Competition and Congestion in Trademark Law

Article - Volume 102 - Issue 3


Imagine that the electronics firm Persimmon, Inc. begins selling a new fitness tracker that it calls ActivTrak. The firm is trying to enter a crowded market for consumer electronics, and it has some new technology that it hopes will set its ActivTrak device apart. Persimmon spends substantial sums of money advertising its device on television and online, to the point that if you took a survey of consumers and asked them about the “ActivTrak,” a large number of them would remember that it’s the new type of fitness tracker made by Persimmon. The firm claims trademark rights in ActivTrak and registers the mark with the United States Patent & Trademark Office (PTO). Now, no other firm can use those words for similar products without violating Persimmon’s trademark rights.[1]

It is presumably good for consumers to have another fitness tracker on the market.[2] But what about Persimmon’s branding and marketing strategy? Will it have positive or negative consequences for consumers in the electronics market? On the one hand, consumers now know that if they see a fitness tracker with the name ActivTrak on it, that device was made by Persimmon.[3] If a consumer thinks that Persimmon generally produces products of high (or low) quality, the trademarks can help her to purchase (or avoid) Persimmon products in the future. And because Persimmon alone will reap the reputational benefits (or bear the reputational costs) of its products, it has an incentive to maintain consistent product quality.

But from a competition perspective, there is a darker side to this branding. Because both “active” and “track” are obvious words to associate with fitness trackers, giving Persimmon the exclusive right to use “ActivTrak” could give it a significant advantage.[4] And that advantage will likely extend beyond being able to prevent competitors from using the term “ActivTrak” as a whole: indeed, there’s a meaningful risk that Persimmon will try to use its rights in that composite mark to bar others from identifying their own devices (or similar products) using the word “active,” and maybe even “track.” Competitors unwilling to risk litigation will have to find other evocative words and phrases to identify their own fitness trackers. But how many other signifiers will have the power of the word “active”? That term immediately summons the idea of fitness, and Persimmon’s ActivTrak may come to the minds of consumers more quickly and easily than the names of competing products. The alternatives available to competitors are not unlimited in number.[5] Once firms have snapped up “active,” “fit,” and “run,” the available store of evocative brand names will diminish relatively rapidly. The latest firm to try to enter this crowded marketplace might find itself climbing a steep hill, so to speak.

The competitive advantages associated with exclusive control of particularly powerful words have real costs for consumers.[6] If Persimmon can limit the words that competitors can use, it can make it harder for those competitors to reach consumers. This may mean that consumers are steered toward Persimmon’s products when they would otherwise prefer another brand, or it might just mean other brands have a harder time gaining traction in the market and therefore can’t compete as effectively with Persimmon. Critically, the advantages to Persimmon do not accrue because its product is better. They accrue because the law gives Persimmon the right to limit others’ use of the word “active” in connection with competing fitness trackers.[7]

Trademark law should be attuned to these competition concerns. Indeed, these are the principal concerns to which trademark law should be attuned. Trademark law exists to promote fair competition, which ultimately benefits consumers. If certain types of trademarks tend to inhibit competition, to the detriment of consumers, those are the types of trademarks that the law should screen out.

And, in fact, trademark law purports to take seriously the idea that some words are more competitively important than others. It does so by making it harder to acquire trademark rights in so-called descriptive terms, specifically by requiring that those terms develop secondary meaning before they are protected as trademarks. That is, before descriptive terms can be protected as trademarks, they must signify to consumers not just what sort of goods they are identifying (the primary meaning), but also the source of those goods (the secondary meaning). Trademark law also purports to make more space for non-trademark use of descriptive terms by recognizing certain uses of those terms as descriptive fair uses. But these doctrines are insufficient.

First, current doctrine draws the most important line between descriptive terms and so-called suggestive terms, treating suggestive terms as inherently distinctive and immediately protectable and only requiring additional evidence of source significance for terms classified as descriptive. But as we explain at length below, experience demonstrates that courts and the PTO can’t consistently distinguish descriptive and suggestive terms. This is much more than just a question of occasional errors; rather, it is a systemic problem. Trademark law draws an essential doctrinal distinction with significant legal consequences on the basis of a factual distinction that borders on arbitrariness.

Making matters worse, trademark law’s mechanism for dealing with anticompetitive descriptive marks is woefully underpowered. As we noted, descriptive terms will be treated as trademarks if the user establishes secondary meaning. But it’s not actually clear what the standard for secondary meaning is or what evidence demonstrates its existence in a particular case. Most often, courts rely on circumstantial evidence of sales or advertising, but that structure does little or nothing to prevent firms from claiming anticompetitive marks. Firms will already be motivated to spend money on advertising, so legal rules that reward them for that spending are likely to have little marginal effect on their willingness to seek anticompetitive marks.

The rules wouldn’t be sufficient even if courts and the PTO were better able to distinguish descriptive and suggestive terms because that distinction misses a fundamental point. Suggestive marks also impose meaningful competitive costs. Terms like “Igloo” for coolers or “Butter” for chardonnay are evocative of their products in much the same way that descriptive terms would be. While the range of available suggestive terms for a product may generally be wider than for descriptive terms, the difference is not nearly as stark as trademark law’s binary distinction treats it. Allowing firms to assert rights in suggestive terms immediately and without any meaningful expense gives them serious competitive advantages that raise costs for consumers.

Thus, in its current form, trademark law fails to effectively screen out competitively harmful marks. This is not inevitable. Indeed, there are a variety of doctrinal changes that could help cure trademark law’s myopia and more effectively focus its gaze on issues of competition, rather than merely source significance. We could, for example, create doctrinal screens that categorically prohibit protection of descriptive or suggestive terms.[8] Or we could return to the doctrinal structure that once prevailed, in which descriptive trademarks were categorically barred, but the doctrine of unfair competition provided more limited remedies when competitors used those descriptive terms in a way that was designed to deceive consumers.[9]

We don’t argue for those changes here. One reason is pragmatic: it’s likely that too much water has passed under the bridge for the old rules of technical trademarks to be implemented. Even more, any new doctrinal screen would presumably only affect yet-to-emerge firms, while firms that have already claimed anticompetitive descriptive or suggestive marks would be allowed to continue using them. This would further bake in existing firms’ advantages.

But the main reason we offer a different approach here is that the optimal number of descriptive or suggestive trademarks may not be zero. For consumers, it’s not the worst thing in the world to have one fitness tracker with “Active” in the name, or one insulated drink container called “Igloo,” or one ride-share company called “Lyft.” It’s easier for consumers to associate these brand names with the relevant goods or services, which makes it easier for consumers to find what they’re looking for.[10] The problem becomes most acute when there is one insulated drink container named Igloo, another named Arctic, a third named Polar, and so forth. The number of useful words available for describing how a given container keeps a drink cold can narrow drastically. Precisely because there are fewer available terms that convey good information, the more of those terms that are taken, the greater the competitive cost associated with each additional term.

What should policymakers do when they want to reduce the amount of some type of activity but don’t want to ban it outright? One option is to raise the price of engaging in that activity. What is called for is a type of “congestion pricing,” similar to what large cities like London use to reduce traffic in the city center. We argue that policymakers should impose a type of tax on descriptive and suggestive marks. This would act as a costly screen, reducing the number of such marks in use, and channeling firms toward arbitrary and fanciful marks that don’t create the same kind of built-in competitive advantages. The result would be to reduce congestion around the most important product-related words.

Using costly screens is a second-best alternative in a world where doctrinal screens are imperfect. But a well-structured costly screen can ultimately aid new entrants and improve competition. It can screen out rights that would be increasingly costly as more firms captured competitively significant terms, reducing linguistic congestion around products that makes life difficult for new entrants and existing businesses alike. Or so we shall argue.

This Article proceeds in four parts. In Part I, we describe the way trademark law is meant to operate. In Part II, we explain and analyze the law’s shortcomings—how its focus on source significance has obscured greater questions of anticompetitive advantage. In Part III, we limn the operation of doctrinal and costly screens. We demonstrate that, despite superficial appearances to the contrary, trademark law lacks a meaningful costly screen that might deter private parties from seeking anticompetitive trademarks. Finally, in Part IV, we explain how these shortcomings of trademark law could be alleviated with more effective doctrinal and costly screens.

One final note is in order before we proceed. This is a work of legal theory, which—like all works of legal theory—is based on a set of factual premises. As is the case with essentially all works of this type, we cannot prove these premises as a matter of irrefutable truth. When we argue, for instance, that the competitive advantages associated with exclusive control of particularly powerful words have real costs for consumers, or that existing trademark doctrines are insufficient to address this problem, we cannot demonstrate these factual claims to be true in the same way that one might prove that 2 + 2 = 4. However, in the pages that follow, we do not merely assert or assume those factual premises. Rather, we offer reasons to believe that they are true. We provide evidence for them and offer arguments for them. If the reader does not believe the factual premises to be true despite our evidence and arguments, then what remains of the Article is not likely to be convincing either. But that is the nature of the genre of legal theory. In that respect, our work is no different than the vast majority of the work that fills the pages of American law reviews. Accordingly, our hope is that the arguments and evidence we muster will be compelling!

With that cautionary note in mind, on to the main event.

I. Trademarks and Their Discontents

A. Trademarks and Competition

The standard account of the way that giving firms[11] or other organizations certain exclusive rights in names, symbols, or designs can improve markets and competition is well known, and we won’t rehearse it at great length here.[12] As is often true in the law, understanding trademark law’s competition benefits is easier when contemplating the doctrine’s nonexistence.

Imagine that a winery has adopted the name SHEEP to identify and label the products that it makes. The winery sells a bunch of bottles so labeled, and people like it and want to buy more SHEEP wine. But when they return to market, they see four different bottles of wine, all labeled SHEEP.[13] The different bottles all look basically the same, and there is no way, short of opening each one, for consumers to determine which, if any, was made by the winery that made the first bottle of SHEEP wine that they tried. Goods like wine are valued primarily for their “experience” qualities—aspects of the goods that consumers can’t discover merely by looking at them but have to actually experience.[14] But when several bottles from different producers all have the SHEEP mark, consumers can’t figure out which of the bottles might offer the same pleasures as the first SHEEP wine, so they choose randomly. Unhappily, three of four consumers won’t get the wine that they want and instead will purchase a wine made by a different winery.

This is unfair competition. When we say that the wineries that bottled the other SHEEP wines did so “unfairly,” what we mean is that they have created a meaningful risk of consumer harm and prevented the first winery from making sales that it had earned. Some of the people who wanted to buy wines with the qualities that they associated with the producer of the first SHEEP wine didn’t get the wines that they wanted. They didn’t necessarily get worse wine; they just didn’t get the wine they were hoping for. They were deceived into purchasing wine from other producers rather than the one with which they had positive associations. And the result was a lost sale for the original SHEEP wine producer.

Trademark law exists to prevent this risk of consumer deception. The law gives the first winery to use the word SHEEP on its products the exclusive right to do so, at least within that category of goods.[15] Now, having formed a positive association with wines labeled SHEEP or having been told to seek out wines so labeled, consumers can reliably obtain the products that they desire. Having identified a particular product with a particular source, consumers can feel confident, in a world with trademark law, that they will be able to use particular symbols as indicators and guarantors of source.[16]

Understood this way, trademark law has significant benefits for consumers.[17] It enables people to buy the products they think they’re buying with less risk of being deceived. But trademark law has benefits for producers too, and indeed courts and commentators frequently describe trademark law in dual-benefit terms.[18] Most directly, giving exclusive rights ensures that the consumer confidence we described translates into sales for the first user and that those sales aren’t diverted to others who act deceptively. Less directly, trademark rights can allow producers to capture the value of investments in product quality.

Consider the winery example again. Imagine that a winery believes that if it ages its wines in new oak barrels the resulting product will be superior to those of its competitors. The barrels and aging cost money, though, and the winery will have to charge more for the wine in order to offset the expense of these improvements. Now, again, the winery sells its wine using the name SHEEP on its labels, and some consumers taste the wine and appreciate the added flavor and richness that the oak barrels contribute. In a market without trademark law, though, other producers could use the same SHEEP mark on their bottles and price them more cheaply because they did not invest in expensive aging.[19] The same problem reemerges. Consumers who are willing to pay more for oak aging cannot use brand names to reliably determine which products are so treated, and knowing this, the wineries are discouraged from making improvements in product quality, because they will not as easily recoup their investments.[20] So trademark law, by helping consumers find the products that they want, can enable socially valuable competition.[21]

But consider a different situation. Because of an error at the winery, after a batch of white grapes is pressed, the juice is left to linger on the grape skins for several weeks. The resulting wine has a much darker amber color than a standard white wine along with a richer, fuller flavor. It turns out that people like the wine, and the winery bottles it, calling it “orange” wine.[22] Because of its popularity, other wineries also want to make this style of wine, and the initial winery doesn’t have a patent that prevents them from doing so.[23] The other wineries also want consumers to be able to find their wines and understand what they are buying, so they too use the term “orange” to describe their wines. From the perspective of the initial winery, allowing others to use the term “orange” feels like unfair competition, because that winery is going to lose sales to its rivals or be forced to reduce its prices. But from consumers’ and rivals’ perspectives, this is the essence of fair competition.[24] Consumers get a range of options in a particular category, and they can choose the one that best aligns with their preferences for quality and price. And competitors are able to accurately describe competing products without having to resort to alternative terms.[25] If competitors were barred from using the word “orange” to describe their orange-tinged wine, they would not be able to accurately convey to consumers that they are making a similar product in a similar way.[26]

Thus, although trademark law can have important pro-competitive benefits, the recognition of exclusive rights to words, symbols, or designs can also have harmful effects on competition and, thus, on consumers if those rights aren’t appropriately tailored. For instance, if a winery were allowed to choose as its mark terms like WINE, RED WINE, or OAK AGED, other wineries would face enormous hurdles trying to describe their products without infringing the initial firm’s rights. These words are so closely tied to the product at issue—red wine—that giving a single firm an intellectual property right over one or more of the words would bestow upon that firm a substantial competitive advantage. The same issue would arise if the firm were to gain trademark rights in a picture of a glass of red wine or a bunch of grapes and those rights were broad enough to prevent others from using similar kinds of pictures.[27]

These are the types of harms that trademark law should be trying to avoid. As we detail below, existing doctrine is not ignorant of these concerns. Indeed, trademark doctrine purports to avoid them. But as we will explain, it has fallen notably short.

B. The Trademark Categorization System

Word marks are the paradigmatic example of how firms and consumers communicate about product source and the paradigmatic form of trademark around which the eligibility rules were developed. But not all word marks are equivalent from a competition perspective. In some cases, a firm’s choice of a particular word mark may give the firm undue competitive advantages over its rivals, enabling it to engage in rent-seeking behavior that lowers consumer welfare.[28] In other circumstances, the sheer volume of claimed word marks may substantially increase the costs of entry for new firms, because too many of the “good” word marks are already claimed in a given field.[29]

It is an axiomatic principle of trademark law that different kinds of word marks have different impacts on competition. Trademark law understands these differential effects primarily through the lens of “distinctiveness.” As a matter of current doctrine, signs that indicate source—that is, signs that indicate to consumers which producer is responsible for the goods—are called “distinctive,” and distinctive signs are, by definition, trademarks.[30] That is, distinctiveness is treated as both necessary and sufficient for trademark validity.[31] But not all words or phrases are treated the same in terms of how the law assesses their potential distinctiveness.

Trademark law has long evaluated words’ distinctiveness by categorizing them along the famous Abercrombie spectrum.[32] That approach treats terms differently depending upon the amount of information the terms provide about the products or services with which they’re used.[33] Arbitrary terms (terms that have existing meanings that are unrelated to the goods, like Apple for computers, Persimmon for fitness trackers, or Sheep for wine) and fanciful terms (coined terms, like Exxon) are considered inherently distinctive, which means they are protected as trademarks immediately upon use, without need for evidence that consumers actually associate the terms with particular producers.[34] That legal treatment is based in part on an empirical assumption that consumers will automatically treat arbitrary and fanciful terms as trademarks.[35] Because those terms provide no information about the nature or characteristics of the goods or services with which they’re used, consumers are presumed to have no way to understand their use except as trademarks. Why else would someone put the word “Apple” on a computer?[36]

Trademark law also treats so-called suggestive terms, like “Igloo” for coolers, as inherently distinctive and protected as marks immediately upon use.[37] Unlike arbitrary and fanciful terms, suggestive terms relate to the goods they’re used with. Though Igloo doesn’t describe coolers, it evokes something cold, so the connection is obvious once the goods are known. But because the information that suggestive terms provide is indirect, trademark law treats them more like arbitrary terms than descriptive ones.[38] Just as with other inherently distinctive terms, the assumption is that consumers will be predisposed to understand suggestive terms as brand names rather than descriptions of the product.[39]

In addition to the assumptions about consumer understanding that underlie the treatment of arbitrary, fanciful, and suggestive terms, courts have long assumed that other parties have no real need to use those terms for their own goods because the terms provide no information about the goods or services with which they’re associated.[40] Indeed, the belief that there was no legitimate explanation for a competitor using an arbitrary or fanciful term previously adopted by another was the justification for treating those terms as trademarks and describing them as property rights.[41] Because competitors could choose from a nearly infinite variety of other terms, the law has similarly assumed that automatically treating suggestive terms as trademarks imposes little to no competitive harm.[42]

Some words convey more information about the goods or services with which they’re used. Descriptive terms convey direct information about the qualities or characteristics of goods.[43] When consumers see the word “tasty” on a food product they might think that word is a trademark, but they might just as easily (and even more plausibly) think that term simply describes the food product as tasty.[44] The same can be said of geographic terms, which are treated like descriptive terms.[45] “Milwaukee’s Best” might be a brand of beer, but, depending on how it’s used, it might also just be a laudatory description of beer from Milwaukee.

Precisely because descriptive terms convey information about the goods or services, protecting them as trademarks also comes at greater cost to competitors who also need to convey information to consumers.[46] Giving one party exclusive rights to use the term “tasty” for food products limits competitors’ ability to communicate that their products are also tasty. Likewise, giving one party exclusive rights to “Milwaukee’s Best” limits other Milwaukee-based beer producers’ ability to inform consumers where their beer is made.

Because consumer understanding is not a given, and because of the potential competitive costs of exclusive rights, descriptive terms (and geographic terms) have to earn their way into trademark status by acquiring distinctiveness.[47] Specifically, those terms are protected as trademarks only when consumers have come to understand that, in addition to conveying information about the goods, the terms identify the source of the goods with which they’re used.[48] Trademark law calls this “secondary meaning”—secondary not in the sense of being of secondary importance, but in the sense of being second in time to the primary, descriptive meaning.[49] “Best Buy” might once simply have described a store with good deals, but over time consumers have come to associate that phrase with a particular company, at least in the context of retail stores selling electronics.

Courts commonly say that secondary meaning may be proven by both direct and circumstantial evidence.[50] But while “[t]he authorities [may be] in agreement that survey evidence is the most direct and persuasive way of establishing secondary meaning,”[51] in fact courts most often focus on circumstantial evidence such as the “amount and manner of advertising, volume of sales, and length and manner of use.”[52] The idea is that consumers’ exposure over time trains them to regard an otherwise descriptive term as having an additional, source-related meaning.[53] When they acquire distinctiveness, descriptive terms become trademarks, and competitors are substantially limited in the ways they can use those terms to identify their own goods.

Finally, terms that simply name a relevant category of product or service are deemed “generic” and can never be trademarks.[54] The primary significance of the term to the relevant consuming public controls,[55] but courts have articulated the test of genericness in several similar ways. In the recent[56] case, the Supreme Court defined a generic term as one that “names a ‘class’ of goods or services, rather than any particular feature or exemplification of the class,” and it emphasized that the only relevant meaning was the meaning to consumers.[57] Hence, “whether ‘’ is generic turn[ed] on whether that term, taken as a whole, signifie[d] to consumers the class of online hotel-reservation services.”[58]

The exclusion of generic terms from protection can also be explained in terms of both consumer understanding and competitive costs. Generic terms do not do the job of a trademark because they do not indicate source.[59] And exclusive rights in generic terms would be particularly harmful to competitors. In fact, courts often characterize protection of a generic term as conferring a sort of monopoly on the claimant.[60] Because a generic term names a relevant product category, giving one producer exclusive rights to that term would give that producer a significant advantage over competitors who would be unable to signal to consumers that their products competed in that same market. That helps explain the concept of “genericide,” in which once-valuable trademarks that clearly did (and likely still do) indicate source to some consumers are nonetheless denied future protection because a majority of the consuming public has come to associate the term with the entire class of goods.[61]

* * *

In sum, the process of determining the distinctiveness of purported word marks is a categorization exercise, where the categories reflect the extent to which the claimed term gives information about the goods. The primary distinction is between terms that are inherently distinctive (fanciful, arbitrary, and suggestive terms) and those that are not. Inherently distinctive terms are automatically treated as trademarks; some non-inherently distinctive terms (descriptive terms, geographic terms, surnames) can earn their way into trademark status by developing secondary meaning. Terms that name a relevant category of products or services are deemed generic and can never serve as trademarks.[62]

The point of these doctrines is to separate words that are likely to signify the source of the goods, and thus deserve protection, from those words that are likely to be understood as describing the goods, and therefore do not. The theory behind these doctrines is that if consumers are likely to regard a sign as identifying the source of the goods, protection of that sign will enhance competition; if consumers are not likely to treat it as a source indicator, then protection will be anticompetitive. But as we will demonstrate below, in its pursuit of these (worthy) goals, trademark doctrine has turned out to be both unsuccessful in operation and profoundly flawed in theory.

II. The Shortcomings of Trademark Doctrine

In the prior Part, we briefly explained both the necessity of trademark rules and the possible pitfalls of granting trademark rights. In theory, well-crafted trademark law doctrine would recognize trademarks that inform consumers and promote competition while screening out signs that award anticompetitive advantages and diminish consumer choice. But that is not how the law has developed, as the following Part explains. Here, we describe how flaws in trademark law’s fundamental assumptions have allowed anticompetitive marks to proliferate.

We make three interlocking arguments below. First (in subpart II(A)), we explain how trademark law’s system of categorization, which is fundamental to the law’s operation, requires unworkable line-drawing. Second (in subpart I(B)), we describe how the doctrines that are meant to protect competitors from the most anticompetitive trademarks systematically fail to do so. And finally (in subpart II(C)), we argue that even if those doctrines were to function correctly, trademark law’s rules cannot bear the weight placed upon them. Words that pass all the various doctrinal tests and achieve trademark status can still create anticompetitive harm greatly in excess of any public value they might offer. If trademark law is meant to promote competition, it is failing.

A. The Thinness of the Suggestive/Descriptive Distinction

As we explained in Part I, a great deal turns on whether a claimed mark is classified as descriptive or suggestive. If it is suggestive, it is deemed inherently distinctive and will be treated as a trademark without further effort or expense; if it is descriptive, its owner must earn trademark status by expending resources so that the mark acquires secondary meaning. Yet given the characterizations of the categories, the line between descriptive and suggestive terms should immediately strike the reader as thin. Both types of terms convey information about the good at issue: descriptive terms do so “directly,” while suggestive terms do so merely “indirectly.” This is not an especially crisp line on which to hang such an important distinction. And, in practice, the line has proven to be notoriously unclear.

Take, for example, the lawsuit filed against Ford by Cruise LLC and General Motors (GM). Cruise and GM claimed that Ford’s use of BlueCruise for an automated driving system infringed their rights in the CRUISE and SUPER CRUISE marks and several CRUISE-formative marks for automated driving technology.[63] It’s hardly a stretch to connect the term Cruise to automatic driving technology—“cruise control” has long been the generic term for the feature that maintains a car’s speed without need for manual operation, and automated driving is simply a technological extension of that feature. Nevertheless, both Cruise and GM have successfully registered several Cruise-formative marks.[64] Indeed, Cruise overcame an initial descriptiveness refusal for the original CRUISE application simply by highlighting that its software for automated driving was not a “device[] used ‘to operate at a predetermined speed by use of a regulating mechanism,’” but instead “cover[ed] far more aspects of driving than speed and the speed of the vehicle is not necessarily constant.”[65] On that basis, it was able to register the mark without evidence of secondary meaning.[66]

All categorization schemes have boundary problems, of course, but trademark law draws a fundamental legal distinction on the basis of a factual distinction that courts cannot consistently make. In that respect, the Cruise case is just one of many examples. In his leading trademark treatise, Professor Thomas McCarthy admits that “lawyers and judges should not deceive themselves into conceiving the descriptive-suggestive dichotomy as some kind of concrete and objective classification system. It is no more objective and free of personal predilections than a test which asks persons to divide all color shades into ‘light’ and ‘dark.’”[67] Professor McCarthy then devotes an entire section to listing marks that were held “nondescriptive and merely suggestive,” including many that convey important information about the nature of the goods or services: CUSTOMER FIRST for community banking services, EQUINE TECHNOLOGIES for horse hoof care products, GOLIATH for large diameter wood pencils, HYGIENT for hygienic mattress covers, MARRIAGE PROPONENTS for prospective marriage partner services, and POM for pomegranate juice.[68]

Critically, in every case in which a court or the Trademark Office classifies an arguably descriptive word like Cruise as suggestive, the consequence is that the term is then protected automatically upon use—secondary meaning is not required. To the extent misclassification is common, trademark law’s primary mechanism for ensuring that the benefits of trademark protection outweigh its costs never even comes into play.

The problem is even worse when we consider the scope of rights often afforded descriptive terms. As the Cruise case demonstrates, descriptive marks are often asserted against non-identical marks that have only the descriptive component in common. Cruise and GM asserted their Cruise marks against Ford’s BlueCruise. More recently, Chipotle sued Sweetgreen for the latter’s naming of its Chipotle Chicken Burrito Bowl—a chipotle-flavored chicken dish.[69] As these examples demonstrate, it’s often more than just control over the descriptive term that is at stake. Protection of a descriptive term often translates into control over adjacent or composite terms with a descriptive component. And that type of control can be particularly harmful to competition precisely because the term is descriptive and thus exactly the sort of word that competitors might well want to use to describe similar types of products.

Enforcement of the other Cruise-related marks also reflects the related problem of granting protection to compound marks. Those marks often contain words that, on their own, would be highly descriptive of the goods with which they’re used. But under existing doctrine, marks must be evaluated as a whole rather than broken into their individual parts.[70] As a result, many longer phrases wind up qualifying for trademark protection despite the risks associated with exclusive control of the descriptive parts. That might be fine if the rights granted those mark owners were limited in scope, such that infringement were only found when the defendant used the entire phrase. But that often does not happen, which means that parties are able to use trademark law to control the descriptive terms within longer phrases.[71]

B. The Failures of Secondary Meaning and Descriptive Fair Use

Trademark law purports to have created two doctrinal bulwarks against anticompetitive marks. The first, discussed above, is that descriptive terms must acquire secondary meaning before they can be protected as marks. The second is the doctrine of descriptive fair use, which is meant to carve out space for competitors to use descriptive trademarks to describe their products, so long as they are not deceiving consumers into believing that the goods are being made by the owner of the trademark. Neither of these doctrines is doing the work it is meant to do.

1. Secondary Meaning

The rules governing secondary meaning have proven to be no more comprehensible or workable than the distinction between descriptive and suggestive terms. To begin with, it’s not even clear that anyone knows what secondary meaning is. According to the Supreme Court, “secondary meaning is acquired when in the minds of the public, the primary significance of a product feature . . . is to identify the source of the product rather than the product itself.”[72] But it’s unclear whether that means that the majority of consumers must recognize the source significance (so that “primary” is a measure of quantity), that the source meaning must have eclipsed the original meaning in the minds of consumers (so that “primary” refers to dominance of meaning), or some combination of both of those things.

Professor McCarthy argues that “the terms have only a temporal significance.”[73] In his view, “‘[s]econdary’ meaning is so called only because it is created second in time, not because it is necessarily secondary in importance or significance in customers’ minds.”[74] He specifically rejects the idea that the claimant must prove that the “most important meaning of the designation is as a source identifier—a trademark or service mark.”[75] “[T]he legal terms of art ‘primary’ and ‘secondary’ meaning do not refer to relative magnitude or prominence, but only to meanings that were created first and second in time.”[76]

But it can’t be true that there is no threshold for this “second in time” meaning, in either the quantitative or qualitative sense.[77] If the three of us, but no one else, came to associate a term with a particular producer, that couldn’t possibly be enough to prove secondary meaning, no matter how clearly that meaning was second in time for us.[78] Nor could it be enough that some people think products bearing the word “tasty” usually come from Frito-Lay, if those people are just very familiar with Frito-Lay and recognize that it uses that term to describe its chips. When trademark status is at stake, primary significance must have more than temporal meaning.

The complete lack of clarity about the relevant thresholds makes secondary meaning a matter of intuition more than evidence. Maybe because they so often infer secondary meaning from circumstantial evidence, courts typically don’t even consider the threshold question (how many consumers must treat the term as a trademark). Alternatively, courts may rely on circumstantial evidence in order to avoid hard threshold questions. But without any identifiable goalposts, even surveys—which would otherwise seem like the best form of secondary meaning evidence, if there were a reliable methodology available[79]—can’t really give clear answers. As Professor McCarthy notes, “[c]ourts have been vague and uncertain in defining what is the minimum acceptable percentage of persons who have a secondary meaning in their minds.”[80] That’s a serious problem for the doctrine that is supposed to be the means of balancing competitive concerns by making it harder for parties to claim descriptive terms.

2. Descriptive Fair Use

Trademark law accounts for the competitive importance of descriptive terms in one other way. Under the doctrine of descriptive fair use, it is a defense that use of the allegedly infringing term is a use “otherwise than as a mark . . . of a term or device which is descriptive of and used fairly and in good faith only to describe the goods or services of such party, or their geographic origin.”[81] But descriptive fair use is notoriously fact intensive and difficult to apply,[82] not least because of the difficulty of determining when a use is “otherwise than as a mark” and how that inquiry differs from the likelihood of confusion test.[83] Firms that should qualify for descriptive fair use are frequently denied the defense—or at least face expensive litigation to establish their right to use the term—due to the confusion and shortcomings of the doctrine.[84] Sweetgreen’s decision to settle with Chipotle is a case in point: Sweetgreen’s naming of its product as a Chipotle Chicken Burrito Bowl very likely would have qualified as descriptive fair use, but it agreed to change the name likely in order to avoid the time and expense of a lawsuit. So there are competitive consequences to awarding trademark rights in descriptive terms that aren’t fully resolved by making room for descriptive fair use.[85] And, of course, that doctrine only makes room for other parties to make non-trademark use of the descriptive terms. It still allows one party to get the benefits of using the descriptive term as a brand.[86]

C. The Inadequacy of Source Significance

While it certainly makes sense for trademark law to approach validity taxonomically rather than trying to adjudicate the competitive effects of each claimed mark, its taxonomy must reflect the realities of language usage and market competition. Yet there are good reasons to believe that the law systematically underestimates the challenges that trademark creates for new market entrants.

Begin with suggestive terms. The law treats suggestive terms as inherently distinctive on the twin assumptions that (a) those terms don’t provide direct information about the goods, and (b) that their exclusive control doesn’t produce significant competitive harms.[87] Both of those assumptions are at least overstated if not flatly wrong in many cases.

First, exclusive use of suggestive terms more frequently imposes meaningful competitive harm than is generally assumed. For example, the Trademark Office treated BUTTER as inherently distinctive for wine and registered it without requiring evidence of secondary meaning.[88] It did so even though white wines are often described as “buttery,”[89] and consumers might therefore be more attracted to a white wine that is branded as BUTTER[90] than they would an otherwise identical product, simply because the name has positive connotations for them.[91] Similarly, the Sixth Circuit held that the mark 5-HOUR ENERGY was suggestive rather than descriptive, because the kind of energy and the method of its transference were ambiguous.[92] But granting the claimant an immediately valid trademark in such a term gives it a substantial competitive advantage.[93] Even if 5-HOUR ENERGY doesn’t describe everything about the relevant product, it clearly identifies some of those product’s most important qualities. Consumers may wonder if competitors’ products will provide an equivalent period of energy.[94]

Again, there is nothing wrong with one party obtaining a competitive advantage by virtue of having created a better product. That is the nature of competition. The problem arises because firms are using government-sanctioned monopolies on language to gain competitive advantages that have nothing to do with the quality of their products. This is a strategic distortion of market competition.

One likely cause of the law’s insensitivity to these issues is its focus on the needs of competitors rather than on the competitive benefits trademarks are conveying to claimants. Courts assume that granting rights in inherently distinctive marks imposes little or no cost on competitors who are free to choose from a nearly infinite dictionary of alternative options to brand their goods. According to Landes and Posner: “There are 450,000 words in Webster’s Third New International Dictionary, and although they are not freely substitutable if one is trying to say something that will be understood, they are freely substitutable if one is uninterested in meaning.”[95] They assume, as do most courts,[96] that word marks at the top of the Abercrombie distinctiveness spectrum are basically fungible and unlimited.

The trouble is, almost no one else thinks so, including the professionals hired to help firms choose brand names.[97] As Jake Linford has recently argued, certain sounds seem to attach more naturally to some goods than to others.[98] Marks like SWIFFER for a sweeping device or VIAGRA for an erectile disfunction medication rely on sound symbolism to create, perhaps unconsciously, associations between a product and its desirable attributes.[99] Because trademark law’s evaluation of those terms focuses exclusively on semantic meaning, it pays no attention to sound symbolism or other ways marks convey information. It therefore underestimates the true competitive effects of exclusive rights in suggestive terms.[100] Although a winery may not need to describe its Chardonnay using the term “butter,” or a caffeine shot may not have to refer to the precise period of energy it provides, being able to use those terms is enormously beneficial.[101]

Consider a mark like “Igloo” for coolers. If we focus on the needs of competitors, we might think that exclusive rights in the term do little to harm them, because they are free to adopt any number of alternative terms that might be available. But if we instead think about the competitive benefits that trademark rights provide, “Igloo” may prove quite beneficial. Consumers might think that Igloo-branded coolers keep their drinks colder than ones with a less evocative name. Igloo might be more memorable and therefore might be front of mind for consumers looking for a cooler, even compared to functionally superior coolers. Thus, competitors will have to work harder and spend more on advertising to offset the advantage of a memorable name. Or they might have to lower their prices to compete on another dimension. If any of those things are true, competitors are at a disadvantage because of the exclusive rights that trademark law bestows cheaply and easily. Courts have been quick to make the illogical leap from the fact that competitors don’t, strictly speaking, need to use the mark (which might be true) to the conclusion that allowing one firm to control the mark imposes no competitive harm. That simply does not follow.[102]

The situation with descriptive marks is worse because trademark law treats secondary meaning as both a necessary and sufficient condition for a descriptive term having net pro-competitive effects.[103] Yet even when a descriptive term acquires secondary meaning and now signifies source, exclusive rights to that term may still confer substantial competitive benefits on the firm that owns the mark. Take, for example, Milwaukee’s Best. Even though that brand name has become associated with a particular producer and acquired secondary meaning, that does not mean that it has lost its descriptive value. The existence of the trademark still prevents competitors from as easily identifying their own beer as the best that is brewed in Milwaukee. Only one company can call its beer Milwaukee’s Best, and that has ramifications for the beer market to the extent that at least some consumers identify Milwaukee with high-quality brewing.

Evidence of the fact that not all marks are equally good comes from recent data that new entrants across a number of industries are facing increasing difficulties finding high quality trademarks. A massive study from Barton Beebe and Jeanne Fromer finds evidence of both trademark “depletion” (a decreasing number of unclaimed trademarks, both in general and even within product categories) and trademark “congestion” (an increasing number of marks claimed by more than one owner in different product categories).[104] According to the authors’ analysis of PTO registration data, many of the “best” trademarks are already claimed, including standard English words, short pronounceable neologisms, and popular American surnames.[105] It’s not just that the “best” marks are claimed by someone in some industry.[106] Rather, Beebe and Fromer demonstrate depletion within product categories, where the competitive consequences of having to choose a “worse” trademark are sharpest.[107] Given the ease with which initial entrants in a market can claim the most valuable suggestive and descriptive terms, later entrants will be shunted off to marks that are less advantageous.

The consequences of congestion—firms increasingly having to accept sharing the same word mark in different markets—are less straightforward.[108] Sharing is simply a result of the limited nature of trademark rights, so congestion may not initially seem like a competition problem.[109] But in a world of ever-expanding trademark rights, where mark owners assert claims against a wider range of uses and where market organization and promotional activity make the boundaries between markets less clear, congestion means conflict. To take one example, when the team formerly known as the Cleveland Indians announced that they would become the Cleveland Guardians, people were quick to criticize them for failing to do their due diligence: there was already a professional roller derby team called the Cleveland Guardians.[110] But as the baseball team made clear, there were legal landmines everywhere: “You’re not going to find a name that someone’s not using today,” said the team president. “You’ve got to work through agreements with others.”[111]

Word mark depletion and congestion across the Abercrombie spectrum impose substantial costs on new entrants to crowded fields. New firms have to select from a competitively weaker range of options—longer or harder to pronounce words, words that are farther from the semantic core of the field, or words that are already used by other parties.[112] Those costs are most significant when trademark law grants firms exclusive rights to descriptive and suggestive terms.

III. Making Trademark Law More Pro-Competitive

A. Should Trademark Law Bar Descriptive Marks?

Our argument to this point might suggest that courts should simply revise the doctrine to ban suggestive and descriptive marks altogether.[113] Perhaps all putative mark owners should be permitted only fanciful and arbitrary marks, which do not create the same types of competition concerns.[114] The supply of those marks is not infinite, but it is much less limited than the supply of useful descriptive and suggestive terms. And arbitrary and fanciful marks generally do not convey inherent competitive advantages in the same manner as descriptive and suggestive marks.

Yet there is a consideration on the other side. Despite the many problems with descriptive and suggestive marks, those marks can have some value to consumers. It is easier for consumers to connect suggestive and descriptive marks to particular products, so those trademarks minimize the cognitive burdens consumers face. Suggestive and descriptive marks reduce the amount of work that consumers must do to find a suitable product or remember what they have purchased in the past. For instance, consumers might have an easier time remembering that Lyft is a ride-sharing company because that name is at least suggestive of the services.[115] It is useful to have a company that provides topical pain relief called IcyHot.[116] This is, of course, why firms desire these marks in the first place—and it’s why exclusive control of those terms provides competitive benefits.

In contrast, the fact that an arbitrary or fanciful mark is not well-connected to the product means that firms must expend resources on advertising to make that connection. We will describe these advertising costs in subpart III(D) below. This advertising is arguably wasteful. Every time Xfinity runs a commercial to remind consumers that it is providing wifi, it is expending resources that might be put to some more socially valuable use. If the firm were instead called “Fast Wifi Company,” it wouldn’t need to spend so many advertising dollars introducing its firm to consumers and explaining what it sells. These advantages of descriptive and suggestive marks, and those marks’ anticompetitive effects, are two sides of the same coin. It is precisely because consumers can more easily find a brand, and precisely because establishing source significance is less expensive, that descriptive and suggestive marks confer anticompetitive advantages.

Critically, however, the competition costs of suggestive and descriptive marks are not constant. Each additional mark in a given industry is more harmful to competition than the last because each additional mark swallows up more of the diminishing linguistic space. For instance, it does not necessarily create significant competitive problems if there is a single cooler company called “Igloo,” or even a second one called “Yeti.” There are still a variety of other terms available to other producers to describe their coolers: “polar,” “arctic,” and so forth. But if fifteen cooler companies snagged all of the most descriptive marks, the sixteenth firm would find itself facing much tougher sledding.[117] This is to say: as the number of descriptive and suggestive marks in a given market increases, the marginal competitive cost of each additional mark increases.

At the same time, the value to consumers of descriptive and suggestive marks decreases as the number of such marks increases. It is useful for there to be one ride-sharing service named “Lyft” so consumers can find it more easily. But once Lyft exists, it’s less important that there be a second service called “Karpool,” a third named “Ridez,” and so forth. In other words, as the number of descriptive and suggestive marks in a given market increases, the marginal benefit of each additional mark decreases. Figure 1 depicts this relationship graphically:

Figure 1: Marginal Costs and Benefits of Descriptive and Suggestive Marks

Ideally, then, policymakers would set the total number of descriptive and suggestive marks in any market at the point where the marginal cost and marginal benefit curves cross. That means that the optimal number of descriptive and suggestive marks is not zero. Thus, a complete bar on suggestive and descriptive marks would likely throw out too much baby with the bathwater. If it were possible to design a mechanism that substantially limited the number of descriptive and suggestive marks but did not completely eliminate them, that might be a better outcome. The next subpart describes such a mechanism.

B. Doctrinal and Costly Screens

As the preceding sections have made clear, trademark law’s goal should be to allow pro-competitive trademarks while eliminating—to the greatest extent possible—anticompetitive ones. That is, law should engage in screening of trademarks. This is, of course, one of the primary functions of law, a function that it performs across many domains. For example, unpatentable inventions are screened out from among the many patent applications that have been filed.[118] Invalid claims to social security disability benefits are screened from among the many claims asserted.[119] Ornamental designs that do not qualify for design patent protection are screened out from amongst the many design patent applications filed.[120] And so forth.

This screening process represents the normal functioning of law, and so it is rarely noticed or remarked upon as such. But it is critical to observe that screening can take place in two different ways. First, law can use substantive doctrinal rules to screen.[121] An invention is patentable only if it is “novel” and “nonobvious.” An individual is entitled to social security disability benefits only if they are unable to engage in “substantial gainful activity.” Etc. The legal rule sets some condition that must be met before the screen is satisfied. We refer to these types of barriers as “doctrinal screens,” in that it is legal doctrine and the decisions of legal actors (i.e., judges, administrators, clerks) that are performing the screening function.[122]

In the alternative, law can use prices to screen.[123] In order to obtain a patent, it is not enough that the applicant’s invention be novel and nonobvious—the applicant must also pay a fee.[124] For that matter, the process of patent examination itself is costly: the applicant is typically represented by an attorney or patent agent who charges thousands of dollars.[125] In order to sponsor an employee for an H-1B visa, a company must pay a substantial fee and must run costly advertisements to determine if there is an American worker who is willing to take the same job.[126] Obtaining benefits from the government typically requires filling out substantial amounts of paperwork, which generates costs for applicants.[127] Design patents require application fees as well, as does registering title to a piece of land.[128]

At first blush, perhaps these fees and expenses would seem to be merely costs to the applicant. They might be necessary (someone has to pay the lawyer!) but otherwise wasteful—to be minimized whenever possible. Yet that is not their only function. The fees and expenses also operate as screens: they will screen out some potential rights claimants who are unwilling or unable to pay them. We refer to them as “costly screens.”[129] Although they share a screening function with doctrinal screens, costly screens shift the locus of decision-making from courts and agencies to individuals. Applicants must choose whether they are willing to pay the necessary costs.

Doctrinal screens and costly screens are thus two mechanisms by which law separates and categorizes. Sometimes they function as complements; sometimes as substitutes; sometimes they work at cross purposes. In the following sections we describe their operation and explain the principles that should guide policymakers in the design of systems of screens. We then examine the role that costly screens might play in the trademark context, as well as the manner in which they actually function.

1. Categories of Rights and Behavior

Any type of action produces both costs and benefits for the person taking the action and costs and benefits for society at large.[130] The costs and benefits for the person taking the action are private costs and benefits, because they are realized by this individual. Some actions have positive private value—they produce net benefits for the individual. Extracting and selling oil that a person has found on their land generates positive private value for that individual. Other actions have negative private value—they produce net costs for the individual. Similarly, some actions have positive social value and others have negative social value. If the three co-authors of this Article were to donate all of their life savings to charity, that would generate positive social value. If the three co-authors of this Article were to form a boy band, that would generate negative social value.[131]

These two dimensions—private value and social value—operate independently.[132] Dumping toxic waste into a waterway (and getting away with it) generates positive private value and negative social value; donating all of one’s money to charity likely generates net negative private value and positive social value. We can thus think of all actions as falling somewhere within a two-by-two matrix: they produce either positive private value and positive social value, negative private value and positive social value, negative social value and positive private value, or negative private value and negative social value. Figure 2 displays this graphically.

Figure 2: Four Possible Types of Actions


Description automatically generated with medium confidence

One of the primary goals of law—if not the principal goal—is to improve human welfare.[133] Law should encourage behaviors that improve welfare for oneself or others and discourage behaviors that diminish welfare for oneself or others.[134] Accordingly, social value is what policymakers—courts, legislatures, and others—should care about.[135] The two sections that follow detail the ways in which doctrinal and costly screens can be used to separate welfare-enhancing activities from welfare-diminishing ones, permitting the former while allowing the latter.

2. Doctrinal Screens

If a principal goal of law is to increase human welfare, in an idealized world—one without information costs—law could be written to permit activities that would enhance welfare and restrict activities that would not. In the context of intellectual property, this means that law would protect trademarks (or copyrights, or patents) where doing so will increase welfare, and deny protection where it will not. Of course, legal doctrine is never written to say, “protect a trademark when doing so will increase welfare.” This would be an impossible standard for any decision-maker to adjudicate.[136]

Rather, substantive legal doctrines—and thus doctrinal screens—can operate as proxies for the ultimate quantity of interest, namely welfare. Consider patent law as an example. It is unlikely that granting a patent on an invention that is well-known or obvious will be welfare-enhancing. The patent didn’t encourage any new innovation—the invention is well-known—but granting it may create a quasi-monopoly and raise prices for consumers.[137] The patent requirements of novelty and nonobviousness thus operate as reasonable doctrinal screens, selecting against welfare-diminishing inventions.[138] Similarly, it is unlikely that giving trademark rights to a generic term (e.g., vacuum) would enhance welfare. It would have strong anticompetitive effects on other market participants while doing little to inform consumers about source.

Accordingly, a well-calibrated doctrinal screen would draw a line separating the positive social value rights in boxes 1 and 3 (from Figure 1) from the negative social value rights in boxes 2 and 4, permitting the former and blocking the latter.[139] Figure 3 displays this graphically.

Figure 3: Idealized Doctrinal Screen

Table, letter

Description automatically generated

In reality, doctrinal screens are never perfect. Novelty and nonobviousness are imperfect proxies for identifying inventions for which granting a patent will generate positive social value;[140] the fact that a worker meets the statutory requirements to qualify for an H-1B visa is an imperfect proxy for whether hiring that worker will generate positive social value.[141] Doctrinal screens often exclude some activities that would generate positive social value and permit some activities that would generate negative social value. That is, they are both overinclusive and underinclusive. Figure 4 displays these imperfections graphically.

Figure 4: Realistic Doctrinal Screen


Description automatically generated

Importantly, doctrinal screens necessarily rely on information that is available to both the private party (in this example, the patent applicant) and the decision-maker (here, the court or the patent examiner).[142] The requirement of novelty would be unworkable if either the patent applicant or the patent examiner could not determine whether the invention was novel. The applicant would never know which patent applications to file or not file; the examiner would never know which applications to reject or not reject.[143]

3. Costly Screens

As we noted above, substantive doctrinal rules are not the only mechanism for sorting activities or rights. Law sometimes also uses prices: a private party is required to pay some amount of money before obtaining a legal right or entitlement.[144] A costly screen is simply a legal rule that uses price to sort private parties. Land registration fees, patent application fees (and attorneys’ fees), fees for H-1B visa applications, the time and effort required to fill out government forms—these are all costly screens.[145] Importantly, a fee or a cost can function as a costly screen even if it was not intended to do so. Any mechanism that has the effect of using price to sort between different types of activities or rights functions as a costly screen.

Costly screens can be employed to curb harmful activities of all types. Take, for instance, the burning of natural gas to produce electricity. Burning natural gas produces harm through the emission of harmful air pollutants, not least of all carbon dioxide.[146] But burning natural gas also produces benefits by yielding electricity. In theory, it is preferable if electricity-generating plants burn natural gas only when the value of the electricity outweighs the harm from the emission of air pollution.[147] One way to manage this tradeoff is through traditional regulation: the Environmental Protection Agency could create a rule that limits the amount of electricity that can be produced using natural gas, or that allows it in only certain circumstances, and so forth.[148]

But an alternative approach would be to simply charge electricity producers a tax equal to the amount of harm they are producing when they burn natural gas. If a ton of natural gas produces air pollution that causes $20 in harm, then electricity producers could be taxed $20 per ton of natural gas they produce.[149] In theory, the electricity producer would elect to burn natural gas only when the electricity produced had a value greater than $20. The tax would thus screen out all of the negative-value uses of natural gas (where the harms outweigh the benefits), leaving only net positive-value uses (where the benefits outweigh the harms).

Costly screens operate along the dimension of private value. A costly screen reduces the private value of taking a particular action, because the party that wishes to take the action must pay for the screen. If the government imposes a fine for dumping toxic waste, that costly screen alters the private costs and benefits for the firm that might engage in dumping.[150] This is a powerful policy tool because most private parties will make decisions largely on the basis of private value, not social value. If a firm thinks that it can get away with dumping toxic waste into a waterway, it will tend to take this action notwithstanding its social impact.[151] But if the firm will be heavily fined for dumping toxic waste, then the action is more likely to generate net negative private value, and the firm is more likely to refrain from engaging in it.[152]

This means that whether a particular action has negative or positive private value can be endogenous to the existence of a costly screen. Dumping toxic waste has positive private value for a firm in the absence of a costly screen, but once the government threatens a heavy fine, it has negative private value. Costly screens can thus be used to turn some actions that would otherwise generate positive private value into actions with negative private value. The effect is to deter private parties from engaging in these actions.[153]

Instead of simply deterring the firm from engaging in the harmful activity entirely, the imposition of a costly screen could also encourage the firm to adopt other options that are less socially costly. In the toxic waste example, a fine for dumping doesn’t just deter the firm from dumping; it also encourages it to invest in technologies that might reduce the amount of waste it produces. That is, the costly screen can help channel actors’ behaviors towards less socially costly activities.

Because they operate on private value, costly screens, unlike doctrinal screens, can rely on information that may only be available to the private party.[154] The policymaker must be able to determine where to set the costly screen—how much additional cost should be added to the activity in question. This depends in part upon the potential harm caused by that activity, as in the natural gas example above. But the other side of the equation is the value to the private party of taking the action (dumping waste, etc.). That value may be entirely nontransparent and unknown to the policymaker. That is not a barrier to using a costly screen. So long as the private party knows the magnitude of the costly screen and can gauge the value that it places on the activity in question, it can act accordingly. Costly screens are thus useful in situations where the regulated party has critical information, such as information regarding their eligibility for a benefit or entitlement, that the government cannot access.[155] Figure 5 displays the operation of a costly screen graphically.

Figure 5: Costly Screen


Description automatically generated

The critical insight from this analysis is that there is no necessary correlation between social value and private value. Policymakers wish to promote positive social value activities. But a costly screen will deter relatively low private value activities by making them net negative. Therefore, the key question is: within a particular legal context, will the activities deterred by a costly screen—those that are transformed into negative private value actions—have positive or negative social value? If there are actions that have low private value but positive social value—such as donating blood, for instance—imposing a costly screen would be counterproductive.[156] The costly screen will make these into negative private-value activities and potentially eliminate them.[157] But if there are actions that have low private value and negative social value, a costly screen could have salutary effects by dampening activity.[158] That is, the question facing a policymaker is whether there are a larger number of cases that fall into box 3 or box 4 of Figure 4.

Finally, doctrinal and costly screens can be used in combination.[159] The costly screen will eliminate negative private value activities; the doctrinal screen will eliminate negative social value activities, with some degree of over- and under-inclusion. Again, it is the possibility of this under- and over-inclusion from the doctrinal screen that makes the costly screen potentially useful. If doctrinal screens were perfect, there would be no need for costly screens—the doctrinal screens would do all of the work. But doctrinal screens will inevitably allow some socially harmful activities to slip through the cracks. Accordingly, costly screens will be a valuable additional tool in the policymaker’s toolkit if the condition described in the prior paragraph is met: if the costly screen will eliminate more negative social value activities than positive social value activities. Figure 6 illustrates the two screens in combination:

Figure 6: Doctrinal and Costly Screens in Combination


Description automatically generated

The application of costly screens thus requires highly context-specific analysis. Before implementing a costly screen, the policymaker must determine whether the doctrinal screen is inadequate, and then whether the screen will affect positive or negative social value activities.[160] The answer to this latter question will itself sometimes depend on the stringency of the screen—that is, how costly it is to overcome the screen—because the decision to implement a screen is not binary. Policymakers must also consider distributional concerns, to the extent that they are relevant, when implementing costly screens. By conditioning rights or resources on parties’ willingness to pay for them, costly screens can also interact with those parties’ ability to pay. Thus, in certain situations, parties might not obtain rights or resources that they would have been willing to pay for if only they had enough money to pay for them. Charging people to vote, for example, might screen out people with weak preferences between candidates, but it would also screen out people with strong preferences but not enough money.[161] Accordingly, policymakers must be sensitive to a costly screen’s wealth effects. The next section undertakes this type of context-specific analysis with respect to trademark law.

C. Trademark Law’s Costly Screens

In Part II, we analyzed a number of trademark law’s doctrinal screens that are meant to exclude negative social value rights from accruing, and we found them wanting. But doctrinal screens aren’t the whole story. Trademark law imposes two expenses on claimants that, at first glance, might appear to be costly screens.

First, trademark law imposes a costly screen for applicants who wish to register their trademarks. Registration costs money, both in the form of government filing fees and (usually) attorneys’ fees.[162] But registration isn’t mandatory, and indeed under modern law, unregistered marks can be protected under federal law on largely the same terms as registered marks.[163] There are benefits to registration, to be sure, but registration remains distinctly less important than in patent and even copyright, where registration isn’t mandatory but one can’t file a lawsuit unless and until the copyright is registered.[164] Thus, trademark registration doesn’t create much of a costly screen.

In addition, trademark law occasionally imposes a separate costly screen that can be both mandatory and expensive: the secondary meaning requirement for marks that aren’t inherently distinctive. Establishing and proving secondary meaning can be very costly for claimants, and, depending on how high those costs are, the secondary meaning requirement could serve to exclude low-private-value descriptive marks. In this subpart, we first explain the costs and benefits of trademark registration before turning to our larger contribution: analyzing secondary meaning as a costly screen. We conclude that the requirement of secondary meaning is similarly imperfect as a costly screen because it imposes only small marginal costs on putative mark holders. In reality, the secondary meaning requirement is more of an uncertain screen than a costly one.

1. Registration

Parties that have made use of a mark may register that mark with the PTO.[165] To be registrable, the claimed mark must be capable of identifying the source of the applicant’s goods and services and must not run afoul of any of the substantive grounds for refusal articulated in § 1052 of the statute.[166] Among other things, the claimed mark must not be likely to cause confusion with a mark previously registered or in use, and it must not be descriptive of the applicant’s goods or services, unless the applicant can also prove secondary meaning.[167] The applicant must also pay registration fees, and in many cases, the fees of the attorneys who prosecute their applications. Registration fees range from $250 to $350, and attorneys’ fees can add an additional thousand dollars or more.[168] Because the statute provides the opportunity for third parties to oppose registration, applicants also sometimes have to pay the costs of opposition proceedings (which operate like federal civil litigation in most respects) to see their application through to registration.[169]

But as we noted, federal registration has never been mandatory in the United States. Indeed, under current law, unregistered marks are enforceable under federal law on largely the same terms as registered marks.[170] That’s not to say that registration doesn’t have significant benefits.[171] Most importantly, federal registration confers nationwide priority,[172] and registrations can become incontestable, a benefit that is particularly valuable for descriptive marks.[173] Those benefits are good incentives for parties to seek registration, at least of marks they know they will be using for longer periods of time. But because unregistered marks are enforceable under federal law, the costs of registration aren’t mandatory, and in many cases, parties do not pay them. Most trade dress, for example, is never registered.[174] The costs of registration thus function only very weakly as a costly screen, at least as currently constituted.[175]

2. Secondary Meaning as Costly Screen

As we explained above, for all signs that are not inherently distinctive, claimants must show that they have developed secondary meaning in order to have trademark rights.[176] But unlike the functionality rules, for example, the secondary meaning requirement isn’t only, and maybe isn’t even really, a doctrinal screen. Instead, making claimants of non-inherently distinctive signs establish and prove secondary meaning is best understood as a costly screen. Here we consider the extent of the rule’s costs and its effectiveness as screening tool.

If we are to believe the standard trademark story, the first cost associated with the secondary meaning requirement is the cost of developing secondary meaning in the first place. This is because, as we are repeatedly told, consumers are not predisposed to treat descriptive words as source indicators.[177] When consumers see a descriptive term like “Fish Fri” on batter for frying fish, or a geographic term like Nantucket Reds on trousers, the law assumes that consumers interpret the terms as merely describing the goods in a non-trademark way.[178] To the extent that this is true, a firm that wants to use a non-inherently distinctive sign must first teach consumers to treat that sign as a mark.[179] Enough consumers must learn to make the semantic switch (at least the addition) from the mark’s primary meaning—these are red trousers made in Nantucket—to the mark’s secondary meaning—these trousers are made by a particular company that is using Nantucket Reds as a trademark.[180] In most cases, this will require substantial expenditures on advertising.[181] The first time consumers see “fish fri,” they presumably won’t think those terms are doing anything other than describing the batter. But after enough exposure to that product and its advertisements, consumers may learn that “Fish Fri” is a Zatarain’s product.[182]

Once a firm believes it has developed secondary meaning for its non-inherently distinctive trademark, it must next prove that it has done so, either to the PTO or to the courts, if it wants to enforce its exclusivity. Proving secondary meaning could make the screen costlier. The best evidence of secondary meaning is direct evidence that a sufficient percentage of relevant consumers treat the mark as having acquired distinctiveness.[183] While direct evidence can come from the testimony of actual buyers,[184] many courts say that the most persuasive evidence of secondary meaning comes from consumer surveys conducted by experts.[185] Courts admit various types of surveys aimed at showing that consumers treat a term as a trademark,[186] but all of those surveys are expensive. Estimates of the cost of a professional trademark survey range from the high five figures well into six figures.[187] Accordingly, the best proof of secondary meaning is also very expensive proof.

Given this expense, most claimants attempt to prove secondary meaning via circumstantial evidence.[188] In such cases, claimants must raise an inference that a non-inherently distinctive sign has acquired distinctiveness by way of proxy.[189] Courts and the PTO consider many kinds of evidence in evaluating circumstantial proof of secondary meaning including:

1. Length and manner of use;

2. Amount and manner of advertising;

3. Amount of sales and number of customers; and

4. Recognition by the trade, media, or potential customers.[190]

Of course, obtaining credible information of these types will be at least somewhat expensive for firms trying to prove secondary meaning. But it is worth noting that many of the most relevant factors require proof that the claimant spent money on its claimed trademark. McCarthy explains courts’ inferences about advertising expenses and secondary meaning: “The seller spent a large amount of money on advertising. The larger the amount spent, the greater the exposure of buyers to this symbol as a trademark: The greater the exposure, the greater the likelihood that buyers will associate this symbol with one seller in a trademark sense.”[191]

It’s impossible to specify the amount of advertising expenditures that courts or the PTO will treat as persuasive of secondary meaning, because it is “the likely effect rather than the effort invested in such activities . . . that is determinative.”[192] That is to say, courts claim to care not just about the sum of advertising expenditures but whether those expenditures raise an inference that the use and advertising of the mark will cause consumers to treat it as source-identifying.[193] Nonetheless, all else being equal, a claimant is more likely to succeed if it spends millions of dollars on advertising than if it spends tens of thousands of dollars.

Relative to other costly screens in IP law, the secondary meaning requirement appears, at first glance, to create an extremely expensive hurdle to the acquisition of exclusive rights. Obtaining a utility patent, for example, typically costs about $35,000, including filing fees and attorneys’ fees.[194] Compare that to the hundreds of thousands or even millions of dollars that firms must spend on advertising to establish and prove secondary meaning.

But when it comes to understanding the magnitude of a costly screen, the issue is not merely the gross sum of money expended. Rather, what matters is the marginal cost added by the costly screen, above and beyond what the right-seeker would otherwise have spent. In this context specifically: how much must a firm pay to obtain trademark rights in a descriptive term, compared with what it would otherwise have spent?

This raises two problems. First, as should be obvious, firms aren’t spending large sums on advertising only in order to establish valid trademarks. They’re spending the money because they believe it will generate private value for the firm by encouraging people to purchase the firm’s products. Tiffany & Co. doesn’t spend hundreds of millions of dollars annually advertising with its blue color solely to obtain valid rights in that color;[195] it spends that money because it hopes that positive associations between the blue color and the brand will give it a competitive advantage in the market.[196] More importantly, even if there were no prospect of acquiring trademark rights in the color blue, and even if that meant Tiffany wouldn’t spend any money trying to build the positive associations specifically with that color, that doesn’t necessarily mean Tiffany would spend less on advertising. It would just advertise differently and focus on different things.

Because firms don’t treat advertising expenses as pure costs, the magnitude of the secondary meaning requirement’s costly screen is substantially lower than it seems. Unlike the filing and attorneys’ fees that a firm must pay to obtain a patent, the costs associated with advertising usually are willfully undertaken irrespective of trademark validity. Of the total that a firm spends on advertising its products, it seems likely that very little money is spent solely for purposes of establishing proof of secondary meaning. Put another way, if trademark law allowed Tiffany & Co. to establish exclusive rights in its blue color automatically with use, we do not think that Tiffany would reduce its advertising expenditures to any meaningful degree.

Second, the costs of establishing an arbitrary or fanciful mark are high as well. This matters for how well the costly screen channels parties towards less socially costly behavior. There is, of course, no statutory requirement that an arbitrary or fanciful mark have secondary meaning, so the issue is not so much connecting the mark to the firm. Instead, the point is that the owner of an arbitrary or fanciful mark will have to spend more than the proponent of a descriptive or suggestive mark to connect its mark to the product being sold.[197] That is, it will likely cost ADT more in advertising to cause consumers to identify its company with home security than it would cost SimpliSafe to achieve the same identification. This is, of course, one of the main reasons why firms often opt for descriptive and suggestive marks in the first place.

Accordingly, the true measure of the secondary meaning screen is reflected in the relationship between two costs: (1) the cost of getting consumers to treat a term as a trademark (source significance), and (2) the cost of getting consumers to associate the term with the product (product significance). The magnitude of the costly screen, then, is the difference between how much the owner of a non-inherently distinctive mark will have to pay to establish both (1) source significance and (2) product significance, and how much the owner of an inherently distinctive mark will have to pay to establish the same. Quantity (1) is likely somewhat larger for descriptive terms because of the doctrinal screen embedded in trademark law. Claiming a non-inherently distinctive mark requires expending money creating and proving secondary meaning. But quantity (2) is likely larger for arbitrary and fanciful marks, because firms need to spend money getting consumers to connect the mark to the product.

It is difficult to know for which type of mark the sum of the costs (1) + (2) will be larger. But our point is not necessarily that descriptive marks will be cheaper to obtain than arbitrary and fanciful ones. Rather, the point is that the gap in cost will often be relatively small, not large enough to establish a costly screen that will effectively eliminate negative social value/low private value marks. That is particularly true when focusing on costs that would not otherwise have been incurred but for the requirement.

IV. Aligning Trademark Validity and Competition

As currently constituted, trademark law’s combination of doctrinal and costly screens is insufficiently sensitive to competition concerns. The law’s doctrinal screen between inherently distinctive and non-inherently distinctive marks is ill-drawn and difficult to apply. Too often, firms can easily obtain trademark rights that produce anticompetitive effects. This wouldn’t be as much of a problem if the law’s costly screens—registration and secondary meaning—were better at excluding negative social value marks. But those screens are simply too inexpensive. In this Part, we offer a series of proposals for each trademark domain that would more clearly align doctrinal and costly screens with trademark law’s competition goals.

A. Screening Trademarks

The problems we have noted with word marks apply particularly to descriptive and suggestive marks. The supply of fanciful and arbitrary marks is much greater than the supply of suggestive and descriptive marks; for any given product, there will always be more invented words or words that have nothing to do with the product than there will be unused words that are associated with the product. We are also less concerned about owners of arbitrary marks gaining anticompetitive advantages by preventing others from referencing the type of product at issue using the arbitrary mark, precisely because arbitrary and fanciful marks often have no connection with the product other than source significance.[198] And so long as arbitrary marks do not resemble one another too closely—which should usually be avoidable, given the array of options available—they are unlikely to create too much clutter or consumer confusion. But descriptive and suggestive marks, by contrast, raise all of these concerns in spades. When marks group around the thing they are meant to describe, concerns of clutter, anticompetitive behavior, and consumer confusion are at their apex.

As we have suggested, one solution is to impose a meaningful costly screen. This costly screen, if appropriately applied, would serve two purposes. First, it would make it costlier for firms to acquire potentially anticompetitive marks. Again, a firm always has the option of electing an arbitrary or fanciful mark, which are less likely to be anticompetitive.[199] If a descriptive or suggestive mark becomes more costly to acquire, a firm seeking a competitive advantage will face a higher bar to acquire that advantage. This will channel more firms toward acquiring arbitrary and fanciful marks, while forcing firms that elect marks that provide competitive advantages to pay for those advantages.[200] Second, and relatedly, imposing a substantial costly screen will act as a type of congestion pricing mechanism. The number of potential mark holders who will be willing to pay the steep fees to obtain the relevant marks will be lower than the number of people who might otherwise try to claim those marks.[201] This will help address the twin concerns of congestion and depletion of word marks.[202]

As we explained above, the usefulness of a costly screen is determined by the proportion of negative social value/low private value activities to positive social value/low private value activities it will screen out. If a costly screen will screen out significant numbers of positive social value activities, compared with the number of negative social value activities it will eliminate, then it is counterproductive and will likely be welfare diminishing. If, on the other hand, the screen will largely eliminate negative social value activities, and the number of positive social value activities that will be affected is low, then the screen should be welfare enhancing. It will curb harmful actions without meaningfully reducing the number of socially beneficial ones.

There is little question that large numbers of negative social value/low private value descriptive and suggestive trademarks could proliferate, absent any sort of screen. That is, the lower-right-hand box (box 4) of Figure 1 could be heavily populated. It would include marks that primarily had informational rather than source-identifying value,[203] as well as descriptive terms that are important for competitors to use. If firms were able to protect those terms as trademarks, they could deprive competitors of the ability to adequately describe their products to consumers, which would lead to competitive advantages that distort competition to the detriment of consumers.[204] The existence of these potential harms, and the inability of doctrinal screens to adequately police this boundary, is the prima facie case for a costly screen.

What about positive social value/low private value trademarks? Would a costly screen bar many of these types of marks? The answer is that marks in this category are unlikely to exist, or at least to exist in meaningful numbers. To understand why, first consider the potential social value of such marks. As we have explained, the social value of a trademark lies in the fact that it signals source to consumers.[205] It is socially valuable for consumers to understand that a particular pair of shoes comes from Nike, or a particular refrigerator from Whirlpool, because it allows them to assess the quality of these goods based on past experiences or existing reviews before purchasing them. That is to say, the value to the consumer of identifying marks is the additional consumer surplus they will obtain from being able to confidently purchase the refrigerator they want, rather than getting some other refrigerator or having to use other information to figure out which refrigerator is which.

The key point is that much of this value is likely to be captured by the firm producing the product. If there are customers who prefer Whirlpool refrigerators and are willing to pay extra to obtain Whirlpool refrigerators, Whirlpool will understand this fact and be able to charge them accordingly. That is, some of the surplus created when a consumer who wants a Whirlpool refrigerator is able to purchase a Whirlpool refrigerator will be shared with the producer. This means that any positive social value mark should have positive private value as well—both the consumer (the “social” component of this analysis) and the firm (the “private” component) will benefit. In turn, that means that there should be vanishingly few positive social value/low private value marks. This is in contrast to many other areas of law, including copyright law, where high social value/low private value works abound.[206] And that, in turn, means that this is an area of law ripe for the imposition of a costly screen.

B. The Mechanics of a Costly Screen

As we explained in Part III, trademark law already employs two price screens that could, in theory, serve to weed out negative social value marks—registration/maintenance fees and the secondary meaning requirement. Registration and maintenance fees are voluntary and very low when they are paid, and they apply irrespective of the anticompetitive risk associated with a particular mark. Secondary meaning, while appearing to set a large costly screen, in fact, imposes relatively little marginal cost on claimants of descriptive terms. The solution is to raise the costly screens against descriptive and suggestive marks. This would have the effect of creating some distance between the costs of acquiring those marks and the cost of acquiring arbitrary and fanciful marks.

To create this necessary distance, the additional cost would need to be some multiple of the existing costs of registration and creating and proving secondary meaning—perhaps many multiples. Here, we propose alternatives to both screens that will align them more closely with competitive concerns. Our goal is to use costly screens to encourage firms to adopt arbitrary or fanciful marks when doing so will produce less social cost.[207] In addition, boosting the costly screens may help cut down on the number of marks that firms are willing to claim.[208]

As a general matter, it is preferable that costly screens are imposed in the form of fees or fines, rather than expenditures made for other purposes. The expense associated with creating secondary meaning is a social cost—once spent on advertising, the dollars cannot be recovered. That is necessary in the context of secondary meaning because, without source significance, there is little social value to trademarks. But here the goal is simply to raise costs on owners of descriptive and suggestive marks in order to deter anticompetitive behavior. Accordingly, it would be preferable if the costly screen were simply paid to the federal government to be used for other purposes, such as improved registration processes.[209] In light of this, we focus first on creating a costly screen with registration fees.

1. Registration

For registration fees to serve as a costly screen they must be far higher than current rates, which allow claimants to register a mark for a few thousand dollars.[210] Further, in order to appropriately incentivize behavior, registration fees should be calibrated by competitive harm. It should cost far more to register a descriptive or suggestive mark than an arbitrary or fanciful mark. The significantly higher fees wouldn’t simply screen out firms with low willingness (or ability) to pay these costs; they would also serve as an explicit signal that other, less socially costly marks can be had much more cheaply.[211] Further study is warranted to determine the level of registration fees that would generate optimal channeling and deterrence, especially in light of potential distributive concerns discussed below.[212]

Registration is, of course, voluntary in the U.S., so firms might simply try to avoid these screens if the prices get too high.[213] There are several solutions available to this concern. First, trademark law might try to further increase the benefits associated with registration. In particular, the law might treat arbitrary and fanciful marks as incontestable at the time of registration, rather than requiring the mark owner to wait five years before applying for incontestable status.[214] While registration gives the mark owner prima facie evidence of exclusive rights to the mark, incontestability gives the owner conclusive evidence thereof, subject to some limited defenses.[215] That benefit would be less valuable to owners of arbitrary or fanciful marks, of course, because descriptiveness challenges aren’t available against those marks by definition. But trademark law could also condition the availability of certain remedies, including injunctive relief[216] or attorneys’ fees,[217] on registration of the mark.

Another possibility is to make registration a condition of federal enforcement of trademarks.[218] That requirement could be tailored to the specific costs of descriptive and suggestive terms by making federal jurisdiction available in cases involving unregistered marks only if they were arbitrary or fanciful.[219] There would be costs to that approach, since it would make categorization of the claimed mark a jurisdictional question even though it would require factual development. But the categorization should be considerably easier when it isn’t necessary for courts to distinguish descriptive and suggestive terms—both would be excluded absent registration. And in any event, the cost of determination might be a virtue from the perspective of costly screens—claimants of potentially descriptive unregistered terms would run the risk of having their cases dismissed after having had to litigate the classification issue, which should do more to encourage registration.[220] The goal is to make it almost essential that firms register marks in order to force those firms that wish to use descriptive or suggestive marks to pay the required costly screen.[221] If enforcement were conditioned on expensive registration, claimants would be discouraged from asserting rights in marks in anticompetitive strike suits if the fees are set high enough to make such actions net negative value.

2. Tying Secondary Meaning to Registration

Making registration more valuable or a precondition to filing suit could also be a way to boost the costly screen associated with proving secondary meaning for descriptive terms. In addition, trademark authorities could apply the same sort of costly screen to suggestive terms by requiring evidence of secondary meaning for registration of those terms as well. As we’ve explained above at length, suggestive terms resemble descriptive terms much more than they do fanciful and arbitrary terms in their anticompetitive effects. They convey competitive advantages to the firms that hold them, and there is a finite supply of the best suggestive terms in any given market. Requiring secondary meaning for registration of suggestive terms would appropriately harmonize the treatment of descriptive and suggestive marks.

As we argued in Part III, in most cases, the secondary meaning requirement adds little to firms’ marginal costs. Because firms can rely on circumstantial evidence of advertising expenditures, and because they would tend to spend that money anyway, the secondary meaning requirement isn’t likely to be generating any new costs.[222] If anything, the current requirement may be increasing costs for competitors whose litigation expenses include contesting secondary meaning. If more descriptive and suggestive marks were required to be registered, however, the PTO could condition validity on better quality and more expensive proof of secondary meaning as a condition of registration, where the costs would fall only on the applicant.

For example, the PTO could require claimants of descriptive and suggestive marks to establish secondary meaning with direct survey evidence rather than with circumstantial evidence. Although survey methods are far from perfect,[223] mandating them for registration would have numerous benefits. First, requiring this evidence for registration might push the Trademark Office to help develop better methodologies with the help of non-partisan experts. If such surveys could be developed, they would almost certainly provide better evidence than the circumstantial evidence that is currently offered for secondary meaning. In addition, from the perspective of costly screening, quality doesn’t matter as much because surveys, unlike circumstantial evidence, impose meaningful marginal costs on claimants. As we mentioned, expert surveys can cost $100,000 or more.[224] The advantage of imposing these additional costs at the time of registration—rather than litigation—is that trademark defendants would not be forced to conduct their own competing surveys and expend their own resources providing evidence of the lack of secondary meaning. The costs would fall entirely on trademark owners. As defendants’ litigation costs go down, so should their willingness to settle anticompetitive lawsuits. At the same time, the downside to this approach is that the PTO would be adjudicating secondary meaning ex parte, without the benefit of the adversarial process. Poorly executed surveys and marks that only debatably have secondary meaning might be more likely to slip through.

Nonetheless, increasing the magnitude of the costly screen for descriptive and suggestive marks would generate benefits across the spectrum of marks. Obviously, by making descriptive and suggestive marks much more expensive, firms will be encouraged to choose marks that won’t face these costs. Thus, a firm considering using the term BUTTER on its Chardonnay would likely anticipate that the PTO would determine that its mark to be descriptive or suggestive, requiring it to pay a higher registration fee and expend substantial sums proving secondary meaning. Instead, it might simply choose an arbitrary or fanciful term, avoiding the higher costs and, importantly for society, imposing less competitive harm on consumers and other firms.

C. The Distributive Consequences of Costlier Screens

Substantially raising registration fees or increasing the expenses associated with proving secondary meaning will, of course, impose greater barriers on smaller and less-well-resourced firms than on wealthier, more established firms. As we mentioned in Part III, wealth effects are unavoidable consequences of using prices as a sorting mechanism. This is one of the reasons why costly screens are an imperfect solution to the limitations of doctrinal screens.

At the same time, it is important to be clear about how these sorts of wealth effects will (or will not) exacerbate existing disparities. In our unacceptably unequal society, what matters is wealth inequality among people, not corporations. The problem is not that Caterpillar is so much larger and wealthier than the Boring Company; the problem is that Elon Musk is so much wealthier than the owners, employees, and customers of both companies.[225] It would be foolish to take steps to help the Boring Company at the expense of Caterpillar, merely because the latter is fifteen times larger than the former, without assessing how this would affect the wealth and income of the people who actually own or work for those companies. Indeed, larger, publicly traded companies are often owned by thousands or millions of people of modest means (via pension plans and mutual funds), while smaller startups can be largely owned by venture capital funds and other, wealthy investors. The size of a company does not determine the wealth of its owners or employees.

Rather, it is critical that policy not disadvantage small businesses and new entrants because these types of firms are often engines for innovation and competition. Without new entrants, existing firms can become stale and acquire undue market power, to the detriment of consumers. And here, we believe that our proposal will tend to help small businesses and new entrants, especially relative to the existing rules. Our animating concern is that current trademark law makes anticompetitive behavior far too easy, limiting the ability of smaller and newer firms to enter markets. By raising the costs of anticompetitive behavior, we hope to even the playing field for new entrants. Relative to the status quo, our proposal should reduce the number of descriptive and suggestive marks in a given field, making it easier for new entrants to reach consumers and sell their goods. Currently, initial market entrants combine first-mover advantages with cheap and easy access to anticompetitive trademarks. Costly screens are an option for addressing the latter concern.

Finally, it is critical to remember that the question about wealth effects is a relative one. While it is true that costly screens will impose a particular kind of cost on would-be claimants of descriptive or suggestive terms, it is already the case that the rules for claiming descriptive terms benefit large companies that can show significant advertising expenditures and that can litigate about secondary meaning. Even if the advertising expenditures that support secondary meaning would be incurred in other forms, a doctrinal screen that emphasizes sales volume and advertising expenditures is at least as bad in terms of its wealth effects. Considering the additional litigation costs imposed on any party that wishes to contest secondary meaning, costly screens likely make the situation better, not worse, for smaller companies and new entrants.


At its core, trademark law is about competition. Trademarks exist to promote fair and effective competition by allowing firms to identify the source of their products to consumers. This enables consumers to purchase the goods they want from the firms they trust, and it creates incentives for firms to build and maintain reputations for quality, which similarly redounds to the benefit of consumers. But the law as it stands has lost track of this goal. The law makes it too easy—and too cheap—for firms to acquire trademarks that afford significant competitive advantages by their very nature. The problem is that the law focuses only on whether a trademark signifies source while ignoring whether the trademark simultaneously affords competitive advantages.

Yet when a firm acquires and asserts a descriptive or suggestive trademark, or functional or aesthetically pleasing trade dress, it gains an edge over competitors who can no longer use that term to describe their own products or that packaging to display and sell them. This advantage has nothing to do with the underlying merits of the products in question. Rather, it is driven by the fact that the government has awarded a monopoly over certain words, pictures, or packaging to one firm and not another. The situation is even worse in crowded areas of commerce where trademarks proliferate and limit the options available to everyone, including new entrants. This type of monopoly-driven power will lead to less robust competition, higher prices, and losses for consumers.

It does not have to be this way. Trademark law’s doctrinal screens are not effective at weeding out anticompetitive marks and preventing the linguistic space from being overrun by too many marks. What is needed is a more robust costly screen, one that imposes substantial costs on putative owners of suggestive and descriptive marks.

Such a costly screen would operate as a form of congestion pricing, reducing trademark overcrowding and freeing up linguistic space for competition. It would also help channel firms away from anticompetitive

marks and toward marks that do not offer competitive advantages. This more robust costly screen could be coupled with doctrinal changes that further limit firms’ ability to strategically leverage trademarks as state-sponsored monopolistic tools. Some of these changes would require legislative action; others could be implemented by the PTO. But all of them would help reorient trademark law back toward the pro-competitive role it was always meant to occupy.

  1. . We address some of trademark’s limiting doctrines below. See infra subpart II(A).
  2. . But see Barry Schwartz, The Paradox of Choice: Why More Is Less 18–20 (2004) (discussing social science evidence that more choices can, in some circumstances, lead to worse outcomes).
  3. . Or at least they should know that all devices bearing those marks come from the same source, even if they don’t know that Persimmon is the source. This is trademark law’s anonymous source rule. Graeme W. Austin, Trademarks and the Burdened Imagination, 69 Brook. L. Rev. 827, 844 n.70 (2004) (citing Manhattan Shirt Co. v. Sarnoff-Irving Hat Stores, Inc., 164 A. 246, 250 (Del. Ch. 1933), aff’d, 180 A. 928 (Del. 1934)).
  4. . Notice, by contrast, that the brand “Persimmon” causes substantially less competitive risk. Because persimmons have nothing to do with electronics, the exclusive use of the term to sell electronics imposes no real harm on competitors while generating considerable benefits for consumers and for the firm itself. See infra subpart II(A).
  5. . The electronic device category is one of the most cluttered fields of trademark registrations. See Barton Beebe & Jeanne C. Fromer, Are We Running Out of Trademarks? An Empirical Study of Trademark Depletion and Congestion, 131 Harv. L. Rev. 945, 1017 (2018) (charting the “steadily increasing congestion” of single-word trademarks in different classes of goods).
  6. . On trademarks and competition more generally, see Mark A. Lemley & Mark P. McKenna, Is Pepsi Really a Substitute for Coke? Market Definition in Antitrust and IP, 100 Geo. L.J. 2055, 2108–11 (2012).
  7. . Depending on the nature of the designs of the firm’s logo or product packaging, the same concerns might apply equally, if not more, to exclusive rights in those features.
  8. . For further elaboration of the concept of doctrinal screens, see Christopher Buccafusco, Mark A. Lemley & Jonathan S. Masur, Intelligent Design, 68 Duke L.J. 75, 87–88 (2018). For a discussion of costly screens, see Jonathan S. Masur, Costly Screens and Patent Examination, 2 J. Legal Analysis 687, 688–91 (2010). For an argument that trademark law focus more on primary meaning and much less on secondary meaning, see Jeanne C. Fromer, Against Secondary Meaning, 98 Notre Dame L. Rev. 211, 215 (2022).
  9. . See Mark P. McKenna, Property and Equity in Trademark Law, 23 Marq. Intell. Prop. L. Rev. 117, 120 (2019) (describing how parties could bring unfair competition claims against competitors who diverted customers by deceptive means even in the absence of trademark rights).
  10. . The overall social value of allowing protection of descriptive and suggestive terms is the net of the benefits to consumers of easier association and the costs of making it more difficult for others to convey similar information.
  11. . Throughout this Article, we refer to “firms” and their behaviors, because firms or other corporate entities are the most important actors in trademark regulation. We recognize, though, that individuals and other entities also use trademarks.
  12. . See Mark A. Lemley & Mark McKenna, Irrelevant Confusion, 62 Stan. L. Rev. 413, 414 (2010) (describing how trademark law promotes competitive markets by permitting accurate communication about product quality and thereby encouraging investment in product quality); William M. Landes & Richard A. Posner, Trademark Law: An Economic Perspective, 30 J.L. & Econ. 265, 269–70 (1987) (noting that trademarks reduce consumer search costs and can incentivize producers to maintain consistent quality).
  13. . This is a serious risk in some wine regions. See Christopher Buccafusco, Jonathan S. Masur & Ryan Whalen, How Many Latours Is Too Many? Measuring Brand Name Congestion in Bordeaux Wine, 16 J. Wine Econ. 419, 422–23 (2021) (discussing word reuse and overlap in Bordeaux producer names).
  14. . On the distinction between search goods and experience goods, see Phillip Nelson, Information and Consumer Behavior, 78 J. Pol. Econ. 311, 312 (1970).
  15. . Thus, for example, a producer of women’s clothing might also adopt the mark SHEEP for its goods. See Sheep, [].
  16. . See Landes & Posner, supra note 12, at 275 (“Trademarks enable the consumer to economize on a real cost because he spends less time searching to get the quality he wants.”).
  17. . See Laura A. Heymann, The Public’s Domain in Trademark Law: A First Amendment Theory of the Consumer, 43 Ga. L. Rev. 651, 654 (2009) (writing that the Supreme Court has justified trademark protections based in part on the idea that trademark law reduces the consumer’s cost of shopping and making purchasing decisions).
  18. . See, e.g., USPTO v. B.V., 140 S. Ct. 2298, 2302 (2020) (“Guarding a trademark against use by others, this Court has explained, ‘secure[s] to the owner of the mark the goodwill’ of her business and ‘protect[s] the ability of consumers to distinguish among competing producers.’” (alterations in original) (quoting Park ’N Fly, Inc. v. Dollar Park & Fly, Inc., 469 U.S. 189, 198 (1985))). Commentators have long debated which interests trademark law primarily serves. Compare Mark P. McKenna, The Normative Foundations of Trademark Law, 82 Notre Dame L. Rev. 1839, 1841 (2007) (noting trademark law’s historical focus on protecting producers from “illegitimate diversions of their trade by competitors”), with Robert G. Bone, Hunting Goodwill: A History of the Concept of Goodwill in Trademark Law, 86 B.U. L. Rev. 547, 555 (2006) (“The major focus of trademark law is protecting the source identification and information transmission function of marks.”). At trademark law’s core, however, consumer and competitor interests will tend to converge.
  19. . See McKenna, supra note 18, at 1864–65 (describing how courts viewed injuries to producers as a central interest of trademark law (citing Boardman v. Meriden Britannia Co., 35 Conn. 402, 414 (1868))).
  20. . Sellers that age their wine could, of course, say so on their labels (describing their wine as “oak aged”), but consumers wouldn’t be able to use the trademark as a shortcut for those characteristics and couldn’t know exactly what characteristics “oak aged” referred to without tasting the wine. And they may not even know that the characteristics they liked were due to the oak aging; they just liked the SHEEP wine.
  21. . See S. Rep. No. 79-1333, at 3 (1946) (stating that trademark statutes aim to “protect the public so it may be confident that, in purchasing a product bearing a particular trade-mark which it favorably knows, it will get the product which it asks for and wants to get”);, 140 S. Ct. at 2302 (“We have recognized that federal trademark protection, supplementing state law, ‘supports the free flow of commerce’ and ‘foster[s] competition.’” (alteration in original) (quoting Matal v. Tam, 582 U.S. 218, 225 (2017))).
  22. . See Madeline Puckette, Everything You Want to Know About Orange Wine, Wine Folly, [] (stating that orange wine is “made by leaving the grape skins and seeds in contact with the juice”).
  23. . In this case, it couldn’t obtain a patent since the methods for making orange wine have been known for centuries. Id.
  24. . See Robert C. Denicola, Freedom to Copy, 108 Yale L.J. 1661, 1663 (1999) (describing the argument that trademark protection beyond prevention of confusion would be an “unjustified intrusion[] into the competitive process”).
  25. . See Landes & Posner, supra note 12, at 290 (“[I]f one producer is allowed to appropriate the word that describes” an attribute that interests buyers, that producer “will obtain rents measured by the higher price [it] receives for [its] brand” because of the high price for rival producers to inform customers of similar attributes of those brands); see also Stacey L. Dogan & Mark A. Lemley, A Search-Costs Theory of Limiting Doctrines in Trademark Law, 97 Trademark Rep. 1223, 1242 (2007) (“When a term has come to signify a class of goods, competitors have the right to explain what they are selling, even when their use of the generic term clearly piggybacks on the efforts of the party that first introduced the product.”).
  26. . This is one of the challenges posed by geographic indication laws that prohibit use of certain geographic terms even when those terms are treated as generic for certain categories of goods. See Kal Raustiala & Stephen R. Munzer, The Global Struggle over Geographic Indications, 18 Eur. J. Int’l L. 337, 362 (2007) (critiquing absolute protection of geographic terms because it can create confusion); Justin Hughes, Champagne, Feta, and Bourbon: The Spirited Debate About Geographical Indications, 58 Hastings L.J. 299, 351 (2006) (arguing that there would be significant confusion if makers of Parmigiano-Reggiano cheese could control global use of the term “Parmesan”).
  27. . See Landes & Posner, supra note 12, at 290 (describing the dangers of permitting trademarks in generic shapes).
  28. . See infra notes 90–101 and accompanying text.
  29. . See infra notes 104–11and accompanying text.
  30. . Barton Beebe, The Semiotic Analysis of Trademark Law, 51 UCLA L. Rev. 621, 669–70 (2004). That distinctive terms qualify as trademarks doesn’t necessarily mean they are protected irrespective of their use. Courts and especially the PTO also insist that the mark be used in a way that it functions as a mark, though as Alex Roberts has detailed at length, the relationship between failure to function and distinctiveness is underdeveloped, and application of the doctrine is often ad hoc. Alexandra J. Roberts, Trademark Failure to Function, 104 Iowa L. Rev. 1977, 1981–82, 1987 (2019). For an exploration of the relationship between distinctiveness and use as a mark in the design context, see Mark A. Lemley & Mark P. McKenna, Trademark Spaces and Trademark Law’s Secret Step Zero, 75 Stan. L. Rev. 1, 5–7 (2023).
  31. . Saying that distinctiveness is both necessary and sufficient for trademark status is not to say that distinctiveness is sufficient for registration. Section 2 of the Lanham Act details the bases on which marks can be denied registration, and those bases include but are not limited to distinctiveness. Lanham Act, Pub. L. No. 79-489, § 2, 60 Stat. 427, 428–29 (1946) (codified as amended at 15 U.S.C. § 1052). The fact that registration can be refused on other grounds does not undermine our point, because registration is not necessary for trademark protection. There are benefits to registration, but unregistered marks are enforceable under section 43(a) of the Lanham Act on largely the same terms. See Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 768 (1992) (“[I]t is common ground that § 43(a) protects qualifying unregistered trademarks and that the general principles qualifying a mark for registration under § 2 of the Lanham Act are for the most part applicable in determining whether an unregistered mark is entitled to protection under § 43(a).”).
  32. . See Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4, 9 (2d Cir. 1976) (“Arrayed in an ascending order which roughly reflects their eligibility to trademark status and the degree of protection accorded, these classes are (1) generic, (2) descriptive, (3) suggestive, and (4) arbitrary or fanciful.”); Two Pesos, Inc., 505 U.S. at 768 (approving of the Abercrombie spectrum as the “classic formulation” of distinctiveness).
  33. . Abercrombie, 537 F.2d at 9.
  34. . Two Pesos, Inc., 505 U.S. at 768; Wal-Mart Stores, Inc. v. Samara Bros., Inc., 529 U.S. 205, 210–11 (2000) (citing Abercrombie, 537 F.2d at 10–11).
  35. . See Qualitex Co. v. Jacobson Prods. Co., 514 U.S. 159, 162–63 (1995) (contrasting color with “fanciful,” “arbitrary,” or “suggestive” words or designs, which “almost automatically tell a customer that they refer to a brand”); id. at 163 (“The imaginary word ‘Suntost,’ or the words ‘Suntost Marmalade,’ on a jar of orange jam immediately would signal a brand or a product ‘source’; the jam’s orange color does not do so.”); Two Pesos, Inc., 505 U.S. at 768 (stating that “because their intrinsic nature serves to identify a particular source of a product,” suggestive, arbitrary, and fanciful marks “are deemed inherently distinctive and are entitled to protection”).
  36. . It does not matter whether consumers know that an APPLE computer is manufactured by the Apple Computer Company, only that they believe all computers that bear the word APPLE come from the same source. See Austin, supra note 3, at 844 n.70 (citing Manhattan Shirt Co. v. Sarnoff-Irving Hat Stores Inc., 164 A. 246, 250 (Del. Ch. 1933), aff’d, 180 A. 928 (Del. 1934)) (describing the anonymous source rule).
  37. . Two Pesos, Inc., 505 U.S. at 768; Wal-Mart Stores, Inc., 529 U.S. at 210–11.
  38. . E.g., Wal-Mart Stores, Inc., 529 U.S. at 212.
  39. . 2 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 11:4 (5th ed. 2023).
  40. . E.g., Zobmondo Ent., LLC v. Falls Media, LLC, 602 F.3d 1108, 1117 (9th Cir. 2010); see also McCarthy, supra note 39, § 11:68 (describing an inverse relationship between the amount of consumer imagination needed to associate a mark with a producer and the need of competitors to use the mark descriptively).
  41. . See, e.g., Goldsmith Silver Co. v. Savage, 229 F. 623, 627 (1st Cir. 1915). Distinguishing between trademark infringement and unfair competition, the First Circuit observed:

    In the case of infringement of a technical trade-mark the intention of the infringer is immaterial, as the essence of the wrong lies in the injury to a property right; while in the case of unfair competition the intention is material, to establish fraud on the part of the defendant in the use of the imitative device to beguile the public into buying his goods as those of his rival.

    Id.; see also, e.g., Scriven v. North, 134 F. 366, 375 (4th Cir. 1904) (“There are certain elements of property right in a technical trade-mark . . . .”); Church & Dwight Co. v. Russ, 99 F. 276, 279 (C.C.D. Ind. 1900) (“It is commonly said that there is a right of property in a technical trade-mark, and an infringement of it is spoken of as a violation of a property right.”). For a more thorough description of the traditional relationship between trademark and unfair competition and its evolution over time, see generally McKenna, supra note 9.

  42. . Landes & Posner, supra note 12, at 289 (“There are 450,000 words in Webster’s Third New International Dictionary, and although they are not freely substitutable if one is trying to say something that will be understood, they are freely substitutable if one is uninterested in meaning.”).
  43. . McCarthy, supra note 39, § 11:16.
  44. . See Wal-Mart Stores, Inc. v. Samara Bros., Inc., 529 U.S. 205, 213 (2000) (“And where it is not reasonable to assume consumer predisposition to take an affixed word or packaging as indication of source—where, for example, the affixed word is descriptive of the product (‘Tasty’ bread) or of a geographic origin (‘Georgia’ peaches)—inherent distinctiveness will not be found.”). In reality, consumer understanding of that term probably depends significantly on how and where the term appears on a package. See Thomas R. Lee, Eric D. DeRosia & Glenn L. Christensen, An Empirical and Consumer Psychology Analysis of Trademark Distinctiveness, 41 Ariz. St. L.J. 1033, 1038 (2009) (finding that non-lexical cues, such as location, can cause consumers to associate descriptive marks with source). But distinctiveness, as opposed to failure to function, considers the meaning of terms in relation to goods and (usually) not based on particular use context.
  45. . McCarthy, supra note 39, § 14:1.
  46. . McCarthy, supra note 39, § 11:18. See also Gimix, Inc. v. JS & A Grp., Inc., 699 F.2d 901, 907 (7th Cir. 1983) (stating that allowing descriptive terms to be protected as trademarks would cause “elements of the language [to] be monopolized in such a way as to impoverish others’ ability to communicate”); Educ. Dev. Corp. v. Econ. Co., 562 F.2d 26, 28 (10th Cir. 1977) (“Merely descriptive terms may not be registered because they do not advise the buyer that the product comes from a single source and because trademark protection would infringe on common speech.”); W. E. Bassett Co. v. Revlon, Inc., 354 F.2d 868, 871 (2d Cir. 1966) (“[S]ince there are only so many words that can accurately describe a type of product, the law does not favor trademark monopolization of such descriptive terms.”); Telechron, Inc. v. Telicon Corp., 198 F.2d 903, 906 (3d Cir. 1952) (stating that trademarks in descriptive terms create “the danger of depleting the general vocabulary available to all for description and denomination of articles of commerce”).
  47. . Those terms used to be excluded from trademark subject matter. Parties claiming that their competitors were misusing descriptive or geographic terms (or surnames) in order to pass off their goods as those of the claimant could seek relief under the common law of unfair competition. McKenna, supra note 9, at 120. Unfair competition remedies were more limited—the claimant typically could not get an injunction against use of the descriptive or geographic term altogether, but only limitations on the manner of the defendant’s use. Id. at 121.
  48. . 15 U.S.C. § 1052(e)–(f); Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 769 (1992); Park ’N Fly, Inc. v. Dollar Park & Fly, Inc., 469 U.S. 189, 194 (1985). The owners of registered marks that are not yet incontestable can also face descriptiveness challenges, though the registration affords the registrant a presumption of validity, putting the burden on the defendant to show descriptiveness. 15 U.S.C. § 1057(b); Park ’N Fly, Inc., 469 U.S. at 196.
  49. . For that reason, “acquired distinctiveness” might be a better term.
  50. . E.g., Yankee Candle Co. v. Bridgewater Candle Co., 259 F.3d 25, 43 (1st Cir. 2001); Nola Spice Designs, L.L.C. v. Haydel Enters., Inc., 783 F.3d 527, 546 (5th Cir. 2015); Herman Miller, Inc. v. Palazzetti Imps. & Exps., Inc., 270 F.3d 298, 315 (6th Cir. 2001); see McCarthy, supra note 39, § 15:30 (describing accepted forms of direct and circumstantial evidence of secondary meaning).
  51. . Zatarains, Inc. v. Oak Grove Smokehouse, Inc., 698 F.2d 786, 795 (5th Cir. 1983).
  52. . Id.; see also McCarthy, supra note 39, § 15:48 (“The easiest and least expensive manner of proving secondary meaning is to introduce evidence of the amount and nature of advertising of the mark; the length of time the mark has been in use; and the amount of goods or services sold under the mark.”).
  53. . McCarthy supra note 39, § 15:48.
  54. . USPTO v. B.V., 140 S. Ct. 2298, 2303 (2020) (“The name of the good itself (e.g., ‘wine’) is incapable of ‘distinguish[ing] [one producer’s goods] from the goods of others’ and is therefore ineligible for registration. Indeed, generic terms are ordinarily ineligible for protection as trademarks at all.”) (alterations in original) (citations omitted) (quoting 15 U.S.C. § 1052)). McCarthy, supra note 39, § 12:1 (“Of course, to properly be called an unprotectable ‘generic name’ in trademark law, the designation must be the name of the same product or service which it is alleged to identify the source of.”).
  55. . 15 U.S.C. § 1064(3) (“The primary significance of the registered mark to the relevant public rather than purchaser motivation shall be the test for determining whether the registered mark has become the generic name of goods or services on or in connection with which it has been used.”).
  56. . USPTO v. B.V., 140 S. Ct. 2298 (2020).
  57. . Id. at 2304 (“A generic term is one that refers to the genus of which the particular product is a species.” (quoting Park ’N Fly, Inc. v. Dollar Park & Fly, Inc., 469 U.S. 189, 194 (1985))).
  58. . Id.
  59. . See id. (noting that protection depends on capacity to distinguish source); Am. Cyanamid Corp. v. Connaught Lab’ys, Inc., 800 F.2d 306, 308 (2d Cir. 1986) (“Trademark protection benefits consumers by enabling them to select products on the basis of their origin. . . . Consumers will not benefit, however, if trademark law prevents competitors from using generic or descriptive terms to inform the public of the nature of their product.”).
  60. . See, e.g., A.J. Canfield Co. v. Honickman, 808 F.2d 291, 305 (3d Cir. 1986) (“[T]he law grants a monopoly over a phrase only if . . . the grant of such a monopoly will not substantially disadvantage competitors by preventing them from describing the nature of their goods.”); Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4, 10 (2d Cir. 1976) (stating that to protect a generic name would be to “confer a monopoly” on one seller in the sale of the named product); see also Deven R. Desai & Sandra L. Rierson, Confronting the Genericism Conundrum, 28 Cardozo L. Rev. 1789, 1819 (2007) (“Modern courts refuse to extend trademark protection to generic words or terms because they are concerned that doing so would negatively impact competition.”).
  61. . See, e.g., King-Seeley Thermos Co. v. Aladdin Indus., Inc., 321 F.2d 577, 579–80 (2d Cir. 1963) (finding “thermos” to be a generic term for vacuum-insulated containers); DuPont Cellophane Co. v. Waxed Prods. Co., 85 F.2d 75, 79–81 (2d Cir. 1936) (finding “cellophane” to be a generic term for transparent cellulose film); Bayer Co. v. United Drug Co., 272 F. 505, 511–12 (S.D.N.Y. 1921) (finding “aspirin” to be a generic term for acetyl salicylic acid).
  62. . Almost never. There are small number of cases in which formerly generic terms have been reclaimed as trademarks, but those are decidedly the exception to the rule. McCarthy, supra note 39, § 12:30.
  63. . Complaint for Federal Trademark Infringement; False Designation of Origin; Common Law Trademark Infringement; and Unfair Competition at 2, Cruise LLC v. Ford Motor Co., No. 3:21-CV-05685 (N.D. Cal. July 23, 2021). According to the complaint, a subsidiary of GM acquired Cruise LLC in 2016, and in 2018 it transitioned to a majority-owned subsidiary of GM. Id. at 6–7.
  64. . Id. at 2.
  65. . Response to Office Action, U.S. Trademark Application Serial No. 87-010961 (Jan. 23, 2017),‌&linkId=35 [].
  66. . See CRUISE, Reg. No. 6,008,158 (registering the CRUISE mark with GM listed as owner).
  67. . McCarthy, supra note 39, § 11:71 (noting the role of the “psychological bent of the observer” in both classifications).
  68. . Id. § 11:72.
  69. . Blake Brittain, Chipotle Sues Sweetgreen for Trademark Infringement over ‘Chipotle Chicken’ Bowl, Reuters (Apr. 6, 2023, 5:44 PM), [
  70. . This is often called the “anti-dissection” rule. McCarthy, supra note 39, § 11:27 (“Under the anti-dissection rule, a composite mark is tested for its validity and distinctiveness by looking at it as a whole, rather than dissecting it into its component parts.”).
  71. . Indeed, courts sometimes struggle even with short terms. In a recent case, the owner of the VAGISIL mark, which it registered for various vaginal-health products, was able to prevent registration of VAGISAN for various similar pharmaceutical preparations. Combe Inc. v. Dr. August Wolff GmbH & Co. KG Arzneimittel, 851 Fed. Appx. 357, 357–58 (4th Cir. 2021). It did so despite the obvious genericness of the “vagi” component of the marks. Id. at 364. Indeed, the court specifically rejected the idea that greater emphasis should be placed on the other parts of the respective marks (“SIL” and “SAN”) for purposes of evaluating similarity. Id. According to the Fourth Circuit, courts should not focus on dominant portions of single word marks; for those marks, “the proper analysis compare[s] whole words, not parts.” Id. (alteration in original) (citation omitted). Comparing the whole words at issue, the court believed the district court correctly concluded that the marks were quite similar—emphasizing the comparable, generic components. Id.
  72. . Qualitex Co. v. Jacobson Prods. Co., 514 U.S. 159, 163 (1995) (quoting Inwood Lab’ys, Inc. v. Ives Lab’ys, Inc., 456 U.S. 844, 851 n.11 (1982) (cleaned up).
  73. . McCarthy, supra note 39, § 15:6.
  74. . Id.
  75. . Id.
  76. . Id.
  77. . In another part of his treatise focused on surveys, McCarthy says that “[t]here would seem to be no logical reason to require any higher percentage to prove secondary meaning than to prove a likelihood of confusion.” 6 McCarthy, supra note 39, § 32:190. In that view, there are no reasons to refuse to recognize trademark rights if 15% of people might be confused.
  78. . McCarthy says: “It is not necessary that each and every member of the buyer class associate the mark with a single source. Nor is it necessary that a majority of that group do so. It is only necessary that a ‘substantial part’ of the buying class make such an association.” McCarthy, supra note 39, § 15:45.
  79. . There probably isn’t a great methodology available. The most widely used methodology asks respondents whether they associate the test term with one or more than one producer and then asks respondents who say “one producer” to identify that producer. See Susan Schwartz McDonald, Secondary Meaning Surveys, in Trademark and Deceptive Advertising Surveys 79, 96–97 (Shari Seidman Diamond & Jerre B. Swann eds., 2d ed. 2022) (discussing the “source-identification question”). That method asks respondents a question they might never have considered on their own (“do you associate this with one or more than one producer”), and it cannot differentiate responses that reflect understanding of a term as a trademark from those that simply reflect the fact that respondents may be most familiar with that producer (what trademark law once called “de facto secondary meaning”).
  80. . McCarthy, supra note 39, § 15:45. Some literature suggests that association should be at least 30% to be considered meaningful, but it is completely unclear what the legal basis is for that threshold. See, e.g., McDonald, supra note 79, at 102–03 (discussing threshold levels above 30% accepted by courts).
  81. . 15 U.S.C. § 1115(b)(4).
  82. . William McGeveran, Rethinking Trademark Fair Use, 94 Iowa L. Rev. 49, 85 (2008).
  83. . Id.; William McGeveran & Mark P. McKenna, Confusion Isn’t Everything, 89 Notre Dame L. Rev. 253, 282–83 (2013).
  84. . See McGeveran & McKenna, supra note 83, at 257 (stating that even if a mark usage would be descriptive fair use, the user still would face “the prospect of long and expensive litigation”).
  85. . See Deborah R. Gerhardt, A Masterclass in Trademark’s Descriptive Fair Use Defense, 52 Akron L. Rev. 739, 748–49 (2018) (“While the senior user’s mark may both describe the product and indicate source, the junior user’s fair use defense depends on the ability to prove that the use is descriptive but not a trademark use that signals source.”).
  86. . Id. Notice, also, that even if Chipotle’s trademark in its restaurant name is suggestive for Mexican-style restaurants, its products and its competitors’ products will often contain chipotle peppers, so enforcement of that trademark limits some valuable forms of competition. We discuss these issues further below.
  87. . See supra notes 37–39 and accompanying text.
  88. . BUTTER, Registration No. 3,999,253.
  89. . White wines that undergo malolactic fermentation, which converts malic acid into lactic acid, may have flavors that people describe as “buttery.” F. Sauvageot & P. Vivier, Effects of Malolactic Fermentation on Sensory Properties of Four Burgundy Wines, 48 Am. J. Enology & Viticulture 187, 187, 190–91 (1997).
  90. . See Maria Dinzeo, Jury Will Decide Winemaker Clash over ‘Buttery’ Chardonnay, Courthouse News Serv. (Apr. 20, 2020), [] (noting that a chardonnay called “Rich & Buttery” has “been a hit, particularly with female wine drinkers”).
  91. . Although not necessarily for the authors of this Article.
  92. . Innovation Ventures, LLC v. N.V.E., Inc., 694 F.3d 723, 730 (6th Cir. 2012); Jake Linford, The False Dichotomy Between Suggestive and Descriptive Trademarks, 76 Ohio St. L.J. 1367, 1380–81 (2015).
  93. . See Glynn S. Lunney, Jr., Trademark Monopolies, 48 Emory L.J. 367, 434–35 (1999) (describing the competitive losses that result from trademark protection).
  94. . See Linford, supra note 92, at 1415 (“For a suggestive mark, the metaphoric connection links mark and product, not mark and source”).
  95. . Landes & Posner, supra note 12, at 289.
  96. . E.g., Zobmondo Enter., LLC v. Falls Media, LLC, 602 F.3d 1108, 1117 (9th Cir. 2010).
  97. . Stephen L. Carter, The Trouble with Trademark, 99 Yale L.J. 759, 770 (1990) (“The immediate difficulty with any theory premised on the IM assumption, however, is that virtually no one involved in the selection and testing of marks seems to think that IM presents a useful picture of the world.”).
  98. . Jake Linford, Are Trademarks Ever Fanciful?, 105 Geo. L.J. 731, 756 (2017) (“Research into sound symbolism has established connections between the sounds that comprise a word and the meanings that the audience will ascribe to the word. . . . It is therefore problematic for courts to assume a fanciful mark bears no inherent meaning.”).
  99. . See Beebe & Fromer, supra note 5, at 966 (“VIAGRA calls to mind, all at once, ‘vigor,’ ‘vitality,’ ‘aggression,’ and ‘Niagara’ (suggesting both water and honeymoons).”).
  100. . See Linford, supra note 98, at 758 (noting that competitors may try to invoke product qualities through the use of similar sounds).
  101. . Linford, supra note 92, at 1385.
  102. . To be more precise, when courts focus on competitive need, they tend to set a high threshold and do not recognize need unless there are very few alternatives available to competitors. Because they ignore important benefits, they misjudge the number of real alternatives and too readily conclude that the benefits to the mark claimant don’t come at the expense of others.
  103. . See supra notes 47–53 and accompanying text.
  104. . Beebe & Fromer, supra note 5, at 951.
  105. . Id. at 951–52.
  106. . Coexistence by users in different markets is a celebrated reflection of the limitations of trademark rights, and there are many well-known examples of the same mark being used by multiple parties for different goods or services (Delta Airlines and Delta Faucets, Apple Computers and Apple Vacations, etc.).
  107. . Beebe & Fromer, supra note 5, at 951; see also Zahr K. Said, Craft Beer and the Rising Tide Effect: An Empirical Study of Sharing and Collaboration Among Seattle’s Craft Breweries, 23 Lewis & Clark L. Rev. 355, 387 & n.171 (2019) (illustrating the “increasing pressure to find viable names” within the craft beer field).
  108. . See Beebe & Fromer, supra note 5, at 952–53 (“Parallel uses may not confuse consumers as to source, but each use destroys the uniqueness and blurs the distinctiveness of the other, particularly for newer entrants.”).
  109. . See supra note 106.
  110. . Avery Williams, Cleveland Guardians Roller Derby Team Files Trademark Application, 19 News (July 31, 2021, 3:00 PM), [].
  111. . Terry Pluto, The Inside Story of How Cleveland Indians Became Cleveland Guardians – Terry Pluto, (July 24, 2021, 8:27 P.M.),
    2021/07/inside-story-of-how-cleveland-indians-became-the-cleveland-guardians-terry-pluto.html []. The “others” included Marvel, with which the team reportedly worked something out. Id. Trademark issues loomed large in consideration of several other names. The team ruled out the Spiders because it’s the nickname of the University of Richmond. Id. (“But a key factor was the University of Richmond has the Spiders nickname. That created branding issues. It could be expensive.”). They also ruled out Rockers in part for legal reasons. Id. (“The Guardians, Spiders and Rockers seem to be the first three names rumored. The Rockers were the name of the former WNBA franchise in Cleveland. Then there are the Colorado Rockies in the National League. All that was a negative.”).
  112. . See Beebe & Fromer, supra note 5, at 952–53 (reporting that applicants are increasingly registering “suboptimal” marks, including more complex words and those used by firms in other sectors).
  113. . “Ban” here might mean denying trademark status to descriptive and suggestive terms but allowing for some more limited unfair competition remedies. That would resemble the system we once had, which distinguished technical trademarks and other signs. See McKenna, supra note 9, at 119–20 (describing the historical concept of “technical trademarks”).
  114. . See Fromer, supra note 8, at 249–50 (arguing for greater emphasis on primary meaning).
  115. . See Daniel J. Hemel & Lisa Larrimore Ouellette, Trademark Law Pluralism, 88 U. Chi. L. Rev. 1025, 1044–45 (2021) (discussing the concept of linguistic distance with respect to trademarks).
  116. . See id. at 1028 (describing “distance costs” to consumers and their relation to trademarks).
  117. . Thus, we are substantially less concerned about congestion in the search engine context, where there are only a few market competitors, than in industries with many competitors, such as clothing, electronics, or brewing.
  118. . Masur, supra note 8, at 700.
  119. . Donald O. Parsons, Self-Screening in Targeted Public Transfer Programs, 99 J. Pol. Econ. 859, 860 (1991).
  120. . Buccafusco et al., supra note 8, at 104–05.
  121. . Id. at 88.
  122. . Id.
  123. . Cf. Michael Spence, Job Market Signaling, 87 Q.J. Econ. 355, 361–63 (1973) (exploring how expected wages can result in individuals selecting education costs to maximize personal benefit).
  124. . See Masur, supra note 8, at 699 n.22 (listing the fees associated with patent prosecution).
  125. . Id. at 699–700.
  126. . Id. at 725–26.
  127. . Claus Thustrup Kreiner & Torben Tranæs, Optimal Workfare with Voluntary and Involuntary Unemployment, 107 Scandinavian J. Econ. 459, 459–60 (2005).
  128. . See Buccafusco et al., supra note 8, at 106 (listing the fees charged by the PTO for a design patent).
  129. . See generally Joseph E. Stiglitz & Andrew Weiss, Sorting Out the Differences Between Signaling and Screening Models, in 1 Selected Works of Joseph E. Stiglitz 232 (2009) (expounding on the concepts of “screening” and “signaling” economic models).
  130. . See David Fagundes & Jonathan S. Masur, Costly Intellectual Property, 65 Vand. L. Rev. 677, 683 (2012) (exploring the ideas of private value and social value in the manufacturing context).
  131. . See Christopher Buccafusco & Jonathan S. Masur, Drugs, Patents, and Well-Being, 98 Wash. U. L. Rev. 1403, 1430–31 (2021) (arguing that spending money on low social value pursuits results in less available investment for high social value pursuits). One of us (McKenna) disputes this claim, or attributes the negative social value to the other two. Indeed, it is possible that one of us has performed as a member of a boy band, at least at a law school talent show.
  132. . Most approaches to considering social value would count the positive private value that an individual gets from an action even if the net effects of the action are negative. See Matthew D. Adler & Eric A. Posner, New Foundations of Cost-Benefit Analysis 12 (2006) (“[W]elfare economics assumes that a person is better off when his preferences are respected, and that society is better off when its members are better off.”).
  133. . See Christopher Buccafusco & Jonathan S. Masur, Intellectual Property Law and the Promotion of Welfare, in 1 Research Handbook on the Economics of Intellectual Property Law 98, 99 (Ben Depoorter & Peter S. Menell eds., 2019) (stating that the “principal consequentialist goal of IP laws should be to maximize social welfare”); John Bronsteen, Christopher Buccafusco & Jonathan S. Masur, Well-Being Analysis vs. Cost-Benefit Analysis, 62 Duke L.J. 1603, 1607–08 (2013) (proposing well-being analysis as a way of answering the critical question of “whether a law will make people’s actual experience of life better or worse on the whole”).
  134. . These aren’t the only goals that law could or does have. Intellectual property law, including trademark law, is generally understood to exist to solve particular economic problems that relate to consumer and producer welfare. See supra Part I. But trademark law might also care about issues of fairness or distributional concerns beyond welfare maximization. We address some of these issues below.
  135. . On the relationship between welfare and social policy generally see Matthew D. Adler & Marc Fleurbaey, Introduction, in The Oxford Handbook of Well-Being and Public Policy 1 (Matthew Adler & Marc Fleurbaey eds., 2016).
  136. . See Buccafusco & Masur, supra note 133, at 104 (outlining criticism of objectivist conceptions of welfare).
  137. . See W. Nicholson Price II, Essay, The Cost of Novelty, 120 Colum. L. Rev. 769, 782–83 (2020) (“[T]he public should suffer the deadweight loss of monopoly only in exchange for information it did not have before.”).
  138. . Buccafusco et al., supra note 8, at 95.
  139. . Id. at 90.
  140. . Id. at 91.
  141. . See H-1B Specialty Occupations, DOD Cooperative Research and Development Project Workers, and Fashion Models, U.S. Citizenship & Immigr. Servs. (Sept. 15, 2023), https:// [] (listing H-1B visa requirements).
  142. . Cf. Stiglitz & Weiss, supra note 129, at 233–34 (describing the consequences for markets with asymmetrical information when an informed or an uninformed party moves first).
  143. . On the quality of granted patents, see Michael D. Frakes & Melissa F. Wasserman, Does the U.S. Patent and Trademark Office Grant Too Many Bad Patents?: Evidence from a Quasi-Experiment, 67 Stan. L. Rev. 613, 618–19, 671 (2015).
  144. . Albert L. Nichols & Richard J. Zeckhauser, Targeting Transfers Through Restrictions on Recipients, 72 Am. Econ. Rev. (Papers & Proc.) 372, 376–77 (1982).
  145. . See, e.g., Andrew Weiss, Human Capital vs. Signalling Explanations of Wages, 9 J. Econ. Persps., Fall 1995, at 133, 134 n.3 (“The relationship between wages and education could be the outcome of either students choosing an education program to signal their ability, or students choosing education levels in response to the relative wage offers of firms, in which case wages would serve to screen workers.”).
  146. . See Steven Ferrey, Unforced Errors, Legal Fulcrum & International Climate, 20 Minn. J.L. Sci. & Tech. 115, 128–29 (2019) (describing the benefits and harms of natural gas as an energy source).
  147. . Jonathan S. Masur & Eric A. Posner, Toward a Pigouvian State, 164 U. Penn. L. Rev. 93, 95 (2015).
  148. . For an example of this type of regulation, though not directed in particular at natural gas emissions, see generally National Emission Standards for Hazardous Air Pollutants From Coal- and Oil-Fired Electric Utility Steam Generating Units, 77 Fed. Reg. 9304 (Feb. 16, 2012) (to be codified at 40 C.F.R. pts. 60, 63).
  149. . This is known as a Pigouvian tax: a tax on a harmful activity, set at a level equivalent to the harm produced by that activity. Masur & Posner, supra note 147, at 95.
  150. . See id. at 100–01 (explaining how a Pigouvian tax can lead to socially optimal production).
  151. . But see Michael E. Porter & Mark R. Kramer, Strategy & Society: The Link Between Competitive Advantage and Corporate Social Responsibility, Harv. Bus. Rev., Dec. 2006, at 78, 78 (asserting that activists and the media pressure companies into socially responsible behavior).
  152. . Masur & Posner, supra note 147, at 101.
  153. . E.g., Vivi Alatas, Abhijit Banerjee, Rema Hanna, Benjamin A. Olken, Ririn Purnamasari & Matthew Wai-Poi, Self-Targeting: Evidence from a Field Experiment in Indonesia, 124 J. Pol. Econ. 371, 372 (2016).
  154. . Joseph E. Stiglitz, The Theory of “Screening,” Education, and the Distribution of Income, 65 Am. Econ. Rev. 283, 287 (1975). As an example in the education context, Professor Stiglitz argues,

    [t]here are some conditions under which even the most able may not be willing to pay for [a] “general screening.” . . . [namely, situations where] individuals are perfectly certain of their ability, and . . . it is possible for their ability to be costlessly observed “on the job[.]” [T]hen the individual would offer to absorb all the risk involved in hiring and training costs. There are obviously instances of this sort, individuals who persuade the employer to hire them at low wages until they can “prove themselves.”

    Id. By “prove themselves,” Professor Stiglitz means these individuals undergo an on-the-job substantive test to objectively and credibly demonstrate their worth.

  155. . See, e.g., Henrik Jacobsen Kleven & Wojciech Kopczuk, Transfer Program Complexity and the Take-Up of Social Benefits, 3 Am. Econ. J. Econ. Pol’y, Feb. 2011, at 54, 59–62 (exploring how complexity impacts benefit program applicant choices when they rely on private information).
  156. . See Fagundes & Masur, supra note 130, at 721 (arguing that a hypothetical $22,000 costly screen for copyrights would lead to a situation where “authors will typically not create the associated work, regardless of whether that outcome is bad for the public”).
  157. . Sometimes even very low costs, like opting out of a default rule, can meaningfully undermine pro-social behavior. See Eric J. Johnson & Daniel Goldstein, Do Defaults Save Lives?, 302 Sci. 1338, 1339 (2003) (finding that organ donation rates increase substantially when donation is the default because of “the cost of changing from the default”).
  158. . See Hugo Benítez-Silva, Moshe Buchinsky, Hiu Man Chan, Sofia Cheidvasser & John Rust, How Large Is the Bias in Self-Reported Disability?, 19 J. Applied Econometrics 649, 652 (2004) (estimating that in the US Social Security context the award error rate is about 20%); Jean-Yves Duclos, Modelling the Take-up of State Support, 58 J. Pub. Econ. 391, 412 (1995) (finding that in the U.K. Supplementary Benefits scheme, providing means-tested cash benefits to the poor, the award error rate was 18.8% and the rejection error rate was 18.1%).
  159. . Buccafusco et al., supra note 8, at 79.
  160. . Masur, supra note 8, at 690–91.
  161. . For a discussion of the effects of income and education on voter turnout, see Raymond E. Wolfinger & Steven J. Rosenstone, Who Votes? 23–26 (1980).
  162. . How Much Does a Trademark Cost?, Richards Pat. L., https://www.richardspatentlaw
    .com/faq/how-much-does-a-trademark-cost/ [].
  163. . See Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 768 (1992) (stating that it is “common ground” that unregistered marks are protected).
  164. . According to the Copyright Act of 1976, a copyright lawsuit cannot be brought until “registration of the copyright claim has been made in accordance with this title.” Copyright Act of 1976, Pub. L. No. 94-553, § 411(a), 90 Stat. 2541, 2583 (1976) (codified as amended at 17 U.S.C. § 411(a)).
  165. . 15 U.S.C. § 1051(a). Parties can also apply to register on the basis of a bona fide intent to use the mark, but those marks will not be registered unless and until the applicant can demonstrate that it has actually made use of the mark. Id. § 1051(b), (d).
  166. . For these grounds for refusal, see id. § 1052.
  167. . Id. § 1052(d)–(f).
  168. . How Much Does a Trademark Cost?, supra note 162 (mentioning $1,500–$2,000 in fees for relatively complex registration issues).
  169. . See 15 U.S.C. § 1063(a) (authorizing third parties to oppose registration); 37 C.F.R. § 2.116(a) (adopting the Federal Rules of Civil Procedure for inter partes proceedings). Registrations can also be cancelled on petition of third parties. 15 U.S.C. § 1064. The grounds for cancellation become more limited after five years, but some grounds (including genericness and functionality but not including descriptiveness) remain forever. Id. § 1065.
  170. . See 15 U.S.C. § 1125(a) (creating civil action for enforcement of unregistered marks). That has not always been true. Every federal trademark statute prior to the Lanham Act made enforcement in federal court contingent on federal registration. Act of March 19, 1920, Pub. L. No. 66-163, § 4, 41 Stat. 533, 534 (creating a cause of action for infringement of registered marks), repealed by Lanham Act, Pub. L. No. 79-489, § 46(a), 60 Stat. 427, 444–45 (1946); Trade-Mark Act of 1905, Pub. L. No. 58-84, § 16, 33 Stat. 724, 728 (same), repealed by Lanham Act, Pub. L. No. 79-489, § 46(a), 60 Stat. 427, 444–45 (1946); see also id. § 17, 33 Stat. at 728–29 (granting federal courts jurisdiction in cases involving use of a registered mark in interstate commerce).
  171. . See, e.g., B&B Hardware, Inc. v. Hargis Indus., Inc., 575 U.S. 138, 142 (2015) (“Registration is significant.”).
  172. . Under the Lanham Act as originally passed, registration served as constructive notice and constructive nationwide use. Lanham Act, Pub. L. No. 79-489, § 22, 60 Stat. 427, 435 (1946) (codified at 15 U.S.C. § 1072) (“Registration of a mark on the principal register provided by this Act . . . shall be constructive notice of the registrant’s claim of ownership thereof.”). As of November 16, 1989, the effective date of the Trademark Law Revision Act, contingent on registration, the application date serves as the constructive notice and constructive use date. Trademark Law Revision Act of 1988, Pub. L. No. 100-667, sec. 109, § 7, 102 Stat. 3935, 3938–39 (codified at 15 U.S.C. § 1057).
  173. . 15 U.S.C. § 1065. Incontestable marks cannot be challenged on certain grounds, including descriptiveness. Park ’N Fly, Inc. v. Dollar Park & Fly, Inc., 469 U.S. 189, 196 (1985).
  174. . Rebecca Tushnet, Registering Disagreement: Registration in Modern American Trademark Law, 130 Harv. L. Rev. 867, 889 (2017).
  175. . The incentive to register may be increasing now, not because of any change in the trademark system itself, but because Amazon’s Brand Registry program requires registration. Ironically, the Brand Registry might particularly increase the incentive to game the system by registering descriptive terms in slightly stylized form, because Amazon only matches text. For a description of the incentives created by Amazon’s Brand Registry, see generally Jeanne C. Fromer & Mark P. McKenna, Amazon’s Quiet Overhaul of the Trademark System (Sept. 11, 2023) (unpublished manuscript) (on file with authors).
  176. . See supra subpart II(A).
  177. . E.g., Wal-Mart Stores, Inc. v. Samara Bros., Inc., 529 U.S. 205, 213 (2000); see also, e.g., Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 769 (1992) (“Marks which are merely descriptive of a product are not inherently distinctive. When used to describe a product, they do not inherently identify a particular source, and hence cannot be protected.”).
  178. . Wal-Mart Stores, Inc., 529 U.S. at 213.
  179. . See Lisa P. Ramsey, Descriptive Trademarks and the First Amendment, 70 Tenn. L. Rev. 1095, 1119–20 (2003) (outlining the factors courts use to evaluate secondary meaning of a descriptive mark).
  180. . See Nantucket Reds Collection, Murray’s Toggery Shop, https://www.nantucketreds.
    com/collections/nantucket-reds-collection/ [].
  181. . In the words of the Seventh Circuit:

    The spread of an advertisement among people is like ever-spreading ripples from a pebble thrown into still water. The ripples go out and out in an ever-increasing circle from a common center, long after the pebble is lost to sight . . . . Throwing pebbles into water is child’s play, but knowledge of a trade-mark, through advertising and as carried by the people, is an important, valuable business asset, gained at much expense.

    Northam Warren Corp. v. Universal Cosm. Co., 18 F.2d 774, 775 (7th Cir. 1927).

  182. . Zatarains, Inc. v. Oak Grove Smokehouse, Inc., 698 F.2d 786, 795–96 (5th Cir. 1983).
  183. . See McCarthy, supra note 39, § 15.30 (discussing direct and circumstantial evidence of secondary meaning).
  184. . Id.
  185. . E.g., Vision Sports, Inc. v. Melville Corp., 888 F.2d 609, 615 (9th Cir. 1989).
  186. . 6 McCarthy, supra note 39, § 32:191; Irina D. Manta, In Search of Validity: A New Model for the Content and Procedural Treatment of Trademark Infringement Surveys, 24 Cardozo Arts & Ent. L.J. 1027, 1036 (2007).
  187. . Lisa Larrimore Ouellette, The Google Shortcut to Trademark Law, 102 Calif. L. Rev. 351, 361 & n.53 (2014) (citing Robert H. Thornburg, Trademark Survey Evidence: Review of Current Trends in the Ninth Circuit, 21 Santa Clara Comput. & High Tech. L.J. 715, 717 (2005)).
  188. . See Barton Beebe, An Empirical Study of the Multifactor Tests for Trademark Infringement, 94 Calif. L. Rev. 1581, 1641 (2006) (finding that only 20% of litigated cases addressed survey evidence).
  189. . McCarthy, supra note 39, § 15:48.
  190. . Id. § 15:30. Manner of advertising includes the existence of “look for” advertising with trade dress. Id.
  191. . Id. § 15:51.
  192. . Restatement (Third) of Unfair Competition § 13 cmt. e (Am. L. Inst. 1995); see also Alyssa Morrison, The Advertising Factor in the Secondary Meaning Instruction, 19 J. Contemp. Legal Issues 296, 300 (2010) (recognizing that courts have found that “even very high advertising expenditures over a long period are insufficient to support a showing of secondary meaning”).
  193. . For various courts’ treatment of advertising expenditures as evidence of secondary meaning, see McCarthy, supra note 39, § 15:52.
  194. . Buccafusco et al., supra note 8, at 117.
  195. . See Julia Faria, Tiffany & Co.’s Advertising Spend Worldwide from 2015 to 2019 (in Million U.S. Dollars), Statista (Jan. 12, 2023), [] (reporting that Tiffany & Co. spent $378.8 million on advertising in 2019).
  196. . Ann Bartow, The True Colors of Trademark Law: Greenlighting a Red Tide of Anti Competition Blues, 97 Ky. L.J. 263, 267–68 (2008).
  197. . See Hemel & Ouellette, supra note 115, at 1040–41 (explaining the concept of distance costs).
  198. . As we discussed in Part II, there are marks that the law treats as arbitrary but, because of sound significance, operate as suggestive marks (e.g., Swiffer and Viagra). See supra note 99 and accompanying text. We would treat those as suggestive marks.
  199. . Though, as we noted earlier, those marks can have clear advantages when taking into account sound symbolism and non-semantic meaning. See supra notes 94–101 and accompanying text.
  200. . In particular, the costly screen will use firms’ private information about the tradeoff between competitively harmful suggestive and descriptive marks and competitively benign arbitrary and fanciful marks.
  201. . See Beebe & Fromer, supra note 5, at 953 (suggesting “various schemes of congestion or peak pricing with respect to application, maintenance, and renewal fees” to compel registrants to internalize more of their own costs).
  202. . Id. at 950–51, 953.
  203. . See Roberts, supra note 30, at 1992 (discussing the distinction between informational use of marks and source-identifying use).
  204. . See supra notes 93–97 and accompanying text.
  205. . See supra notes 15–21 and accompanying text.
  206. . See Fagundes & Masur, supra note 130, at 720–21 (discussing how authors produce numerous works that may generate significant but diffuse social value that they cannot fully and easily capture).
  207. . As Hemel and Ouellette note, forcing firms to choose marks far from what they call the “semantic core” can also impose costs on society, as firms must spend more money to create associations and because consumers may struggle to recognize unusual marks. Hemel & Ouellette, supra note 115, at 1041. One major benefit of a costly screen is that it relies on firms’ private knowledge about the costs and benefits of choosing marks that are further from the semantic core.
  208. . Sometimes firms may desire the opportunity to claim exclusive rights in additional marks, but those additional marks may create little consumer value. Consider the situation with craft beer names. Yuengling Brewing may want to claim the term “golden” as a brand name for its pilsner style beer. But if Yuengling only makes one pilsner style beer, the additional name isn’t giving consumers any additional information, but it is affecting the ability of other breweries to describe their beers. See Said, supra note 107, at 392 (recounting a controversy between two breweries over the term “IPA”).
  209. . On the role of user fees at the PTO, see generally Michael D. Frakes & Melissa F. Wasserman, The Failed Promise of User Fees: Empirical Evidence from the U.S. Patent and Trademark Office, 11 J. Empirical Legal Stud. 602 (2014).
  210. . See How Much Does a Trademark Cost?, supra note 162 (describing the costs of trademark registration).
  211. . The U.S. is allowed to substantially raise its registration fees and remain in compliance with its international obligations. Article 6 of the Paris Convention gives each country the right to determine the conditions for filing and registering trademarks in its domestic legislation. Appellate Body Report, United States—Section 211 Omnibus Appropriations Act of 1998, ¶ 132, WTO Doc. WT/DS176/AB/R (adopted Jan. 2, 2002). According to the World Trade Organization, “an applicant who chooses to seek registration of a trademark in a particular foreign country under Article 6 must comply with the conditions for filing and registration specified in that country’s legislation.” Id. ¶ 133.
  212. . See infra section IV(A)(3).
  213. . In some respects, of course, that is our goal. We want firms to instead choose arbitrary and fanciful marks rather than suggestive or descriptive ones.
  214. . Under current law, a registrant can seek incontestability by filing an affidavit “within one year after the expiration of any such five-year period setting forth those goods or services stated in the registration on or in connection with which such mark has been in continuous use for such five consecutive years and is still in use in commerce,” 15 U.S.C. § 1065(3), and meeting the other requirements of 15 U.S.C. § 1065.
  215. . 15 U.S.C. § 1115. For example, a defendant can still challenge an incontestable mark as functional and thus invalid. Id. § 1115(b)(8).
  216. . Injunctive relief is standard in trademark infringement cases irrespective of registration. See Mark A. Lemley, Did eBay Irreparably Injure Trademark Law?, 92 Notre Dame L. Rev. 1795, 1797 (2017) (“Trademark cases similarly presumed that an injunction was the appropriate remedy in the large majority of cases.”).
  217. . Attorneys’ fees are currently available in “exceptional” cases under the Lanham Act. Christopher P. Bussert, Interpreting the “Exceptional Cases” Provision of Section 1117(a) of the Lanham Act: When an Award of Attorney’s Fees Is Appropriate, 92 Trademark Rep. 1118, 1118 (2002). Typically, this means willful or intentionally deceptive infringement. Id. at 1119. We could relax that requirement to treat all infringements as exceptional if we wanted to further incentivize registration.
  218. . That was the way things worked before courts reinterpreted the Lanham Act to find unregistered marks enforceable under section 43(a). Mark P. McKenna & Brittany Von Rueden, Registration and Federalization: 75 Years of the Lanham Act, 39 Cardozo Arts & Ent. L.J. 987, 996 (2021). Claimants still might bring lawsuits alleging violations of state unfair competition laws, but those will be much less valuable than federal litigation.
  219. . This approach is not dissimilar to copyright law’s treatment of registration, where registration is necessary to file a federal lawsuit even though a copyright is valid from the moment of fixation. See 17 U.S.C. § 411(a) (“[N]o civil action for infringement of the copyright in any Unites States work shall be instituted until preregistration or registration of the copyright claim has been made.”); see also Fourth Est. Pub. Benefit Corp. v., LLC, 139 S. Ct. 881, 887–88 (2019) (“Before pursuing an infringement claim in court . . . a copyright claimant generally must comply with § 411(a)’s [registration] requirement.”).
  220. . And registration has substantial social benefits generally because it improves competitors’ notice of claimed marks.
  221. . A different approach would require firms to pay a huge fee before filing a federal lawsuit on descriptive or suggestive terms. There might still be an adjudication challenge, however, and we would lose the notice benefits that registration provides.
  222. . See supra subpart III(C).
  223. . See supra note 79. For an argument that survey methods are imperfect in the dilution context, see Barton Beebe, Roy Germano, Christopher Jon Sprigman & Joel H. Steckel, Testing for Trademark Dilution in Court and the Lab, 86 U. Chi. L. Rev. 611, 619–20 (2019).
  224. . Ouellette, supra note 187, at 361 & n.53.
  225. . Compare Matt McFarland, Elon Musk’s The Boring Company Now Valued at Nearly $5.7 Billion, CNN Business (Apr. 22, 2022, 9:31 AM), [] (reporting the Boring Company’s valuation at $5.7 billion), with Noah Bolton, 3 Reliable Manufacturing Stocks for Steady Growth, InvestorPlace (Aug. 23, 2023, 8:30 AM), [] (reporting Caterpillar’s market cap of $140 billion).