Data-Driven Constitutional Avoidance

I.     Introduction

Epitomizing Oliver Wendell Holmes’s truism “[g]reat cases like hard cases make bad law”1 is the U.S. Supreme Court’s decision in Matal v. Tam.2 Tam isn’t “great” because it involves especially momentous questions of public policy. To the contrary: It centers on the quest of an obscure Asian-American rock band to secure trademark protection for its name, “The Slants.” Rather, Tam is a great case because, to quote Holmes, it “appeals to the feelings and distorts the judgment.”3 The government refused to trademark the band’s name on the theory that the term “slants” was insulting to Asian Americans and therefore would violate a federal statutory bar on the registration of “disparaging” trademarks (the “disparagement bar”).4 The government ignored the band’s rationale for wanting the trademark: not to disparage Asian Americans for their “slanted” eyes but to transform the slur into a badge of pride—just as some gays, feminists, and blacks have attempted to reappropriate5 derogatory labels (such as “queer,” “bitch,” and even the N-word).6

If Tam is a “great case,” it also made “bad law.” In striking down the federal ban on disparaging trademarks, the Court invoked the most powerful taboo in First Amendment speech law: the nearly absolute rule against restricting speech because of its viewpoint.7 Taking this step had the dual and unnecessary effects of expanding the First Amendment’s scope to invalidate a long-standing federal economic regulation and of broadening, by elaboration, the rule against viewpoint-based discrimination.

We say “unnecessary” in the face of empirical evidence we developed in an experiment focused on the Tam case. From the data, we learned that Americans (Asians and non-Asians alike) construe the term “Slants” differently depending on the context: They are more likely to believe a band’s reappropriation motives when the band is Asian than when the band is not Asian. Put another way, had The Slants been a white band, Americans would have concluded that the band was out to disparage Asians, but Americans can recognize when an Asian band deploys “Slants” to help Asians.

The natural conclusion from these data is that the bar on disparaging trademarks simply lacked internal coherence. Had the Justices understood this, they would have seen an alternative, less constitutionally momentous ground for striking down the bar: vagueness. For two reasons, such a holding would have transformed Tam from a “great case” exemplifying judicial overreaching into a great decision promoting judicial modesty. First, declaring the disparagement bar unconstitutionally vague would not have pushed the boundaries of substantive First Amendment law into an uncharted area of economic regulation. Second, grounding the decision in vagueness would have eliminated the need to elaborate on the difficult concept of viewpoint discrimination. A vagueness holding in Tam would have exemplified constitutional avoidance: minimizing a decision’s constitutional impact rather than making important new constitutional law.

These conclusions pertain to Tam, but the lessons of our study are larger. Constitutional avoidance axiomatically serves to promote judicial modesty. Conversely, courts’ forays into factual analysis often register as judicial overreaching. Our Tam study complicates that dichotomy by showing how courts can achieve constitutional avoidance by grounding their decisions in empirical data. In cases that arguably present novel or important constitutional issues, reviewing courts might productively encourage litigants to develop and present empirical data relevant to the dispute, with the goal of finding grounds for decision that don’t require substantial constitutional lawmaking. By expanding the grounds for the decision and encouraging fact-specific adjudication, empirical analysis can help to avoid arguably unnecessary judicial action and to encourage informed, authoritative legislative policymaking.

Part II of this Article introduces the Tam decision as a case study. We set forth the First Amendment stakes in the case and then explain the doctrinal importance of the Court’s decision. Part III describes our study of how different people perceive the term “Slants” and shows the diversity and complexity of responses the term elicits. Part IV explains how our data could have supported a narrower, vagueness-based legal analysis in Tam and suggests how similar routes from data to avoidance might help resolve future constitutional disputes.

II.     The First Amendment Precipice of Matal v. Tam

The Slants are an Asian-American rock band led by Simon Tam.8 According to Tam, the band chose its name to reappropriate a term often understood in other contexts as derogatory to Asians and people of Asian descent.9 Tam applied with the U.S. Patent and Trademark Office (“PTO”) to register the band name as a federal trademark. Federal registration is not necessary to establish and legally protect a trademark, but registration carries important degrees of legal advantage for the trademark holder.

The PTO denied Tam’s application, citing a long-standing provision in § 2(a) of the Lanham Act that permits the PTO to deny registration of trademarks that “may disparage . . . persons, living or dead, institutions, beliefs, or national symbols, or bring them into contempt, or disrepute.”10 To decide whether a trademark was “disparaging,” the PTO first assessed “the likely meaning of the matter in question,” informed by “the relationship of the matter to the other elements in the mark, the nature of the goods or services, and the manner in which the mark is used in the marketplace in connection with the goods or services.”11 The PTO then asked “whether that meaning may be disparaging to a substantial composite of the referenced group.”12 Applying that standard, the PTO determined that the name “Slants” disparaged Asians and Asian–Americans. The PTO based its conclusion on dictionary definitions of “Slants” as a derogatory term, as well as objections that bloggers, commenters, and a performance venue had expressed about the band’s name.

A.     The Constitutional Stakes

Matal v. Tam is an important First Amendment case for several reasons. The immediate legal holding about registration of disparaging trademarks may affect other cases, like applications to register sports team names and logos thought to demean Native Americans.13 Tam also resonates strongly with other aspects of Lanham Act § 2(a), most obviously the bar on registering “scandalous” trademarks.14 More broadly, Tam implicates two crucial and fundamental problems in First Amendment law: the extent of the amendment’s coverage, particularly its reach into the realm of economic regulation, and the force of the prohibition on viewpoint-based speech regulations.

1.     Covering “Uncovered Speech” in the Economic Realm

Part of the impact of Tam comes from the case’s very novelty and obscurity. Intellectual property represents a largely new frontier for First Amendment law. Even in the copyright field, which deals entirely with the creation and use of expressive material, the Supreme Court has declared a First Amendment-free zone. Copyright protection, in the Court’s view, provides an essential economic incentive for creators to express themselves. At the same time, the Court acknowledges that enforcement of copyrights restricts communications that seek to use copyrighted works. The Court has answered these concerns with faint assurances that doctrines built into the federal Copyright Act—the allowance for “fair use” of copyrighted materials and the restriction of copyright protection to “expressions” rather than “ideas”—adequately represent the interests the First Amendment protects elsewhere.15 Trademarks convey information, but their function, in contrast to expressive material protected by copyright, is intrinsically and entirely commercial. Prior to Tam, the Supreme Court had not applied the First Amendment to the trademark field.16

Frederick Schauer has identified a broad range of speech “uncovered” by the First Amendment.17 As distinct from speech categories like fighting words18 and obscenity,19 which the Court has explicitly excluded from First Amendment protection, uncovered speech has avoided the attention of First Amendment law altogether. Examples of uncovered speech include representations in corporate reports subject to securities laws, certain antitrust violations, and various restrictions on employees’ and unions’ speech in labor disputes.20 Uncovered varieties of speech tend to share two characteristics: limited relevance for public debate and regulation by well-established, broadly accepted government regimes.21 Schauer ascribes the noncoverage phenomenon to a continuous political and cultural discourse about the “constitutional salience” of different forms and instances of speech.22 That discourse ultimately shapes the development of legal doctrine. Opportunistic lawyers seek out openings for expanding First Amendment doctrine that will serve their clients’ interests.23 They press those openings until courts either reject or embrace the plea for expanded coverage.24

Schauer posits that litigants, mostly businesses, are pushing ever more aggressively to impose First Amendment limits on previously uncovered types of speech.25 The Roberts Court has already shown those efforts great favor.26 The Court has extended First Amendment coverage to commercial transactions that do not implicate the flow of information for consumers,27 to public sector unions’ collection and expenditure of nonmember “agency fees” for nonpolitical purposes,28 and to several previously unproblematic types of campaign finance regulation.29 In two cases involving businesses that sell expressive materials, the Court has restricted the doctrinal basis for formally recognizing new categories of speech as not protected.30 The Court’s gestures toward toughening the core First Amendment restriction on regulating the content of speech31 could generally ease the process of expanding the First Amendment to formerly noncovered speech.

Extending First Amendment coverage to further precincts of economic regulation could entail the most significant expansion of free speech law’s scope since the Court in the 1970s extended full First Amendment protection to political campaign spending32 and limited protection to commercial advertising.33 Beyond the immediate significance of expanded coverage, Schauer warns that broadening First Amendment law to protect previously uncovered speech can distort established doctrine by putting new strains on doctrinal mechanisms.34 The most serious concern about the current calls for expansion goes to the endlessly contentious balance of constitutional norms against government power. Imposing broad First Amendment rights on the commercial realm would threaten to revive the Lochner era’s infamous constitutional constraints on the government’s power to regulate economic matters in the public interest.35 Indeed, Justice Breyer has warned that the Roberts Court’s free speech jurisprudence is already flirting with Lochnerism.36

Trademark law prior to Tam resisted First Amendment coverage for reasons embedded in the logic and norms of economic regulation. A few trademark scholars have raised First Amendment concerns about § 2(a).37 A more substantial body of trademark scholarship, responding to the Tam litigation, contends that subjecting § 2(a) to First Amendment constraints would overlook and undermine essential functions and features of trademark law.38 Rebecca Tushnet, for example, acknowledges that trademark registration is undeniably a form of speech regulation, but she maintains that imposing serious First Amendment scrutiny on the disparagement bar and the other bars in Lanham Act § 2(a) would threaten the entire system of trademark registration.39 The Lanham Act regulates the use of trademarks so that consumers can identify products and distinguish among commercial offerings. Subjecting that regulatory scheme to First Amendment scrutiny, Tushnet argues, would seriously threaten the entire regulatory scheme save restrictions on deceptive marks.40 Tushnet echoes Schauer by situating free speech limits on trademark registration in a broader movement toward “misguided application of tough First Amendment scrutiny to the modern regulatory state.”41 The Supreme Court’s imposition in Tam of First Amendment restrictions on trademark registration both changes trademark law and casts a long constitutional shadow over commercial regulation in general.

2.     Content and Viewpoint Discrimination

The cornerstone principle of First Amendment speech law is a strong presumption against the government’s power to regulate speech based on its content. A corollary to that principle is the even stronger presumption against the government’s power to regulate speech based on its viewpoint. Usually the force of the content principle makes the viewpoint corollary an afterthought. Both content and viewpoint regulations generally prompt strict First Amendment scrutiny, which a law can survive only if the government can show the law is the least restrictive means to achieve a compelling interest. However, the Supreme Court has identified two circumstances in which the government may regulate based on content but not viewpoint: government properties that amount to “nonpublic forums,”42 and commercial advertising.43 Because trademarks arguably fall into the commercial advertising category,44 the viewpoint corollary mattered for Tam.

First Amendment scholars have disputed the conceptual integrity of the strong presumption against content discrimination.45 Two recent Supreme Court decisions underscore both the centrality and the difficulty of the content principle. In Reed v. Town of Gilbert, a city ordinance imposed dramatically different limits on signs with different kinds of messages.46 The Supreme Court unanimously struck down the ordinance as impermissible content discrimination. The majority reiterated and firmly enforced the strong First Amendment rule against content regulation.47 Concurring Justices, however, warned that the content principle’s lack of nuance might force the Court to strike down common-sense preferential treatment for informational signs that provide public benefits.48 Town of Gilbert shows how the content discrimination bar in general remains complex and fraught. In McCullen v. Coakley, the Court struck down a state law that created a “buffer zone” barring foot traffic around facilities that provide abortion services, on the ground that the state had drawn the law too broadly.49 The majority, however, found that the law did not target anti-abortion speech and was therefore content neutral.50 The concurring Justices strongly disagreed with that finding. In their view, a law that effectively regulates speech on a single topic must be guilty of content-based discrimination.51 McCullen shows how the mechanics of the content discrimination bar in particular cases remain uncertain and contestable.

Scholars have criticized the corollary presumption against viewpoint discrimination as conceptually incoherent and practically ineffectual.52 The basic problem is one of baselines: How, across different cases, can we distinguish a subject of discussion (“content”) from a distinct perspective on the subject (“viewpoint”)? It’s a tricky problem because of the different levels of generality at which speech can operate. A concept (say, democracy) may amount to content in one setting (a debate between Democratic and Republican political candidates) but animate a viewpoint in a different setting (an argument about the relative merits of different political systems). The content-viewpoint distinction has vexed the Supreme Court. In one case, five Justices struck down a university’s denial of student activities funds to religious publications as viewpoint discrimination.53 The four dissenters, though, argued that evenhanded treatment of religious perspectives could not possibly discriminate against any viewpoint, but merely regulated content.54 In another case, the majority asserted that an ordinance understood to ban certain bigoted “fighting words” discriminated based on viewpoint, because it unfairly advantaged “those arguing in favor of . . . tolerance and equality.”55 A concurring Justice disagreed, calling the ordinance viewpoint neutral because it restricted bigoted speech no matter what person or group the speech targeted.56 The Court has recently complicated the problem even further by introducing a concept of “speaker-based” discrimination that resembles, but holds itself out as distinct from, viewpoint discrimination.57

The legal battle over the Lanham Act disparagement bar presented the problem of content and viewpoint discrimination in the socially and politically combustible context of racially derogatory speech.58 The most prominent and contentious disparagement battle has involved the effort by the Washington, D.C. professional football franchise to register its name, which many have criticized as a slur against Native Americans.59 The Patent and Trademark Office’s rejection of the Slants’ application triggered the same sort of dispute. A defining axiom of U.S. free speech law, especially in comparison to our closest liberal democratic neighbors, is that the First Amendment does not blink even at the most appalling racist invective.60 However, First Amendment doctrine on “hate speech” reflects significant ambivalence and uncertainty. The Supreme Court held in the 1950s that the First Amendment did not protect “group libel,” invective directed at a particular race but not a particular person.61 Twenty years later, in the celebrated case of a Nazi group that sought to rally in a town home to many Holocaust survivors, a lower federal court suggested that the Supreme Court’s allowance for regulating “group libel” likely had fallen into illegitimacy.62

The Court’s most emphatic decision about racist speech came in R.A.V. v. St. Paul.63 The defendant in that case had burned a cross in an African-American family’s yard. He was convicted of violating a local ordinance that punished the display of any symbol that “arouses anger, alarm or resentment in others” based upon race or certain other immutable characteristics.64 Justice Scalia’s majority opinion presumed that the ordinance reached only “fighting words,” a speech category unprotected by the First Amendment.65 Justice Scalia also validated a government strategy of selecting the most inflammatory fighting words for special legal attention.66 The Court nonetheless struck down the ordinance for singling out bigoted fighting words rather than restricting fighting words more generally.67 Despite R.A.V., the Court has found no impermissible content or viewpoint discrimination when states enhance penalties for crimes motivated by racial animus68 or ban specifically racist cross-burnings in circumstances where a reasonable person would view the cross-burning as conveying a threat of violence.69 This line of racist speech cases, into which Tam fits, spotlights both the importance and the indeterminacy of the First Amendment content and viewpoint doctrines.

B.     The Supreme Court’s Tam Decision

A mostly harmonious but formally divided Supreme Court in Tam held that the Lanham Act’s disparagement bar violated the First Amendment by discriminating against particular viewpoints. The decision has a very curious structure. Justice Alito’s partial majority opinion states, in one sentence of its introductory section, the central conclusion that the disparagement bar discriminated against speech based on viewpoint.70 Justice Alito then describes the case’s facts and procedural posture (Part I)71 and rejects a statutory argument that could have obviated the First Amendment issue (Part II).72 The most substantial portion of Justice Alito’s opinion, Part III, refutes the government’s efforts to characterize registered trademarks or trademark registration as something other than private speech: government speech (Part III.A),73 a government subsidy (Part III.B),74 or a “government-program” that straddles or blends the government speech and subsidy categories (Section III.C).75 Likewise, in Part IV, Justice Alito finds irrelevant the question whether registered trademarks are commercial speech.76 Only in that commercial speech discussion does Justice Alito briefly elaborate the core holding on viewpoint discrimination, and all he does there is reject as viewpoint-based the government’s attempted justifications for the bar.77 Justice Alito garnered only four votes (including his own) for Sections III.B and III.C and Part IV, limiting most of the opinion’s legal analysis to mere plurality status.78

Justice Kennedy’s partial concurring opinion, joined by Justices Ginsburg, Sotomayor, and Kagan, says more than Justice Alito’s opinion about the Court’s viewpoint discrimination holding. Justice Kennedy states the legal test for viewpoint discrimination as “whether—within the relevant subject category—the government has singled out a subset of messages for disfavor based on the views expressed.”79 He goes on to reject the government’s argument that the disparagement bar was viewpoint neutral because it equally barred any sort of disparaging trademark. That argument, says Justice Kennedy, ignores the central sin of the disparagement bar: “mandating positivity,” which “might silence dissent and distort the marketplace of ideas.”80 Likewise, Justice Kennedy rejects the government’s argument that the disparagement bar was constitutional because it turned on the potential audience reaction to trademarked names, not the government’s subjective distaste for the speaker’s message or motives.81 Finally, Justice Kennedy chides the government for even construing “The Slants” as “a negative comment.”82 This is some of the Supreme Court’s most focused explication of the viewpoint discrimination doctrine, although the discussion’s relegation to a concurrence denies it the full force of precedent. Justice Kennedy echoes Justice Alito by characterizing trademark registration as a government effort to encourage the expression of diverse viewpoints rather than promote the government’s own viewpoint.83

Despite the decision’s muddled majority status, Tam makes important law on both the First Amendment’s coverage and the nature of viewpoint-based discrimination. A core theme of the decision is that trademarks contribute ideas to public discourse, debate, and education, just like political arguments or art. Justice Alito states that the disparagement bar must fall because “[s]peech may not be banned on the ground that it expresses ideas that offend.”84 Justice Kennedy goes further, calling the disparagement bar an “attempt[ ] to remove certain ideas or perspectives from a broader debate” and extolling “the trademark’s potential to teach.”85 This is a remarkable, far-reaching conception. Frustratingly, the Court fails to develop the conception in any meaningful way, let alone defend it. If trademarks centrally express what we might call abstract ideas, then the bedrock presumption of trademark law—that trademarks serve a primarily commercial, instrumental function—is substantially wrong. More broadly, if trademarks deserve First Amendment protection because they contribute abstract ideas to public discourse, then virtually any action or projection with semantic content must do so. That proposition carries potent implications for both the First Amendment’s coverage and the viewpoint discrimination doctrine.

Tam significantly expands the scope of the First Amendment’s coverage. Only once in the past decade, in Sorrell v. IMS Health,86 has the Court invoked the First Amendment to strike down what most people would call a commercial regulation.87 IMS Health, which barred a state from restricting the commercial sale of information about which medications individual doctors prescribed, has prompted substantial controversy.88 The disparagement bar in Tam diverged from the law struck down in IMS Health in two ways that would seem to insulate the bar from First Amendment danger: substantial longevity, and a federal rather than state pedigree. The disparagement bar has figured in federal law for seven decades, since the Lanham Act’s adoption in 1946. The Roberts Court has stated that tradition-tested allowances for regulation of speech deserve some measure of judicial respect.89 In Tam, though, the Court did not hesitate to strike down a long-standing federal economic regulation, one that the Court itself had ignored for seven decades. Regulated entities should now reasonably expect their lawyers to raise First Amendment challenges to all manner of federal (let alone state) economic regulations, no matter how seemingly ingrained. If those challenges bear fruit, Tam will have dramatically accelerated First Amendment law’s encroachment on economic regulation.

The Tam Court’s rejection of the government’s arguments that registration amounts to either government speech, a permissibly restricted government program, or a properly conditioned speech subsidy raise the decision’s stakes for the scope of First Amendment coverage. The Court’s attempts to sort out contexts in which the First Amendment leaves the government with heightened regulatory latitude have grown thorny and consequential, especially as the Justices have warmed to the notion that some seemingly private speech is really government speech.90 By and large, the Court’s navigation of the private speech, subsidy, and government speech categories has been haphazard and seemingly outcome driven, generating lines of decisions that prioritize palatable results over coherent doctrine.91 The Tam Court’s deflection of the government’s categorical arguments for heightened regulatory latitude makes much more sense as a defense of a preferred result than as a promotion of coherent, generally applicable doctrine. To take just one issue, Justice Alito’s denial that trademark registration amounts to a government subsidy92 does not square easily with his insistence that trademark registration confers a substantial benefit.93 Nothing about Tam inspires confidence that the Court will remain highly skeptical of government control arguments the next time a litigant challenges a regulatory regime the Court favors. At the same time, the Tam Court’s skepticism of the government’s categorical arguments could easily buttress future efforts to impose First Amendment constraints on economic regulations.

Justice Kennedy’s attempt to flesh out Justice Alito’s skeletal statement that the disparagement bar discriminated against the viewpoint of speech makes an important, contestable intervention in the Court’s viewpoint jurisprudence. Justice Kennedy resolves the content-viewpoint baseline problem by declaring that the government may not “mandat[e] positivity.”94 In essence, the government may not in any context make beneficiaries of public support for expression speak civilly rather than derisively to one another. Civility mandates, Justice Kennedy asserts, “silence dissent.”95 No doubt that’s substantially true and important. Justice Kennedy’s declaration, however, digs a shallow well into a deep free speech problem: the elusive balance that robust public discussion requires between civility norms and expressive autonomy, which Robert Post calls the “paradox of public discourse.”96 Civility norms enable the common discursive frequencies that make dissent (and other expressive interactions) meaningful and effective. Justice Kennedy’s one-size-fits-all conception of the content-viewpoint distinction ignores that important dimension of free speech theory.

In the particular context of Tam, Justice Kennedy’s discussion of viewpoint discrimination carries uncertain, potentially important implications for the Court’s First Amendment jurisprudence on racist speech. The Court’s earlier decisions lurch from one outcome to another, barring limits on racist speech here, permitting them there.97 Like the government speech and subsidized speech cases, the racist speech cases seem driven more by preferred results than by coherent principles. Perhaps Justice Kennedy’s rigid account of viewpoint discrimination in Tam, if a majority of the Court someday adopts it, will resolve the contradictory strains of racist speech doctrine into a harmonious chord. However, Justice Kennedy’s complete failure in Tam to work through those contradictions—he omits any mention of the earlier racist speech decisions—suggests that his sweeping pronouncements will likelier pour more heat onto this First Amendment problem than shed any light on it.

Justice Kennedy’s Tam concurrence condemns viewpoint discrimination more aggressively than any Supreme Court opinion in recent memory. Even the Court’s contentious holding that speech about religion constitutes a class of viewpoints rather than a subject matter category showed modesty by abjuring any unified field theory of the content-viewpoint distinction.98 The Tam viewpoint holding, in contrast, resembles the Court’s recent reiteration in Town of Gilbert of the basic First Amendment presumption against content-based regulation, where Justices Breyer and Kagan warned about the danger of overreach.99 In turbo-charging a familiar doctrinal engine, the Court creates a serious risk of seemingly unintended, certainly unexamined consequences. May the government no longer impose a bare baseline of “positivity” in any circumstance where it provides incentives for autonomous speech?100 Must courts now block any linkage of legal burdens to racist viewpoints?101 The unknown consequences loom larger in Tam than in Town of Gilbert because the viewpoint principle carries even more force than the content principle; and Tam is less definite in its implications than Town of Gilbert because the Tam viewpoint analysis resides almost entirely in a concurring opinion.

III.     The Empirical Complexity of a “Disparaging” Term

First Amendment law depends on all sorts of empirical premises. At a theoretical level, our legal system has long emphasized the notion “that the best test of truth is the power of the thought to get itself accepted in the competition of the market.”102 At a practical level, courts routinely both grant and deny First Amendment claims based on empirical assertions. States may not provide extra public funding to opponents of unusually well-financed candidates because “an advertisement supporting the election of a candidate that goes without a response is often more effective than an advertisement that is directly controverted.”103 Cities may geographically scatter stores that sell sexually explicit books and films “because a concentration of [such stores] in one locale draws . . . a greater concentration of adult consumers to the neighborhood, and a high density of such consumers either attracts or generates criminal activity.”104

Unfortunately, those examples of causal logic in First Amendment cases illustrate a norm of failing to ground empirical assertions in actual data. Constitutional law is slowly catching up to other legal fields in its use of empirical analysis.105 In analyzing First Amendment issues, courts106 and scholars107 occasionally develop and assess data. Too often, however, court decisions and law review articles blithely toss off, without substantiating, key empirical premises for First Amendment conclusions.

At the core of the Tam dispute lies a set of questions about what the term “Slants,” in this context, means. The parties’ contentions about meaning—disparagement vs. empowerment—highlight the particular question of how the term’s objects, a subset of Asian Americans, understand the term. To answer this and other questions, we designed a study that draws on interviews conducted in February and March 2017 with Americans in two national probability samples: (1) 511 Asian Americans;108 and (2) 2065 non-Asian Americans. Because our questionnaire asked all respondents the same questions, we can directly compare the two samples.

The questions we asked relate both to the term “slants” and to the band The Slants. We began by assessing the validity of the government’s claim that “slants” disparages persons of Asian ancestry. To do so, we asked respondents (in three separate questions) whether they thought the term “slants” (1) offends, (2) ridicules, or (3) insults Asian people. As Figure 1 shows, all respondents—Asians and non-Asians alike—agree with the government: The term is insulting, offensive, or ridiculing. (The difference is statistically significant for “Insults,” with significantly more non-Asians believing that “slants” is an insulting term.)

Figure 1 MagarianBut that is not the end of the matter. Recall that one reason the band took issue with the PTO’s refusal to register their name was the PTO’s failure to consider the context of the speech: an Asian-American band’s effort to reappropriate a slur, not to insult Asian people. To assess the band’s argument, we asked respondents why they thought the band named itself The Slants. Embedded in this question was an experiment: half the sample responded to this question while viewing a photo of an Asian band (actually The Slants); the other half responded while viewing a photo of a non-Asian band. See the Appendix for the photos.

As Figure 2 shows, The Slants make a good case for the importance of considering the band’s motive in the context of its Asian identity. Respondents—again, both Asian and non-Asian—clearly attribute different motives to the band depending on whether they viewed a photo of The Slants or the non-Asian band.

Figure 2 Magarian

Specifically, for the “badge of Asian pride” motive, we asked respondents whether they thought the band wanted “to use ‘The Slants’ as a badge of Asian pride rather than ridicule.” Note, in Figure 2, that most Asian (52.9%) and non-Asian (51.4%) respondents agreed that this was a possible motive of The Slants, but only a small fraction attributed that motive to the non-Asian band (15.4% of the Asian and 23.5% of the non-Asian respondents). For the “slur right back” motive, we asked respondents whether the band “wanted to throw the slur right back in the faces of those prejudiced against people of Asian ancestry.” Again, note the difference between the reactions to the Asian versus the non-Asian band regardless of whether the respondents were Asian or non-Asians.

Put in slightly different terms: Had the PTO registered “The Slants,” our results suggest that Americans who were made aware of the band members’ racial background would have ascribed non-disparaging motives to the band, and not disparaging motives. Not so, however, had the band’s members been of European ancestry. Also contributing to different respondents’ understanding of the term were demographic and psychological attributes.109

IV.     Empirical Analysis as a Means to Constitutional Avoidance

The results of our study render the Lanham Act’s disparagement bar incoherent in practice. Armed with our data, the Supreme Court in Matal v. Tam could have found the bar unconstitutionally vague. That approach would have supported the holding in Tam while avoiding weightier First Amendment issues.

Constitutional avoidance is the judicial practice of not making a constitutional ruling where a lesser legal ruling can resolve a case. In Justice Brandeis’ canonical formulation, “[t]he Court will not pass upon a constitutional question although properly presented by the record, if there is also present some other ground upon which the case may be disposed of.”110 In statutory cases, the Court has framed avoidance as a substantive canon of construction, under which courts should prefer a reasonable interpretation of a statute that avoids calling the statute’s constitutionality into doubt.111 The avoidance principle reflects judicial attention to the separation of powers, particularly the limitation of the judicial power to resolving cases and controversies.112

Legal scholars have questioned the logical foundations and normative justifications of the avoidance principle, expressing concern that courts can deploy the rhetoric of avoidance to disguise the very sort of judicial overreaching that avoidance is supposed to prevent.113 We take no position here on whether or to what extent constitutional avoidance is normatively desirable or even conceptually coherent. We simply posit that courts have asserted the value of avoidance, which suggests that empirical analysis as a means to advance avoidance could serve as a valued tool in our legal system.

A.     Incoherence, Vagueness, and Avoidance in Tam

Our empirical findings, summarized in Part III, support finding the Lanham Act § 2(a) disparagement bar unconstitutionally vague. That conclusion follows from the context-sensitive variance in people’s understandings of the term “slants,” particularly the understanding that members of a putatively disparaged group can seek to reappropriate a disparaging term, which renders incoherent the bar’s treatment of certain words as inherently disparaging. Finding the bar vague would have accomplished constitutional avoidance by deciding a less momentous question of constitutional law than the Tam Court’s First Amendment viewpoint discrimination analysis.

1.     Incoherence as Vagueness

Courts have long held that the Fifth and Fourteenth Amendments prohibit the government from enforcing a statute that speaks in terms too vague to “give the person of ordinary intelligence a reasonable opportunity to know what is prohibited.”114 This prohibition on vague regulations has roots in the separation of powers, because vague legislation improperly empowers the executive and judicial branches to make broad policy determinations. As Justice Brennan explained, “[t]he requirement that government articulate its aims with a reasonable degree of clarity ensures that state power will be exercised only on behalf of policies reflecting an authoritative choice among competing social values.”115 Accordingly, courts frequently base findings of unconstitutional vagueness on patterns of arbitrary or capricious law enforcement.116 Courts have most commonly found criminal statutes vague, but the doctrine applies to civil statutes as well.117

The vagueness doctrine aims to prevent the erosion of constitutional rights.118 The Supreme Court has specially adapted the vagueness principle for First Amendment law to strike down speech restrictions whose lack of clarity might cause speakers to self-censor, chilling protected speech.119 “Because First Amendment freedoms need breathing space to survive,” the Court has explained, “government may regulate [speech subject to lawful limitation] only with narrow specificity.”120 The acute danger of chilling protected speech requires “a more stringent vagueness test” in First Amendment cases than in other settings.121

A handful of commentators have condemned the Lanham Act’s disparagement bar as unconstitutionally vague122 or criticized the PTO’s inconsistent application of the bar.123 In In re Tam, two members of the en banc Court of Appeals would have held, as an alternative to the substantive First Amendment holding, that the disparagement bar was unconstitutionally vague.124 Judge O’Malley wrote that the statutory term “disparage” gave no real notice of what terms would run afoul of the bar. The PTO’s “substantial composite” gloss on the statute, she maintained, “compounds the confusion [because] . . . . a mark need only potentially disparage a subset of any group as long as that group can be ‘identifi[ed].’”125 She argued further that the PTO’s history of applying the disparagement bar inconsistently and arbitrarily bore out the bar’s vagueness.126 Finally, she discussed a Sixth Circuit decision that found a university’s discriminatory harassment policy unconstitutionally vague because of the subjectivity of terms like “de-meaning or slurring.”127 The Court of Appeals majority did not find the disparagement bar unconstitutionally vague but merely noted the uncertainty of the statutory language.128

The Supreme Court in Tam said nothing about vagueness, instead jumping to the substantive First Amendment holding that the disparagement bar discriminated based on viewpoint. Perhaps the Court affirmatively wanted to call out and cut down viewpoint discrimination in the disparagement bar. That explanation, though, uncharitably presumes that the Justices did not care about judicial restraint. Alternatively, either of two features of Tam may have discouraged a vagueness approach. First, courts most commonly find vagueness in cases that involve weighty legal consequences.129 The incremental administrative burden of the PTO’s refusal to register a trademark may have seemed too trivial to warrant a vagueness finding. That reasoning, however, rests on the premise that the Court’s substantive First Amendment analysis showed greater judicial restraint than a vagueness holding would have, a premise we refute below. Second, federal courts more easily find state laws vague than federal laws because federal judges have greater latitude to craft saving constructions for federal laws.130 In Tam, however, the Court attempted no saving construction.

The simplest explanation for the Tam Court’s reliance on viewpoint discrimination rather than vagueness is that the Justices doubted the basis in the record for finding the disparagement bar vague. They may have found the meaning of “disparage” in Lanham Act § 2(a) reasonably clear. Indeed, the logic of viewpoint discrimination presumes that a law speaks clearly enough to discriminate against some identifiable viewpoint.

Our empirical data undercut that view and greatly strengthen the case for finding the disparagement bar unconstitutionally vague. The complexity of survey respondents’ understanding of the term “slants” in our study renders § 2(a)’s formulation of trademarks that “may disparage . . . persons” 131 flatly incoherent and thus unconstitutionally vague. To set the stage, a helping verb could hardly muddy the semantic waters more than “may” does here. By connoting an unquantified range of possibilities, “may” immediately calls into question the likelihood or frequency of “disparagement” required for refusing an application.132

Our data expose a similar, deeper problem with “disparage” itself. Our survey respondents expressed a range of understandings of the term “slants.” Of course, the term has an innocuous geometric meaning that could make sense as a rock band’s name.133 That ambiguity probably accounts for some initial divergence of understandings. Understandings varied somewhat further with respondents’ ethnicities. Understandings split sharply based on the term’s context. In particular, background information that an Asian-American speaker was using the term dramatically changed respondents’ understanding, suggesting an effort to reappropriate or defuse the term’s historically negative connotation. In short, the meaning of “slants” is a moving target.134 Other words known for their use as identity slurs might lack some of the qualities that disperse people’s understandings of “slants,” but all at a minimum would be subject to the contextual variation in our survey results.

The word “disparage” sometimes has a discernible meaning. In the context of § 2(a), however, our data expose “disparage” as a referent without any real antecedent. Neither “slants” nor, our data strongly suggest, any other term simply or definitely disparages when employed as a trademark. A term may be understood by a subset of people in a subset of circumstances as disparaging, but even that penumbral subset changes with exposure to some of the vortex of semantic information that swirls around words in their social contexts.135 In light of the complexities our data illuminate, the only way to imbue § 2(a)’s “disparage” with definite meaning is to note that people may understand a vast range of terms as disparaging a target group in some circumstances. At that point, however, the statutory language becomes unconstitutionally overbroad.136

The PTO’s efforts to clarify the disparagement bar’s meaning fail to cure the bar’s incoherence. First, the PTO has explained that an application fails the disparagement bar if the trademark’s “meaning may be disparaging to a substantial composite of the referenced group.”137 “Substantial” speaks with barely more precision than the statutory “may.” The Supreme Court could not make sense of the word “composite” in context and therefore took the very unusual step of presuming that the PTO meant to use a different word, “component.”138 Even if we attach an arbitrary value to “substantial”—ten percent, one third, take your pick—and ascribe sense to “composite,” our data show that the PTO could not coherently count the offended heads. As we have shown, individuals’ understandings of “Slants” vary depending on who uses the term and in what context.139 Second, the PTO has explained that it may find trademarks disparaging if they “dishonor by comparison with what is inferior, slight, deprecate, degrade, or affect or injure by unjust comparison.”140 That gloss breaks down into two genres: verbs that fail to improve on the deficiencies of “disparage,” and two different iterations of the concept of unfavorable comparison. That concept contributes to a fairly intuitive functional definition of “disparage,” but it does nothing to cure the defect revealed by our data: the impossibility of finding in the term “Slants” any consistent, fixed disparaging character without reference to contextual variables. The PTO’s refusal to recognize any safe harbor for “good intentions” washes away the disparagement bar’s last, faint hope for greater coherence.

To illustrate the general problem that our data pose for the disparagement bar, consider the prominent legal conflict over registering the name of the Washington, D.C. professional football team: “Redskins.” That name has less immediate semantic play in its joints than “Slants.”141 No one would seriously argue that the football team is trying to reappropriate or defuse a racially derogatory term. On the other hand, it seems unlikely that the team specifically intends to use the slur for its derogatory impact. The team presumably wants to register the trademark in order to increase the profitability of a name with a distinctive set of connotations to football fans. Meanwhile, other people presumably invoke the team’s name for a disparate set of reasons: to talk about football but also to insult Native Americans; to illustrate and rally activism against anti-Native racism; and to help place Native American experiences in a political, cultural, and historical context. Our data show how the dynamic character of language supports a judgment that the disparagement bar speaks in hopelessly vague terms rather than discriminating against any viewpoint.

2.     Vagueness as Avoidance

Finding a speech-restrictive statute vague can serve as a form of constitutional avoidance. Because vagueness is a constitutional doctrine, finding a law vague does not avoid constitutional adjudication altogether. A robust conception of avoidance, however, includes not just absolute avoidance of constitutional holdings but relative avoidance: resolution of a dispute based on a constitutional ground narrower in some meaningful sense than other possible constitutional grounds.

Judges and scholars have long acknowledged vagueness as a vehicle for relative avoidance in First Amendment and other constitutional cases. The avatar of judicial minimalism, Alexander Bickel, counted vagueness among the “passive virtues” because a judicial finding of vagueness “withholds adjudication of the substantive issue in order to set in motion the process of legislative decision.”142 Cass Sunstein has argued along the same lines that finding a speech-restrictive statute vague, rather than determining exactly what speech affected by the statute the First Amendment protects, can exemplify “democracy-forcing [judicial] minimalism.”143 William Eskridge has suggested that vagueness may serve courts especially well as a tool for defusing high-stakes political cases.144 Judges frequently reject vagueness analysis to set up constitutionally significant endorsements of government policies145 or, conversely, embrace vagueness analysis as a less constitutionally significant ground for striking down government policies.146 In a recent concurring opinion, for example, Justice Alito found a state prohibition on selling violent video games to children unconstitutionally vague and therefore saw “no need to reach the broader First Amendment issues addressed by the Court.”147

At least where a First Amendment claimant’s speech is not clearly proscribable, the remedy for a vague statute is facial invalidation.148 That potent remedy reflects the central goal of the First Amendment vagueness doctrine to prevent the government from chilling protected speech.149 The facial invalidation remedy might appear to disqualify vagueness as a relatively narrow ground for dealing with a free speech claim. However, even where the remedy for a substantive First Amendment violation would be an as-applied injunction rather than facial invalidation, a vagueness finding may entail a more modest exercise of judicial power than a substantive First Amendment analysis. To determine which ground is narrower, a court must measure vagueness, which entails totally invalidating a particular statute, against the substantive First Amendment analysis, which entails establishing a constitutional precedent that may affect many future disputes. Vagueness works as avoidance when invalidating the statute vents a lesser degree of judicial power than establishing the precedent.150

In Tam, the remedy for either vagueness or a substantive First Amendment violation was the same: facial invalidation of the disparagement bar.151 The Court held that the disparagement bar discriminated against speech based on its viewpoint.152 For reasons we have discussed, that is a potent constitutional holding. It both helps to validate First Amendment challenges to other sorts of federal economic regulations and broadens the Court’s conception of impermissible viewpoint discrimination. A finding that the disparagement bar was unconstitutionally vague would address the particular word “disparaging” in the particular context of trademarks. For this simple reason, a vagueness finding that our data support in Tam would have worked as a means of relative constitutional avoidance.

Rebecca Tushnet has argued that the logic of finding the disparagement bar vague would apply to other parts of the Lanham Act as well, because the PTO’s practice of individually evaluating applications creates inconsistency of results along many or most Lanham Act criteria.153 “[T]he registration system,” she avers, “is not highly predictable except at the probabilistic level.”154 If Tushnet is right, then a vagueness holding in Tam would have had greater impact than we’re suggesting. Tushnet, however, can’t show that the level of uncertainty or inconsistency in the enforcement of other Lanham Act provisions even approaches the well-documented arbitrariness of decisions about the disparagement bar. Moreover, substantial probabilistic predictability might well imbue terms with sufficiently stable meanings to clear the vagueness bar. In any event, our empirical basis for finding the disparagement bar vague establishes the bar’s essential incoherence, a more elemental and powerful ground for finding vagueness than just showing inconsistent enforcement.

A finding of vagueness, unlike the Tam Court’s finding of viewpoint-based discrimination, would not categorically doom efforts to deny Lanham Act registration for racially charged trademark applications. Congress might try to redesign the disparagement bar with greater nuance, most importantly attending to contextual distinctions in the meanings of terms. Perhaps a firm, fleshed-out grounding in a concept like racial subordination would give the PTO adequate guidance and trademark applicants adequate notice to avoid the vagueness trap.155 Congress, in any event, would have the opportunity to engage in the kind of dialogic response to constitutional concerns that avoidance is meant to enable but that rigid judicial impositions of constitutional mandates foreclose. More broadly, a vagueness analysis would have averted the Tam decision’s two significant constitutional moves: expanding the First Amendment’s application to economic regulation, and pulling the Court further into the doctrinal minefield of racially derogatory speech and viewpoint discrimination.

B.     Propagating Data-Driven Avoidance

Our data-based vagueness analysis in Tam provides one illustration of what can be a broader practice. Empirical analysis provides a fruitful basis for constitutional avoidance. Of course, every constitutional case presents distinctive legal and factual problems. The specific path we have traced here—data to vagueness to avoidance—probably won’t appear in many cases. Our central point is that the broader linkage—data to avoidance—frequently will arise. Avoidance depends on the availability of alternative paths to decision. Alternative paths emerge from increased knowledge. One rich and, in constitutional law, underutilized method of increasing knowledge is empirical analysis.

Data-driven avoidance can operate in constitutional disputes that present three preconditions. First, the dispute must prominently implicate a constitutional right or principle. Second, the constitutional issue must be, at some level, an issue of first impression, such that judicial resolution of the constitutional question must make new law. These first two preconditions for data-driven avoidance are axiomatic of any constitutional avoidance case. The distinctive utility of data-driven avoidance emerges from the third precondition: the enacting legislature and implementing agency must have failed to pursue an available course of empirical analysis relevant to the lawfulness (at least in the context of the immediate dispute) of the challenged regulation. This sort of empirical gap creates the basis for data-driven avoidance.

The major benefit of data-driven avoidance, in most cases, will come from the specificity of empirical analysis. An empirical study answers a particular set of questions in a particular context. That specificity contrasts with much conventional legal reasoning, which typically seeks to establish general principles that can govern a broad range of problems. Judicial resort to data as a means of constitutional avoidance may seem counter-intuitive, because we generally expect legislatures rather than courts to assess the facts relevant to policy decisions.156 In a data-driven avoidance case, however, the legislature has failed to make that sort of assessment, and the judge has grounds for rejecting the law on constitutional grounds. In that confluence of circumstances, the specificity of data-driven avoidance enables a more modest, restrained decision than the constitutional alternative. The judge, in order to do the least possible damage to the legislature’s policy choice, invades only that empirical space the legislature has neglected to fill.

Predicting future disputes that may raise constitutional problems, let alone the substance of empirical analyses useful for resolving those disputes, is difficult. However, two First Amendment problems—one recently settled by the Supreme Court, the other speculative—illustrate how data-driven avoidance might work in other settings where constitutional issues turn on empirical premises.

Union Agency Fees. The Supreme Court’s recent decision in Janus v. American Federation of State, County, and Municipal Employees, Council 31 announced a major new First Amendment restriction on public employee labor unions.157 Some states let public employee unions negotiate contracts under which the unions collect “agency fees” from state employees who choose not to join the union. The Court in the 1970s worried that agency fees might violate the First Amendment by forcing nonunion members to pay for unions’ advocacy of political causes the nonmembers opposed. In a curious instance of partial constitutional avoidance, the Court first interpreted federal labor laws as entitling private sector unions to collect agency fees only to cover costs of collective bargaining and related activities, not political advocacy.158 Then, in Abood v. Detroit Board of Education, the Court invoked the First Amendment to mandate that bargaining-advocacy split for public sector unions.159 The Roberts Court in recent years chipped away at Abood by restricting the purposes for which160 and the categories of employees from whom public employee unions could collect agency fees.161 Finally, after a brief delay caused by Justice Scalia’s death,162 Janus overruled Abood and imposed a flat First Amendment ban against collection of agency fees for any purpose from any nonconsenting public employee.163

The Janus holding that public sector agency fees violate the First Amendment depends on a sweeping empirical presumption: that nonunion employees pervasively oppose the substantive political positions unions advance through their political advocacy.164 If nonunion members generally oppose unions’ political stands and actions, then the need for First Amendment intervention gains urgency, and the imperfect Abood boundary between collective bargaining and political advocacy expenses may underprotect nonmembers’ rights. However, the Roberts Court’s decisions cite no empirical data to validate the Court’s presumption about nonunion employees’ political attitudes. That presumption is highly contestable. Employees might choose not to join unions for a wide range of reasons, most obviously a preference to pay the union only an agency fee rather than full union dues and then free ride on union activities (potentially including both collective bargaining and political advocacy) for which union members pay more.165

Empirical analysis could substantially falsify or verify the Roberts Court’s presumption about nonunion employees’ political attitudes. A survey of nonunion state employees could determine why they opt not to join unions and to what extent they oppose unions’ political stands. If the nonmembers’ opposition to the unions’ political activities proved relatively weak, the data would validate a relative avoidance strategy. In this instance, relative avoidance would likely take the form of reaffirming the Abood Court’s First Amendment compromise. In effect, empirical analysis would backfill the information gap that subsequent litigation exposed in the Abood treatment of agency fees.166 Of course, an empirical study might instead show that nonunion employees passionately detested unions’ political positions and rejected union membership for largely political reasons. In that event, knowing that nonmembers shunned unions for political reasons could bring benefits that diverged from but also complemented judicial restraint: substantiating the unavailability of an avoidance path and buttressing the Janus Court’s justification for overruling legislative choices to permit public sector agency fees.

Public Securities Disclosures. Section 13(f) of the Securities Exchange Act of 1934 requires institutional investment managers to disclose to the Securities Exchange Commission (“SEC”) the names, shares, and value of securities they control.167 The Act requires the SEC to make § 13(f) disclosures publicly available unless the SEC concludes that an investment manager satisfies a statutory ground for exemption from public disclosure.168 Investment managers have argued that the publication of § 13(f) disclosures compels them to speak in violation of the First Amendment.169

The Supreme Court has invoked the First Amendment to stop the government from compelling speech in two sorts of situations: where the compulsion would cause the speaker to violate some conscientious belief, like forcing students with religious objections to salute the flag and recite the Pledge of Allegiance;170 and where the compulsion would expose the speaker to a strong likelihood of official reprisal, like forcing a civil rights group to disclose its membership list to a racist state government.171 The investment managers’ First Amendment argument against § 13(f) disclosures falls into neither of those categories. Thus, courts might reject the argument as simply outside the boundaries of First Amendment doctrine. Like the Lanham Act disparagement bar, however, compelled public disclosure of investment managers’ holdings clearly regulates speech based on its content. The present Supreme Court’s hard line against content discrimination and increasing willingness to subject even federal economic regulations to First Amendment scrutiny could plausibly cause it to view public securities disclosures as raising a constitutional problem.

Instead of concerns about violations of conscience or official reprisals, investment managers ground their compelled speech argument against § 13(f) on a theory of economic harm. Making institutional investment managers disclose their holdings to the public, the argument goes, will drive up the prices of the securities under a manager’s control, because disclosure of the manager’s holdings will show other investors the manager’s strategy.172 Unlike conscientious injury, as to which courts defer to individuals’ accounts of their beliefs, and dangers of official reprisal, as to which courts presume governments threaten individual rights, the investment managers’ economic injury argument depends on empirical premises. The argument only works if elements of the public (a) can draw certain inferences about investment managers’ strategies from knowledge of their holdings, (b) have the means to act on their insights about the managers’ strategies, and (c) actually take such actions. Empirical study could provide useful data about each of those premises. In this context, analysis of the data could support the most conventional form of constitutional avoidance: reading statutory language narrowly to avert constitutional harms. Depending on what the data revealed about investors’ behavior, and assuming the requisite semantic flexibility at the relevant margin in the language of § 13(f), the Court could read the statute to let the SEC publicize only those disclosures that posed an acceptably low risk of financial harm to investment managers.

As these examples suggest, the vagueness path to constitutional avoidance that our study opens in Matal v. Tam is no universal archetype of data-driven avoidance. Still, the Tam case study usefully shows how avoidance can take forms other than statutory narrowing constructions. The agency fee example shows another such form, while the securities disclosure example presents a classic narrowing construction. Data can enable avoidance where avoidance might not otherwise be available by giving courts more information about putative constitutional disputes, thereby increasing decisional latitude. Empirical analysis will tell different stories about different constitutional cases, and avoidance will work differently based on how those stories play out. In some cases, data would presumably support rather than discourage constitutional lawmaking.173 Even in those instances, however, data might encourage narrower, more fact-specific judicial reasoning, which might in turn leave legislatures with greater opportunities than abstract constitutional reasoning would afford to reconcile legislators’ policy preferences with constitutional concerns. Empirical analysis, then, can enable even nonavoidance of constitutional lawmaking to advance judicial modesty values.

V.     Conclusion

Judges and many commentators advocate avoiding unnecessary constitutional adjudication as a prudent limitation on the exercise of judicial power. Constitutional avoidance can take various forms. The key ingredient for avoidance is the decisional latitude to resolve a dispute of potential constitutional magnitude without reaching the constitutional issue. Our principal aim in this Article has been to show how empirical analysis can enable constitutional avoidance by providing information that increases decisional latitude. Judges apply legal principles to material conditions. Empirical analysis, by illuminating material conditions, affords opportunities for different legal reasoning. Our empirical examination of the great variance in how different people understand the term “slants” in different contexts could have allowed the Supreme Court in Matal v. Tam to reject the Lanham Act’s bar on registering disparaging trademarks under the vagueness doctrine, rather than perform the highly consequential First Amendment lawmaking of the actual decision. In other cases, empirical analysis might give courts grounds for the more traditional mode of constitutional avoidance: parsing statutes to impose narrowing constructions and thus dodging constitutional issues altogether. We hope this initial account of data-driven constitutional avoidance will encourage litigants, courts, and scholars to expand the range of settings in which empirical analysis contributes to sound constitutional adjudication.



  1. [1]. N. Sec. Co. v. United States, 193 U.S. 197, 400 (1904) (Holmes, J., dissenting).

  2. [2]. See generally Matal v. Tam, 137 S. Ct. 1744 (2017) (holding that the disparagement clause of the Lanham Act violated the Free Speech Clause of the First Amendment).

  3. [3]. N. Sec. Co., 193 U.S. at 400–01.

  4. [4]. See infra notes 10–11 and accompanying text.

  5. [5]. By “reappropriate,” we refer to “the process of taking possession of a slur previously used exclusively by dominant groups to reinforce a stigmatized group’s lesser status.” Adam D. Galinsky et al., The Reappropriation of Stigmatizing Labels: The Reciprocal Relationship Between Power and Self-Labeling, 24 Psychol. Sci. 2020, 2020 (2013).

  6. [6]. See generally Randall Kennedy, Nigger: The Strange Career of a Troublesome Word (2002) (explaining the history of the N-word); Galinsky et al., supra note 5, at 2020 (noting that study shows “self-labeling with a derogatory label can weaken the label’s stigmatizing force”); Todd Anten, Note, Self-Disparaging Trademarks and Social Change: Factoring the Reappropriation of Slurs into Section 2(A) of the Lanham Act, 106 Colum. L. Rev. 388 (2006) (analyzing trademarks that include reappropriated racial slurs).

  7. [7]. See infra notes 42–57 and accompanying text.

  8. [8]. This discussion of the facts and salient law draws from Tam. Matal v. Tam, 137 S. Ct. 1744, 1751–54 (2017).

  9. [9]. Id. at 1754. This slur recently marred the 2017 Major League Baseball World Series, when the Houston Astros’ Yuli Gurriel made a gesture simulating slanted eyes after hitting a home run off Japanese Los Angeles Dodgers pitcher Yu Darvish. See David Waldstein, Astros’ Yuli Gurriel Escapes World Series Ban, but Will Miss 5 Games in 2018, N.Y. Times (Oct. 28, 2017),

  10. [10]. 15 U.S.C. § 1052(a) (2012).

  11. [11]. USPTO, Trademark Manual of Examining Procedure § 1203.03(b)(i) (2017).

  12. [12]. Id.

  13. [13]. See, e.g., Pro-Football, Inc. v. Harjo, 415 F.3d 44, 46 (D.C. Cir. 2005) (discussing Native Americans’ petitioning of the TTAB to cancel the Washington Redskins’ trademarks). Interestingly, the Cleveland professional baseball franchise decided shortly after the Court decided Tam to end on-field use of the team’s much-criticized “Chief Wahoo” logo. See David Waldstein, Cleveland Indians Will Abandon Chief Wahoo Logo Next Year, N.Y. Times (Jan. 29, 2018),

  14. [14]. See 15 U.S.C. § 1052(a) (authorizing refusal to register a trademark that “[c]onsists of or comprises immoral, deceptive, or scandalous matter”).

  15. [15]. See Golan v. Holder, 565 U.S. 302, 328 (2012); Eldred v. Ashcroft, 537 U.S. 186, 221 (2003).

  16. [16]. See, e.g., S.F. Arts & Athletics, Inc. v. U.S. Olympic Comm., 483 U.S. 522, 540–41 (1987) (rejecting a First Amendment challenge to a federal restriction on commercial or promotional uses of the word “Olympic” and related images).

  17. [17]. See Frederick Schauer, The Boundaries of the First Amendment: A Preliminary Exploration of Constitutional Salience, 117 Harv. L. Rev. 1765, 1777–84 (2004) [hereinafter Schauer, Boundaries].

  18. [18]. See Chaplinsky v. New Hampshire, 315 U.S. 568, 572 (1942).

  19. [19]. See Miller v. California, 413 U.S. 15, 23–24 (1973).

  20. [20]. See Schauer, Boundaries, supra note 17, at 1778–83.

  21. [21]. Id. at 1805–07.

  22. [22]. See generally id. (linking political and cultural acceptance of the speech at issue with judicial determinations that the speech is protected under the First Amendment).

  23. [23]. Id. at 1795.

  24. [24]. See id. at 1787–800.

  25. [25]. See Frederick Schauer, The Politics and Incentives of First Amendment Coverage, 56 Wm.
    & Mary L. Rev. 1613, 1633–34 (2015) [hereinafter Schauer, Politics and Incentives].

  26. [26]. For a thorough analysis and critique of the Roberts Court free speech decisions discussed in this Article, see generally Gregory P. Magarian, Managed Speech: The Roberts Court’s First Amendment (2017).

  27. [27]. See Sorrell v. IMS Health Inc., 564 U.S. 552, 566 (2011).

  28. [28]. See Janus v. Am. Fed’n of State, Cty., & Mun. Emps., Council 31, 138 S. Ct. 2448, 2486 (2018).

  29. [29]. See generally McCutcheon v. Fed. Election Comm’n, 135 S. Ct. 1434 (2014) (striking down the federal aggregate limits on campaign contributions); Ariz. Free Enter. Club’s Freedom PAC v. Bennett, 564 U.S. 721 (2011) (striking down part of a state’s public campaign financing system); Citizens United v. Fed. Election Comm’n, 558 U.S. 310 (2010) (striking down the federal bar on independent electoral expenditures by corporations and unions).

  30. [30]. See Brown v. Entm’t Merchs. Ass’n, 564 U.S. 786, 821 (2011); United States v. Stevens, 559 U.S. 460, 482 (2010).

  31. [31]. See Reed v. Town of Gilbert, 135 S. Ct. 2218, 2232 (2015).

  32. [32]. See Buckley v. Valeo, 424 U.S. 1, 143–44 (1976).

  33. [33]. See Va. State Bd. of Pharmacy v. Va. Citizens Consumer Council, 425 U.S. 748, 771–73 (1976).

  34. [34]. See Schauer, Politics and Incentives, supra note 25, at 1634–36.

  35. [35]. See Lochner v. New York, 198 U.S. 45, 53–54 (1905) (invoking a constitutional right of contract to strike down a state labor law). The Court emphatically repudiated the Lochner doctrine beginning with Nebbia v. New York, 291 U.S. 502, 538–39 (1934).

  36. [36]. See Sorrell v. IMS Health, Inc., 564 U.S. 552, 591–92, 602–03 (2011) (Breyer, J., dissenting); see also Julie E. Cohen, The Zombie First Amendment, 56 Wm. & Mary L. Rev. 1119, 1126–28 (2015) (arguing that the Roberts Court has used First Amendment law to protect property rights).

  37. [37]. See Kristian D. Stout, Terrifying Trademarks and a Scandalous Disregard for the First Amendment: Section 2(A)’s Unconstitutional Prohibition on Scandalous, Immoral, and Disparaging Trademarks, 25 Alb. L.J. Sci. & Tech. 213, 217–30 (2015); Russ VerSteeg, Blackhawk Down or Blackhorse Down? The Lanham Act’s Prohibition of Trademarks That “May Disparage” & the First Amendment, 68 Okla. L. Rev. 677, 698–700 (2016); see also Ron Phillips, A Case for Scandal and Immorality: Proposing Thin Protection of Controversial Trademarks, 17 U. Balt. Intell. Prop. L.J. 55, 66–68, 71–78 (2008) (proposing “thin protection” for scandalous and immoral trademarks as an accommodation to First Amendment concerns).

  38. [38]. See Christine Haight Farley, Stabilizing Morality in Trademark Law, 63 Am. U. L. Rev. 1019, 1020–22 (2014); Michael Grynberg, A Trademark Defense of the Disparagement Bar, 126 Yale L.J. F. 178, 180–87 (2016); Ned Snow, Free Speech & Disparaging Trademarks, 57 B.C. L. Rev. 1639,
    1639–43 (2016); Rebecca Tushnet, The First Amendment Walks Into a Bar: Trademark Registration and Free Speech, 92 Notre Dame L. Rev. 381, 384–87 (2016); see also Lisa P. Ramsey, A Free Speech Right to Trademark Protection?, 106 Trademark Rep. 797, 797–801 (2016) (advocating a restrained First Amendment analysis sensitive to the distinctive functions and values of trademark law). Megan M. Carpenter, while advocating abandonment of the § 2(a) bars on disparaging and scandalous trademarks, tracks the critics of First Amendment review by basing her argument on the harms of morally grounded registration limits for trademark law’s commercial functions. See Megan M. Carpenter, Contextual Healing: What to Do About Scandalous Trademarks and Lanham Act 2(a), 68 Hastings L.J. 1, 4–5 (2016).

  39. [39]. See Tushnet, supra note 38, at 381–84.

  40. [40]. See id. at 406–17.

  41. [41]. Id. at 383.

  42. [42]. See Perry Educ. Ass’n v. Perry Local Educators’ Ass’n, 460 U.S. 37, 49 (1983).

  43. [43]. The limitations on First Amendment protection for commercial speech that the Court first announced in Va. State Bd. of Pharmacy v. Va. Citizens Consumer Council, Inc. 425 U.S. 748, 770–73 (1976), presuppose the constitutionality of substantial content-based regulation of commercial speech.

  44. [44]. See Matal v. Tam, 137 S. Ct. 1744, 1763–65 (2017) (Alito, J., plurality opinion).

  45. [45]. See generally Seth F. Kreimer, Good Enough for Government Work: Two Cheers for Content Neutrality, 16 U. Pa. J. Const. L. 1261 (2014) (offering a qualified defense of the content principle). But see generally Erwin Chemerinsky, Content Neutrality as a Central Problem of Freedom of Speech: Problems in the Supreme Court’s Application, 74 S. Cal. L. Rev. 49 (2001) (criticizing strict enforcement of the content principle).

  46. [46]. Reed v. Town of Gilbert, 135 S. Ct. 2218, 2224 (2015).

  47. [47]. See id. at 2227–32; see also Nat’l Inst. of Family & Life Advocates v. Becerra, 138 S. Ct. 2361, 2370–72 (2018) (applying the Gilbert content discrimination analysis to state requirements of professional disclosures by anti-abortion “crisis pregnancy centers”).

  48. [48]. See Reed, 135 S. Ct. at 2234–36 (Breyer, J., concurring); id. at 2237–39 (Kagan, J., concurring); cf. id. at 2233–34 (Alito, J., concurring) (attempting to limit the impact of the majority’s reasoning); see also Nat’l Inst. of Family & Life Advocates, 138 S. Ct. at 2380–83 (Breyer, J., dissenting) (reiterating and elaborating criticisms of the Gilbert content discrimination analysis).

  49. [49]. McCullen v. Coakley, 134 S. Ct. 2518, 2536–40 (2014).

  50. [50]. See id. at 2530–34.

  51. [51]. See id. at 2543–48 (Scalia, J., concurring). Justice Scalia charged the buffer zone law with discriminating not just against the content of abortion-related speech but also against the specific viewpoint of anti-abortion speakers. Id.

  52. [52]. See Marjorie Heins, Viewpoint Discrimination, 24 Hastings Const. L.Q. 99, 104 (1996); Steven J. Heyman, Spheres of Autonomy: Reforming the Content Neutrality Doctrine in First Amendment Jurisprudence, 10 Wm. & Mary Bill Rts. J. 647, 665–67 (2002); Kreimer, supra note 45, at 1314–17.

  53. [53]. See Rosenberger v. Rector & Visitors of the Univ. of Va., 515 U.S. 819, 830–32 (1995).

  54. [54]. See id. at 893–99 (Souter, J., dissenting).

  55. [55]. See R.A.V. v. City of St. Paul, 505 U.S. 377, 391 (1992).

  56. [56]. See id. at 434–35 (Stevens, J., concurring in the judgment).

  57. [57]. See Sorrell v. IMS Health Inc., 564 U.S. 552, 563 (2011).

  58. [58]. For scholarly discussions of race and trademark law, see generally Jasmine Abdel-khalik, Disparaging Trademarks: Who Matters, 20 Mich. J. Race & L. 287 (2015) (discussing the 1946 Lanham Act’s disparaging trademark prohibition); K.J. Greene, Trademark Law and Racial Subordination: From Marketing of Stereotypes to Norms of Authorship, 58 Syracuse L. Rev. 431 (2008) (discussing the racial dynamics of trademark law); Sonia K. Katyal, Trademark Intersectionality,
    57 UCLA L. Rev. 1601 (2010) (examining the competing approaches to governance of trademark law through intersectionality).

  59. [59]. See, e.g., Pro-Football, Inc. v. Harjo, 415 F.3d 44, 46 (D.C. Cir. 2005).

  60. [60]. See, e.g., Floyd Abrams, The Soul of the First Amendment 38–49 (2017) (contrasting U.S. constitutional protection of hate speech with nonprotection in other democratic systems).

  61. [61]. See Beauharnais v. Illinois, 343 U.S. 250, 263–67 (1952).

  62. [62]. See Collin v. Smith, 578 F.2d 1197, 1204 (7th Cir.), cert. denied, Smith v. Collin,
    439 U.S. 916 (1978).

  63. [63]. R.A.V. v. City of St. Paul, 505 U.S. 377, 387–88 (1992).

  64. [64]. Id. at 380 (quoting Minn. Stat. § 292.02 (1990)).

  65. [65]. Id. at 386.

  66. [66]. Id. at 388–89.

  67. [67]. Id. at 396.

  68. [68]. See Wisconsin v. Mitchell, 508 U.S. 476, 488–90 (1993).

  69. [69]. See Virginia v. Black, 538 U.S. 343, 354 (2003).

  70. [70]. See Matal v. Tam, 137 S. Ct. 1744, 1751 (2017) (plurality opinion). Justice Kennedy’s opinion says that he joins only “Parts I, II, and III–A” of the Court’s opinion. Id. at 1765 (Kennedy, J., concurring in part and concurring in the judgment). Because Justice Alito states his conclusion about viewpoint discrimination in the introductory section of his opinion that precedes Part I, that statement does not speak for a majority of the Court, although Justice Kennedy emphatically endorses the substance of Justice Alito’s conclusion. See id.

  71. [71]. See id. at 1751–55.

  72. [72]. See id. at 1755–57. Justice Thomas alone declined to join Part II, based on his view that the respondent had not properly raised the statutory argument in the Court of Appeals. See id. at 1769 (Thomas, J., concurring in part and concurring in the judgment).

  73. [73]. See id. at 1757–60.

  74. [74]. See id. at 1760–61 (plurality opinion).

  75. [75]. See id. at 1761–63.

  76. [76]. See id. at 1763–65.

  77. [77]. See id. at 1764–65.

  78. [78]. Id. at 1751.

  79. [79]. Id. at 1766 (Kennedy, J., concurring in part and concurring in the judgment).

  80. [80]. Id.

  81. [81]. See id. at 1766­–67.

  82. [82]. Id. at 1767.

  83. [83]. See id. at 1767. Justice Kennedy never fully explains why his foursome joined Justice Alito’s discussion of the government speech argument but not Justice Alito’s discussions of the subsidy, government program, and commercial speech arguments. Justice Kennedy’s opinion seems roughly to track Justice Alito’s reasoning about the subsidy and commercial speech issues, see id. at 1765­–69, but apparently he believed Justice Alito wrote more than the case required about those issues.

  84. [84]. See id. at 1751 (plurality opinion).

  85. [85]. Id. at 1767 (Kennedy, J., concurring in part and concurring in the judgment).

  86. [86]. Sorrell v. IMS Health Inc., 564 U.S. 552, 580 (2011).

  87. [87]. Here we set aside regulations of expressive material that moves in commerce, such as the bar on sale of violent video games to minors that the Court struck down in Brown v. Entm’t Merchs. Ass’n, 564 U.S. 786, 804–05 (2011).

  88. [88]. See, e.g., Tamara R. Piety, “A Necessary Cost of Freedom”? The Incoherence of Sorrell v. IMS, 64 Ala. L. Rev. 1, 2–5 (2012).

  89. [89]. See United States v. Stevens, 559 U.S. 460, 468–73 (2010).

  90. [90]. See, e.g., Pleasant Grove City v. Summum, 555 U.S. 460, 464 (2009) (holding that placement of a monument in a city park is government speech).

  91. [91]. The most recent Supreme Court decision to engage with these categories and illustrate the difficulties they present for the Court is Agency for Int’l Dev. v. All. for Open Soc’y Int’l, Inc.,
    570 U.S. 205 (2013).

  92. [92]. See Matal v. Tam, 137 S. Ct. 1744, 1760–61 (2017) (plurality opinion).

  93. [93]. See id. at 1752–53.

  94. [94]. Id. at 1766 (Kennedy, J., concurring in part and concurring in the judgment).

  95. [95]. Id.

  96. [96]. See Robert C. Post, Constitutional Domains: Democracy, Community, Management 147 (1995); see also Robert C. Post, Viewpoint Discrimination and Commercial Speech, 41 Loy. L.A. L. Rev. 169, 173–74 (2007) (arguing that the legal substance of “viewpoint discrimination” necessarily depends on contestable theoretical commitments about the meaning of the First Amendment).

  97. [97]. See supra notes 61–69 and accompanying text.

  98. [98]. See Rosenberger v. Rector & Visitors of the Univ. of Va., 515 U.S. 819, 830–32 (1995).

  99. [99]. See Reed v. Town of Gilbert, 135 S. Ct. 2218, 2234–36 (2015) (Breyer, J., concurring in the judgment); id. at 2236–39 (Kagan, J., concurring in the judgment). Interestingly, Justice Kagan joined Justice Kennedy’s Tam concurrence, but Justice Breyer did not.

  100. [100]. Cf. Nat’l Endowment for the Arts v. Finley, 524 U.S. 569, 580–86 (1998) (rejecting a First Amendment challenge to a federal requirement that recipients of NEA grants satisfy “general standards of decency and respect for the diverse beliefs and values of the American public”). Justice Alito in Tam purported to distinguish Finley based on the absence in Tam of a cash subsidy. See Tam, 137 S. Ct. at 1761 (plurality opinion).

  101. [101]. See generally Wisconsin v. Mitchell, 508 U.S. 476 (1993) (holding that a Wisconsin statute providing for a sentence enhancement when a defendant selected a victim based on race did not violate a defendant’s free speech right).

  102. [102]. Abrams v. United States, 250 U.S. 616, 630 (1919) (Holmes, J., dissenting). That formulation, of course, faces all sorts of hard-hitting objections, both empirical and theoretical. See, e.g., Daniel E. Ho & Frederick Schauer, Testing the Marketplace of Ideas, 90 N.Y.U. L. Rev. 1160,
    1167–75 (2015); Stanley Ingber, The Marketplace of Ideas: A Legitimizing Myth, 1984 Duke L.J. 1, 16–48.

  103. [103]. Ariz. Free Enter. Club’s Freedom Club PAC v. Bennett, 564 U.S. 721, 747 (2011).

  104. [104]. City of Los Angeles v. Alameda Books, Inc., 535 U.S. 425, 436 (2002) (plurality opinion).

  105. [105]. See generally Lee Epstein et al., Foreword: Testing the Constitution, 90 N.Y.U. L. Rev. 1001 (2015) (introducing a symposium dedicated to expanding the use of empirical analysis in constitutional litigation and scholarship).

  106. [106]. See, e.g., Harris v. Quinn, 134 S. Ct. 2618, 2634 (2014); id. at 2657 (Kagan, J., dissenting) (dueling empirical arguments in a case about labor unions’ and nonunion workers’ speech); see also John Monahan & Laurens Walker, Social Science in Law: Cases & Materials 225–48 (8th ed. 2014) (discussing the use of empirical evidence in U.S. Supreme Court cases about sexually explicit speech).

  107. [107]. See generally Rebecca L. Brown & Andrew D. Martin, Rhetoric and Reality: Testing the Harm of Campaign Spending, 90 N.Y.U. L. Rev. 1066 (2015) (presenting and discussing an empirical analysis of the link between political spending and voters’ faith in democratic processes); Ho
    & Schauer, supra note 102 (presenting and discussing an empirical analysis of connections in public forums between speech restrictions and expressive activity).

  108. [108]. These are panelists identifying themselves as of Chinese, Filipino, Japanese, Korean, or Vietnamese ancestry.

  109. [109]. For a more extensive discussion of our research methods and findings, see generally James L. Gibson et al., Taming Uncivil Discourse, Version 90 (draft) (on file with the authors).

  110. [110]. Ashwander v. Tenn. Valley Auth., 297 U.S. 288, 347 (1936) (Brandeis, J., concurring).

  111. [111]. See U.S. ex rel. Attorney Gen. v. Del. & Hudson Co., 213 U.S. 366, 418 (1909).

  112. [112]. See U.S. Const. art. III, § 2, cl. 1.

  113. [113]. For leading critical assessments of constitutional avoidance, see generally Henry J. Friendly, Mr. Justice Frankfurter and the Reading of Statutes, in Benchmarks 196 (1967) (criticizing the practice of constitutional avoidance for its overreach and inconsistent application); Lisa A. Kloppenberg, Avoiding Constitutional Questions, 35 B.C. L. Rev. 1003 (1994) (discussing and criticizing “the justifications for the general avoidance doctrine”); John Copeland Nagle, Delaware & Hudson Revisited, 72 Notre Dame L. Rev. 1495 (1997) (explaining that Delaware
    & Hudson
    wrongly depended solely on the doubts canon when dealing with an ambiguous statute).

  114. [114]. Village of Hoffman Estates v. Flipside, 455 U.S. 489, 498 (1982) (quoting Grayned v. City of Rockford, 408 U.S. 104, 108–09 (1972)).

  115. [115]. Roberts v. U.S. Jaycees, 468 U.S. 609, 629 (1984).

  116. [116]. See generally Kolender v. Lawson, 461 U.S. 352 (1983) (striking down an anti-loitering law whose vagueness had led to inconsistent and arbitrary enforcement); Coates v. City of Cincinnati, 402 U.S. 611 (1971) (striking down as vague a statute that allowed criminal punishment of a peaceable assembly if a police officer found the assembly “annoying”).

  117. [117]. See Hoffman Estates, 455 U.S. at 498–99.

  118. [118]. See Note, The Void-for-Vagueness Doctrine in the Supreme Court: A Means to an End,
    109 U. Pa. L. Rev. 67, 75–76 (1960).

  119. [119]. See, e.g., Smith v. California, 361 U.S. 147, 151 (1959) (advocating rigorous vagueness review of “a statute having a potentially inhibiting effect on speech”).

  120. [120]. NAACP v. Button, 371 U.S. 415, 432–33 (1963).

  121. [121]. Hoffman Estates, 455 U.S. at 499.

  122. [122]. See VerSteeg, supra note 37, at 737–48 (conflating the vagueness and overbreadth doctrines); Robert H. Wright, Today’s Scandal Can Be Tomorrow’s Vogue: Why Section 2(a) of the Lanham Act Is Unconstitutionally Void for Vagueness, 48 How. L.J. 659, 676–81 (2005).

  123. [123]. See Megan M. Carpenter & Kathryn T. Murphy, Calling Bulls**t on the Lanham Act: The 2(a) Bar for Immoral, Scandalous, and Disparaging Marks, 49 U. Louisville L. Rev. 465, 473–78 (2011); Jessica M. Kiser, How Dykes on Bikes Got It Right: Procedural Inequities Inherent in the Trademark Office’s Review of Disparaging Trademarks, 46 U.S.F. L. Rev. 1, 10, 16–17 (2011).

  124. [124]. See In re Tam, 808 F.3d 1321, 1358 (Fed. Cir. 2015) (O’Malley, J., concurring). A prior District Court decision had declined to reach a vagueness challenge to the disparagement bar on grounds—ironic for the present discussion—of constitutional avoidance. See Pro-Football, Inc. v. Harjo, 284 F. Supp. 2d 96, 100–01 (D.D.C. 2003).

  125. [125]. Tam, 808 F.3d at 1359 (O’Malley, J., concurring) (emphasis omitted).

  126. [126]. See id.; see also Wright, supra note 122, at 678–81 (documenting arbitrary enforcement of § 2(a)).

  127. [127]. See Tam, 808 F.3d at 1362 (quoting Dambrot v. Cent. Mich. Univ., 55 F.3d 1177
    (6th Cir. 1995)).

  128. [128]. See id. at 1342 (Moore, J., majority opinion).

  129. [129]. See Village of Hoffman Estates v. Flipside, 455 U.S. 489, 498–99 (1982) (identifying the relative severity of criminal or civil penalties as a reason to deploy the vagueness doctrine).

  130. [130]. See Rex A. Collings, Jr., Unconstitutional Uncertainty—An Appraisal, 40 Cornell L.Q. 195, 223 (1955).

  131. [131]. 15 U.S.C. § 1052(a) (2012).

  132. [132]. Cf., e.g., Johnson v. United States, 135 S. Ct. 2551, 2558 (2015) (finding vagueness where the statutory language “le[ft] uncertainty about how much risk it takes for a crime to qualify as a violent felony”).

  133. [133]. See 1,000+ Artist Search Results for Square, AllMusic,
    artists/square (last visited Dec. 21, 2018) (documenting numerous bands with “square” as or in their names); 540 Artist Search Results for Circle, AllMusic,
    search/artists/circle (last visited Dec. 21, 2018) (“circle”); 146 Artist Search Results for Triangle, AllMusic, (last visited Oct. 6, 2018) (“triangle”); 1,000+ Artist Search Results for Line, AllMusic, (last visited Dec. 21, 2018) (“line”).

  134. [134]. Cf., e.g., Lanzetta v. New Jersey, 306 U.S. 451, 454, 458 (1939) (finding vagueness where the meanings of a statutory term “indicated in dictionaries and in historical and sociological writings are numerous and varied”).

  135. [135]. Cf., e.g., Winters v. New York, 333 U.S. 507, 519–20 (1948) (finding vagueness where “the specification of publications, prohibited from distribution, [was] too uncertain and indefinite to justify the conviction of [the] petitioner”).

  136. [136]. See, e.g., Broadrick v. Oklahoma, 413 U.S. 601, 612–15 (1973).

  137. [137]. U.S. Patent & Trademark Office, Trademark Manual of Examining Procedure
    § 1203.03(b)(i) (2017).

  138. [138]. Matal v. Tam, 137 S. Ct. 1744, 1754 n.3 (2017).

  139. [139]. See supra Part II. For arguments that the PTO should not enforce the disparagement bar against members of disparaged groups who seek to reclaim or reappropriate disparaging terms, see Farley, supra note 38, at 1044–48; Ingrid Messbauer, Beyond “Redskins”: A Source-Based Framework for Analyzing Disparaging Trademarks and Native American Sports Logos, 25 Fed. Cir. B.J. 241, 254–55 (2015).

  140. [140]. Pro–Football, Inc. v. Harjo, 284 F. Supp. 2d 96, 124 (D.D.C. 2003) (quoting Harjo v. Pro-Football, Inc., 50 U.S.P.Q.2d 1705, 1999 WL 375907, at *35 (T.T.A.B. 1999)), remanded,
    415 F.3d 44 (2005).

  141. [141]. It has some, though. Consider another illustration from the music world: a 1980s British rock band whose members chose the name “Redskins” because it distinctively signaled their allegiances to socialist politics and skinhead culture. See Redskins, AllMusic, http://www.all (last visited Dec. 21, 2018).

  142. [142]. Alexander M. Bickel, Foreword: The Passive Virtues, 75 Harv. L. Rev. 40, 62–63 (1961).

  143. [143]. Cass R. Sunstein, Foreword: Leaving Things Undecided, 110 Harv. L. Rev. 4, 25 (1996).

  144. [144]. See William N. Eskridge, Jr., Pluralism and Distrust: How Courts Can Support Democracy by Lowering the Stakes of Politics, 114 Yale L.J. 1279, 1316 (2005).

  145. [145]. For example, the Roberts Court in Holder v. Humanitarian Law Project, 561 U.S. 1 (2010), rejected a vagueness challenge to a federal statute that prohibited “service” to designated foreign terrorist groups. See id. at 16–17. The Court then made substantial First Amendment law in upholding the statute’s application to peace activists who sought to counsel groups in nonviolent conflict resolution. See id. at 7–10, 40.

  146. [146]. See, e.g., United States v. Five Gambling Devices, 346 U.S. 441, 453­­­­–54 (1953) (Black, J., concurring) (urging the Court to find a federal statute vague rather than use a different avoidance analysis, which implicated the Commerce Clause, to invalidate indictments under the statute).

  147. [147]. Brown v. Entm’t Merchs. Ass’n, 564 U.S. 786, 806–07 (2011) (Alito, J., concurring).

  148. [148]. See, e.g., Village of Hoffman Estates v. Flipside, 455 U.S. 489, 497 (1982).

  149. [149]. See, e.g., Coates v. City of Cincinnati, 402 U.S. 611, 614–15 (1971).

  150. [150]. The Supreme Court has stated that “economic regulation is subject to a less strict vagueness test because its subject matter is often more narrow, and because businesses, which face economic demands to plan behavior carefully, can be expected to consult relevant legislation in advance of action.” Hoffman Estates, 455 U.S. at 498 (citations omitted). That broad observation does not address the utility of vagueness analysis for avoiding more legally significant constitutional grounds for invalidating a law.

  151. [151]. Matal v. Tam, 137 S. Ct. 1744, 1765 (2017).

  152. [152]. Id.

  153. [153]. See Tushnet, supra note 38, at 416.

  154. [154]. Id.

  155. [155]. See Greene, supra note 58, at 433–40.

  156. [156]. See, e.g., United States v. Morrison, 529 U.S. 598, 628–36 (2000) (Souter, J., dissenting) (objecting strongly to the majority’s second-guessing of extensive congressional factual findings in striking down the civil damages provision of the Violence Against Women Act).

  157. [157]. Janus v. Am. Fed’n of State, Cty., & Mun. Emps., Council 31, 138 S. Ct. 2448, 2460 (2018) (holding that unions may not force public employees to pay union fees).

  158. [158]. See generally Commc’ns Workers of Am. v. Beck, 487 U.S. 735 (1988) (interpreting the National Labor Relations Act); Int’l Ass’n of Machinists v. Street, 367 U.S. 740 (1961) (interpreting the Railway Labor Act).

  159. [159]. Abood v. Detroit Bd. of Educ., 431 U.S. 209, 229–36 (1977).

  160. [160]. See Knox v. Serv. Emps. Int’l Union, Local 1000, 567 U.S. 298, 317–22 (2012) (tightening procedural constraints on unions’ separation of “chargeable” collective bargaining expenses from “nonchargeable” political advocacy expenses).

  161. [161]. See Harris v. Quinn, 134 S. Ct. 2618, 2638 (2014) (exempting certain state-funded home health care workers from the Abood allowance for unions to collect agency fees). For a critique of these decisions, see Magarian, supra note 26, at 203–20.

  162. [162]. See generally Friedrichs v. Cal. Teachers Ass’n, 136 S. Ct. 1083 (2016) (per curiam) (rejecting, by a 4–4 vote, a First Amendment challenge to Abood).

  163. [163]. See Janus v. Am. Fed’n of State, Cty., & Mun. Emps., Council 31, 138 S. Ct. 2448,
    2460–61 (2018).

  164. [164]. See id. at 9–10 (treating unions’ collection of agency fees as compelling objectors to subsidize unions’ political speech).

  165. [165]. See Abood, 431 U.S. at 222–24 (citing this free rider problem as a justification for public sector agency fees).

  166. [166]. The Roberts Court has criticized the lack of an empirical basis for other reasoning in Abood. See Harris v. Quinn, 134 S. Ct. 2618, 2634 (2014).

  167. [167]. See 15 U.S.C. § 78m(f)(1) (2012).

  168. [168]. Id.

  169. [169]. See Schauer, Politics and Incentives, supra note 25, at 1614; Schauer, Boundaries, supra note 17, at 1778–80.

  170. [170]. See W. Va. State Bd. of Educ. v. Barnette, 319 U.S. 624, 641–42 (1943).

  171. [171]. See NAACP v. Alabama ex rel. Patterson, 357 U.S. 449, 450–54, 466–67 (1958).

  172. [172]. See, e.g., Full Value Advisors, LLC v. SEC, 633 F.3d 1101, 1104 (D.C. Cir. 2011).

  173. [173]. See, e.g., City of Los Angeles v. Alameda Books, Inc., 535 U.S. 425, 457–60 (2002) (Souter, J., dissenting) (suggesting that empirical data would weaken the basis for the Court’s rejection of a First Amendment claim).


Professor of Law at Washington University in St. Louis.


Ethan A.H. Shepley Distinguished University Professor at Washington University in St. Louis.


Sidney W. Souers Professor of Government at Washington University in St. Louis.

We are deeply indebted to Steven S. Smith, Director of the Weidenbaum Center at Washington University in St. Louis, for his support of this research, both for allowing us to put our “Slants” module on The American Panel Survey and for funding an oversample of people of Asian ancestry. We also appreciate the research assistance of Dan Sicorsky.