University of Colorado Law Review > Printed > Volume 93 > Issue 1 > Text Is Not Enough

Text Is Not Enough

In Bostock v. Clayton County, the Supreme Court held that Title VII of the Civil Rights Act of 1964 protects gay and lesbian individuals from employment discrimination. The three opinions in the case also provided a feast for Court watchers who study statutory interpretation. Commentators across the ideological spectrum have described the opinions as dueling examples of textualism. The conventional wisdom is thus that Bostock shows the triumph of textualism. The conventional wisdom is wrong. Instead, Bostock shows what those who have studied statutory interpretation have known for decades: judges are multimodalists, drawing from a panoply of forms of legal argumentation. In particular, Bostock shows that judges are inevitably common-law thinkers, even when interpreting statutes.

PDF: Anuj C. Desai,* Text Is Not Enough


Bostock v. Clayton County,[1] 2020’s blockbuster Title VII gay rights case, has nothing to do with textualism. Nothing! Nada! Rien! Nichts! Everyone who says otherwise (and that’s pretty much everyone, no matter where they come out on the case[2]) is wrong. Okay, now I have your attention. Let me be more precise. The case has nothing to do with what many textualists claim is the core of the textualist method: the semantic meaning of statutory language. Title VII’s text has little to say about how to resolve the case. Bostock shows instead what anyone who has carefully studied statutory interpretation both before and after the supposed new textualist turn knows: in difficult, contested cases, statutory interpretation is unavoidably a multimodal enterprise that involves consideration of, at least, text, semantic context, statutory purpose, history (statutory, legislative, social, and political), social context, precedent, moral judgment, and consequentialist reasoning.[3]

For cases that reach the Supreme Court of the United States, this is almost inevitable. Moreover, because this multimodal process necessarily encompasses extratextual context, there is simply no Archimedean point for determining how much or what extratextual context a judge should consider.

Yet Bostock has been held up as the Platonic ideal for textualism, the case that would test the textualists on their textualist bona fides.[4] Indeed, the case is supposedly so textualist that it is said to reveal varieties of “textualisms.”[5] And, on the surface, perhaps there is a sliver of truth to all this. Perhaps “we are,” in Justice Kagan’s famous (infamous?) words, “all textualists now.”[6] But it is far more accurate to say “we are all multi-modalists now,” just as we all long have been. Even this supposedly textualist‑of‑all‑textualist case cannot be resolved with semantic meaning: to decide the case requires multiple modalities of analysis, including analogic, common-law-like reasoning. If textualism is supposed to entail determining the semantic meaning of the words the legislature enacted, textualism simply cannot resolve a case like Bostock or many of the difficult interpretive questions courts face.

This essay proceeds in three parts. Part I begins with a brief description of the three opinions in Bostock—Justice Gorsuch’s majority opinion and the dissents by Justices Alito and Kavanaugh—and how all three claim to be textualist.[7] I then briefly explain the conventional understanding of textualist interpretive methodology, focusing on two crucial aspects of that methodology: (1) textualists’ focus on the reader’s understanding of the statutory text; and (2) textualists’ view that statutory interpretation differs from common-law, analogic reasoning.

In Part II, I analyze the three principal arguments discussed by the parties and lower courts and at oral argument. Only one of the three purports to be a textualist argument, while the other two rely on analogies rather than text.

In Section II.A, I turn first to the argument the Court framed as “textualist”—compare the male plaintiff with a female to determine whether the male plaintiff’s discharge was “because of [his] sex.” Justice Gorsuch’s majority and Justice Alito’s dissent each chose different “comparators,” a straight woman versus a lesbian. Each Justice viewed his comparator as the correct application of the text of Title VII, yet the two came to opposing conclusions simply by virtue of choosing different comparators. Before examining the two other principal arguments, I take a brief interlude in Section II.B. There, I explain how the principle that statutes are to be interpreted “in context”—a principle with which all textualists agree—embeds a crucial ambiguity in any difficult interpretive question, an ambiguity that helps explain the differences between the majority’s and Justice Alito’s choice of comparator.

In Sections II.C and II.D, I then turn to the two analogical arguments. Section II.C addresses the sex-stereotyping argument, while Section II.D discusses the so-called “associational” argument. Both arguments—based as they are on precedent—rely on analogies, rather than directly on text, and thus embody common-law reasoning. Laying these arguments out in detail is necessary to demonstrate one core aspect of my thesis: there is no “textualist” way to choose between the majority’s textualist argument and Justice Alito’s response. Moreover, all three principal arguments in Bostock are logical equivalents, and to decide Bostock, textualism thus required analogic, common-law-like reasoning. To resolve the dispute in the case thus required a reliance on tools outside the textualist’s ordinary toolbox.

In Part III, I draw on some Title VII examples to explain further why the attempt to analyze the problem raised by Bostock through bare linguistic analysis will inevitably fail without an assist from other interpretive modalities. There are undoubtedly legal questions that text alone can resolve, but the question Bostock raises—like almost all cases that reach the Supreme Court—isn’t one of them.

I. Bostock and Textualism

In Section I.A, I briefly summarize the Bostock and Zarda cases, including the reasoning in each of the Supreme Court’s three opinions. In Section I.B, I provide a brief explanation of the conventional understanding of textualism, with an emphasis on two fundamental components of textualist methodology: (1) textualists’ focus on statutory readers rather than statutory drafters, and (2) textualists’ view that interpreting statutes (or, for that matter, any text) should be a distinct enterprise from common-law reasoning.

A. The Bostock and Zarda Cases

The relevant allegations in the Bostock and Zarda cases are pretty straightforward.[8] In each case, an employer allegedly fired a long-time employee simply for being gay. In Bostock, the plaintiff had

worked for Clayton County, Georgia, as a child welfare advocate . . . . After a decade with the county, Mr. Bostock began participating in a gay recreational softball league. Not long after that, influential members of the community allegedly made disparaging comments about Mr. Bostock’s sexual orientation and participation in the league. Soon, he was fired for conduct “unbecoming” a county employee.[9]

Zarda, also gay, had “worked as a sky-diving instructor . . . . As part of his job, he regularly participated in tandem skydives, strapped hip-to-hip and shoulder-to-shoulder with clients.”[10] Just before a tandem dive with one female client, Zarda told her that he was gay and that he “‘ha[d] an ex-husband to prove it.’ . . . [T]he client alleged that Zarda inappropriately touched her and disclosed his sexual orientation to excuse his behavior.”[11] Zarda’s boss then fired him shortly thereafter. “Zarda denied inappropriately touching the client and insisted he was fired solely because of his reference to his sexual orientation.”[12]

Both plaintiffs sued under Title VII of the Civil Rights Act of 1964. The interpretive question in the cases was whether the plaintiffs’ discharges were “unlawful employment practice[s]” under section 703(a)(1), which reads as follows:

It shall be an unlawful employment practice for an employer — (1) to fail or refuse to hire or to discharge any individual, or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual’s race, color, religion, sex, or national origin.[13]

The core question the cases raised, then, was whether an employer who fires someone because he is gay has done so “because of . . . sex” and has thus violated Title VII.

The Supreme Court held 6-3 that the answer is yes.[14] Speaking through Justice Gorsuch, the Court determined that an employer that fires an individual merely for being gay violates Title VII.[15] The core of the Court’s reasoning was that the “ordinary public meaning”[16] of the phrase “because of . . .” embodies a “but-for” causation standard[17] and that, because a person’s sexual orientation cannot be understood apart from that person’s sex, sex necessarily plays some role when an employer discharges an employee based on sexual orientation.[18] Importantly, the Court concluded that the text of the statute necessitated this result.[19]

The majority’s “textual” analysis compares Bostock with a hypothetical woman who is the same as Bostock in every other way, what I will call a comparator logic. The Court determined that if Bostock had been a woman and everything else had been the same (including Bostock’s attraction to men), Clayton County would not have fired him. Therefore, his sex was a “but-for cause” of his discharge, and his discharge was thus “because of [his] sex” within the meaning of Title VII. This comparator logic is the Court’s only “textual” argument, the only argument that purports to tell us what the “ordinary public meaning” of the statute is.[20]

The majority then turns to Title VII jurisprudence but claims that it does so only for “more support.”[21] It relies on prior cases to draw three lessons: (1) “it’s irrelevant what an employer might call its discriminatory practice, how others might label it, or what else might motivate it”; (2) “the plaintiff’s sex need not be the sole or primary cause of the employer’s adverse action”; and (3) “an employer cannot escape liability by demonstrating that it treats males and females comparably as groups.”[22]

Finally, the Court addresses the employers’ arguments head-on. Justice Gorsuch starts with the employers’ arguments rooted in text, concluding that “each of these arguments turns out only to repackage errors we’ve already seen and this Court’s precedents have already rejected.”[23] He then turns to the employers’ arguments based on legislative purpose and pragmatic consequentialism. His responses to these arguments rely on staples of the textualist’s rhetorical toolkit.

As for the employers’ argument about legislative purpose, Justice Gorsuch says that “legislative history can never defeat unambiguous statutory text.”[24] He then explicitly rejects the employers’ argument that Title VII should not protect gays and lesbians because the 1964 Congress would have expected the language not to protect them.[25] In particular, he concludes that this argument “proves too much. If we applied Title VII’s plain text only to applications some . . . group expected in 1964, we’d have more than a little law to overturn.”[26] After then describing decades of caselaw and administrative decisions that would also have been unexpected in 1964, Justice Gorsuch asks rhetorically, “Would the employers have us undo every one of these unexpected applications too?”[27]

The textualist rhetoric heats up even more when Justice Gorsuch responds to the employers’ consequentialist arguments:

With that, the employers . . . fall back to the last line of defense for all failing statutory interpretation arguments: naked policy appeals. . . . Gone here is any pretense of statutory interpretation; all that’s left is a suggestion we should proceed without the law’s guidance to do as we think best.[28]

No, no, no, Justice Gorsuch seems to say, our job is law, not policy: “[T]hat’s an invitation no court should ever take up. The place to make new legislation, or address unwanted consequences of old legislation, lies in Congress.”[29] Key to this claim that statutory interpretation must ignore consequences is what Justice Gorsuch calls “judicial humility”: “As judges we possess no special expertise or authority to declare for ourselves what a self-governing people should consider just or wise.”[30]

The decision drew two dissents, one from Justice Alito and one from Justice Kavanaugh. Although the two dissents differ in tone, they do share one common theme: what the majority did was “legislation,” not interpretation.[31] Some members of Congress have tried to amend Title VII to include sexual orientation, but no bill has ever passed.[32] Plus, numerous discrimination statutes treat “sex” and “sexual orientation” as distinct terms, and by negative implication, the former thus does not encompass the latter; that is, to include “sexual orientation” discrimination within the concept of “sex” discrimination is to conflate these two distinct forms of discrimination.[33] Moreover, by reading “sexual orientation” into Title VII, the Court’s decision undermines the political process[34] and is directly contrary to what the legislature did in 1964 when adopting Title VII’s prohibition on sex discrimination. In response to Justice Gorsuch’s claim that the majority exercised “judicial humility,”[35] Justice Alito couldn’t disagree more: he argues that the “arrogance of [the majority’s] conclusion is breathtaking.”[36] As to the majority’s claim to textualism, Justice Alito writes, “The Court’s opinion is like a pirate ship. It sails under a textualist flag, but what it actually represents is a theory of statutory interpretation that Justice Scalia excoriated—the theory that courts should ‘update’ old statutes so that they better reflect the current values of society.”[37]

On its face, then, Bostock looks like a debate about textualism. It thus seems to raise the question of which opinion in the case best represents the true textualism, the correct answer to the question of “What Would Justice Scalia Do?”[38] But in fact, as I make clear in Part II, there is no correct “textualist” resolution of the case. Applying “textualist” principles to the question Bostock raised cannot decide the case. Without assistance from some other modality of analysis, there is no correct “textualist” resolution of the case. Before we get there, though, let’s take a look at some of the tenets of textualism.

B. Textualism as Reader-Focused and as Deductive Reasoning

As a theory of legal interpretation, the standard account of textualism[39] is that interpreters should determine the “ordinary public meaning” of the text of a legal document, such as a statute.[40] Textualism tells interpreters to determine the statute’s “communicative content” and to emphasize the semantic meaning of the text.[41] Let me emphasize two points about textualism as a theory of interpretation. First, the standard account of textualism is a reader-focused approach to interpretation: interpretation should focus on the semantic meaning of the text as it “would reasonably be understood to mean” rather than on “what it was intended to mean.”[42] Second, textualism is generally understood to be based on deductive reasoning and thus to be distinct from common-law adjudication based on inductive reasoning.

First, textualists purport to focus on the statutory reader. By seeking the “ordinary public meaning” of a statutory text, the emphasis is on how the reader would understand the text rather than on how the writer intended it. To the extent that the textualist seeks “intent,” it is an “objectified” intent.[43] In particular, this “objectified intent” is premised on the notion that statutory “meaning” should be sought in what linguists refer to as “sentence meaning” rather than “speaker’s meaning.” Without getting too far into the weeds here, “speaker’s meaning” is the meaning an author intends the reader to glean, while “sentence meaning” is the meaning the text would have to a reader who is unaware of the speaker’s intention. It is as if, as Professor Solum has put it, “we were imagining a sort of generic speaker” who wrote the text “in a generic context.”[44]

Second, textualists purport to view statutory interpretation as fundamentally different from common-law judging. Statutory interpretation starts with a text whereas common-law decision-making does not. This is why Justice Scalia famously referred to judging in modern-day federal courts, a system of written law, as being “common-law courts in a civil-law system.”[45] He wanted courts in statutory cases to act more like civil-law courts and less like the common-law courts they had traditionally been.[46] In particular, the core approach embedded in textualism—take a text and apply its linguistic meaning to a set of factual circumstances—purports to be deductive: the judge is supposed to start with the objective meaning of the words of a statute (“ordinary public meaning”) and then apply that meaning to the case’s facts. In contrast, common-law decision-making depends on analogic reasoning: Are the facts of the case at bar relevantly similar to a previous one?[47]

My core claim is that Bostock cannot be understood through textualist reasoning alone. Using deductive logic to apply the semantic meaning of Title VII’s words to the facts of a case like Bostock cannot resolve the interpretive question. Instead, the case depends on analogic reasoning⎯the core of common-law judging. This, in turn, requires decision-makers to consider factors that textualists purport to eschew in the interpretation of statutory texts.[48] In particular, it requires judgment about similarities and differences in the real world, the bread and butter of the common law.

II. Analyzing the Principal Arguments in Bostock

My principal claim about the case is descriptive: Bostock shows that, notwithstanding the textualist rhetoric that pervades the case, cases involving statutory interpretation that reach the Supreme Court will inevitably require multiple modes of analysis. In such cases, statutory interpretation involves something akin to Professor Bobbitt’s notion of the “modalities” of constitutional argumentation.[49] I will refer to this multimodal approach to statutory interpretation as determining the “social meaning” or “social understanding” of the statute. The social meaning of a statute thus contrasts with the linguistic meaning.[50] All three opinions in Bostock ostensibly treat the goal of statutory interpretation as finding the “ordinary public meaning” of the statute.[51] By this, they purport to apply the statutory text’s linguistic meaning to the facts of the case before them. On the surface, then, they disagree simply about how much “context” to consider, not about the ultimate goal. The problem Bostock raised, though, makes it impossible to achieve that goal. Instead, Bostock required the Court to think about Title VII’s “social meaning.”

All the opinions may well have been premised on intuitions about the “ordinary public meaning” of the text, but such an inquiry simply cannot resolve the case. More importantly, a closer look at the reasoning of all three opinions shows that choosing between the majority and the dissents requires thinking about the development of the law since Title VII’s passage, including analogies with other caselaw—in other words, common-law thinking. Bostock shows us the common-law nature of the Supreme Court’s approach to statutory interpretation, notwithstanding all three opinions’ claims to being premised on textualist methodology.[52]

Understanding why ordinary textualist tools cannot resolve the case requires a deeper dive into several arguments, including some the majority purported to ignore. The employees made three principal arguments, and the multiple briefs, the lower court opinions, and oral argument engaged significantly with all three.[53] The majority adopted the first, the comparator argument: if Bostock had been a woman and everything else had been the same (including Bostock’s attraction to men), he would not have been discharged; therefore, his sex was a “but-for cause” of his discharge, and his discharge was thus “because of [his] sex” within the meaning of Title VII. Justice Alito’s response challenges the majority’s choice of comparator: he argued that the proper female comparator is a lesbian, not a straight woman. I address this argument in Section II.A.

In Section II.B, I look at an ambiguity in what it means for a textualist to interpret a statute “in context”: Should a textualist’s “context” include the social context surrounding a statute or just the linguistic context? This ambiguity helps us better understand Justice Alito’s opinion, including his claim that textualism requires consideration of social context.

I then turn to the employees’ second and third arguments, both of which draw on Title VII jurisprudence and thus require common-law, analogic reasoning. Although the majority did not address either argument, the lower court opinions and briefs discussed them extensively, and they were a focus of oral argument at the Supreme Court. The employees’ second argument, which I address in Section II.C, is the sex-stereotyping argument: as a gay man, Bostock failed to meet society’s expectations of what it means to be male, and firing him for this reason violates Title VII. The third, which I address in Section II.D, is the associational argument: discharging Bostock due to his same-sex relationship is analogous to discharging him for being in an interracial relationship, and because the latter obviously violates Title VII, so too must the former.

As the Court and most commentators have framed it, only the first argument, the comparator argument, is “textualist.” No one views either of the other arguments as “textualist.” Instead, the sex-stereotyping and associational arguments both rely on modalities other than textualism and, in particular, analogies drawn from precedent.

Yet, at the same time, the sex-stereotyping and associational arguments are logical equivalents of the comparator argument.[54] The comparator argument seems more like a “textual” argument than the other two because the other two sound in analogy and precedent. Yet the fact that they are logical equivalents to the comparator argument means that the comparator argument isn’t really a purely textual argument either. To decide whose comparator is “correct” thus requires common-law thinking. The text cannot resolve the dispute between the majority and Justice Alito’s dissent about the appropriate comparator. Both sides need arguments beyond a purely semantic analysis: text is just not enough.

A. The “Textualist” Argument: Compare Plaintiff with a Woman

The majority’s comparator argument goes like this: Bostock is a gay man. If Bostock had been a woman and everything else about him remained the same, he would not have been fired. Put another way, “but for” the fact that Bostock is a man, he would not have been fired. The employer is thus treating Bostock differently “because” he is a man. Ergo, the employer discriminated against him “because of . . . [his] sex.”[55] Importantly, to make the majority’s comparator argument work, the sex of Bostock’s partner must remain the same. In other words, the comparison is to a woman whose romantic partner is a man.

Justice Alito’s dissent says, in effect, “Wait a minute, not so fast. Your comparison is wrong. Although you say you are only changing Bostock’s sex, you have also changed Bostock’s sexual orientation: in your hypothetical, he’s now both a woman and straight. So, everything else about him did not stay the same.” For Justice Alito, then, the proper comparator is a lesbian, not a heterosexual woman.[56]

Justice Alito used a hypothetical that encapsulates this way of thinking perfectly: Consider an employer behind the veil of ignorance—one who does not know the sex of a job candidate but does know that the candidate is gay and who refuses to hire the candidate for that reason. Does that violate Title VII?[57] At oral argument, Bostock’s lawyer, the indomitable Pam Karlan, said no, but she then added that it didn’t matter because there had never been such a case.[58]

Justice Alito’s veil-of-ignorance hypothetical places the problem through the lens of what I will call a “single‑discrimination” framework: discrimination against an individual gay person of either sex is a form of discrimination against gay people in general. It is all the same form of discrimination. In contrast, the majority views the question through the lens of what Professor Koppelman has called “parallel discriminations”:[59] discrimination against a gay male is distinct from discrimination against a gay female, and so there are two different discriminations. So, even if an employer discriminates against all gay people, that just means that both gay males and gay females are being discriminated against “because of . . . sex.”

Notice how these two different characterizations of the proper comparator do not depend on the semantic meaning of Title VII’s language (which, again, in relevant part, is the phrase “discharge an individual [or discriminate against an individual with respect to compensation, terms, conditions, or privileges of employment] . . . because of such individual’s . . . sex”).[60] Neither opinion appeals to the definition of any word or set of words to defend its comparator against the competing comparator. Both make a claim about the semantic meaning of the words; both depend on a claim that “because of” embodies but-for causation. Yet, to go from one argument to the other requires almost a gestalt shift in thinking. Nothing in the structure of the sentence or the definition of “sex” or “because of” or “discriminate” or even “individual” helps us choose which of the two comparators (straight woman v. gay woman) to select. In other words, even the comparator argument, the one that everyone seems to agree is a “textualist” argument, cannot be resolved with text.

Many readers obviously find one of these comparator arguments more persuasive than the other. If you are in either camp—and some extremely smart and sophisticated thinkers are[61]—then you might stop reading right now.[62] But my claim requires that such readers recognize that the opposite comparator argument is at least a plausible frame through which to interpret the text as a semantic matter. To accept my argument, you don’t have to concede that the opposing argument is the best reading of the text. You need only accept that someone whose sole goal is to apply a semantic approach to interpretation cannot, on purely semantic grounds, choose between the straight woman and the lesbian as comparator. Nothing in the linguistic meaning of the words inherently tells us which of these conflicting arguments is correct.[63]

In a recent article, Professors Berman and Krishnamurthi argue that the phrase “because of” requires choosing a lesbian as the proper comparator.[64] Their argument rests on the idea that even a but-for causal framework requires distinguishing between “a valid, or plausible counterfactual” and “an invalid, implausible, or downright silly one.”[65] They note that choosing the comparator in a but-for causation analysis—what they refer to as the proper “counterfactual”—requires determining how “close” the hypothetical world (here, the chosen comparator: lesbian or straight woman) is to the actual world (here, the fact that Bostock is a gay man).[66] For assessing Title VII causation, they posit what they refer to as a “principle of conservation in motivational analysis,” and applying that principle, they suggest that the better comparator for Bostock is a lesbian, not a straight woman.[67]

The claim rests on the following logic. Bostock has three facts of relevance: he is (1) male; (2) gay; and (3) attracted to men. By changing Bostock’s sex to female but leaving him attracted to men, the Bostock majority has not left everything else constant: it has changed fact 2 as well as fact 1. Indeed, there is no way to change fact 1 without also changing either fact 2 or fact 3. Since, they argue, the employer’s avowed motivation is that Bostock is “gay,” the better approach is to hold fact 2 constant and make the comparator a lesbian, rather than a straight woman.[68]

The “principle of conservation in motivational analysis” strikes me as Professors Berman and Krishnamurthi’s attempt to formulate an approach to what Professor Schauer has referred to as “attributive” causation in the context of employment discrimination.[69] The crux of their argument rejects the idea that but-for (deterministic or logical) causation in the literal sense could possibly be the actual law. Rather, any legal analysis of causation requires some version of “attributive” causation: it requires choosing from among the many “but-for” (i.e., logical) causes the one(s) that really matter.[70]

Their formulation is not unreasonable, but it does appear just to be another way—a far more sophisticated way, to be sure—of restating Justice Alito’s chosen comparator. It depends on an assumption about a social construction of “sexual orientation”—that is, that certain individuals in our society, of both sexes, fall into a single category, “gay,” and that that category is meaningful. Without that assumption, what they refer to as fact 2 (Bostock is gay) cannot be treated as distinct from the combination of fact 1 and fact 3 (Bostock is male and is attracted to men). And without “Bostock is gay” as a distinct fact, there is no way to claim a lesbian should be the proper comparator. This is, of course, the very assumption that Justice Alito brings to his chosen comparator.[71] In essence, then, applying their “principle of conservation in motivational analysis” to Bostock simply reformulates the choice-of-comparator problem. It does not solve it.[72]

B. Textualism, “Context,” and the Unexpected Application of Text

Textualists generally agree that judges should interpret statutes in context. But the phrase “in context” is ambiguous: “context” could mean linguistic context (looking at surrounding language, the Whole Act Rule, or even the Related-Statutes Canon[73]), but it can also mean the social context[74] (ranging from statutory purpose to the broad moral, political, or social assumptions that drafters and readers would have shared at the time of the statute’s passage).[75] If “context” is limited to linguistic context, then the primary focus of interpretation can be limited to semantic analysis.[76] If not, though, then the interpreter needs to make some kind of social judgment: the interpreter needs to decide which aspects of the social context of the language should count and which should not.[77]

Justice Alito explicitly says that textualism requires consideration of “societal norms.”[78] He explains, “when textualism is properly understood, it calls for an examination of the social context in which a statute was enacted because this may have an important bearing on what its words were understood to mean at the time of enactment.”[79] Although Justice Alito does have his own comparator argument,[80] most of his opinion has nothing to do with the comparator. Instead, his core claim is that the social context surrounding the 1964 Civil Rights Act makes it unimaginable that the statute protects gays and lesbians. A competent reader of the English language in 1964, he argues, would not have understood Title VII to encompass the plaintiffs’ claims.[81] This is clear, Justice Alito says, in part, because “[a]ny such notion would have clashed in spectacular fashion with the societal norms of the day.”[82] He contends that true textualism demands considering social context.[83] Implicitly, then, he recognizes that semantic meaning is insufficient to resolve the case.

Justice Alito’s focus on the social context of 1964 directly raises another fundamental question about statutory interpretation that has no preordained answer: When does a statute apply to circumstances not contemplated by its drafters? The answer is obviously not never. As the Court, speaking through Justice Scalia in a Title VII case involving allegations of male-on-male sexual harassment famously—and unanimously—put it, “statutory prohibitions often go beyond the principal evil to cover reasonably comparable evils, and it is ultimately the provisions of our laws rather than the principal concerns of our legislators by which we are governed.”[84]

But neither Justice Alito’s appeal to social context nor Justice Gorsuch’s refusal to do so tells the textualist how to answer this question. The principal reason a competent speaker of the English language in 1964 may not have understood Title VII’s language to encompass Zarda’s claim is that most people at that time would have shared certain social and cultural assumptions about what the language could have meant. The employees’ comparator argument just wasn’t on many people’s radar.[85] Justice Alito then goes one step further to argue that most readers would have categorically rejected the employees’ argument if someone had raised it.[86] The problem with this argument, though, is that the same argument could plausibly be made about a lot of Title VII jurisprudence, including hostile-workplace environment claims,[87] male-on-male sexual-harassment claims,[88] interracial relationship claims,[89] and sex-stereotyping claims.[90] That fact does not help us beyond the ever-present problem of applying statutes to situations not addressed (or perhaps not even contemplated) at the time of a statute’s passage. Everyone agrees that statutes are not limited solely to their original, expected applications and that statutes always have the potential to raise what Professor Nelson has helpfully called the problem of “[u]nexpected [a]pplications of an [i]nvariant [d]irective.”[91]

But Justice Alito’s confident conclusion about “how Americans in 1964 would have understood Title VII’s prohibition of discrimination because of sex”[92] conflates the distinction between linguistic drift and drift in readers’ moral, political, or social assumptions about what the statutory language could possibly mean. His claim about 1964 seems stronger because of the unexpected real-world consequences of the plaintiffs’ comparator argument. The semantic meaning of the words has not relevantly changed since 1964, but social understandings of the world have.[93]

This assumption about a competent speaker of the English language in 1964 just takes us one step down a logic rabbit hole: if the judge’s job is to determine the original ordinary public meaning of the statutory text, we need then to ask how much social context the ordinary reader in 1964 would bring to the text[94] rather than how much social context the judge should bring to the text. According to Justice Alito, if the ordinary reader in 1964 brings his or her attitudes about homosexuality to the interpretive task, that reader would likely reject the plaintiffs’ claims. If not, however, we just don’t know. Either way, that too is a question that the textualist’s ordinary toolkit—dictionaries, linguistic canons, etc.—cannot answer.[95]

Justice Alito’s reliance on social context also dovetails with one of his primary accusations against the majority: that the Court’s decision amounts to “legislation.”[96] As Justice Alito sees it, the textualism practiced by the majority, divorced as it was from social context, can lead to results that the legislators who adopted the statute—and the public that engaged with it at the time of adoption—would have unequivocally rejected.[97] If one core value of textualism is supposed to be judicial modesty and deference to the legislature,[98] then a textualism shorn of social context presumably undermines that core value.

But Justice Alito has not articulated any explicit basis for distinguishing appropriate from inappropriate unexpected applications. It seems as though he just views the majority’s conclusion here as so unexpected as to render the conclusion illegitimate. The social context of the world in 1964, he seems to say, can help us understand when a judicial decision has gone too far, so far that it amounts to “legislation.” Yet Justice Alito has not provided any basis for distinguishing appropriate applications of Title VII from inappropriate applications based on the text of the statute alone.

In the next two sections, I turn to the employees’ two other principal arguments⎯arguments based on common-law reasoning⎯that played prominent roles in the lower courts, briefs, and oral arguments. In doing so, I want to emphasize two important points about both arguments: (1) logically, both arguments map precisely onto the comparator argument; and (2) both arguments depend on analogic, rather than deductive, reasoning. Once we see the connection between the comparator argument and these two other arguments, we can see that analogic reasoning is effectively doing the real work in the comparator argument. To choose between the two comparator arguments thus requires deciding about the appropriateness of particular analogies, the bread and butter of the common law. If the comparator argument is “textualist,”[99] there is no way to be a textualist in a case like Bostock without resorting to common-law reasoning.

C. Sex-Stereotyping Argument

The employees’ second argument is based on nonconformity with a sex stereotype. According to society’s traditional stereotypes, men who are not attracted to women are not sufficiently “masculine,” and women who are not attracted to men are not sufficiently “feminine.” An employer that fires a gay employee is thus acting on the basis of a sex stereotype, which is thus in turn “because of . . . sex.”

Key here is that the sex-stereotyping argument is not rooted directly in text but instead in precedent. The argument stems from a long line of Title VII jurisprudence, relying particularly on a plurality opinion in the seminal 1989 Price Waterhouse v. Hopkins case.[100] In Hopkins, there was evidence that Ann Hopkins’ employer, Price Waterhouse, had refused to promote her to a partnership in the firm because she walked like a man, did not “wear make-up,” was too aggressive for a woman, and was insufficiently “feminine.”[101] A plurality of the Court concluded that this evidence was enough to go to a jury on a sex-discrimination claim if Price Waterhouse would have promoted a man who had acted in the same way.[102]

Under this argument, sexual-orientation discrimination is just an example of the exact same kind of discrimination Hopkins experienced. The basic idea was for, say, a lesbian plaintiff to frame an employer that discriminates against her as discriminating against her for being too “masculine” rather than for being a lesbian. Of course, prior to Bostock (and its immediate Second and Seventh Circuit predecessors Zarda[103] and Hively[104]), it didn’t violate Title VII to discriminate against an employee for being a lesbian. So, courts (and juries) were forced to distinguish between discrimination based on the sex-stereotyping (being too “masculine”) and discrimination based on sexual orientation, a task that at times required significant mental gymnastics.[105] In 2017, in Hively v. Ivy Tech Community College, the Seventh Circuit en banc finally concluded that there was no distinction between the two types of claims: “Our panel described the line between a gender nonconformity claim and one based on sexual orientation as gossamer-thin; we conclude that it does not exist at all.”[106] By this point, everyone’s social understanding of Title VII was that it prohibited employers from firing a woman for being too “masculine” or a man for being too “effeminate.” By characterizing an employer’s attitudes about gays and lesbians as simply an instantiation of sex stereotyping, the Seventh Circuit thus used precedent—and common-law analogic reasoning—rather than text to resolve the question of whether Title VII prohibits discrimination based on sexual orientation.

Although the Bostock majority did not rely on the sex-stereotyping argument, it did allude to the argument a couple of times.[107] More importantly, Justice Alito explicitly rejected it. How he did so, though, tells us something about the pull that precedent—here, the decades of jurisprudence on sex stereotyping—had on the textualist approach he claimed to adopt.[108] Recall the core logic of Justice Alito’s comparator argument: compare Zarda with a lesbian, and since a lesbian would have also been fired, Zarda was not discharged because of his sex. Moreover, Justice Alito’s argument relies on a rejection of the majority’s comparator logic: you cannot compare Zarda with a heterosexual female, because if you do that, you are comparing Zarda with an employee “who differ[s] in two ways—sex and sexual orientation.”[109]

This same logic, however, could apply equally to a sex-stereotyping claim. How so? Think for a moment about the logic of a claim based on sex-stereotyping evidence: it too can rely on a comparator logic. If Ann Hopkins had been a man who acted the way she did, she would have made partner. In other words, hold everything else constant but change her sex, and Price Waterhouse would have made her partner. That, as Justice Gorsuch might have put it, satisfies Title VII’s but-for causation standard. But the sex-stereotyping comparator argument has a counterargument that runs directly parallel to the counterargument to it in the sexual-orientation cases: you haven’t held everything else constant. By changing the employee’s sex, you have also changed the employee’s sex conformity.[110] A man who acted as Ann Hopkins did would have made partner, but he would also have been a sex-conforming individual. So, you’ve changed two things, not just one: Hopkins’ sex and her sex conformity. The employer could thus characterize the disparate treatment as due to sex conformity, not sex: we want feminine females and masculine males, or sex-conforming employees. Key is that sex conformity is just like sexual orientation, a non-sex-specific trait, because it can be applied to both sexes.[111] Conforming to the stereotypes of one’s sex is no more a sex-specific trait than is being attracted to those of the opposite sex.[112] That is why the sex-stereotyping argument is logically identical to the comparator argument.[113]

But Justice Alito did not make that argument, presumably because he could not do so without contradicting Hopkins. Instead he danced around the fact that the same logic he applied in Bostock could easily apply to Hopkins, and if it had been, Hopkins would have come out differently. He begins his analysis of Hopkins by noting that “‘Title VII creates no independent cause of action for sex stereotyping’, but . . . that ‘[e]vidence of use by decisionmakers of sex stereotypes is . . . quite relevant to the question of discriminatory intent.’”[114] He then argues that, in cases involving discrimination based on sexual orientation, “the grounds for the employer’s decision . . . apply equally to men and women.”[115] He then concedes that there may nonetheless “be cases where traits or behaviors that some people associate with gays [or] lesbians . . . are tolerated or valued in persons of one biological sex but not the other.”[116] But that, he then concludes, “is a different matter.”[117]

But why is that a different matter? It depends on characterizing sexual-orientation discrimination as a single form of non-sex-specific discrimination and characterizing sex-stereotype discrimination as two different forms of discrimination, one against men and another against women. This should sound familiar. It is exactly the single-discrimination versus parallel-discriminations disagreement between Justice Alito and the majority that the two comparator arguments raised.[118] Yet here Justice Alito seems to acknowledge that evidence about sex-stereotyping could be sex-specific discrimination. It could be two distinct forms of discrimination: discrimination against a woman because she is, say, “aggressive” (and thus acts counter to the stereotype of a woman) would be distinct from discrimination against a man because he is, say, “passive” (and thus acts counter to the stereotype of a man). Yet one could also characterize these two discriminations as a single discrimination based on, say, “sex conformity based on level of assertiveness.”

The key point here is that there is nothing inherent in the text of Title VII that tells us that sex and “sexual orientation” should be treated as distinct characteristics for purposes of understanding the semantic meaning of the words (as Justice Alito would have held) while sex and “sex conformity based on level of assertiveness” should not be (as Title VII’s sex-stereotyping jurisprudence had long held).[119] A purely semantic understanding of the words cannot tell us one way or the other. Only a social understanding of what constitutes “sex discrimination”⎯a concept that has evolved over time through common-law development⎯is able to tell us (courtesy of the sex-stereotyping jurisprudence) that “sex conformity based on level of assertiveness” is not a sufficiently distinct characteristic from sex that discrimination on that basis warrants being treated as sex discrimination. Likewise, only a social understanding of sex discrimination can tell us whether sexual orientation is a sufficiently distinct characteristic from sex that discrimination on that basis should not be treated as sex discrimination.

D. Associational Argument

The employees’ third argument, the associational argument, was that discriminating against Bostock was no different from discriminating against an employee in an interracial relationship.[120] As with the sex-stereotyping argument, the majority ignored the associational argument altogether, while Justice Alito’s dissent rejected it. The argument depended on two claims: (1) the fact that Title VII’s operative language treats “sex” and “race” identically, and (2) a belief that it would surely violate Title VII’s prohibition on race discrimination for an employer to fire someone for being in an interracial relationship. The argument relies in part on text—a parallel structure in the statutory text—but mostly on an analogy, an analogy between same-sex and interracial relationships. The importance of that analogy to the persuasiveness of the associational argument is what again demonstrates the pull of common-law reasoning on this supposedly textualist case.

The associational argument takes the principle from Loving v. Virginia,[121] incorporates it into Title VII, and then applies it to same-sex relationships. The argument goes like this:

Premise A: Section 703(a)(1)’s language makes no distinction between race and sex, and if a race-related adverse employment action qualifies as being “because of race” under section 703(a)(1), then an equivalent sex-related adverse employment action must likewise qualify as being “because of sex” under section 703(a)(1).[122]

Premise B: It qualifies as “because of race” (and thus violates Title VII) to take an adverse employment action against an employee because that employee is in an interracial relationship.

Conclusion: It qualifies as “because of sex” (and thus violates Title VII) to take an adverse employment action against an employee because that employee is in a same-sex relationship.

To start, notice that this argument does not directly sound in text. It does not seem to be based on semantics. Rather, it depends on an analogy to Loving v. Virginia, a case about the Equal Protection Clause, which has a completely different linguistic formulation from Title VII.[123] It could, of course, be that deductively applying the text of both the Equal Protection Clause and Title VII happens to yield the same result, but that is not how the argument is formulated. Rather, the associational argument depends on analogic reasoning.[124] In that sense, it seems quite different from the comparator argument.

But if the comparator argument is a “textual” argument, so too is the associational argument; both Premise A and Premise B sound in “text.” Indeed, the argument has a much closer connection to the comparator argument than first appears. Premise B is identical in structure to the Bostock majority’s claim that the proper comparator for Bostock is a straight woman.[125] Moreover, Premise A is at core a “textual” argument too: it derives from the fact that the words “race” and “sex” are both objects of the same prepositional phrase in the statute.[126] So, the associational argument amounts to a combination of the majority’s comparator argument applied to race, plus the fact that the semantic structure of the statute requires that race and sex cases be treated the same. We could thus call the employees’ associational argument a purely “textual” argument, despite the fact that what gives it rhetorical power is the analogy between same-sex and interracial relationships.

Let us now look at each of the premises a little more closely. I want to start with Premise B. As we will see, although Premise B involves race, not sex, it is virtually identical to Justice Gorsuch’s comparator argument.

1. Discrimination Against an Interracial Relationship Violates Title VII

Premise B is the claim that Title VII prohibits an employer from discriminating against an individual for being in an interracial relationship. Some lower courts had so explicitly held,[127] but this is almost irrelevant. That Title VII couldn’t possibly be interpreted to permit such discrimination is treated like a premise in the literal sense, much as no constitutional theory today could conclude that Brown v. Board of Education was wrongly decided.[128] Everyone accepted that discrimination against an individual for an interracial relationship violates Title VII.

But before we turn to the heart of the dispute between the employees and Justice Alito’s dissent⎯whether the connection between interracial relationships and race discrimination should be treated like the connection between same-sex relationships and sex discrimination⎯we need to look a little closer at how Premise B directly parallels the Bostock majority’s comparator argument. Recall the core of the dispute between the majority and Justice Alito: for a gay male employee, the majority’s comparator is a straight woman, while Justice Alito’s is a lesbian. Again, the dispute is about what changes. The majority changes the sex of the employee and the sexual orientation of the employee, while Justice Alito’s dissent changes the sex of the employee and the sex of the employee’s partner. So, how can we think of the two sides of the comparator-argument dispute as to race?

Majority: A person of race (chromosome?) XX who is in a relationship with a person of race XY is treated differently from a person of race XY who is in a relationship with a person of race XY. Holding everything else constant, the person of race XX is being treated differently from an equivalent person of race XY.

Dissent: Wait a minute, that’s not the right comparison. You’ve changed the person’s race from XX to XY, but you’ve also changed the person’s racial orientation from hetero-race to homo-race. The proper comparison is to change the person’s race (from XX to XY) but to keep the person’s racial orientation the same (namely, hetero-race). That would require changing the employee’s race and the partner’s race. The employer isn’t discriminating against the employee because of “such individual’s race” but is instead discriminating because of that individual’s “racial orientation.” Those are different concepts, and the fact that the statute didn’t use the term “racial orientation” means, by negative implication, that discrimination based on racial orientation is excluded from the statute’s coverage.[129]

In a recent article, Professor Koppelman builds on his decades of scholarship and an amicus brief in Bostock to draw this out explicitly. To counter the dissenters’ comparator argument—that men and women are being treated similarly because the employer discriminates against gay men and lesbians equally—Koppelman draws the Loving analogy using a wonderfully evocative term:

Suppose an employer who rejects employees who are in interracial relationships claimed that it was merely discriminating against “miscegenosexuals,” and that the law’s protection of African-Americans should not be extended to an entirely different category of people? The only difference between the two responses is that here the neologism is unfamiliar. The flaw in both responses is the same: in any individual case, a person is discriminated against for being the wrong race or sex.[130]

But we can take Professor Koppelman’s point and frame the question differently. We have a term that covers the category of people of both sexes in same-sex relationships—namely, “gay” or “homosexual.” But for interracial relationships, we do not have such a term; that is, “miscegenosexual” simply isn’t a word. The question, then, is why not? The answer has nothing to do with linguistics or the semantic meaning of the words “race” or “sex” or any of the other words in Title VII. The answer depends on our social understanding of the world. Gays exist as a distinct category with a social and political identity that we recognize as making that identity socially salient, whereas those in interracial relationships do not. Professor Koppelman is implicitly correct that any defense of the Bostock dissent requires an explicit recognition of that distinction. At the same time, though, any defense of the Bostock majority needs to address directly why that is a distinction without a difference.

In short, everything about both sides of Premise B can be mapped logically onto the parallel side of the comparator argument. One can reject the employees’ arguments, but to reject the associational argument requires rejecting it as to race or distinguishing race and sex. But everyone, including Justice Alito, agreed with Premise B, that an adverse employment action taken against someone in an interracial relationship qualifies as discrimination “because of race.”

2. Title VII Requires Same Treatment of Race and Sex

The key dispute, therefore, is Premise A—whether race and sex cases must be governed by the same comparator analysis. Although the majority did not explicitly address the question, it obviously would have said yes.

Justice Alito, in contrast, said no. To distinguish between an interracial relationship and a same-sex relationship, he frames discrimination against an interracial relationship through what amounts to an anti-subordination lens:[131] he writes that an employer who fires an employee in an interracial relationship “is discriminating on a ground that history tells us is a core form of race discrimination.”[132] In contrast, “[d]iscrimination because of sexual orientation is different,” he says, because it “is not historically tied to a project that aims to subjugate either men or women.”[133]

In other words, when an employer fires an employee because of an interracial relationship, we understand the social meaning of that act: it is part of a badge of inferiority that the employer is imposing on one race—namely, those who are Black—because everyone in American society understands it as such.[134] Importantly, this argument about interracial relationships applies to Black and White employees alike. Either way, it’s discrimination against one race rather than discrimination against the association (i.e., the connection) per se.[135] Thus, discrimination against someone in an interracial relationship is discrimination against those who are Black, the very group that was the primary target of the 1964 Civil Rights Act’s protection.[136] In contrast, discrimination against those in a same-sex relationship is not discrimination against one sex or the other. Why not? Because, at least according to Justice Alito, our social understanding of discrimination against gays and lesbians is not discrimination against one sex or the other.[137] Rather, it’s discrimination against the association itself.

Yet there is an obvious response to Justice Alito that explicitly incorporates social context, one that the original panel in Hively v. Ivy Tech highlighted. Discrimination on the basis of sexual orientation is sex discrimination for the same reason sex stereotyping is sex discrimination: treating an employee differently because of the sex of the employee’s partner is inextricably bound up with preserving traditional gender hierarchies, even if it happens to both sexes equally.[138] It may be, as Justice Kavanaugh put it in his dissent, that “Seneca Falls was not Stonewall,”[139] but that doesn’t mean that discriminating against employees in same-sex relationships—gay males and lesbians alike—doesn’t perpetuate discrimination against women. My point is not to make that argument or say that it is unequivocally correct; rather, my point is simply that one needs to make an argument of that sort to defend the majority’s conclusion. The majority’s conclusion cannot be defended solely based on the argument that the semantic meaning of the text dictates the result. It requires a social understanding of the meaning of sex discrimination.

The Bostock majority, though, fails to make that argument. Rather, by relying solely on its comparator logic—Bostock is being treated differently because he is a man—the majority’s approach seems implicitly to rely on an anti-classification theory. But, of course, as I explained in Section II.A, the problem is that an anti-classification theory cannot resolve the case. An anti-classification mindset just returns us to the core dispute about the comparator. Nothing about an anti-classification theory can tell us whose comparator is correct, the majority’s or the dissent’s. This is why the majority’s characterization of its method betrays a fundamental misunderstanding of how to decide the case.[140]

Moreover, there is a textual counterargument to Justice Alito’s anti-subordination premise. The statute doesn’t talk about “races” or “sexes” in the plural, and it prohibits certain employment actions against an individual because of “such individual’s” race or sex.[141] Justice Alito’s anti-subordination theory would seem to run into trouble with the White employee in an interracial relationship: if the social import of the employer’s action is Black inferiority, then the action was not “because of” the employee’s race at all, but solely because of the race of the employee’s partner. This same problem would similarly arise with the argument I raised in the previous paragraph—that discrimination based on sexual orientation is anti-subordination as it applies to women—if the employee were a man.

3. Title VII’s Text Alone Cannot Distinguish the Race and Sex Associational Arguments

Just as importantly, race and sex are treated the same in the text itself, the basis of what I referred to earlier as Premise A.[142] As a purely textual matter, if the race-associational argument works, so too must the sex-associational argument. Put another way, there is no semantic reason why sex and race should be treated differently. The Bostock dissenters are thus faced with a choice if they want to prioritize semantics. They can apply their own comparator argument to both sex and race cases, but if they did this, it would allow employers to discriminate against persons in both interracial and same-sex relationships. Or they can accept the Bostock majority’s comparator argument for both sex and race cases, which would prohibit both types of discrimination.

A semantic analysis of the text does not compel one choice or the other, and so it does not compel the majority’s choice. But it does mean that if the dissenters want to claim textualism’s mantle, they cannot have it both ways. Title VII either prohibits discrimination against persons in both interracial and same-sex relationships, or it prohibits neither of these forms of discrimination. But nothing in the text tells us which of these two options is correct. In other words, since nothing in the text tells us whether an employee who is discharged because of an interracial relationship violates the statute,[143] nothing in the text tells us that Title VII prohibits discrimination against persons in same-sex relationships either.[144]

Thus, whether or not one believes race and sex discrimination under Title VII must be treated the same,[145] there is no textual or semantic way to resolve either Bostock or an interracial relationship case. Without an understanding of the social meaning of what race discrimination and sex discrimination are and how similarly the law ought to treat them, there is no way to decide whether (1) both interracial relationships and same-sex relationships are protected by Title VII, (2) neither is protected by Title VII, or (3) one is protected and the other is not. Separate-but-equal race-segregated bathrooms probably “discriminate against” a Black employee “with respect to his . . . terms, conditions, or privileges of employment, because of such individual’s race.”[146] In contrast, at least in 2021, courts will probably not treat separate-but-equal sex-segregated bathrooms in the workplace as “discriminat[ing] against” either a male or a female employee with respect to the employee’s “terms, conditions, or privileges of employment, because of such individual’s . . . sex.”[147] There is no way to distinguish these based on semantics alone. The interpreter needs to have some understanding of the social meaning of the differential treatment.

Sophisticated textualists argue that semantic analysis must be enriched by pragmatics.[148] The empirical questions raised by such an approach will almost inevitably yield inconclusive results, as they do here.[149] That’s why cases like Bostock are hard. If the dictionaries, corpus, and surveys all pointed unequivocally in one direction, we almost certainly would not have significant debate about the semantic meaning of the text. But when semantics yield plausible competing interpretations, the question of how much social context to consider will come to the fore. And because there is no way to reconcile the competing semantic interpretations without social context, and there is no preordained way to determine how much and what social context to consider, courts have to rely on, and thus inevitably will rely on, other modes of analysis.

In short, there is no single correct semantic meaning of Title VII’s language without claims about the social understanding of the statute: the text is underdeterminate. Just as importantly, choosing which of the two readings to adopt requires making implicit claims about not just sex discrimination but race discrimination too. It requires making a claim about the social understanding of same-sex and interracial relationships. There is simply no way to do a “textual” interpretation of Title VII without some extratextual judgment.

III. Textualism and the Common-Law Method: Variations on a Theme

One other way to see how textualist tools cannot resolve Bostock is to return to what I referred to earlier as the “single-discrimination” versus “parallel‑discriminations” debate that shapes the choice of comparator.[150] Recall that Justice Alito’s dissent effectively treats the discharge of a male employee because he is gay the same as the discharge of a female employee because she is a lesbian: both are examples of a single form of discrimination, discrimination because of sexual orientation, and are therefore not discrimination because of sex. The majority, in contrast, sees them as distinct: the gay male is discharged, at least in part, because of his sex, and the lesbian is likewise discharged, again at least in part, because of her sex. As I noted, nothing about the semantic meaning of the words of Title VII can resolve that dispute.

This becomes even clearer if we think about any number of now-paradigmatic examples of sex discrimination through the single‑discrimination versus parallel‑discriminations lens. Let’s start with a prototypical quid pro quo sexual‑harassment claim: A heterosexual, male employer says to a female employee, “Sleep with me, or I’ll fire you.” If she doesn’t and he fires her, that would constitute the “discharge” of an “individual because of that individual’s . . . sex.”[151] The same principle applies to a heterosexual female employer who does the same thing to a male employee. Oncale v. Sundowner Offshore Services makes clear that if the employers are gay in either of those scenarios and the employee is the same sex as the employer, the statute requires the same result.[152] Why? Because of but-for causation. In a male-on-male quid pro quo sexual-harassment claim, the theory presumably is that if the employee had been female (i.e., holding all else equal), the gay employer would not have harassed the employee. Likewise, in the female-on-female sexual-harassment claim, if the female employee had been male (again, holding all else equal), the lesbian employer would not have harassed the employee. This is why plaintiffs in these kinds of cases have to distinguish between sexual harassment and a more generalized abusiveness by the employer.[153]

But it should not take too much imagination, then, to consider the bisexual employer. Let’s call this employer Pat.[154] Imagine that Pat is what we might call an “equal opportunity harasser,” but importantly, Pat only wants to sleep with the employees Pat finds attractive. Pat says, “Sleep with me, or I’ll fire you,” to male and female employees alike but limits these unwanted demands to the good-looking ones.[155] Those employees that Pat views as unattractive escape Pat’s harassment completely. Now, imagine employee Roberta, who has just been fired because she wouldn’t sleep with Pat. Imagine also, however, that another employee, Robert, similarly attractive to Pat, has also just been fired because he wouldn’t sleep with Pat either. Has Roberta been discharged “because of [her] sex” or not? Pat says, “Of course not. If Roberta had been Robert and just as good-looking, I would have fired him too. In fact, I did!” Ergo, no but-for causation, and thus no satisfying of the phrase “because of sex.”[156]

Importantly, Pat can go further and point out that Roberta was fired because of a completely different quality, her attractiveness, and sex and attractiveness are two distinct qualities. Moreover, attractiveness is similar to sexual orientation in that it applies to both sexes. One can thus argue that discrimination on that basis is simply not covered by Title VII. The similarity with sexual orientation goes even further for purposes of an argument analogous to Justice Alito’s. Some jurisdictions explicitly prohibit discrimination on the basis of attractiveness.[157]

But, in 2021, doesn’t it seem likely that Roberta would prevail under Title VII and that Pat’s defense would be rejected? Irrespective of the lack of but-for causation, we would likely treat Pat’s firing of Roberta as “because of sex” under Title VII. So too was Pat’s firing of Robert. Each of those acts on Pat’s part was separate: there were, in other words, two parallel discriminations. Roberta was fired “because of [her] sex,” and Robert was fired “because of [his] sex.” These two acts of discrimination do not fall into the single category of discrimination “because of attractiveness.” The analogy with sexual orientation should be clear, and it obviously supports the Bostock majority. If Roberta was fired because she was a lesbian and Robert was fired because he was gay, the argument goes, then we should understand it as two parallel discriminations “because of sex,” not two examples of discrimination “because of sexual orientation.”

But why do we view Roberta’s and Robert’s firings as strong cases for the employees? There is nothing in the semantic meaning of the term “because of sex” that tells us this.[158] Instead, I suspect, as Judge Lynch might have put it, we understand that firing an employee because that employee won’t sleep with an employer is socially embedded in our historically patriarchal (and hence, “sex”ist) society as an act tied inherently to sex.[159] Even if Pat fires Robert because Robert won’t sleep with Pat, that assertion of power fits into the broader purpose of Title VII’s sex-discrimination provision, to eliminate the barriers for women in the workplace. Courts likely now view quid pro quo sexual harassment as so central to our understanding of Title VII that no court would bother to parse the text or frame this scenario through the logic of but-for causation. It is not an abstract equality, nor even but-for causation, that is driving what I suspect almost any court today would conclude. It can’t be; otherwise Pat’s defense that Roberta and Robert were treated the same would prevail. Instead, we can see quid pro quo sexual harassment in the workplace as historically connected, as a social practice, to discrimination against women. And so, firing Robert for not sleeping with Pat is of a piece with discrimination against women, even though Robert is a man.

This becomes even clearer if we tweak the hypothetical ever so slightly. Imagine now that Pat’s preference for attractive employees does not manifest itself in direct requests for sex. Instead, Pat’s preference for good looks expresses itself simply by choosing to hire Roberta and Robert over Johanna and John, who are equally qualified but unattractive. Everything in the semantic meaning of the words of Title VII remains the same. Has Pat “failed to hire [Johanna] because of [her] sex”? Has Pat “failed to hire [John] because of [his] sex”?[160] Again, Pat defends his act by saying, “I chose Roberta over Johanna because she’s better looking, and I chose Robert over John because he’s better looking. I did not ‘fail to hire’ either of them ‘because of sex.’”[161]

Here, the argument that Pat chose based on a characteristic independent of sex (in the biological sense) and did not discriminate “because of sex” is far stronger. Why? Because we know that in our society the “beauty bias” exists.[162] Perhaps it is sexist as a general matter, but we know it can benefit (and disadvantage) both men and women in the employment context, though in different ways perhaps.[163] But we still understand, as a socially embedded matter, that choosing to hire one candidate rather than another because of attractiveness (assuming it happens to men and women alike) is not as deeply intertwined with our conception of what Title VII was meant to prohibit. The semantic meaning of the words applies equally in the firing example, but the social meaning is different.[164]

In short, there is no way to choose between the two comparator arguments without making a judgment about the social meaning of an employer’s act and how that act fits into a socially constructed understanding of Title VII’s language. There is no way to decide, based on semantic meaning alone, whether some (quite hypothetical)[165] employer policy of discriminating against both gay men and lesbians is a single form of discrimination “because of sexual orientation” or two parallel but distinct forms of discrimination “because of sex.”


Professor Post once noted that “[t]here is . . . a strong impulse” within what he referred to as the “dominant perspective” on Title VII[166] to imagine the law “as standing in neutral space outside of history and of the contingent social practices of which history is comprised.”[167] That impulse is nowhere more on display than in the arguments from “text” and “logic” that both sides have brought to bear in Bostock. That the semantic meaning of the fifty words of section 703(a)(1) of the Civil Rights Act of 1964 can alone decide these cases is now a mantra that has infected our thinking about statutory interpretation. The Court’s textualist turn has too easily rendered us all incapable of seeing that, even for textualists, statutory interpretation has in practice always been, and always will be, multimodal.[168]

So perhaps the Court’s failure to come to agreement on the ordinary public meaning of Title VII will, in the long run, prove salutary⎯it exposes the fallacy that semantic meaning is the only object of inquiry when courts face a difficult question of statutory interpretation. The fact that both sides can equally appeal to text and logic should unmask the reality that statutory interpretation is inescapably a multimodal and necessarily socially embedded practice. Maybe “we are all textualists now,”[169] but courts are most certainly not linguists, focused solely on the “original public meaning” or “ordinary public meaning” of a statute’s text. The semantic meaning of the words is certainly not unimportant, but it is a broader understanding of those words—shaped by all the factors that go into the many modalities of interpretation, including prior precedent, other statutes, normative judgments, and historical understandings and assumptions—not their semantic meaning, that determines the outcome. As any comprehensive review of statutory interpretation shows, we are most decidedly not all textualists now, even if we often talk as though we were and even if text no doubt plays an important role. If the empirical data have not yet brought this fallacy to our collective consciousness, maybe the claim to textualism’s mantle in all of the Bostock opinions will finally get us to see that text is just not enough.


*Anuj C. Desai, Voss-Bascom Professor of Law, University of Wisconsin Law School. Thanks to those who provided comments at faculty workshops at the University of Wisconsin Law School and Sturm College of Law, University of Denver, as well as for helpful comments from BJ Ard, Eric Berger, Mitch Berman, Aaron-Andrew Bruhl, Andy Coan, Katie Eyer, Tara Grove, Guha Krishnamurthi, Jamie Macleod, Nina Mendelson, Caleb Nelson, Robert Post, David Schwartz, Naomi Schoenbaum, Brian Soucek, Kevin Tobia, and Nina Varsava. I am especially indebted to Andy Coan, who made suggestions that both helped clarify my thinking on some points and improved the exposition significantly. Thanks also to Natalia Gess and Nathan Watson for helpful research assistance; Myra Sun for careful word-processing; and Nick Steverson and Stacey Weber for superb editing. Thanks finally to the University of Wisconsin Law School for funding.

  1. Bostock v. Clayton County, 140 S. Ct. 1731 (2020).
  2. See Rena M. Lindevaldsen, Bostock v. Clayton County: A Pirate Ship Sailing Under a Textualist Flag, 33 Regent U. L. Rev. 39 (2020); Andrew Koppelman, Bostock, LGBT Discrimination, and the Subtractive Moves, 105 Minn. L. Rev. Headnotes 1, 16 (2020); Tara L. Grove, Which Textualism?, 134 Harv. L. Rev. 265, 269 (2020); Shelby Sternberg Moylan, Context to Overcome Definition: How the Supreme Court Used Statutory Interpretation to Define “Person” and “Sex”, 69 U. Kan. L. Rev. 171 (2020); Anita S. Krishnakumar, Three Lessons About Textualism from the Title VII Case, Yale J. on Regul.: Notice & Comment (June 24, 2020), []; Katie Eyer, Progressive Textualism and LGBTQ Rights, SCOTUSblog (June 16, 2020, 10:23 AM), []; Guha Krishnamurthi & Peter Salib, Bostock and Conceptual Causation, Yale J. on Regul.: Notice & Comment (July 22, 2020), []; Ryan Anderson, The Simplistic Logic of Justice Neil Gorsuch’s Account of Sex Discrimination, SCOTUSblog (June 16, 2020, 1:28 PM), []; Jonathan Skrmetti, The Triumph of Textualism: “Only the Written Word Is the Law”, SCOTUSblog (June 15, 2020, 9:04 PM), []; Sarah Rice, The Strength of the Written Word Fulfills Title VII’s Promise, SCOTUSblog (June 15, 2020, 6:32 PM), []; Michael Dorf, Does Justice Gorsuch’s Magnificent Opinion in the Title VII Sexual Orientation and Gender Identity Cases Redeem Textualism?, Dorf on L. (June 16, 2020), []; George T. Conway III, Why Scalia Should Have Loved the Supreme Court’s Title VII Decision, Wash. Post (June 16, 2020), []; Nelson Lund, Unleashed and Unbound: Living Textualism in Bostock v. Clayton County, 21 Federalist Soc’y Rev. 158, 167 (2020); Steven D. Smith, The Mindlessness of Bostock, Law & Liberty (July 9, 2020), []; Ed Whelan, A ‘Pirate Ship’ Sailing Under a ‘Textualist Flag’, Nat’l Rev. (June 15, 2020, 1:01 PM), []; Josh Blackman & Randy Barnett, Justice Gorsuch’s Halfway Textualism Surprises and Disappoints in the Title VII Cases, Nat’l Rev. (June 26, 2020, 6:30 AM), []. But cf. Mitchell N. Berman & Guha Krishnamurthi, Bostock Was Bogus: Textualism, Pluralism, and Title VII, 97 Notre Dame L. Rev. (forthcoming 2021), [] (arguing that if a textual approach is correct, then Bostock was wrongly decided).
  3. See Nina A. Mendelson, Change, Creation, and Unpredictability in Statutory Interpretation: Interpretive Canon Use in the Roberts Court’s First Decade, 117 Mich. L. Rev. 71 (2018); Anita S. Krishnakumar, Statutory Interpretation in the Roberts Court’s First Era: An Empirical and Doctrinal Analysis, 62 Hastings L.J. 221 (2010); Jane S. Schacter, The Confounding Common Law Originalism in Recent Supreme Court Statutory Interpretation: Implications for the Legislative History Debate and Beyond, 51 Stan. L. Rev. 1, 32 (1998); see also Miranda McGowan, Do as I Do, Not as I Say: An Empirical Investigation of Justice Scalia’s Ordinary Meaning Method of Statutory Interpretation, 78 Miss. L.J. 129, 134–35 (2008) (after reviewing a large random sample of Justice Scalia’s dissents, concluding that “[Justice Scalia’s] practice . . . resembles Legal Process methodology;” he “follows the ‘ordinary’ meaning only about a third of the time, even if ‘common law statutes’ are excluded;” “[a]bout a fourth of the time he . . . follows controlling precedent;” and “in nearly three quarters of the issues in my sample, Justice Scalia considers and weighs the purpose of the statute or the consequences and incentives created by different interpretations”).
  4. See, e.g., Michael C. Dorf, SCOTUS LGBT Discrimination Case Will Test Conservative Commitment to Textualism, Justia: Verdict (May 1, 2019), [] (arguing that if the Court’s conservatives “keep faith with their textualist commitment, they will rule in favor of the plaintiffs”).
  5. Grove, supra note 2, at 267 (emphasis added).
  6. Harvard Law School, The Antonin Scalia Lecture Series: A Dialogue with Justice Elena Kagan on the Reading of Statutes, YouTube (Nov. 25, 2015), [].
  7. Although the Court’s opinions made little distinction between the cases involving gay plaintiffs on the one hand (Bostock and its companion case, Altitude Express v. Zarda, 140 S. Ct. 34 (2019)) and the Stephens case involving a transgender plaintiff on the other (R.G. & G.R. Harris Funeral Homes, Inc. v. Equal Emp. Opportunity Comm’n, 140 S. Ct. 35 (2019)), my discussion throughout will focus exclusively on the cases with gay plaintiffs. I do this in large part because the various arguments I will probe arguably play out differently in the two types of cases. I will refer to the cases either as “Bostock” or “the Title VII cases,” but when I do, I mean to include both Bostock and Zarda but to exclude Stephens. Recognizing that the language used to describe the plaintiffs may be contested and in flux, I will track the phrasing the Court adopted and refer to them as “gays and lesbians.”
  8. In both cases, the lower courts decided the case without addressing potential factual disputes: Zarda was decided on summary judgment while Bostock was decided on a motion to dismiss. Compare Zarda v. Altitude Express, Inc., 883 F.3d 100, 109 (2d Cir. 2018) (en banc), cert. granted, 139 S. Ct. 1599 (2019), with Bostock v. Clayton County, No. 16-CV-001460, 2016 WL 9353356, at *8 (N.D. Ga. Nov. 3, 2016), aff’d sub nom., 723 F. App’x 964 (11th Cir. 2018), rev’d and remanded sub nom., 819 F. App’x 891 (11th Cir. 2020).
  9. Bostock v. Clayton County, 140 S. Ct. 1731, 1737–38 (2020).
  10. Zarda, 883 F.3d at 108.
  11. Id.
  12. Id. at 109.
  13. 42 U.S.C. § 2000e-2(a)(1) (2018).
  14. Bostock, 140 S. Ct. at 1734.
  15. Id. at 1754.
  16. Id. at 1738.
  17. Id. at 1739.
  18. Id. at 1741 (“[I]t is impossible to discriminate against a person for being homosexual . . . without discriminating against that individual based on sex.”); id. at 1737 (“Sex plays a necessary and undisguisable role in [a] decision [to fire an individual for being gay].”).
  19. Id. at 1737 (noting that “the express terms of a statute give us” the answer).
  20. Id. at 1741–43.
  21. Id. at 1743.
  22. Id. at 1744.
  23. Id. at 1744–45.
  24. Id. at 1750; see generally John F. Manning, What Divides Textualists from Purposivists?, 106 Colum. L. Rev 70 (2006) (textualists generally reject the use of legislative history).
  25. Bostock, 140 S. Ct. at 1750–51.
  26. Id. at 1751 (emphasis added).
  27. Id. at 1752.
  28. Id. at 1753.
  29. Id.
  30. Id.
  31. Id. at 1754 (Alito, J., dissenting) (“There is only one word for what the Court has done today: legislation.”); id. at 1822–23 (Kavanaugh, J., dissenting) (“[T]he responsibility to amend Title VII belongs to Congress and the President in the legislative process, not to this Court. . . . Our role is not to make or amend the law. As written, Title VII does not prohibit employment discrimination because of sexual orientation.”); id. at 1836 (“In judicially rewriting Title VII, the Court today cashiers an ongoing legislative process, at a time when a new law to prohibit sexual orientation discrimination was probably close at hand.”).
  32. Id. at 1755 (Alito, J., dissenting); id. at 1824, 1830 (Kavanaugh, J., dissenting).
  33. Id. at 1830–32 (Kavanaugh, J., dissenting).
  34. Id. at 1755 (Alito, J., dissenting) (“For the past 45 years, bills have been introduced in Congress to add ‘sexual orientation’ to the list . . . . But to date, none has passed both Houses.”); id. at 1824 (Kavanaugh, J., dissenting) (“For several decades, Congress has considered numerous bills to prohibit employment discrimination based on sexual orientation. . . . But . . . although Congress has come close, it has not yet shouldered a bill over the legislative finish line.”).
  35. Id. at 1753.
  36. Id. at 1757 (Alito, J., dissenting).
  37. Id. at 1755–56.
  38. For those who need to “txt,” that would be “WWJSD”!
  39. The arguments in favor of or against textualism are numerous, and because they are not important to my argument, I do not intend to catalogue them all here. Since I’ve tackled this briefly elsewhere with numerous citations, I refer the reader to my own typology of arguments. See Anuj C. Desai, The Dilemma of Interstatutory Interpretation, 77 Wash. & Lee L. Rev. 177, 209–10 (2020) (enumerating rationales for textualism, along with citations).
  40. See, e.g., Bostock, 140 S. Ct. at 1738.
  41. See generally Lawrence Solum, Legal Theory Lexicon 030: Textualism, Legal Theory Blog (Jan. 21, 2018), []; Manning, supra note 24.
  42. Antonin Scalia, Response, in A Matter of Interpretation: Federal Courts and the Law 144 (Amy Gutmann ed., 1997); see also William N. Eskridge & Victoria F. Nourse, Textual Gerrymandering: The Eclipse of Republican Government in an Era of Statutory Populism, N.Y.U. L. Rev. (forthcoming 2021) (manuscript at 2), [] (discussing the “producers” and “consumers” of statutes).
  43. Eskridge & Nourse, supra note 42, at 17. Professor Nelson has argued that the distinction between a reader-focused and a writer-focused mindset is not in fact what distinguishes textualists from intentionalists in practice. Caleb Nelson, What Is Textualism?, 91 Va. L. Rev. 347 (2005). While he has persuaded me, most avowed textualists still appear to think otherwise. See, e.g., John F. Manning, Textualism and Legislative Intent, 91 Va. L. Rev. 419 (2005); Brett M. Kavanaugh, Fixing Statutory Interpretation, 129 Harv. L. Rev. 2118 (2016). Since I am making a claim about textualism as its proponents and the courts currently understand it, I describe the standard view.
  44. Solum, supra note 41.
  45. Antonin Scalia, Common-Law Courts in a Civil-Law System: The Role of United States Federal Courts in Interpreting the Constitution and Laws, in A Matter of Interpretation: Federal Courts and the Law 3 (Amy Gutmann ed., 1997) (emphasis added) (giving the lecture the title “Common-Law Courts in a Civil-Law System”).
  46. Id. at 9 (“[T]his system of making law by judicial opinion, and making law by distinguishing earlier cases, is what every American law student, every newborn American lawyer, first sees when he opens his eyes. And the impression remains for life. His image of the great judge—the Holmes, the Cardozo—is the man (or woman) who has the intelligence to discern the best rule of law for the case at hand and then the skill to perform the broken-field running through earlier cases that leaves him free to impose that rule: distinguishing one prior case on the left, straight-arming another one on the right, highstepping away from another precedent about to tackle him from the rear, until (bravo!) he reaches the goal—good law. That image of the great judge remains with the former law student when he himself becomes a judge, and thus the common-law tradition is passed on. . . . All of this would be an unqualified good, were it not for a trend in government that has developed in recent centuries, called democracy.”).
  47. See Edward Levi, Introduction to Legal Reasoning 27 (2d ed. 2013) (“It is customary to think of case-law reasoning as inductive and the application of statutes as deductive.”); see also Richard A. Posner, Legal Formalism, Legal Realism, and the Interpretation of Statutes and the Constitution, 37 Case W. Rsrv. L. Rev. 179, 190 (1986) (noting the “traditional view that common law reasoning resembles induction and statutory interpretation, deduction”); cf. Jean Louis Goutal, Characteristics of Judicial Style in France, Britain, and the U.S.A., 24 Am. J. Compar. L. 43, 45 (1976) (“French courts, with amazing regularity, have practiced deduction and nothing but deduction.”).
  48. See, e.g., Blackman & Barnett, supra note 2 (textualist scholars criticizing Justice Gorsuch’s majority opinion in Bostock for failing to start with “first principles” and treating “decades of precedent as part of the ‘law’s ordinary meaning’ in 1964” (emphasis added)).
  49. See generally Philip Bobbitt, Constitutional Fate: Theory of the Constitution (1982).
  50. When I use the phrase “linguistic meaning” here, I include both a purely semantic meaning and one enriched by pragmatics. See Lawrence B. Solum, Communicative Content and Legal Content, 89 Notre Dame L. Rev. 479, 486–89 (2013).
  51. Bostock v. Clayton County, 140 S. Ct. 1731, 1738 (2020); id. at 1825 (Kavanaugh, J., dissenting). Textualists sometimes also refer to “original public meaning” rather than “ordinary public meaning.” The two have different implications, potentially affecting the outcomes in cases. The word “original”, of course, reminds the interpreter that Congress adopted the relevant language in 1964, while the word “ordinary” tells the interpreter to think consciously of a lay reader. Most of the time, though, regardless of whichever phrase is used, both “ordinary” and “original” are embedded into the textualist’s approach to interpretation. Professor Eyer castigates those who use “original” in statutory interpretation, particularly in the context of the application of Title VII to claims based on sexual orientation. See generally Katie R. Eyer, Statutory Originalism and LGBT Rights, 54 Wake Forest L. Rev. 63 (2019). For those who have seen the trend in constitutional interpretation since the 1980s, incorporating an “original” understanding of a statute should not be particularly controversial. As I discuss below, the dispute is largely about what gets to count in that “original” understanding. See infra Section II.B.
  52. See generally Schacter, supra note 3.
  53. For discussion of the employee’s comparator argument, see, for example, Brief for Petitioner at 13–18, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (No. 17-1618), 2019 WL 2763119, at *13–18; Zarda v. Altitude Express, Inc., 883 F.3d 100, 116–19 (2d Cir. 2018); Transcript of Oral Argument at 43–48, 61–63, 67–71, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (Nos. 17-1618, 17-1623).For the employee’s sex-stereotyping argument, see, for example, Brief for Petitioner at 23–29, 51–57, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (No. 17-1618), 2019 WL 2763119, at *23–29; Zarda v. Altitude Express, Inc., 883 F.3d 100, 119–23 (2d Cir. 2018); Transcript of Oral Argument at 8–10, 65–66, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (Nos. 17-1618, 17-1623).For the employee’s associational argument, see, for example, Brief for Petitioner at 18–23, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (No. 17-1618), 2019 WL 2763119, at *18–23; Zarda v. Altitude Express, Inc., 883 F.3d 100, 124–28 (2d Cir. 2018); Transcript of Oral Argument at 38–41, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (Nos. 17-1618, 17-1623).
  54. In a short but very insightful piece written after the Seventh Circuit’s decision on this question, Professor Soucek made this very point. See Brian Soucek, Hively’s Self-Induced Blindness, 127 Yale L.J.F. 115, 118–19 (2017).
  55. See Bostock, 140 S. Ct. at 1741 (“[C]onsider, for example, an employer with two employees, both of whom are attracted to men. The two individuals are, to the employer’s mind, materially identical in all respects, except that one is a man and the other a woman. If the employer fires the male employee for no reason other than the fact he is attracted to men, the employer discriminates against him for traits or actions it tolerates in his female colleagues. Put differently, the employer intentionally singles out an employee to fire based in part on the employee’s sex, and the affected employee’s sex is a but-for cause of his discharge.” It is thus “impossible to discriminate against a person for being homosexual . . . without discriminating against that individual based on sex.”); see also id. at 1742 (“Imagine an employer who has a policy of firing any employee known to be homosexual. The employer hosts an office holiday party and invites employees to bring their spouses. A model employee arrives and introduces a manager to Susan, the employee’s wife. Will that employee be fired? If the policy works as the employer intends, the answer depends entirely on whether the model employee is a man or a woman. To be sure, that employer’s ultimate goal might be to discriminate on the basis of sexual orientation. But to achieve that purpose the employer must, along the way, intentionally treat an employee worse based in part on that individual’s sex.”).
  56. Id. at 1763 (Alito, J., dissenting). Of course, Justice Alito’s comparator also changes two things: the sex of both the employee and the employee’s partner. See infra Section II.D.
  57. Bostock, 140 S. Ct. at 1758 (Alito, J., dissenting); see also Berman & Krishnamurthi, supra note 2, at 24 n.147 (using a similar example: “[I]t is quite easy and common to know whether somebody is gay without knowing their sex. If your friend tells you ‘my cousin Lee is homosexual,’ then you know (or have reason to believe) that Lee is gay without knowing or taking account of Lee’s sex.”).
  58. Bostock, 140 S. Ct. at 1759; see also Brief for Philosophy Professors as Amici Curiae in Support of the Employees at 9, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (Nos. 17-1618, 17-1623, 18-107), 2019 WL 2915039, at *9; Amanda Shanor, Sex Discrimination Behind the Veil Is Still Sex Discrimination, Take Care Blog (Oct. 11, 2019), []; Dothard v. Rawlinson, 433 U.S. 321 (1977) (holding that a policy of prohibiting prison guards of one sex from guarding inmates of the opposite sex violated Title VII). I discuss this in more detail below when addressing the associational argument. See infra Section II.D.
  59. Andrew Koppelman, Is Marriage Inherently Heterosexual?, 42 Am. J. Juris. 51, 54 (1997); see also Andrew Koppelman, Why Discrimination Against Lesbians and Gay Men Is Sex Discrimination, 69 N.Y.U. L. Rev. 197, 208 (1994). Perhaps more accurately, one might refer to this way of thinking as “parallel distinct discriminations” to emphasize that the two discriminations are distinct from each other, but “parallel distinct discriminations” doesn’t really roll off the tongue.
  60. Justice Alito does have another textual argument: the fact that the statutory phrase is “discriminate against,” not just “discriminate,” means that the employer must show animus based on the protected trait (i.e., must show animus based on “sex”). Bostock, 140 S. Ct. at 1769 n.22 (Alito, J., dissenting) (citing James Phillips, The Overlooked Evidence in the Title VII Cases: The Linguistic (and Therefore Textualist) Principle of Compositionality (May 11, 2020) (unpublished manuscript) (on file with author), []). While that argument may strengthen his overall claim that the employees should not prevail, the argument is logically unnecessary to accept his comparator argument. The logic of his comparator argument works just as well, even if the statute does not require the employee to show animus. I should also note that that argument depends on a flawed reading of Title VII’s text and grammatical structure, a point I explain elsewhere. See Anuj C. Desai, Is Title VII an Anti-Discrimination Statute?, 93 U. Colo. L. Rev. F. (forthcoming 2022).
  61. See, e.g., Brief of William N. Eskridge Jr. and Andrew M. Koppelman as Amici Curiae in Support of Employees at 9, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (Nos. 17-1618, 17-1623 and 18-107), 2019 WL 2915046, at *9; Robert P. George, Counterfeit Textualism, Nat’l Rev. (Nov. 19, 2019, 5:06 PM), [].
  62. See generally James Macleod, Finding Original Public Meaning, 56 Ga. L. Rev. (forthcoming 2021) (manuscript at 14), [] (discussing the role of intuition in causal reasoning).
  63. Cf. Berman & Krishnamurthi, supra note 2, at 18–19 (after criticizing the Bostock majority’s choice of comparator, conceding that choosing either comparator—what they refer to as “counterfactual liberality”—might be permissible).
  64. Id. at 37.
  65. Id. at 36 (quoting Robert N. Strassfeld, If . . . : Counterfactuals in the Law, 60 Geo. Wash. L. Rev. 339, 343–44 (1992)).
  66. Id.
  67. Id. at 37.
  68. See id. at 37–41.
  69. See Frederick Schauer, Causation Theory and the Causes of Sexual Violence, 12 Am. Bar Found. Rsch. J. 737, 747.
  70. See id.; see generally H. L. A. Hart & Tony Honoré, Causation in the Law 26–61 (2d ed. 1985). Causation questions of this sort raise an endemic problem in law and policy, one that I’ve addressed elsewhere. Anuj C. Desai, Attacking Brandenburg with History: Does the Long-Term Harm of Biased Speech Justify a Criminal Statute Suppressing It?, 55 Fed. Comm’n L.J. 353, 367–72 (2003).
  71. See generally infra text following note 145.
  72. One way to see how it fails to solve the problem is to apply the “principle of conservation in motivational analysis” to someone in an interracial relationship. The logic of the Berman and Krishnamurthi approach means that an employer that discharges an employee for being in an interracial relationship would not be acting “because of race.” See Berman & Krishnamurthi, supra note 2, at 47–48. But that can’t possibly be right, can it? Or, more precisely, it’s neither right nor wrong as a textual matter: nothing in the text can tell us whether discriminating against someone in an interracial relationship is “because of such individual’s race” or the race of the individual’s partner. See infra Section II.D. Relatedly, their argument runs up against the empirical evidence suggesting that a significant number of ordinary people view a discharge of a person for being gay as being “because of sex.” See generally Macleod, supra note 62; Kevin Tobia & John Mikhail, Two Types of Empirical Textualism, Brook. L. Rev. (forthcoming 2021), []. While they rightly conclude that the empirical evidence fails to demonstrate that the Bostock majority’s interpretation is correct, see Berman & Krishnamurthi, supra note 2, at 24–27, the evidence does undermine Justice Alito’s claim that his interpretation better captures “ordinary meaning.” To be sure, Professors Berman and Krishnamurthi’s argument is that the but-for causation test depends upon a specialized legal (i.e., technical) meaning of the phrase “because of.” They may well be correct that Justice Gorsuch’s majority opinion veers over from “ordinary meaning” to “technical meaning” at times. See Bostock v. Clayton County, 140 S. Ct. 1731, 1739 (2020) (using the phrase “[i]n the language of law” to describe the but-for test). But it seems clear that all three opinions seek to claim not just textualism’s mantle but also the rhetorical power of the phrase “ordinary meaning” or “ordinary public meaning.” See, e.g., id. at 1738–39 (“With this in mind, our task is clear. We must determine the ordinary public meaning of Title VII’s command . . . . To do so, we orient ourselves to the time of the statute’s adoption, here 1964, and begin by examining the key statutory terms in turn before assessing their impact on the cases at hand and then confirming our work against this Court’s precedents.”); id. at 1739 (“[T]he ordinary meaning of ‘because of’ is ‘by reason of’ or ‘on account of.’”) (internal citations omitted); id. at 1750 (“[T]he law’s ordinary meaning at the time of enactment usually governs . . . .”); id. at 1766 (Alito, J., dissenting) (“How would the terms of a statute have been understood by ordinary people at the time of enactment?”); id. at 1767 (“In 1964, ordinary Americans reading the text of Title VII would not have dreamed that discrimination because of sex meant discrimination because of sexual orientation, much less gender identity. The ordinary meaning of discrimination because of ‘sex’ was discrimination because of a person’s biological sex, not sexual orientation or gender identity. The possibility that discrimination on either of these grounds might fit within some exotic understanding of sex discrimination would not have crossed their minds.”); id. at 1772 (“Without strong evidence to the contrary (and there is none here), our job is to ascertain and apply the ‘ordinary meaning’ of the statute. And in 1964, ordinary Americans most certainly would not have understood Title VII to ban discrimination because of sexual orientation or gender identity.”) (internal citation omitted); id. at 1825 (Kavanaugh, J., dissenting) (“The ordinary meaning that counts is the ordinary public meaning at the time of enactment—although in this case, that temporal principle matters little because the ordinary meaning of ‘discriminate because of sex’ was the same in 1964 as it is now.”).
  73. See, e.g., William N. Eskridge Jr., Interpreting Law: A Primer on How to Read Statutes and the Constitution 85–138 (2016).
  74. See Manning, supra note 24, at 79–80.
  75. Compare Bostock, 140 S. Ct. at 1766–67 (Alito, J., dissenting) (referring to the social context in which a statute was embedded when passed), with Manning, supra note 24, at 91 (arguing that what divides textualists from purposivists is that textualists gives priority to semantic context over “policy context”), and Solum, supra note 50, at 479 (referring to “communicative content” as “the linguistic meaning communicated by a legal text in context”) (emphasis added). This debate largely tracks the debate about judicial discretion: permitting more social context could be seen as giving judges more discretion, while permitting less social context could be seen as giving them less discretion. We could usefully frame the question through the algorithm-versus-human debate, a longstanding frame for the problem of judging. See, e.g., John Dewey, Logical Method and Law, 10 Cornell L. Rev. 17, 23 (1924) (“The problem is not to draw a conclusion from given premises; that can best be done by a piece of animate machinery, by fingering a key-board.”). Professor McGinnis criticizes the majority for acting too much like a computer, arguing that Justice Alito’s dissent represents a better understanding of the words in context. John O. McGinnis, Errors of Will and of Judgment, L. & Liberty Blog (June 25, 2020), [https://]. But the problem is deeper than that. Even a computer could not choose between the two comparator arguments based solely on a set of predefined algorithmic tools. Interpreting Title VII requires deciding what kind of social context to incorporate into one’s interpretation, no matter which side of the case one comes out on. Neither reading of the statute is “correct” from a purely semantic perspective. Like Professor McGinnis, Professor Grove has also recently argued that the Bostock majority used a more “formalistic” textualism while Justice Alito’s dissent used a more “flexible” textualism. See Grove, supra note 2. In contrast to Professor McGinnis, however, Professor Grove argues that the more formalistic textualism is normatively preferable because it better comports with the federal judicial role. See Grove, supra note 2, at 269. Judges, she seems to imply, ought to be as close to computers as possible, presumably to limit judicial discretion. See id. at 269–70, 281 (referring to the majority opinion as having an “almost algorithmic feel”). But Professor Grove ignores the fact that Justice Alito makes a completely formalistic argument too (i.e., his veil-of-ignorance comparator argument) and that her formalistic notion of textualism simply cannot decide which of the two comparator arguments to choose. See Grove, supra note 2. To be sure, Justice Alito also argues that social context supports his interpretation of the statute. Bostock, 140 S. Ct. at 1777–78 (Alito, J., dissenting). But that doesn’t render his formalistic textualist argument any less formalistic than the majority’s, nor does it make the majority’s conclusion any more correct or necessary as a formalistic matter.
  76. I should probably be a bit more precise here. For more sophisticated scholars, linguistic meaning incorporates both semantics and pragmatics and so necessarily allows for the consideration of some social context. Indeed, even the semantic meaning of words is a function of social practice. See Solum, supra note 50. So, the question is not really semantic context versus social context; rather, it is what types of social context should count for a textualist. As I explain in detail below, the types of social context necessary to resolve the sex-stereotyping or associational arguments are not the types ordinarily associated with textualist reasoning. Because the two comparator arguments map directly onto the sex-stereotyping and associational arguments, the comparator argument cannot be resolved solely with textualist reasoning either. Dean Manning notes that one core distinction between textualists and non-textualists is that “[t]extualists give precedence to semantic context” over “policy context.” See Manning, supra note 24, at 76. This may well accurately describe what textualists attempt to do. My principal point in this essay is that what Dean Manning refers to as “semantic context” yields no correct answer to the interpretive question the Court faced in Bostock.
  77. See generally Eskridge & Nourse, supra note 42.
  78. Bostock v. Clayton County, 140 S. Ct. 1731, 1769 (2020) (Alito, J., dissenting).
  79. Id. at 1767 (emphasis added). But see Eyer, supra note 51, at 96 (2019) (arguing that the focus on “original public meaning” is a smokescreen for importing anti-gay prejudices).
  80. See supra Section II.A.
  81. Bostock, 140 S. Ct. at 1769 (Alito, J., dissenting).
  82. Id.; see also Hively v. Ivy Tech Cmty. Coll. of Ind., 853 F.3d 339, 362–63 (7th Cir. 2017) (en banc) (Sykes, J., dissenting).
  83. Bostock, 140 S. Ct. at 1767.
  84. Oncale v. Sundowner Offshore Servs., 523 U.S. 75, 79 (1998). One might reasonably ask what constitutes a “reasonably comparable evil,” but that, of course, would be analogic, not deductive, reasoning. See also Henson v. Santander Consumer USA, 137 S. Ct. 1718, 1725 (2017) (rejecting “speculation about what Congress might have done had it faced a question that, on everyone’s account, it never faced”); Pa. Dep’t of Corrs. v. Yeskey, 524 U.S. 206, 211–12 (1998) (The fact that prisons and prisoners were not mentioned, or perhaps even contemplated, in the drafting of the Americans with Disabilities Act “is irrelevant. . . . [T]he fact that a statute can be applied in situations not expressly anticipated by Congress does not demonstrate ambiguity. It demonstrates breadth.” (internal quotation marks and citations omitted)); cf. Richard A. Posner, Statutory Interpretation―In the Classroom and in the Courtroom, 50 U. Chi. L. Rev. 800, 817 (1983) (positing as an interpretive technique the notion of “imaginative reconstruction” to “imagine how [the legislators] would have wanted the statute applied to the case at bar”). Not surprisingly, Justice Gorsuch’s majority opinion in Bostock plays this point up. Bostock, 140 S. Ct. at 1737 (“Those who adopted the Civil Rights Act might not have anticipated their work would lead to this particular result. Likely, they weren’t thinking about many of the Act’s consequences that have become apparent over the years, including its prohibition against discrimination on the basis of motherhood or its ban on the sexual harassment of male employees. But the limits of the drafters’ imagination supply no reason to ignore the law’s demands. When the express terms of a statute give us one answer and extratextual considerations suggest another, it’s no contest. Only the written word is the law, and all persons are entitled to its benefit.”).
  85. But see Bostock, 140 S. Ct. at 1750–51 (noting that “[n]ot long after the law’s passage, gay . . . employees began filing Title VII complaints, so at least some people foresaw this potential application” of the statute); Eskridge & Nourse, supra note 42, at 51 (noting that the relevant time for a case involving a government defendant like Clayton County was after the 1972 amendments to Title VII, when “it was hardly unthinkable that ‘homosexuals’ would be protected by a sex discrimination prohibition”); cf. Cary Franklin, Inventing the “Traditional Concept” of Sex Discrimination, 125 Harv. L. Rev. 1307 (2012) (concluding that the legislative history of Title VII’s prohibition on sex discrimination failed to support the “traditional” anticlassificationist approach that, for example, assumed the statute did not protect gays and lesbians).
  86. Bostock, 140 S. Ct. at 1769 (Alito, J., dissenting).
  87. See Meritor Sav. Bank v. Vinson, 477 U.S. 57 (1986).
  88. See Oncale, 523 U.S. 75. Indeed, perhaps any sexual-harassment claim. See generally Vicki Schultz, Reconceptualizing Sexual Harassment, 107 Yale L.J. 1683, 1701–02 (1998) (noting that, in the early years of Title VII, courts tended to reject even quid pro quo sexual-harassment claims, “reasoning that the women’s adverse treatment occurred because of their refusal to engage in sexual affairs with their supervisors and not ‘because of sex’ within the meaning of the statute”).
  89. See Holcomb v. Iona Coll., 521 F.3d 130, 139 (2d Cir. 2008); Deffenbaugh-Williams v. Wal-Mart Stores, Inc., 156 F.3d 581, 589 (5th Cir. 1998), vacated in part on other grounds en banc, 182 F.3d 333 (5th Cir. 1999); Parr v. Woodmen of the World Life Ins. Co., 791 F.2d 888, 892 (11th Cir. 1986).
  90. See generally Price Waterhouse v. Hopkins, 490 U.S. 228 (1989) (plurality opinion); Bostock, 140 S. Ct. at 1751–52 (making this point about male-on-male sexual harassment, sex-segregated job advertising, and quid-pro-quo sexual harassment and asking, “Would the employers have us undo every one of these unexpected applications too?”); id. at 1752 (“[T]hanks to the broad language . . . many, maybe most, applications of Title VII’s sex provision were ‘unanticipated’ at the time of the law’s adoption.”).
  91. Caleb Nelson, Statutory Interpretation 905 (2011); see also Mark D. Greenberg & Harry Litman, The Meaning of Original Meaning, 86 Geo. L.J. 569, 584–86 (1998) (giving examples of similar phenomena).
  92. Bostock, 140 S. Ct. at 1767.
  93. See Bostock, 140 S. Ct. at 1756 (Alito, J., dissenting); see also Eyer, supra note 51, at 76 (describing this argument as based on “our gut intuition” that “the ‘original public’ would not have imagined that [the LGBT community] would be covered” by Title VII). See id. at 83 (citing William N. Eskridge, Jr., Title VII’s Statutory History and the Sex Discrimination Argument for LGBT Workplace Protections, 127 Yale L.J. 322, 335–36, 352 (2017)), for the view that the normative backdrop of the statute at the time of its passage made the formal argument seem normatively absurd).
  94. Macleod, supra note 62, at 7–8. Professor Macleod refers to this as “extratextual considerations.”
  95. It strikes me as at least possible that, even in 1964, readers might have grasped and agreed with the logic of the majority’s comparator argument. We of course don’t know, but recent work in experimental jurisprudence sheds some light on the question. In two recent articles investigating ordinary readers’ understanding of Title VII’s language, Professor Macleod and Professors Tobia and Mikhail asked survey participants to apply Title VII’s language to, among other scenarios, employees who had been fired for being gay or in same-sex marriages. See Macleod, supra note 62; Tobia & Mikhail, supra note 72. Both studies found that a significant number of respondents had no difficulty characterizing a discharge based on sexual orientation as being “because of sex.” Macleod, supra note 62, at 20–22; Tobia & Mikhail, supra note 72, at 19–24. Of course, because these studies were conducted in 2020, they don’t directly tell us anything about ordinary readers in 1964. But Professors Tobia and Mikhail have one finding suggesting that their survey respondents really are trying to do a linguistic analysis, rather than bringing their social, moral, or political views to bear on the problem: the study found that fewer respondents thought Title VII’s language protected those in an interracial relationship than thought that the language protected those in a same-sex relationship. Tobia & Mikhail, supra note 72, at 19–24. Given what I suspect is a more widespread social condemnation of an employer who fires someone in an interracial relationship, this suggests that survey participants really were trying to do linguistic analysis, rather than making a social, cultural, or moral judgment about the employer’s act. If that is right and assuming no change in the semantic meaning of Title VII’s language between 1964 and today, Justice Alito’s empirical claim about ordinary readers in 1964 may not be as strong as he thinks. Compare Bostock, 140 S. Ct. at 1750 (“[T]he employers and dissents merely suggest that, because few in 1964 expected today’s result, we should not dare to admit that it follows ineluctably from the statutory text.”), with id. at 1756 (Alito, J., dissenting) (“[I]n 1964, it was as clear as clear could be that this meant discrimination because of genetic and anatomical characteristics that men and women have at the time of birth.”), and id. at 1757 (noting that every circuit court to have addressed the question until 2017 rejected the majority’s argument); see also Macleod, supra note 62, at 11–12.
  96. Bostock, 140 S. Ct. at 1754 (Alito, J., dissenting).
  97. Or, at least, results that, if understood by the legislators and the public at the time, would have led the legislature either to reject the statute altogether or include language clarifying that the statute did not apply to such claims.
  98. See, e.g., Scalia, supra note 45, at 17–18 (explaining that a non-textualist approach allows judges to interpret a statute based on “their own objectives and desires”).
  99. See Bostock, 140 S. Ct. at 1741–43 (analytically jumping directly from “the ordinary public meaning of the statute’s language at the time of the law’s adoption” to the comparator argument). Let me emphasize here that I do not view the comparator argument as a purely textualist argument. It’s just that everyone seems to think it is. It is, of course, textualist in the sense that the idea of comparing the plaintiff to someone of a different sex depends on some words in the statute. But, as I noted, nothing in the text compels a particular choice of comparator. See supra Section II.A.
  100. Price Waterhouse v. Hopkins, 490 U.S. 228, 231 (1989) (plurality opinion).
  101. Id. at 235.
  102. See id. at 258.
  103. Zarda v. Altitude Express, Inc., 883 F.3d 100 (2d Cir. 2018) (en banc).
  104. Hively v. Ivy Tech Cmty. Coll. of Ind., 853 F.3d 339 (7th Cir. 2017) (en banc).
  105. See generally Brian Soucek, Perceived Homosexuals: Looking Gay Enough for Title VII, 63 Am. U. L. Rev. 715, 726 (2014) (“The challenge facing the lower courts since Price Waterhouse is finding a way to protect against the entire spectrum of gender stereotyping while scrupulously not protecting against the stereotype that people should be attracted only to those of the opposite gender.”).
  106. Hively, 853 F.3d at 346 (citing Hively panel); see also Hively v. Ivy Tech Cmty. Coll. of Ind., 830 F.3d 698, 709 (7th Cir. 2016) (“Because we recognize that Title VII in its current iteration does not recognize any claims for sexual orientation discrimination, this court must continue to extricate the gender nonconformity claims from the sexual orientation claims. We recognize that doing so creates an uncomfortable result in which the more visibly and stereotypically gay or lesbian a plaintiff is in mannerisms, appearance, and behavior, and the more the plaintiff exhibits those behaviors and mannerisms at work, the more likely a court is to recognize a claim of gender non-conformity which will be cognizable under Title VII as sex discrimination.”).
  107. See Bostock v. Clayton County, 140 S. Ct. 1731, 1741 (2020) (“So an employer who fires a woman, Hannah, because she is insufficiently feminine and also fires a man, Bob, for being insufficiently masculine may treat men and women as groups more or less equally. But in both cases the employer fires an individual in part because of sex. Instead of avoiding Title VII exposure, this employer doubles it.”); id. at 1742–43 (“So just as an employer who fires both Hannah and Bob for failing to fulfill traditional sex stereotypes doubles rather than eliminates Title VII liability, an employer who fires both Hannah and Bob for being gay or transgender does the same.”).
  108. Interestingly enough, Justice Alito never says explicitly that he adheres to Justice Scalia’s brand of textualism. He simply critiques the majority for misapplying it. Still, Justice Alito’s dissent largely takes Justice Scalia’s textualist approach as its premise. See, e.g., id. at 1766–67 (Alito, J., dissenting). I am indebted to Caleb Nelson for this astute observation.
  109. Id. at 1762.
  110. By “sex conformity,” I mean simply a person’s conformity to the stereotypical characteristics of their sex (or, for a transgender employee, perceived sex).
  111. This argument would of course have to be premised on treating the “too-masculine” female and the “too-feminine” male equally.
  112. See Raelynn J. Hillhouse, Reframing the Argument: Sexual Orientation Discrimination as Sex Discrimination Under Equal Protection, 20 Geo. J. Gender & L. 49, 59 (2018) (“Heterosexuality is the ultimate rule of gender conformity: the social expectation that women should be intimately associated with men and vice versa.”); see also Stephen Clark, Same Sex-But Equal: Reformulating the Miscegenation Analogy, 34 Rutgers L.J. 107, 116–17 (2002).
  113. As I noted earlier, Professor Soucek has made this very point. See Soucek, supra note 54, at 118–19 (noting that the “gender stereotyping argument . . . sounds like the same argument as before, and it invites the same response: if nonheterosexual men and nonheterosexual women both violate the gender norm, the norm must not be sex-specific”).
  114. Bostock, 140 S. Ct. at 1764 n.17 (Alito, J., dissenting) (quoting Price Waterhouse v. Hopkins, 490 U.S. 228, 294 (1989) (Kennedy, J., dissenting)).
  115. Id. at 1764.
  116. Id.
  117. Id.
  118. Recall that Justice Alito’s “comparator” for Zarda and Bostock (gay males) is a lesbian, and he thus thinks of the discrimination against Zarda and Bostock as part of a single form of discrimination based on sexual orientation. In contrast, the majority thinks of the “comparator” for Zarda and Bostock as a straight woman (and in turn the “comparator” for a lesbian as a straight man), which leads the majority to treat discrimination against gay males as a distinct discrimination from the discrimination against lesbians. See supra text accompanying note 60.
  119. See generally Berman & Krishnamurthi, supra note 2, at 20–22.
  120. The argument played a prominent role in both the Seventh and Second Circuits and in briefs and oral argument before the Supreme Court but barely made a cameo appearance in the Court’s opinions: only Justice Alito addressed it, and he did so only briefly. Compare Reply Brief for Petitioner at 9–10, Bostock v. Clayton County, 140 S. Ct. 1731, (No. 17-1618), 2019 WL 4464221, at *9–10, and Brief for Wisconsin Advocacy Organizations as Amici Curiae Supporting Petitioners at 28–31, Bostock v. Clayton County, 140 S. Ct. 1731, (Nos. 17-1618, 17-1623, 18-107), 2019 WL 3061209, at *28–31, and Oral Argument at 38-40, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (No. 17-1618), [], with Bostock, 140 S. Ct. at 1764–65 (Alito, J., dissenting).
  121. 388 U.S. 1 (1967). Loving is the 1967 case in which the Supreme Court invalidated antimiscegenation laws under the Equal Protection Clause.
  122. Title VII does have a “bona fide occupational qualification” (“BFOQ”) exception for sex that is inapplicable to race. 42 U.S.C. § 2000e-2(e)(1). But, this would not be relevant to the interpretation of section 703(a)(1), and in any event, none of the employers in these cases argued that being straight was a BFOQ.
  123. U.S. Const. amend. XIV, § 1 (“No State shall . . . deny to any person within its jurisdiction the equal protection of the laws.”).
  124. Cf. Bostock, 140 S. Ct. at 1764 (Alito, J., dissenting) (describing the argument as “analogiz[ing] discrimination against gays and lesbians to discrimination against a person who is married to or has an intimate relationship with a person of a different race”).
  125. Again, Professor Soucek has beat me to the punch and made this very point. See Soucek, supra note 54, at 119 (“Call this an associational claim if you like, but it is really just the same comparator claim as before: if the employee had been black rather than white, the boss would not have held his marriage against him.”).
  126. Recall again that the statute reads as follows: “It shall be an unlawful employment practice for an employer—(1) to fail or refuse to hire or to discharge any individual, or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual’s race, color, religion, sex, or national origin . . . .” 42 U.S.C. § 2000e-2(a)(1) (emphases added).
  127. See, e.g., Holcomb v. Iona Coll., 521 F.3d 130, 132 (2d Cir. 2008); Parr v. Woodmen of World Life Ins. Co., 791 F.2d 888, 889 (11th Cir. 1986).
  128. See, e.g., Michael W. McConnell, The Originalist Case for Brown v. Board of Education, 19 Harv. J.L. & Pub. Pol’y 457, 464 (1996).
  129. See generally Hillhouse, supra note 112, at 84 (responding to the Hively dissent’s claim in response to the gender-stereotyping argument by replacing all the language about sex with language about race. As one might imagine, the language sounds to our current ear to be racist: the final line, “[m]iscegenation does not classify people according to invidious or idiosyncratic white or colored stereotypes” is used as the parallel for Judge Sykes’ language, “[s]exual orientation discrimination does not classify people according to invidious or idiosyncratic male or female stereotypes.”).
  130. See Koppelman, supra note 2, at 19; see also Brief of William N. Eskridge Jr. and Andrew M. Koppelman as Amici Curiae in Support of Employees at 19, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (Nos. 17-1618, 17-1623, 18-107), 2019 WL 2915046, at *19 (describing the hypothetical employer defense as arguing “that the law’s protection of African Americans should not be extended to an entirely different category of people, namely, white interracial-sexuals”). The term “miscegenosexual” appears to be Professor Marcosson’s. See Samuel A. Marcosson, Harassment on the Basis of Sexual Orientation: A Claim of Sex Discrimination Under Title VII, 81 Geo. L.J. 1, 6 (1992); see also Koppelman, supra note 2, at 19 n.95; William N. Eskridge, Gaylaw: Challenging the Apartheid of the Closet 220 (2002).
  131. An anti-subordination framing would see anti-discrimination law through the lens of “the secondary social status of historically oppressed groups.” Jack M. Balkin & Reva B. Siegel, The American Civil Rights Tradition: Anticlassification or Antisubordination?, 58 U. Miami L. Rev. 9, 9 (2003). Under an anti-subordination approach, the point of Title VII’s sex-discrimination provision was to remedy discrimination against women and so the statute should be interpreted with that goal in mind. In contrast, the anti-classification principle simply prohibits the classification of people “on the basis of a forbidden category,” id. at 10, such as sex. Under an anti-classification approach, differential treatment alone would violate Title VII.
  132. Bostock v. Clayton County, 140 S. Ct. 1731, 1765 (2020) (Alito, J., dissenting). He then goes on to quote Judge Lynch’s dissent in the Second Circuit:

    It would require absolute blindness to the history of racial discrimination in this country not to understand what is at stake in such cases . . . . A prohibition on ‘race-mixing’ was . . . grounded in bigotry against a particular race and was an integral part of preserving the rigid hierarchical distinction that denominated members of the black race as inferior to whites.

    Id. (quoting Zarda v. Altitude Express, Inc., 883 F. 3d 100, 158–59 (2d Cir. 2018) (Lynch, J., dissenting)).

  133. Id.
  134. Cf. Brown v. Bd. of Educ., 347 U.S. 483, 495 (1954) (“Separate educational facilities are inherently unequal.”).
  135. I realize that much interracial dating/marriage/sex does not even involve someone who is Black. Part of my point here is that this paradigm shaped Justice Alito’s social understanding of it because of the social connotations of the concept of miscegenation in American society.
  136. See generally Charles W. Whalen & Barbara Whalen, The Longest Debate: A Legislative History of the 1964 Civil Rights Act (1989).
  137. In particular, it is not discrimination against women. Cf. Phillips, supra note 60, at 7 (“As a matter of linguistically sound textual interpretation, sex discrimination under Title VII will always rest on unfair beliefs or attitudes about women in particular, or about men in particular. In a word, on sexism.”); see Bostock, 140 S. Ct. at 1828–29 (Kavanaugh, J., dissenting) (arguing that animosity based on sex and animosity based on sexual orientation are distinct, and “to treat one as a form of the other . . . misapprehends common language, human psychology, and real life”); Hively v. Ivy Tech Cmty. Coll. of Ind., 853 F.3d 339, 368 (7th Cir. 2017) (en banc) (Sykes, J., dissenting) (unlike miscegenation laws, which are inherently racist, “[s]exual orientation discrimination . . . is not inherently sexist. No one argues that sexual orientation discrimination aims to promote or perpetuate the supremacy of one sex.”); Brief for the United States as Amicus Curiae Supporting Affirmance in No. 17-1618 and Reversal in No. 17-1623 at 19–20, Bostock v. Clayton County, 140 S. Ct. 1731 (2020), (Nos. 17-1618 & 17-1623), 2019 WL 4014070, at *19–20 (“[I]f an employer treats gay men and women the same, it has not engaged in sex discrimination.”).
  138. Hively, 830 F.3d at 706 (“Lesbian women and gay men upend our gender paradigms by their very status—causing us to question and casting into doubt antiquated and anachronistic ideas about what roles men and women should play in their relationships. . . . In this way the roots of sexual orientation discrimination and gender discrimination wrap around each other inextricably.”); see also Soucek, supra note 54, at 121–26; cf. Andrew Koppelman, Note, The Miscegenation Analogy: Sodomy Law as Sex Discrimination, 98 Yale L.J. 145, 158 (1988) (“Just as the prohibition of miscegenation preserved the polarities of race on which white supremacy rested, so the prohibition of sodomy preserves the polarities of gender on which rests the subordination of women.”); Marty Lederman, Thoughts on the SG’s “Lesbian Comparator” Argument in the Pending Title VII Sexual-Orientation Cases, Balkinization (Sept. 6, 2019), [].
  139. Bostock, 140 S. Ct. at 1828 (Kavanaugh, J., dissenting).
  140. The majority writes,

    This Court normally interprets a statute in accord with the ordinary public meaning of its terms at the time of its enactment. After all, only the words on the page constitute the law adopted by Congress and approved by the President. If judges could add to, remodel, update, or detract from old statutory terms inspired only by extratextual sources and our own imaginations, we would risk amending statutes outside the legislative process reserved for the people’s representatives. And we would deny the people the right to continue relying on the original meaning of the law they have counted on to settle their rights and obligations.

    Bostock, 140 S. Ct. at 1738. But, even if we put aside the fact that the Court’s holding pretty obviously “den[ied] the people the right to continue relying on the original meaning of the law they have counted on to settle their rights and obligations,” the resolution of the case required “extratextual sources,” and even the majority needed such sources, its protestations notwithstanding. Id.

  141. 42 U.S.C. § 2000e-2(a)(1) (2018) (emphasis added). See, e.g., Koppelman, supra note 2, at 16 (noting that “Title VII does not regulate by categories of people. It bars discrimination on the basis of certain classifications.”) (emphases added). But see Oncale v. Sundowner Offshore Servs., Inc., 523 U.S. 75, 80 (1998) (“The critical issue, Title VII’s text indicates, is whether members of one sex are exposed to disadvantageous terms or conditions of employment to which members of the other sex are not exposed.”) (emphases added).
  142. See supra text accompanying note 122.
  143. What I referred to earlier as “Premise B.”
  144. In their recent piece exploring the “ordinary meaning” of Title VII, Professors Tobia and Mikhail found that a majority of survey participants thought that an employee who was fired because of an interracial relationship was not fired “because of his race.” Tobia & Mikhail, supra note 72, at 21 fig. 2a. Apparently, most “ordinary people” reading the words of Title VII, even in 2020, think the statute does not prohibit an employer from discriminating against an employee in an interracial relationship. A majority of ordinary readers seems to believe, to borrow Professor Koppelman’s phrase, that equal treatment of all “miscegenosexuals” is not “obviously race discrimination,” despite Justice Alito’s concession that it is and the lower courts’ unanimous agreement (and my sneaking suspicion that not a single federal judge in the nation would disagree). See Koppelman, supra note 2, at 19.
  145. What I earlier referred to as “Premise A.” See supra text accompanying note 122.
  146. 42 U.S.C. § 2000e-2(a)(1) (2018).
  147. Id. § 2000e-2(a)(1). I am putting to one side here the question of transgender individuals and how an employer might enforce its rules about sex-segregated bathrooms as to such individuals.
  148. See, e.g., Solum, supra note 50, at 486–89.
  149. Tobia & Mikhail, supra note 72; Macleod, supra note 62.
  150. See supra text among notes 58 and 60.
  151. 42 U.S.C. § 2000e-2(a)(1) (2018).
  152. Oncale v. Sundowner Offshore Servs., 523 U.S. 75 (1998); see supra text accompanying notes 84 and 85.
  153. Oncale, 523 U.S. at 80 (“Title VII does not prohibit all verbal or physical harassment in the workplace; it is directed only at ‘discrimination . . . because of . . . sex.’”); Harris v. Forklift Servs., 510 U.S. 17, 25 (1993) (“The critical issue, Title VII’s text indicates, is whether members of one sex are exposed to disadvantageous terms or conditions of employment to which members of the other sex are not exposed.”); e.g., Holman v. Indiana, 211 F.3d 399, 403 (7th Cir. 2000) (stating that Title VII does not apply to the “equal opportunity” harasser); Fitzpatrick v. Winn-Dixie Montgomery, Inc., 153 F. Supp. 1303, 1305–06 (M.D. Ala. 2001) (no sexual harassment where two employees of different sexes brought sexual-harassment claims against supervisor); see also Henson v. Dundee, 682 F.2d 897, 904 (11th Cir. 1982) (no sexual harassment under Title VII when supervisor makes “sexual overtures” to both sexes). But see Steiner v. Showboat Operating Co., 25 F.3d 1459, 1464 (9th Cir. 1994) (rejecting “equal opportunity” harasser defense).
  154. Or, as Professor Karlan might have put it, “Saturday Night Live Pat.” Oral Argument at 51:23, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (No. 17-1618), [https://]. Of course, in my example, Pat is now the employer, not the employee.
  155. It may be important here for me to characterize “good looks” as purely subjective to Pat and not limited to stereotypical features of one sex or the other. More on this point below. I am indebted to a conversation with Pam Karlan for this point.
  156. Early arguments that a male employer who engaged in quid-pro-quo sexual harassment of a female employee constituted discrimination “because of sex” took pains to exclude the bisexual employer. See Schultz, supra note 88, at 1702–04.
  157. E.g., Santa Cruz, Cal., Mun. Code § 9.83.020(12); D.C. Code §§ 2–1401.02(22), 2–1402.11(a); Madison, Wis., Code of Ordinances § 39.03; see generally Deborah Rhode, Beauty Bias (2010); Robert Post, Prejudicial Appearances: The Logic of American Antidiscrimination Law, 88 Cal. L. Rev. 1 (2000). So, the same sort of negative-implication argument can be made in my hypothetical as can be made from the presence of the term “sexual orientation” in other statutes, both state and federal. See Bostock v. Clayton County, 140 S. Ct. 1731, 1754–55 (2020) (Alito, J., dissenting); id. at 1823–24 (Kavanaugh, J., dissenting); see also Hively v. Ivy Tech Cmty. Coll. of Ind., 853 F.3d 339, 360–61 (7th Cir. 2017) (Sykes, J., dissenting); Zarda v. Altitude Express Inc., 883 F.3d 100, 152–53 (2d Cir. 2018) (Lynch, J., dissenting).
  158. One framing might be that the firings can be characterized as “sexual misconduct” and therefore “because of . . . sex.” See generally David Schwartz, When Is Sex Because of Sex? The Causation Problem in Sexual Harassment Law, 150 U. Pa. L. Rev. 1697 (2002). But here, “sexual misconduct” is using the word “sexual” as in “having to do with sexual activity,” not “biological sex,” and courts have generally treated the latter as the sole meaning of “sex” in Title VII.
  159. See Zarda, 883 F.3d at 158–59 (Lynch, J., dissenting).
  160. See 42 U.S.C. § 2000e-2(a)(1) (2018).
  161. See id. Cf. Wilson v. Sw. Airlines, 517 F. Supp. 292, 293 (N.D. Tex. 1981) (rejecting airline’s BFOQ defense that it hired only “attractive female flight attendants” because its “sexy image” was “crucial to the airline’s continued financial success”). Of course, in the real world, the given justification would be something about “personal compatibility,” but hang with me on this.
  162. See Deborah L. Rhode, The Beauty Bias (2010).
  163. See Ana Swanson, Why So Many Women Spend So Much Time Getting Ready, Chi. Trib. (May 20, 2016), [] (commenting on the relationship between attractiveness, personal grooming, and salaries, with an emphasis on distinctions between men and women); Jaclyn S. Wong & Andrew M. Penner, Gender and the Returns to Attractiveness, 44 Res. Soc. Stratification & Mobility 113 (2016) (empirical study finding that grooming accounts for the entirety of a woman’s “attractiveness premium” but only half of a man’s “attractiveness premium”).
  164. Now, some readers may view my hiring example as sufficiently bad that it too violates Title VII and creates two parallel but distinct forms of discrimination “because of sex” (one against John because he is male, and one against Johanna because she is female), rather than just two instances of discrimination “because of beauty.” For such readers, I ask only that you be willing to agree that John’s and Johanna’s cases would be weaker than Roberta’s and Robert’s in the firing example.
  165. Transcript of Oral Argument at 6, Bostock v. Clayton County, 140 S. Ct. 1731 (2020) (No. 17-1618); Appellant’s Reply Brief at 9–11, Zarda v. Altitude Express Inc., 883 F.3d 100 (2d Cir. 2018) (No. 15-3775), 2017 WL 3536425, at *9–11; Lederman, supra note 138.
  166. Post characterizes the “dominant perspective” as the view that Title VII “typically requires employers . . . to make decisions as if their employees did not exhibit forbidden characteristics, as if, for example, employees had no race or sex.” Post, supra note 157, at 11.
  167. Id. at 30.
  168. Mendelson, supra note 3; Krishnakumar, supra note 3; Schacter, supra note 3; McGowan, supra note 3.
  169. Harvard Law School, supra note 6.