in trade-mark cases”).
  28. .See Nims, supra note 27, § 367, at 648–49 (noting that, in contrast to a case involving “a fanciful word or term,” “[w]here the case involves a name that is publici juris the injunction should not be too general in terms but should show that it is the trade meaning of the word or name and not its general meaning that is involved.”); Handler & Pickett II, supra note 23, at 781 (“In trade-mark infringement, the approach is decidedly different. Where the infringement is . . . practically a duplicate, the courts have sustained the broad right of plaintiff not to have the mark used denominatively in any context.”); Handler, supra note 22, at 184 (“A restraint against the infringement of a trademark prohibits any use of the mark. A decree in unfair competition merely requires the cessation of the deceit—the tradename can still be used if care is taken to avoid confusion.”); see also Restatement (First) of Torts, § 744, cmt. d (discussing qualified injunctions with regard to personal names, geographic names, or descriptive words).
  29. .However, at least one early twentieth-century commentator complained about “the weakness of the unfair competition remedy” ordering disclaimers on the ground that it “may engender more confusion than the bare use of the tradename” and observed that “many, but not all, of the courts have framed their decrees so as to provide more effective relief.” Handler, supra note 22, at 184–85. Also, it is worth noting that courts tended to err on the side of broad injunctions when it was too difficult to enjoin unlawful uses without also enjoining lawful ones, thereby leaving it to the defendant to figure out how to cure the confusion while still using the symbol lawfully.
  30. .See 1 McCarthy, supra note 1, § 1:16 (describing the fluid nature of unfair competition law). Indeed, the late nineteenth-century account of unfair competition was tied to flexibility and expansion. See Bone, Hunting Goodwill, supra note 8, at 566–67 (noting that many at the time viewed unfair competition as a more-or-less ad hoc effort by courts of equity to block increasingly clever passing off strategies by “applying intuitive notions of fairness”).
  31. .See, e.g., Hopkins, supra note 10, §§ 67–69, at 153–55 (including trade secret in the law of unfair trade); Nims, supra note 27, at x–xiv, xvi-xvii (listing in the table of contents tradename infringement, trade dress infringement, trade disparagement, interference with contractual or business relations, trade secret infringement, false advertising, and possibly other wrongs, all in a treatise devoted to unfair competition and trademark); see generally 1 McCarthy, supra note 1, § 1:10 (providing examples of unfair competition).
  32. .Nims, supra note 27, at vi.
  33. .Johnson & Johnson v. Quality Pure Mfg., 484 F. Supp. 975, 979 (D.N.J. 1979) (“When competition is engaged in beyond the boundaries of fair play, there is unfair competition.”); Dutcher v. Harker, 377 S.W.2d 140, 145 (Mo. Ct. App. 1964). In 1995, the Restatement (Third) of Unfair Competition summed up the state of unfair competition in the following way: “It is impossible to state a definitive test for determining which methods of competition will be deemed unfair . . . . Judicial formulations [of unfair competition] have broadly appealed to principles of honesty and fair dealing, rules of fair play and good conscience, and the morality of the marketplace.” Restatement (Third) of Unfair Competition § 1, cmt. g (Am. Law Inst. 1995).
  34. .See, e.g., Nims, supra note 27, § 4, at 12–16 (collecting definitions of unfair competition).
  35. .See William H. S. Stevens, Unfair Competition: A Study of Certain Practices 1 (1917) (“The term ‘unfair competition’ is very difficult to define, and it is scarcely less difficult to explain.”); Oliver R. Mitchell, Unfair Competition, 10 Harv. L. Rev. 275, 275 (1896) (“To most lawyers, it is safe to say, the title [unfair competition] carries no very definite meaning . . . .”).
  36. .See Hopkins, supra note 13, § 1, at 2 (discussing how development of trademark law led to more inventiveness in “stealing another’s trade” without infringing a trademark and how courts responded by extending the law to cover these new methods); Rogers, supra note 23, at 551–52.
  37. .One well known commentator, Milton Handler, remarked at the time that “[s]ome judges have gone so far as to assert that ‘unfair competition consists in selling goods by means which shock judicial sensibilities,’ but this broader meaning of the term has not been commonly accepted.” Handler, supra note 22, at 187 (citation omitted). He went on to clarify: “[w]hile the term [unfair competition] has been loosely used in judicial decisions to connote various practices, its chief meaning has been and continues to be passing off. Other types of competitive injury have been governed by separate rules and vindicated, in the main, in different forms of action.” Id.
  38. .Perhaps, the most famous example is Int’l News Serv. v. Associated Press, 248 U.S. 215 (1918), in which the Supreme Court used unfair competition to grant a misappropriation remedy that enjoined the defendant from copying and reporting news collected by the plaintiff. Unable to find any established basis for the remedy, the Court invoked unfair competition, relying in part on a general fairness norm against free riding. Id. at 239 (noting that the defendant is “endeavoring to reap where it has not sown”).
  39. .For an account of this history and the legal realist attack on nineteenth- and early twentieth-century formalism, see Bone, Hunting Goodwill, supra note 8, at 585–89.
  40. .See, e.g., E. I. du Pont de Nemours Powder Co. v. Masland, 244 U.S. 100, 102 (1917) (Holmes, J.) (“The word property as applied to trade-marks and trade secrets is an unanalyzed expression of certain secondary consequences of the primary fact that the law makes some rudimentary requirements of good faith.”); Premier-Pabst Corp. v. Elm City Brewing Co., 9 F. Supp. 754, 758 (D. Conn. 1935) (pointing out that “to say that a right to a name has value; and therefore it is a property right within the protection of the law . . . is to argue a priori”); Edward S. Rogers, New Directions in the Law of Unfair Competition, 74 N.Y. L. Rev. 317, 319–20 (1940) (arguing that “the way to find out if a right exists is to determine if a wrong has been done,” and therefore, the unfairness of the defendant’s actions is what matters, not the existence of a property right); Floyd A. Wright, The Nature and Basis of Legal Goodwill, 24 Ill. L. Rev. 20, 24 (1930) (maintaining that goodwill is property only “in so far as we may correctly predict that the courts will afford it protection”).
  41. .For a particularly sharp critique of trademark law along these lines, see Felix Cohen, Transcendental Nonsense and the Functional Approach, 35 Colum. L. Rev. 809, 815 (1935). According to Cohen, the contention that a mark should be protected because it has acquired value through advertising is circular: “It purports to base legal protection upon economic value, when, as a matter of actual fact, the economic value of a sales device depends upon the extent to which it will be legally protected.” Id.
  42. .Handler & Pickett I, supra note 21; Handler & Pickett II, supra note 23.
  43. .See Handler & Pickett I, supra note 21, at 189, 191 (noting the lack of differences between the rules as applied); Handler & Pickett II, supra note 23, at 762, 768, 787–88 (same).
  44. .Zechariah Chafee, Jr., Unfair Competition, 53 Harv. L. Rev. 1289, 1315–21 (1940).
  45. .Also, the federal trademark registration statute in effect before the Lanham Act reflected aspects of this distinction.
  46. .At first, courts struggled with the idea that unfair competition could apply when there was no competition. But this problem was quickly resolved. See, e.g., Vogue Co. v. Thompson-Hudson Co., 300 F. 509, 512 (6th Cir. 1924) (“[T]here is no fetish in the word ‘competition.’ The invocation of equity rests more vitally upon the unfairness.”).
  47. .Even when technical trademarks were involved.
  48. .See, e.g., Harold F. Baker, Editorial Note, The Monopoly Concept of Trade-Marks and Trade-Names and “Free Ride” Theory of Unfair Competition, 17 Geo. Wash. L. Rev. 112, 118–20 (1948) (observing that in many of the broadest holdings, “the courts find confusion where it is extremely doubtful or de minimis” and that these holdings really rest on an antifree-riding rationale); Bone, Hunting Goodwill, supra note 8, at 596 (giving examples of decisions that mention a risk of confusion but seem primarily driven by antifree-riding intuitions).
  49. .This was particularly pronounced during the 1940s in the Second Circuit when the court was split on the issue of how far to extend protection for tradenames. See Robert G. Bone, Taking the Confusion Out of “Likelihood of Confusion”: Toward a More Sensible Approach to Trademark Infringement, 106 Nw. U. L. Rev. 1307, 1320–29 (2012) [hereinafter Bone, Taking the Confusion Out] (discussing Second Circuit judges’ competing views on product market expansion).
  50. .See Bone, Hunting Goodwill, supra note 8, at 598–99 (describing judicial willingness to provide plaintiffs with relief largely on the ground of appropriation of plaintiffs’ goodwill). Later expansions along the temporal dimension, such as postsale confusion, were also facilitated by judicial willingness to extend protection under the umbrella of unfair competition and influenced at least in part by a sense that free riding was wrong. See Mastercrafters Clock & Radio Co. v. Vacheron & Constantin-Le Coultre Watches, Inc., 221 F.2d 464, 466–67 (2d Cir. 1955) (“Plaintiff’s intention thus to reap financial benefits from poaching on the reputation of the Atmos clock is of major importance.”).
  51. .1 McCarthy, supra note 1, § 4:6.
  52. .Today, for example, false advertising is codified in § 43(a)(1)(B) of the Lanham Act, which gives much broader protection than the common law of unfair competition ever did. See 15 U.S.C. § 1125(a)(1)(B); 5 McCarthy, supra note 4, §§ 27:1, 27:9. Moreover, trade secret law today is treated as a separate tort with its own cause of action, and civil remedies are even provided by a federal statute, the Defend Trade Secrets Act. 18 U.S.C. § 1836 (2012). See Nat’l Conference of Comm’rs on Unif. State Laws, Unif. Trade Secrets Act with 1985 Amendments (1985); see also Handler, supra note 22, at 187 (noting that as early as the 1930s “other types of competitive injury” not involving passing off have been “vindicated, in the main, in different forms of action” distinct from unfair competition).
  53. .See, e.g., William McGeveran & Mark P. McKenna, Confusion Isn’t Everything, 89 Notre Dame L. Rev. 253, 301–11 (2013) (proposing that some types of use be excluded from liability for trademark infringement with the caveat that confusion can be dealt with through limited unfair-competition injunctions).
  54. .See, e.g., K Mart Corp. v. Cartier, Inc., 485 U.S. 176, 185–86 (1988) (noting that trademark law confers private rights that “are themselves rights of exclusion”); Elvis Presley Enters. v. Capece, 950 F. Supp. 783, 790 (S.D. Tex. 1996) (noting that the goal of trademark law is “to provide the holder of a trademark the exclusive right to use a phrase, word, symbol, image, or device”), rev’d and remanded on other grounds, 131 F.3d 188 (5th Cir. 1998); Jean Patou, Inc. v. Jacquelin Cochran, Inc., 201 F. Supp. 861, 863 (S.D.N.Y. 1962) (noting that the essential element of a trademark is “the exclusive right of its owner to use a word or device to distinguish his product”).
  55. .See Louis Altman & Malla Pollack, Callmann on Unfair Competition, Trademarks, and Monopolies § 17:7 (4th ed. 2019) (stating that a trademark right “assures the trademark owner the right to its exclusive use; and the right to challenge the use of the same or a similar mark by others, or any other disturbance of the owner’s right to the exclusive use thereof”); 1 McCarthy, supra note 1, § 2:10 (noting that a trademark is a kind of property and confers a “right to exclude” which amounts to a right to prevent consumer confusion).
  56. .Parameter Driven Software, Inc. v. Mass. Bay Ins. Co., 25 F.3d 332, 337 (6th Cir. 1994) (quoting 1 McCarthy, supra note 1, § 2.02); see also Indus. Indem. Co. v. Apple Computer, 95 Cal. Rptr. 2d 528, 536 (Cal. Ct. App. 1999) (noting that “in trade mark actions the question to be decided is the narrower and more artificial one of whether the defendant’s mark is confusingly similar to the registered mark, and many factors are not relevant which are relevant in passing-off actions”).
  57. .Elvis Presley Enters., 950 F. Supp. at 790 (quoting 1 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition, § 2.02 (3d ed. 1992)).
  58. .Id. (quoting Professional Golfers Ass’n of America v. Banker’s Life & Casualty Co., 514 F.2d 665, 670 (5th Cir. 1975)).
  59. .See supra notes 10–29 and accompanying text.
  60. .See 1 McCarthy, supra note 1, § 4:6 (calling the distinction between trademark and unfair competition law “archaic”).
  61. .2 McCarthy, supra note 4, § 16:1.
  62. .See 2 McCarthy, supra note 1, § 16:34 (describing the requirement of secondary meaning).
  63. .Id. § 11:4.
  64. .See Murphy Door Bed Co. v. Interior Sleep Sys., Inc., 874 F.2d 95, 102 (2d Cir. 1989) (finding unfair competition based on the defendant having “intentionally represented his product as plaintiff’s wall bed” by, among other things, advertising those products as “Original Wall Bed Systems”).
  65. .See 15 U.S.C. § 1125(a)(1)(B) (2018).
  66. .Trademark law developed separately from false advertising in the late nineteenth and early twentieth centuries—at a time when the false advertising tort was extremely limited—and it is treated independently today. One commentator has argued, however, that the connection between trademark law and false advertising supports importing false advertising’s materiality requirement into trademark law. Rebecca Tushnet, Running the Gamut from A to B: Federal Trademark and False Advertising Law, 159 U. Pa. L. Rev. 1305, 1352–68 (2011).
  67. .It is important to note, however, that this difference is not always easy to discern. A judge is not likely to announce that a broad injunction is warranted because the case sounds in trademark infringement even though a more limited injunction would have been appropriate had the case been one for unfair competition. But the difference is apparent to anyone who reads enough trademark opinions.
  68. .See Mark P. McKenna, Back to the Future: Rediscovering Equitable Discretion in Trademark Cases, 14 Lewis & Clark L. Rev. 537, 543–46 (2010) (“Notwithstanding their clear authority—and obligation—to do so, courts have been quite reluctant of late to limit the scope of injunctive relief, instead tending simply to enjoin the defendant’s use without qualification.”). As one court put it:

    [I]n “ordinary trademark infringement actions . . . complete injunctions against the infringing party are the order of the day. The reason is simple: the public deserves not to be led astray by the use of inevitably confusing marks—even in cases in which more than one entity has a legal right to use the mark.”

    Angel Flight of Ga., Inc. v. Angel Flight America, Inc., 522 F.3d 1200, 1209 (11th Cir. 2008) (quoting SunAmerica Corp. v. Sun Life Assur. Co. of Can., 77 F.3d 1325, 1336–37 (11th Cir. 1996)).

  69. .See McKenna, supra note 68, at 543 (noting that “courts have continued reasonably consistent practices of entering limited injunctive relief only in cases involving generic terms, surnames, and geographic terms”). It is worth noting that the section of the McCarthy trademark treatise that mentions equitable discretion to limit injunctive relief features mostly cases involving personal name marks, geographic marks, and descriptive marks, all of which were the province of unfair competition law historically. See 5 McCarthy, supra note 1, § 30:3 (describing injunctions that “balance the rights of the parties”). McCarthy also mentions creeping genericity cases where a generic mark has a dual meaning, and I discuss these cases below. See infra notes 80–88 and accompanying text.
  70. .4 McCarthy, supra note 4, §§ 23:1–4.
  71. .Registration makes a bigger difference in those countries that require registration and employ a property-type theory of infringement for registered marks.
  72. .See 3 McCarthy, supra note 4, § 19:3 (noting that registration does not create a trademark).
  73. .I do not mean to minimize the benefits of registration, which can be substantial, including nationwide protection and enhanced immunity from challenge when a registered mark becomes incontestable. See 5 id. § 26:31 (nationwide protection); 6 id. § 32:141 (incontestability). My point is only that the registration itself does not change the basic rules for qualifying trademarks as protectable and for finding liability.
  74. .Trademark law has always been concerned with preventing reputational harm to sellers, and more recently, harm to celebrities from false endorsement. However, there is some dispute about whether and when those harms trigger moral concerns. See, e.g., Bone, Taking the Confusion Out, supra note 49, at 1356–60 (discussing moral justifications for trademark protection). In particular, it is not at all clear that appropriation of goodwill is in itself unfair or that trademark law is the best legal instrument to redress any unfairness that might exist. Id. at 1355–58.
  75. .4 McCarthy, supra note 4, §§ 23:104, 23:106.
  76. .Courts do this by stretching sponsorship confusion and dilution theories.
  77. .See, e.g., Mark A. Lemley & Mark P. McKenna, Owning Mark(et)s, 109 Mich. L. Rev. 137, 176–80 (2010) (arguing against property-type protection of brands).
  78. .Id.
  79. .I do not mean to suggest that there are no sensible ways to divide up the field of trademark law. My focus here is on the particular division that distinguishes between trademark infringement and unfair competition. Later in this Article, I recommend, on policy grounds, broad liability and absolute injunctions for the use of fanciful or arbitrary word marks in direct competition with the plaintiff and categorical exclusions for certain socially desirable uses. See infra notes 107, 114–122 and accompanying text.
  80. .King-Seeley Thermos Co. v. Aladdin Indus., Inc., 321 F.2d 577, 579 (2d Cir. 1963).
  81. .2 McCarthy, supra note 1, §§ 12:1–2.
  82. .King-Seeley, 321 F.2d at 579.
  83. .See Murphy Door Bed Co. v. Interior Sleep Sys., Inc., 874 F.2d 95, 100–01 (2d Cir. 1989) (citing ASPIRIN and CELLOPHANE as examples of marks that became generic over time and holding that MURPHY BED has become generic for beds that fold up into the wall).
  84. .Indeed, it is possible that a single consumer might hold both meanings.
  85. .See 2 McCarthy, supra note 1, §§ 12:1, 12:6 (explaining that the “primary significance of the designation is controlling” and that “[t]he result of the primary significance rule is that majority usage controls”).
  86. .See, e.g., Blinded Veterans Ass’n v. Blinded Am. Veterans Found., 872 F.2d 1035, 1041–43 (D.C. Cir. 1989) (holding that Blinded Veterans is generic and others are free to use it, but that users can still be liable for unfair competition if they fail to take reasonable steps to distinguish themselves and prevent confusion); 2 McCarthy, supra note 1, § 12:51; Restatement (Third) of Unfair Competition, supra note 33, § 15 cmt. d (noting that while a generic term is “not protectable as a trademark, tradename, collective mark or certification mark” and thus others are free to use it, nevertheless subsequent users might be required “to take reasonable precautions to prevent a likelihood of confusion”).
  87. .Some of the cases involve active misrepresentations in addition to use of the generic term. But it is enough if the defendant uses only the generic mark without taking reasonable steps to mitigate consumer confusion. See Blinded Veterans, 872 F.2d at 1043, 1045 (“A second manufacturer may increase the risk of confusion by, for example, using a similar label, similar packaging, misleading advertisements, or simply by failing to state the product’s source” (emphasis added)).
  88. .The current approach to creeping genericity is also in tension with how trademark law handles abandonment, and the discrepancy, I believe, is due to the same mistaken assumption that trademarks are property and trademark law enforces exclusive rights. I have discussed this issue in other writing. See Robert G. Bone, Of Trolls, Orphans, and Abandoned Marks: What’s Wrong with Not Using Intellectual Property?, 42 Colum. J.L. & Arts 1, 47–48 (2018).
  89. .819 F.3d 697 (4th Cir. 2016).
  90. .Id. at 713.
  91. .Id. at 708–10 (“It is important to emphasize that this is an unfair competition case, not a trademark infringement case.”); see also ITC Ltd. v. Punchgini, Inc., 518 F.3d 159, 160, 164 (2d Cir. 2008) (recognizing the New York state court’s advisory opinion that state unfair competition law might provide relief to an owner of a foreign mark with substantial secondary meaning in the United States despite the territoriality principle, but holding that the requirements are not satisfied on the facts of the case); Empresa Cubana Del Tabaco v. Culbro Corp., 399 F.3d 462, 477–79 (2d Cir. 2005) (noting the possibility that the plaintiff might obtain relief on an unfair competition claim, but treating the claim brought in the case as one for trademark infringement and denying relief on the basis of the territoriality principle). For a recent historical account of the role of unfair competition in treaties and in the Lanham Act, especially § 43(a) and § 44, see Christine Haight Farley, The Lost Unfair Competition Law 25–65 (Feb. 20, 2020) (unpublished manuscript) (on file with author).
  92. .For a discussion of these issues, see Graeme B. Dinwoodie, Trademarks and Territory: Detaching Trademark Law from the Nation-State, 41 Hous. L. Rev. 885, 956–71 (2004).
  93. .See, e.g., Grupo Gigante v. Dallo & Co., Inc., 391 F.3d 1088, 1094 (9th Cir. 2004) (recognizing a famous marks exception to the territoriality principle); 5 McCarthy, supra note 4, § 29:61 (explaining the meaning of “famous mark” and its impact on rules related to foreign marks).
  94. .5 McCarthy, supra note 1, § 30:1; Mark A. Lemley, Did eBay Irreparably Injure Trademark Law?, 92 Notre Dame L. Rev. 1795, 1797 (2017).
  95. .Lemley, supra note 94, at 1807–08 (concluding that “unlike other IP cases, it is not only possible but common to win your [trademark] case and still not be awarded money”).
  96. .The Supreme Court recently held that willfulness is not an absolute requirement for recovering profits, but at the same time, it noted that “a trademark defendant’s mental state is a highly important consideration in determining whether an award of profits is appropriate.” Romag Fasteners, Inc. v. Fossil, Inc., 140 S. Ct. 1492, 1497 (2020).
  97. .547 U.S. 388 (2006).
  98. .See Adidas Am., Inc. v. Skechers USA, Inc., 890 F.3d 747, 756–57, 759–61 (9th Cir. 2018) (requiring plaintiff to provide “concrete evidence” of harm to reputation or goodwill in order to support a finding of likely irreparable injury for a preliminary injunction); Herb Reed Enters., LLC v. Florida Entm’t Mgmt., Inc., 736 F.3d 1239, 1249–50 (9th Cir. 2013) (holding that eBay applies to trademark suits and rejecting any presumption of irreparable harm, insisting instead on actual evidence).
  99. .The social benefits of enjoining a use are measured in terms of the policies trademark law serves: reducing search costs, maintaining product quality, achieving allocative efficiency, and preventing serious harm to consumers, sellers, and others. The social costs include reducing competition in the product market, impairing First Amendment values associated with expressive use of marks, and interfering with the optimal coordination of distinct intellectual property regimes. Trademark law should aim to strike an optimal balance between these benefits and costs.
  100. .See, e.g., AMF v. Sleekcraft, 599 F.2d 341, 354–55 (9th Cir. 1979) (issuing a limited injunction).
  101. .See, e.g., KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 543 U.S. 111, 124 (2004) (applying descriptive fair use).
  102. .See, e.g., Rogers v. Grimaldi, 875 F.2d 994, 996 (2d Cir. 1989).
  103. .See, e.g., TrafFix Devices, Inc. v. Mktg. Displays, Inc., 532 U.S. 23, 26 (2001) (trade dress); Toyota Motor Sales, U.S.A., Inc. v. Tabari, 610 F.3d 1171, 1179 (9th Cir. 2010) (online broker for Lexus cars).
  104. .See Robert G. Bone, Notice Failure and Defenses in Trademark Law, 96 B.U. L. Rev. 1245, 1263–65 (2016) [hereinafter Bone, Notice Failure] (identifying these types of uses as having special social value even if they also create confusion); McGeveran & McKenna, supra note 53, at 257–58 (noting that many of these types of uses are valuable and should receive more effective protection).
  105. .Examples include special secondary meaning requirements for product-design trade dress, Wal-Mart Stores, Inc. v. Samara Bros., 529 U.S. 205, 215–16 (2000); the functionality doctrine (either as an element of the prima facie case or as an affirmative defense), TrafFix Devices, Inc., 532 U.S. at 35; descriptive fair use, KP Permanent Make-Up, Inc., 543 U.S. at 121–22; nominative fair use, Toyota Motor Sales, U.S.A., Inc., 610 F.3d at 1179, and the Rogers test for expressive use, Rogers, 875 F.2d at 999.
  106. .See, e.g., Toyota Motor Sales, U.S.A., Inc., 610 F.3d at 1177, 1182 (directing district courts to enter limited injunctions when they find that the defendant’s nominative use is excessive).
  107. .See generally Bone, Enforcement Costs, supra note 4 (explaining various trademark doctrines by considering error and administrative costs). To be sure, this exception comes close to replicating the contours of the late nineteenth-century trademark infringement tort. See supra notes 10–29 and accompanying text. However, late nineteenth-century trademark infringement was justified by formalistic reasoning tied to a natural law conception of property rights. By contrast, the exception to case-specific balancing proposed here is justified functionally by reference to the policy goals of trademark law.
  108. .Am. Intell. Prop. L. Ass’n, Report of the Economic Survey 43–44 (2017). These figures are not as high as for patent cases, but they are still very substantial. See id. at 41 (reporting median figures for patent cases that are 1.5 to 3 times those for trademark cases). The survey is conducted every two years, and the 2017 publication includes the results for previous years. The highest figures were reported in 2015, when the median figure for trademark litigation ranged from a low of $325,000 for suits with less than $1 million at stake to a high of $1.6 million for suits with more than $25 million at stake. Id. at 43–44.
  109. .See Bone, Notice Failure, supra note 104, at 1256 (noting outcome uncertainty associated with the multifactor test).
  110. .See 4 McCarthy, supra note 1, at §§ 23:1–4 (discussing the likelihood-of-confusion test); Mark strength, mark similarity, product proximity, defendant’s intent, and evidence of actual confusion (if available) are factors that appear in most circuits’ multifactor tests. Bone, Taking the Confusion Out, supra note 49, at 1315.
  111. .Barton Beebe, An Empirical Study of the Multifactor Tests for Trademark Infringement, 94 Calif. L. Rev. 1591, 1596–98 (2006).
  112. .To be sure, many trademark cases settle after the preliminary injunction stage, but settlement is negotiated in light of the costs and risks of trial, which include the cost and risk of crafting a remedy.
  113. .The magnitude of the chilling effect is unclear. Today, judges are free to consider granting a limited injunction in any trademark case, and defendants who lose on liability have strong incentives to push for one. However, it seems reasonable to suppose that judges will be less likely to dismiss defendants’ arguments in a trademark system that invites remedial flexibility across the board.
  114. .Bone, Notice Failure, supra note 104, at 1274–80.
  115. .Others have proposed categorical exclusions, but the exclusions they propose are not absolute. See McGeveran & McKenna, supra note 53 (describing categorical exclusions while making room for limited remedies under unfair competition law). They would allow limited injunctions in those rare cases where there is a substantial risk of consumer confusion. My exclusions are absolute. The cost of confusion in rare cases does not justify exceptions that undermine the benefits of a categorical approach.
  116. .See Bone, Notice Failure, supra note 104, at 1274–77 (justifying a categorical exclusion for expressive use).
  117. .These include the nominative fair use defense and the Rogers test. See Toyota Motor Sales, U.S.A., Inc. v. Tabari, 610 F.3d 1171, 1179 (9th Cir. 2010) (applying the nominative fair use test); Rogers v. Grimaldi, 875 F.2d 994, 999 (2d Cir. 1989) (construing the Lanham Act to apply only when “the public interest in avoiding consumer confusion outweighs the public interest in free expression”).
  118. .This increases costs and makes it harder for a defendant to exit a lawsuit early.
  119. .A court must still classify a use as expressive in order to apply any categorical exclusion. Parties are likely to contest the classification issue, and this will generate litigation cost and risk. However, classification should be relatively straightforward for those purely expressive uses lying at the core of the First Amendment. There will be some difficult cases, but that is inevitable for any classification scheme. Moreover, if limits are imposed on the set of expressive uses covered by the categorical exclusion, those limits should be defined as clearly as possible. See Bone, Notice Failure, supra note 104, at 1276 & n.148 (distinguishing core First Amendment expression, such as critical commentary, artistic creation, political speech, and the like from expression not in the core).
  120. .The audience is not likely to care whether the trademark owner authorized the expressive use; the expression usually has the same value either way. The trademark owner does not lose sales in most cases. The only exception is a case in which the owner licenses expressive uses of its mark, but in that case, the social interest in competition and the right to freedom of expression weigh against protection. The trademark owner might suffer a reputational harm if the expressive use is critical, but that is not a harm that trademark law should prevent. Indeed, criticism is a form of expression that has special First Amendment value.
  121. .In addition, expressive use is particularly sensitive to litigation cost and risk. Prospective users are often risk-averse individuals, and in many cases, an expressive use is not profitable enough to justify taking the risk of litigation in the face of a cease-and-desist letter.
  122. .A similar argument applies to product-design trade dress. See Bone, Notice Failure, supra note 104, at 1278–80. Competitors copy product features in order to compete in the product market. In general, this is perfectly consistent with trademark law. Trademark law is not supposed to impede product market competition; indeed, from an economic point of view, trademark is supposed to promote competition by facilitating the transmission of product information to consumers. Problems arise, however, when features that competitors copy also have source-identifying meaning. Under these circumstances, trademark law must balance the benefits of preventing consumer confusion against the costs of impeding competition. Trademark law strikes this balance primarily through the functionality doctrine. But many courts apply the functionality doctrine in fact-sensitive, case-specific ways, which can create chilling effects for lawful copying and use. A categorical exclusion makes sense as a solution to this problem because of the relatively low probability of harmful consumer confusion in the long run. The key here is that sellers always use some other mark—e.g., a word mark or a logo—to sell their products. If competitors were free to copy product-design trade dress, consumers would soon learn not to rely on product features as source-identifiers and focus on the word mark or logo instead. And knowing that consumers will rely on word marks and logos, sellers would have strong incentives to advertise and promote those symbols, thereby further reinforcing consumer reliance on them.