Note from the Editor:

The Federalist Society takes no positions on particular legal and public policy matters. Any expressions of opinion are those of the author. Whenever we publish an article that advocates for a particular position, as here, we offer links to other perspectives on the issue, including ones opposed to the position taken in the article. We also invite responses from our readers. To join the debate, please email us at


In Bostock v. Clayton County, Georgia, the Supreme Court held that Title VII of the Civil Rights Act of 1964 prohibits—and has always prohibited—discrimination by employers on the basis of homosexuality or of what the Court called transgender status.[1] The statute forbids employers to intentionally discriminate against any individual “because of such individual’s . . . sex.”[2] The Court concluded that discrimination on the basis of homosexuality or of being transgender violates the unambiguous text of the statute.

The decision was immediately controversial, especially among critics who see it as part of a campaign by elements of the elite legal establishment to impose a cultural agenda that Congress has failed to enact. The result in this case would not have been much of a surprise in the period during which Justice Anthony Kennedy held the controlling vote on issues dealing with sex, and especially with homosexuality.[3] But the 6-3 majority opinion in Bostock was written by Justice Neil Gorsuch and joined by Chief Justice John Roberts. The majority opinion has virtually no policy analysis or political rhetoric, and it lacks the kind of inflated pseudo-philosophic pontification that Kennedy favored. Instead, the Bostock opinion presents itself as nothing more than a straightforward application of the legally binding text of the statute. Gorsuch conspicuously casts himself as the true intellectual successor to the man whom he literally succeeded: the high priest of statutory textualism, Justice Antonin Scalia.[4]

Bostock invites unconfirmable speculation, even cynical speculation, about the motives of Gorsuch and the other members of the majority. I propose instead to take the Court’s opinion seriously, looking for its underlying assumptions and its legal implications. The opinion’s analytical approach resembles a theory known as “living originalism.” During the last decade, this approach to constitutional interpretation has been gaining steam in the legal academy. Bostock seems implicitly to extend that approach beyond the academy, beyond the field of constitutional law, and even beyond the limits recognized by its academic adherents.[5]

Bostock presents itself as something quite different, namely a particularly uncompromising version of what one might call static or non-living originalism. Taken on its own terms, the opinion is analytically flawed and the case was wrongly decided. But whatever one thinks of Bostock’s resolution of the precise issue in the case, it lays the groundwork for the Court to correct one of its most egregiously mistaken lines of case law.

The text of Title VII that Congress enacted in 1964 unambiguously forbids employers to discriminate on the basis of race or sex, yet the Court has upheld quotas and preferences explicitly based on the race or sex of people in favored groups. In 1991, Congress adopted a new provision outlawing employment practices in which race or sex is a motivating factor, even if not the only factor. The courts should have recognized that the 1991 amendment reaffirmed Title VII’s ban on race- and sex-based preferences. They have not done so, but Bostock now requires the enforcement of Title VII’s ban on these employment practices.

I. Living Originalism

As an academic theory, originalism arose in opposition to a modern jurisprudence—often called “living constitutionalism”—that seeks to make constitutional law ever more consistent with what Chief Justice Earl Warren called “the evolving standards of decency that mark the progress of a maturing society.”[6] Living constitutionalism liberates the Court from the text, history, and original purpose of the Constitution’s written provisions, as well as from precedents established when our society was putatively less mature.

The seminal decision was probably Brown v. Board of Education, in which the Court held that segregated public schools violate the Fourteenth Amendment.[7] Rather than develop a legally plausible argument for this appealing conclusion, as academic commentators later did,[8] the Court’s unanimous opinion was based entirely on speculation about segregation’s psychological effects, along with a bit of bogus social science. Although or because it contained barely a whiff of legal analysis, Brown has been a sensational political success. Had the Court confined itself to using this jurisprudential approach as a weapon against Jim Crow, it might never have provoked lasting opposition. But that’s not what happened. In subsequent cases, the Court aggressively expanded legal protections for criminal suspects, defendants, and convicts.[9] This happened at a time when violent crime was dramatically increasing, which generated significant political dissatisfaction. Dissatisfaction turned into a durable political fury when the Justices used living constitutionalism to invent a constitutional right to abortion,[10] which many millions of Americans regard as murder.

In response to these developments, a few academic commentators and judges began defending originalism as an alternative to the free-wheeling exercise of judicial review that became prominent in the Warren and Burger Courts.[11] Academic defenders of living constitutionalism then attacked originalism on various grounds.[12] Originalists responded,[13] and a robust and complex debate ensued during the following decades. Notwithstanding subtle variations within both camps, they seemed to be mutually exclusive. Originalists advocated sticking to the text of the written Constitution and interpreting vague or ambiguous provisions in light of the enactors’ purposes. Advocates of a “living Constitution” held that the original meaning was frequently unknowable, and in many cases should not be controlling even when it can be reliably identified.

Several years ago, Professor Jack Balkin purported to eliminate the opposition between these approaches by arguing that living constitutionalism is originalism. According to his theory, the text of the Constitution, understood as its semantic content, is binding on judges, but that only means that an interpretation must be one that the words of the text “can bear”; apart from that narrow constraint, judges should be free to adopt whatever interpretation will produce what the judge thinks will give the nation the best possible Constitution.[14] Evolving standards of decency, justice, wisdom, and sound policy—visible to judges if not to the people’s elected representatives—thus provide a trump card that will always be at hand when the Constitution’s text contains so much as a hint of ambiguity or vagueness. And a determined interpreter will almost always be able to find such hints.

The genius of Balkin’s theory arises in part from the fact that shrewd judges will not need to play the trump card very often. In most cases, they can reach their favored outcomes by applying or colorably interpreting precedent. In many other cases, they can apply traditionally originalist methods to build a respectable case for the result they want to reach. In others, political prudence may dictate restraint. But the card is always waiting to be used, and it is in a sense the quintessence of living originalism.

In order to see how radical Balkin’s theory is, consider two examples, both of which are based on this provision of the Fourteenth Amendment: “No State shall . . . deny to any person within its jurisdiction the equal protection of the laws.”[15]

According to Balkin, the original meaning of the Fourteenth Amendment protects a right to abortion, which happens to be the same right that living constitutionalism created in the Roe/Casey line of cases.[16] The overwhelming majority of states restricted abortion in 1868, and the subject never came up in debates about the Amendment. But the words “equal protection of the laws” could conceivably mean that governments must guarantee “equal citizenship,” and a judge could believe that restrictions on abortion deprive women of equal citizenship. That reading may be adopted as the original meaning of the Amendment, whether or not anyone believed at the time that equal protection of the laws means equal citizenship.[17] Q.E.D.

Living originalism could just as easily be used to get a completely different result. The word “person” will easily bear an interpretation that includes unborn children (and will certainly do so at least as easily as a meaning that includes corporations). These children are a politically powerless minority, and laws allowing them to be killed for the convenience of adults literally deprive them of the equal protection of the laws. This conclusion follows much more readily from the text than Balkin’s claim that abortion restrictions undermine the equal citizenship of women, who are an electoral majority. Therefore, laws permitting abortion are unconstitutional. Q.E.D.

As this example suggests, living originalism can use the Equal Protection of the Laws Clause to reach almost any conceivable result, and few conclusions will follow by necessity from the actual original meaning of the constitutional text.

As applied by Balkin himself, living originalism consistently produces results agreeable to the political left. But it has been widely accepted as a legitimate form of originalism by theorists who do not necessarily share those political views. And it has been endorsed and applied even by someone as conservative as Professor Steven Calabresi.[18] Calabresi has concluded, for example, that the original meaning of the Fourteenth Amendment protects a right to same-sex marriage. The core argument is straightforward:

State laws that ban same-sex marriage formally discriminate on the basis of sex in the same way that State laws that banned interracial marriage discriminated on the basis of race. Same-sex marriage laws allow a man to marry a woman, but not another man. This is, again as a formal matter, sex discrimination—plain and simple.[19]

Of course, the matter is not quite so plain and simple. In a different and more obvious sense, traditional marriage laws do not discriminate as a formal matter against either men or women on the basis of their sex: all members of both sexes are forbidden to marry a person of their own sex. But as with the abortion example, it is true that the bare words of the Equal Protection of the Laws Clause can accommodate multiple interpretations, including the one that Calabresi chooses. Because all laws treat some people differently than others, and thus unequally, any law could be declared unconstitutional without adopting a linguistically impossible interpretation of the Fourteenth Amendment.

Applying the theory of living originalism, a judge or other interpreter decides which laws violate the Constitution by reference to principles or policies that are congenial to the interpreter. Accordingly, Calabresi approves of laws that he thinks are for the general good of the whole people, apparently including laws against polygamy, which he thinks “arguably leads to sex discrimination.”[20] Of course, billions of people for thousands of years have believed that traditional marriage laws promote the general good, and polygamy has been practiced in many cultures, including that of the Biblical patriarchs. Living originalism makes it easy to declare either practice unconstitutional. Or both. Or neither.[21]

II. Living Textualism

The Bostock majority opinion consists essentially of a two-step argument. Step one is to find an interpretation of the phrase “because of an individual’s sex” that the words can bear. Justice Gorsuch assumes, though only arguendo, that “sex” has its ordinary meaning, which refers to the biological classes of male and female.[22] More significantly—indeed crucially—he says that the term “because of” incorporates into Title VII what he regards as the traditional standard of but-for causation, which “is established whenever a particular outcome would not have happened ‘but for’ the purported cause.”[23]

Step two is to find a hypothetical that illustrates why an individual’s sex is necessarily a but-for cause of every adverse employment decision resulting from a rule that discriminates against homosexual or transgender individuals. And that, Gorsuch thinks, is easily done. Assume, for example, that an employer has two employees, one of each sex, both of whom are sexually attracted to men. If the employer discriminates based on sexual orientation and discharges the male but not the female employee, the discharged employee would have kept his job but for his sex.[24] Ergo, Title VII was violated.

Gorsuch supplements this argument with a discussion of a few Supreme Court precedents and with responses to several arguments advanced by the employers in their briefs. But he insists that the argument I just summarized is conclusive.[25] The text is the law and that’s that. Most importantly, evidence about the understanding of the text held by those who voted for the statute in 1964, or by the public that authorized those legislators to act, is simply irrelevant. Why? Because the text is so completely clear and unambiguous that it cannot possibly mean anything other than what Gorsuch says it means:

The people are entitled to rely on the law as written, without fearing that courts might disregard its plain terms based on some extratextual consideration. Of course, some Members of this Court have consulted legislative history when interpreting ambiguous statutory language. But that has no bearing here. “Legislative history, for those who take it into account, is meant to clear up ambiguity, not create it.” And as we have seen, no ambiguity exists about how Title VII’s terms apply to the facts before us.[26]

Justice Samuel Alito’s dissent (in which Justice Clarence Thomas joined) does not mince many words. The majority’s claim that it is merely enforcing the terms of the statute is “preposterous.”[27] The majority opinion is “like a pirate ship,” flying the false flag of textualism.[28] The majority attempts to “pass off” its decision as the inevitable result of Justice Scalia’s interpretive method, while actually doing what he excoriated, namely updating a statute to better reflect evolving values that Congress has not enacted.[29] “A more brazen abuse of our authority to interpret statutes is hard to recall.”[30]

Alito’s critique has two main components. First, the linchpin of the majority’s argument is that the statutory language unambiguously forbids discrimination based on homosexuality or the characteristic of being transgender. A core principle of textualism does indeed hold that unambiguous statutory language should almost always be followed, without regard to what Gorsuch calls “extratextual considerations” such as the legislative history of the statute. But, Alito says, only through “breathtaking” arrogance can the majority claim that its reading of the language is unambiguously clear.[31] There is not a shred of evidence that anyone who voted for the statute perceived this supposedly unambiguous meaning, which the majority implies is a sign that the legislators were not “smart enough” to understand what it plainly says.[32] Every Court of Appeals until 2017 failed to perceive the same supposedly unambiguous meaning. And for 48 years, so did the statute’s enforcement agency, the EEOC.[33]

The second component of Alito’s response is that the majority’s interpretation went undiscovered for half a century because it is unambiguously wrong. “Sex” does not mean “sexual orientation,” nor does it mean “sexual identity.” An employer can have and enforce a policy against employing homosexuals or transgender individuals without knowing the sex of the individuals adversely affected by the policy. That cannot possibly be intentional discrimination against an individual “because of such individual’s sex.”[34] And if the employer happens to know the sex of an individual adversely affected by the policy, that cannot transform intentional discrimination because of sexual orientation or because of being transgender into intentional discrimination because of the employee’s sex. Thus, the majority is demonstrably wrong to claim that “it is impossible to discriminate against a person for being homosexual or transgender without discriminating against that individual based on sex.”[35]

Recall the majority’s supposedly dispositive hypothetical. An employer has two employees, one of each sex, both of whom are sexually attracted to men. If the employer discharges the male for being attracted to men but not the female, the discharged employee would have kept his job but for his sex, and he has therefore necessarily been discriminated against because of his sex. But has he been intentionally discriminated against because of his sex? To show why not, Justice Alito offers a different hypothetical. Rather than assume just two employees, assume four: a heterosexual of each sex and a homosexual of each sex. When the employer enforces a policy against employing homosexuals, it results in the discharge of the two employees whose descriptions are crossed out:

Man attracted to men

Woman attracted to men

Woman attracted to women

Man attracted to women

The discharged employees have something in common, but it is not their sex. Nor is it an attraction to men, or an attraction to women. Both individuals were discharged because of their homosexuality. Neither was discharged because of being a man or because of being a woman, or because of any characteristic of the sex to which they belong.[36]

Contrary to the majority’s claim, Alito insists, its approach is not the textualism adopted by Justice Scalia, which holds that the words of a law “mean what they conveyed to reasonable people at the time they were written.”[37] Alito piles up mountains of evidence confirming that the phrase “because of sex” would not have been understood in 1964 to include “because of sexual orientation or because of being transgender.” In sum, “[e]ven if discrimination based on sexual orientation or gender identity could be squeezed into some arcane understanding of sex discrimination, the context in which Title VII was enacted would tell us that this is not what the statute’s terms were understood to mean at the time.”[38]

Justice Brett Kavanaugh’s solo dissent makes a similar point in a somewhat different way. He observes that courts presumptively accept the ordinary meaning of a phrase in a statute as its legal meaning. The Bostock majority, he says, adopts “the literal meaning,” i.e. but-for causation, thereby “latching on to a novel form of living literalism” to remake American law.[39] His conclusion resembles Alito’s, but he missed the most significant part of the argument. Words and phrases frequently have multiple literal (i.e. non-metaphorical) meanings, and the crucial move in the majority’s argument is the substitution of one arguably literal meaning of the term “because of” for the indubitably literal ordinary meaning.[40]

As the majority acknowledges, the meaning of the term “because of” is “by reason of” or “on account of.”[41] That definition does not imply that but-for causation is either necessary or sufficient. In fact, Gorsuch never says that “because of” always encompasses any and every but-for cause, which would be patently false. On the contrary, he implicitly agrees that it does not when he admits that in “ordinary conversation,” discrimination on the basis of homosexuality or transgender status would not be regarded as sex discrimination.[42]

Taking “because of” to mean “motivated by” is every bit as literal as Gorsuch’s but-for causation interpretation. Nevertheless, Gorsuch says that “in the language of law” the term “because of X” can only mean “X was a but-for cause of.” [43] But the text of the statute says no such thing. And Gorsuch points to nothing in the statute that implies or even suggests any such thing.[44] In addition, he simply waves away the part of Title VII’s text specifying that liability is established when sex is a “motivating factor” for an employment practice, even if it is not a but-for cause.[45] An employer that dismisses an individual because of a policy motivated by disapproval of homosexuality or of transgender people or behavior is obviously not motivated by the individual’s sex.

Note how far the majority has gone beyond the form of living originalism promoted by Professors Balkin and Calabresi. The professors stress that the text of the Constitution is often open to multiple interpretations that its words can bear. They then choose the interpretations they prefer and defend their choices with non-linguistic arguments based on factors such as the Constitution’s historical background, the nation’s evolving traditions, justice, prudence, or anything else that supports their choice. Gorsuch, on the other hand, rests the majority’s decision solely on the bare words of the statutory text. In particular, he does not defend his conclusion on the ground that it is consistent with the purpose of Title VII. Justice Alito’s dissent demonstrates that it would be almost impossible to do so with a straight face, but the important point here is that Gorsuch does not even try.[46]

In the end, the majority’s core claim—that the statutory text is unambiguous—is indefensible. What Bostock implicitly suggests is that any interpretation of a statutory text that its words can bear is a legally sufficient basis for adopting that interpretation. No further explanation is needed, and evidence of any kind for a different interpretation, no matter how overwhelming the evidence is, may simply be ignored.

Living originalism has thus been unleashed from the academy and unbound from the rather minimal limitations that its academic promoters have acknowledged. The unacknowledged theory underlying the result in Bostock is that statutory language can simply be declared unambiguous so long as the imputed meaning is not linguistically impossible. This form of textualism thus operates as a super trump card.

III. The Road Ahead

A. Applying Bostock

In one of the majority’s few efforts to put some limit on the reach of its decision, Justice Gorsuch declares (without textual support) that Title VII applies to traits or actions that are “inextricably bound up with sex,” but not to traits or actions “related to sex in some vague sense or because discrimination on these bases has some disparate impact on one sex or another.”[47] If there is one action in human life that is inextricably bound up with sex, it is sexual intercourse. It is therefore not surprising that people frequently refer to this as “sex,” as in the phrase “having sex with one’s spouse.” Nor is it surprising that dictionaries recognize this as one of the word’s literal meanings.[48] Now suppose that an executive is fired because she had consensual sex with a subordinate. Has the employer violated Title VII? The employee was certainly fired “because of” her sex. But for her sex, i.e. the sex she had with her subordinate, she would not have been fired.[49] As a linguistic matter, the text of the statute can bear this interpretation at least as easily as it can bear the interpretation of the statute adopted in Bostock.

The executive’s legal claim would be rejected by the courts, of course, just as they may refuse, at least for a while, to apply Bostock to such practices as single-sex locker rooms and single-sex sporting competitions.[50] But on what grounds? Gorsuch himself suggests that the Court might, in another case, impute to the statute a meaning of “sex” that is different from the one applied in Bostock.[51] It is undoubtedly true that one of the meanings of “sex” is “sexual intercourse,” and sexual intercourse is undoubtedly “inextricably bound up with sex” in the biological sense of the term. Thus, in order to avoid holding the executive liable, it would seem that courts will have no choice except to recur to what Bostock calls extratextual considerations. Why would that be justified in my hypothetical, but not in Bostock itself?

B. Implications for Race and Sex Preferences

As this example may suggest, Bostock’s form of textualism will be impossible to apply consistently across the full range of statutory construction cases that courts must decide. But there is one case in which the language of Title VII is truly unambiguous: When members of one race or one sex are given preferential treatment because of their race or sex, the employer has intentionally discriminated against other individuals because of their race or sex. Although the Supreme Court has authorized such discrimination, this is the easiest case that could possibly arise under Bostock’s form of textualism. The text of the statute unambiguously forbids such disparate treatment, and the Supreme Court has never pretended otherwise.

Today, the only legal obstacle to following the unambiguous text of the statute arises from the existence of longstanding precedents that upheld such quotas and preferences. That obstacle was removed by a 1991 amendment to Title VII. The courts have refused to recognize the implications of the amendment, but Bostock has now established a new precedent that demands the restoration of the law that Congress has twice enacted. To show why Bostock requires that the courts stop ignoring the law, we need to examine several somewhat complex and interrelated developments that have taken place since 1964.

With certain textually specified exceptions, Title VII by its terms forbids employers “to discriminate against any individual . . . because of such individual’s race, color, religion, sex, or national origin.” The Supreme Court has never doubted that this language prohibits employers from intentionally discriminating against applicants or employees because they are black or white, or because they are male or female.[52]

In 1971, Griggs v. Duke Power[53] held that the statute also prohibits the unintentional discrimination that can arise when the qualifications for a given job are not randomly distributed among various racial groups. Chief Justice Warren Burger’s opinion for a unanimous Court contains an astonishing number of factual misstatements and other errors.[54] These include a refusal to acknowledge the existence of a textual provision that expressly warns against interpretations that would require preferential treatment on account of racial imbalances in an employer’s workforce.[55]

Burger ignored or misrepresented the text of the statute and rested the decision on three blatant ipse dixits. First, he assumed that the statute prohibits unintended disparate effects as well as intentional discrimination. Second, he claimed that the statute contains a massive unstated exception to this prohibition for practices that result in racial disparities smaller than some unspecified large magnitude. Third, he said that even if there is a sufficiently large disparate impact, the job qualifications set by the employer do not violate the statute if they have some unspecified kind of business justification. There is not so much as a hint of these exceptions in the statute, which is not surprising since the disparate impact rule to which they are exceptions is not itself in the statute. Nor was there any support for any of this in the legislative history. The Supreme Court just made the whole thing up. The Justices then spent many years trying, without much success, to clarify the doctrinal mess that Griggs had created.

Having radically expanded the reach of the statute by inventing a new theory of liability with no basis in the law, the Court went on to narrow the scope of the prohibition that Congress actually had enacted. This was not irrational. As Justice Harry Blackmun pointed out, the combination of the literal language of the statute and the Griggs decision put employers in a precarious position: potentially liable for past discrimination against blacks, they faced liability to whites for voluntary preferences adopted to mitigate the effects of such prior discrimination.[56] But this was not how the Court chose to justify its narrowing of the statutory ban on racial discrimination.

United Steel Workers v. Weber[57] involved a program, adopted in a collective-bargaining agreement, through which the employer decided to train some of its unskilled employees for higher paying skilled jobs at the company. Slots in the program were limited, and the openings were allocated by seniority, with one exception. Under pressure from the federal Department of Labor, the employer and the union agreed to impose a 50 percent quota for black workers. The plaintiff in the case was a white employee who would have been selected on the basis of his seniority but for the operation of the racial quota.

Writing for the majority, Justice William Brennan acknowledged at the outset that the quota violated what he called the statute’s “literal” language, which forbids discrimination because of race.[58] Nevertheless, he maintained, the overt intentional racial discrimination entailed in this quota scheme was consistent with what he called the “spirit” of the statute.[59] Brennan purported to find evidence of this spirit in the legislative history. Then-Justice William Rehnquist’s dissent systematically demolished the majority’s arguments, but it is equally important to observe that all of the legislative history Brennan cited pertained to private employers and to what was called “the plight of the Negro in our economy.”[60] That is, it was all about eliminating the economic barriers created by the notoriously widespread discrimination against blacks in the labor markets.

Eight years later, Johnson v. Transportation Agency[61] approved a preference given to a white woman by a public employer. Once again, Justice Brennan wrote the majority opinion. He purported to rely on Weber, although Weber’s holding had expressly extended only to what the Court had called “affirmative action plans designed to eliminate conspicuous racial imbalance in traditionally segregated job categories.”[62] Because the holding in Weber was expressly limited to racially segregated job categories, the result in Johnson was not based on stare decisis, and Brennan did not say that it was. Instead, he maintained that because Congress had not overruled Weber, “we therefore may assume that our interpretation of [Title VII] was correct.”[63] And what was that interpretation? Not the one actually adopted in Weber, but rather a much broader rubric under which the statute should be read to favor voluntary efforts to further what Brennan vaguely called “the objectives of the law.”[64]

Justice Scalia’s dissenting opinion skillfully refuted all of Brennan’s arguments, much as Rehnquist had done in Weber. Scalia persuasively showed, for example, that “vindication by congressional inaction is a canard.”[65] But I want to focus here on the fact that Johnson’s decision to reaffirm and extend Weber rested entirely on the proposition that Congress had tacitly endorsed Weber’s anti-textual interpretation by failing to overrule it. That means that Weber and Johnson would necessarily lose their precedential value if Congress were to remove the tacit endorsement that Brennan attributed to the legislature’s post-Weber inaction. And that is just what the Civil Rights Act of 1991 did.

This statute resulted from a lengthy and bitter debate in Congress about a series of employment discrimination decisions from the Supreme Court in 1989.[66] Most of the controversy arose from efforts to codify a stringent and expansive disparate-impact rule that the Supreme Court had implicitly rejected in the line of cases that began with Griggs. Opponents objected that doing so would force employers to use racial preferences to avoid the statistical imbalances that could trigger disparate impact liability. In the end, Congress enacted a compromise disparate impact rule that was sufficiently ambiguous to allow both sides to claim victory in public.

But the statute made a number of other changes to Title VII as well. Most importantly for our purposes, Congress added the following new provision to the statute:

Except as otherwise provided in [Title VII], an unlawful employment practice is established when the complaining party demonstrates that race, color, religion, sex, or national origin was a motivating factor for any employment practice, even though other factors also motivated the practice.[67]

On its face, this section of the 1991 Act unambiguously overrules Weber and Johnson. Race and sex were indisputably motivating factors in both cases, as they are in every practice that grants preferences to certain racial groups or to members of one sex. Under Bostock’s approach to statutory interpretation—under which the unambiguous text of the law controls, regardless of external evidence of congressional intent or policy implications—preferences based on race or sex would necessarily be held to violate this provision.

One might try to escape this conclusion by appealing to a savings clause in the 1991 statute, which provides: “Nothing in the amendments made by this [Act] shall be construed to affect court-ordered remedies, affirmative action, or conciliation agreements, that are in accordance with the law.”[68] This clause, however, cannot save the kinds of affirmative action at issue in Weber and Johnson. Under the traditional series-qualifier canon in statutory construction, the qualifying term “court-ordered” would apply to affirmative action and conciliation agreements, as well as to remedies. The canon fits because the statute expressly provides for only one defined form of “affirmative action,” namely court-ordered equitable relief for individual victims of Title VII violations,[69] and because conciliation agreements are enforceable through court orders.[70]

One might try instead to invoke the rule of the last-antecedent,[71] so that the qualifying term “court-ordered” would apply only to remedies. But the statute contains no definition of affirmative action other than court-ordered equitable relief, so this interpretation would turn the savings clause into a vague reference to something that might or might not include preferences based on race or sex.

Even if one adopted that reading, however, the savings clause would still apply only to actions and programs that are “in accordance with the law.” If Weber and Johnson were “the law,” the 1991 statute would be hopelessly self-contradictory because it would mean that the statute should not be construed to mean what it unambiguously does mean: that race- and sex-based preferences are unlawful.[72] But the statute does not contradict itself. The savings clause cannot save Weber and Johnson because judicial opinions interpreting the law are themselves law only in a metaphorical sense. The text of Title VII, however, is literally the law, as Bostock forcefully emphasizes. Although lawyers sometimes adopt the colloquial shortcut of calling judicial opinions “law,” these opinions cannot be a form of law superior to an actual statute. The text of the Constitution itself tells us that validly enacted federal statutes are “the supreme Law of the Land,” and it nowhere so much as suggests that judges may set their own opinions up as a form of law superior to the supreme law of the land.[73]

Thus, the unambiguous text of the motivating-factor provision of the 1991 Act—which is not altered by the savings clause—bans race- and sex-based preferences even more clearly than the language in the 1964 Act did. Because Bostock was a case of first impression in the Supreme Court, Justice Gorsuch did not discuss the possibility of tension between his form of textualism and the Court’s precedents. Nevertheless, Bostock provides strong support for rejecting Johnson’s acquiescence theory. Implicitly rejecting Brennan’s claim that legislative inaction can constitute legislative endorsement, Bostock declares that “speculation about why a later Congress declined to adopt new legislation offers a ‘particularly dangerous’ basis on which to rest an interpretation of an existing law a different and earlier Congress did adopt.”[74]

Although the 1991 Act overrules Weber and Johnson on its face, the legislative history suggests that the new provision about motivating factors was directly aimed at overruling a different Supreme Court decision, Price Waterhouse v. Hopkins.[75] Nor does there seem to be anything in the legislative history expressly indicating that it was also meant to overrule Weber and Johnson. Needless to say, the approach to legislative history adopted by Justice Brennan in Weber could be used to conclude that the unambiguously plain meaning of the motivating factor provision is trumped by the “spirit” of the statute. Bostock, however, rules that out because the Court now emphatically insists that legislative history is irrelevant when the text is unambiguous. The unambiguous meaning of the motivating factor provision therefore controls, notwithstanding the Court’s misguided use of legislative history in the past.

But even Justice Brennan’s own approach—applied consistently—would not be able to save Weber and Johnson. One of the strangest features of the long and tangled process that led to the 1991 Act was the absence of any extended discussion of those cases in the legislative history. This is not what one might expect in a debate that focused largely on disputes about whether disparate impact liability would lead employers to adopt racial preferences.[76] It is as though there was a conspiracy of silence about Weber and Johnson. But that does not mean that Congress has confirmed Brennan’s assumption in Johnson that congressional inaction after Weber (and Johnson itself) constitutes ongoing approval of those decisions.

On the contrary, if Congress can confer approval by inaction, as Johnson claimed it can, the legislature can also withdraw that approval by inaction, at least if it enacts language that unambiguously overrules prior decisions and fails to enact an exception that preserves those decisions. When Congress simultaneously acts to endorse some Supreme Court decisions while declining to endorse other closely related decisions, it follows a fortiori from Johnson’s premises that any implied approval by the legislature has been withdrawn. And that is just what happened in 1991, when Congress added or amended at least 22 statutory provisions dealing with employment discrimination. In doing so, it expressly endorsed the controversial Griggs decision (and others), but it chose not to endorse Weber and Johnson.

This choice was specifically confirmed in the legislative history. A memorandum submitted for the record on behalf of 14 Senators who supported the final compromise, as well as the Administration, specifically said that nothing in the new statute approved or disapproved of Weber and Johnson.[77] Nobody in either House contradicted that statement.[78] Even if one accepts Brennan’s theory of approval by inaction, Congress in 1991 even more clearly indicated that it was not approving of Weber and Johnson. The new statute outlaws race- and sex-based preferences on its face, and uncontradicted legislative history confirms that nothing in the statute constitutes approval of Weber and Johnson. Justice Brennan’s sole justification for affirming and extending Weber to approve race and sex preferences, despite the plain meaning of Title VII, has therefore been invalidated by Congress. Even if Justice Gorsuch’s Bostock approach would defer to on-point precedents because of stare decisis, these precedents have lost any force they might have had before the 1991 Act.[79] Cases interpreting language in the original 1964 Act could not possibly be regarded as binding precedents that control the interpretation of a new provision enacted after those cases were decided.

Nor can Weber and Johnson be saved by the kind of argument that Justice Blackmun advanced in his Weber concurrence. It is true that the codification of the disparate impact theory in the 1991 statute created an excruciating tension with Title VII’s basic prohibition against intentional discrimination: an employer that balances its workforce in order to avoid disparate impact liability risks being held liable for disparate treatment. In Ricci v. DeStefano,[80] the Supreme Court addressed this Catch-22. The Court assumed, quite properly, that it had a responsibility to reconcile the conflicting textual provisions as best it could. It did so by allowing employers to practice intentional discrimination when, but only when, there is strong evidence that such discrimination is necessary to avoid disparate impact liability.

Similarly, continued adherence to Weber and Johnson creates an intolerable tension with the plain language of the motivating-factor provision in the 1991 Act (as well as with the prohibition against intentional discrimination originally adopted in 1964). Unlike the problem addressed in Ricci, however, this tension is not in the statute itself, so it is easy to eliminate—the Court need only apply the statute that Congress enacted. Any practical problems that may have loomed before Ricci was decided have been adequately addressed by that decision.

Because Congress has deprived Weber and Johnson of whatever precedential force they may have had, and because they were clearly wrong when decided, the courts are free to apply the 1964 Act and the 1991 amendment as written. In truth, they are obliged to do so, and Bostock removes any plausible excuse for shirking that obligation. As the Supreme Court has now pointedly declared, legislative history is not law. Neither are judicial opinions. “Only the written word is the law, and all persons are entitled to its benefit.”[81]

IV. Conclusion

Bostock is an outlandish judicial performance. Its argument that the words of Title VII’s prohibition of sex discrimination unambiguously outlaw discrimination based on homosexuality and on being transgender is analytically untenable. Bostock’s textualism has a kinship with the academic theory of “living originalism,” at least in the sense that they are both simulacra of the real thing. But Bostock is even more unrestrained in the license it takes with the actual original meaning of the text it purports to apply.

Although Bostock’s particular application of textualist principles is fatally flawed, those principles can and should be faithfully applied in other cases. Title VII’s anti-discrimination language does not apply, unambiguously or otherwise, to discrimination on the basis of sexual orientation or of being transgender. But the application of the statute’s language to preferences based on race or sex is indisputably clear. The use of such preferences undeniably constitutes discrimination “because of” race or sex, and race or sex are undoubtedly “motivating factors” for the preferences even if other factors also motivated the practice. The Supreme Court has never even pretended that the text says anything else.

The Supreme Court precedents upholding such preferences pre-date the 1991 amendment that reconfirmed and strengthened the statute’s unambiguous textual ban on discrimination because of race or sex. Because the precedential force of those judicial decisions has rested, according to the Court itself, entirely on the absence of legislation like the 1991 amendment, they must now be judicially overruled. To preserve their holdings after Bostock would require a transparent abandonment of any commitment to logic, consistency, or fidelity to the rule of law. Surely that will not happen.

[1]  140 S. Ct. 1731 (2020).


[2]  42 U.S.C. §2000-e2(a):

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; or

(2) to limit, segregate, or classify his employees or applicants for employment in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee, because of such individual’s race, color, religion, sex, or national origin.

[3]  See, e.g., Obergefell v. Hodges, 135 S. Ct. 2584 (2015); Lawrence v. Texas, 539 U.S. 558 (2003); Romer v. Evans, 517 U.S. 620 (1996). Cf. Planned Parenthood v. Casey, 505 U.S. 833 (1992).


[4]  See 140 S. Ct. at 1748, 1749.


[5]  I am not asserting that anyone on the Court consciously saw the Bostock opinion as an extension of the academic theory, or that its academic proponents believe the theory should necessarily be applied to statutory construction cases. But I think “living originalism” offers a heuristic that is useful for understanding what the Court did in Bostock.


[6]  Trop v. Dulles, 356 U.S. 86, 101 (1958).


[7]  347 U.S. 483 (1954).


[8]  See, e.g., Michael W. McConnell, Originalism and the Desegregation Decisions, 81 Va. L. Rev. 947 (1995).


[9]  Most of the criminal procedure provisions in the Bill of Rights, for example, were made applicable for the first time to the states. And some of them were expansively interpreted. See, e.g., Miranda v. Arizona, 384 U.S. 436 (1966); Gideon v. Wainwright, 372 U.S. 335 (1963).


[10]  See Roe v. Wade, 410 U.S. 113 (1973).


[11]  See, e.g., Robert Bork, Neutral Principles and Some First Amendment Problems, 47 Ind. L.J. 1 (1971); William H. Rehnquist, The Notion of a Living Constitution, 54 Tex. L. Rev. 693 (1976).


[12]  See, e.g., Paul Brest, The Misconceived Quest for the Original Understanding, 60 B.U. L. Rev. 204 (1980); Richard B. Saphire, Making Noninterpretivism Respectable: Michael J. Perry’s Contributions to Constitutional Theory, 81 Mich. L. Rev. 782 (1983).


[13]  See, e.g., Antonin Scalia, Originalism: The Lesser Evil, 57 U. Cin. L. Rev. 849 (1989).


[14]  Jack M. Balkin, Living Originalism 254, 267-68 (2011). Some legal theorists, including Balkin, have begun to call the identification of a text’s semantic content “interpretation,” and choices about how to interpret textual ambiguity and vagueness “construction.” See, e.g., Lawrence B. Solum, Originalism and Constitutional Construction, 82 Fordham L. Rev. 453 (2013). Nothing in my argument depends on accepting or rejecting this terminological innovation. For a short discussion of the role it plays in living originalism, see Nelson Lund, Living Originalism: The Magical Mystery Tour, 3 Tex. A&M L. Rev. 31, 33-34 (2015).


[15]  U.S. Const. amend. XIV.


[16]  See Jack M. Balkin, Abortion and Original Meaning, 24 Const. Comment. 291, 313-25 (2007).


[17]  “[T]here could be other constitutional principles [i.e. other than “equal citizenship”] embodied by the Equal Protection Clause that no particular person living in 1868 intended but that we come to recognize through our country’s historical experience.” Jack M. Balkin, Original Meaning and Constitutional Redemption, 24 Const. Comment. 427, 498 (2007).


[18]  See, e.g., Steven G. Calabresi, Two Cheers for Professor Balkin’s Originalism, 103 Nw. U. L. Rev. 663 (2009).


[19]  Steven G. Calabresi & Hannah M. Begley, Originalism and Same-Sex Marriage, 70 U. Miami L. Rev. 648, 700 (2016) (emphasis added). Calabresi and Begley also rely on legislative history to argue that the Fourteenth Amendment bans hereditary class legislation like the Hindu Caste system and European feudalism. Id. at 653. This interpretation is eminently defensible, as is its application to Jim Crow anti-miscegenation laws. Homosexuals, however, do not constitute a legally defined hereditary class like the Indian dalits, medieval serfs, or black Americans during Jim Crow, and there is no evidence that the public thought they did in 1868.


[20]  Id. at 691-92, 702.


[21]  For a more detailed discussion of Calabresi’s arguments for the unconstitutionality of traditional marriage laws, see Lund, Living Originalism, supra note 14, at 37-43.


[22]  See 140 S. Ct. at 1739 (“[W]e proceed on the assumption that ‘sex’ [in 1964] signified what the employers suggest, referring only to biological distinctions between male and female.”) (emphasis added).


[23]  Id. (citing Gross v. FBL Financial Serv., 557 U.S. 167, 176 (2009)).


[24]  Id. at 1741-42.


[25]  In his dissent, Justice Alito distinguishes the precedents. Whether one is persuaded by Alito’s distinctions or not, the more important point is that the majority does not actually rely on those precedents. In the majority’s view, the outcome of this case would be the same even if none of those precedents existed.


[26]  Bostock, 140 S. Ct. at 1749 (citations omitted).


[27]  Id. at 1755 (Alito, J., dissenting).


[28]  Id.


[29]  Id. at 1755-56.


[30]  Id. at 1755.


[31]  Id. at 1757.


[32]  Id. (quoting Hively v. Ivy Tech Community College, 853 F.3d 339, 357 (7th Cir. 2017) (Posner, J., concurring)).


[33]  Id. at 1757-58.


[34]  “At oral argument, the attorney representing the employees, a prominent professor of constitutional law, was asked if there would be discrimination because of sex if an employer with a blanket policy against hiring gays, lesbians, and transgender individuals implemented that policy without knowing the biological sex of any job applicants. Her candid answer was that this would ‘not’ be sex discrimination. And she was right.” Id. at 1759 (footnote omitted).


[35]  Id. at 1758 (quoting the majority opinion).


[36]  Id. at 1763.


[37]  Id. at 1755 (Alito, J., dissenting) (quoting A. Scalia & B. Garner, Reading Law: The Interpretation of Legal Texts 16 (2012) (emphasis added by Alito)).


[38]  Id. at 1772.


[39]  Id. at 1824-25, 1836 (Kavanaugh, J., dissenting).


[40]  I say “arguably literal” though I seriously doubt that the literal meaning of “because of” includes all but-for causes, no matter how remote they may be. Would one really be speaking literally, for example, in saying that an employee was fired “because of the Big Bang” or “because of Christopher Columbus’s voyage across the Atlantic Ocean” or “because of her parents’ decision to marry and have a child”? But even assuming that the majority has adopted a literal interpretation of the statute, it certainly has not adopted what Kavanaugh calls the literal meaning.


[41]  Bostock, 140 S. Ct. at 1739 (majority opinion). Every dictionary I have consulted gives the same definition.


[42]  Id. at 1745.


[43]  Id. at 1739. See also id. at 1745 (“You can call the statute’s but-for causation test what you will—expansive, legalistic, the dissents even dismiss it as wooden or literal. But it is the law.”).


[44]  Instead, he offers citations to two cases, neither of which dealt with Title VII’s anti-discrimination provisions. Id. at 1739 (citing Univ. Tex. Southwestern Med. Ctr. v. Nassar, 570 U.S. 338, 350 (2013) and Gross, 557 U.S. at 176). In any event, the cases merely construed those provisions to make but-for causation a necessary minimum element of a plaintiff’s proof, so that a plaintiff would not be able to prevail simply by showing that the challenged practice was a motivating factor for an adverse employment decision. Neither case held that every but-for cause is always sufficient to establish legal causation.


[45]  Id. at 1739-40 (dismissing relevance of 42 U.S.C. § 2000e-2m).


[46]  One politically appealing argument that could be made in favor of the majority’s ruling is based on an analogy with discrimination on the basis of interracial intimate relationships. Although the Court has not ruled that such discrimination would violate Title VII, Justice Alito assumes that it might. He rebuts the analogy by noting that history tells us that such discrimination is a core form of racial discrimination used in a system designed to subjugate one race of Americans as a class, whereas discrimination on the basis of homosexuality or of being transgender was never a part of a project to subjugate either men or women. See Id. at 1764-65 (Alito, J., dissenting). Even if one is not persuaded by Alito’s rebuttal, it is significant that Gorsuch does not mention the analogy. Why would he, given that the majority considers all “extratextual considerations” irrelevant? See id. at 1749 (majority opinion).


[47]  Id. at 1742 (majority opinion). As Alito pointedly notes, the text of the statute forbids discrimination because of sex itself, not because of things “inextricably bound up with sex,” such as sexual harassment or sexual assault. “Does the Court really think that Title VII prohibits discrimination on all these grounds? Is it unlawful for an employer to refuse to hire an employee with a record of sexual harassment in prior jobs? Or a record of sexual assault or violence?” Id. at 1761 (Alito, J., dissenting). There is no response to these questions in the majority opinion.


[48]  As Appendix A to Justice Alito’s dissent documents, dictionaries from the era in which Title VII was adopted, as well as more recently, include sexual intercourse or terms including sexual intercourse (such as “phenomena of sexual instincts and their manifestations”) among the literal meanings of the word “sex.” One dictionary even gives “sexual activity, especially sexual intercourse” as the first of several literal meanings. American Heritage Dictionary 1605 (5th ed. 2011) (quoted in Alito’s dissent, 140 S. Ct. at 1791).


[49]  Even under the ordinary meaning of “because of,” the employer would be liable: the firing was motivated by the executive’s sex, which she had had with her subordinate.


[50]  Bostock was careful to stress that the Court has not outlawed single-sex bathrooms and locker rooms, or required girls and women to compete against male athletes, or forbidden sex-based dress codes. 140 S. Ct. at 1753. What the Court really means is “not yet.” Even that isn’t quite true since one of the three cases decided in Bostock itself held an employer liable for discharging a male employee who decided to dress as a woman while at work. See EEOC v. R.G & G.R. Harris Funeral Homes, 884 F.3d 560, 566 (6th Cir. 2018), aff’d Bostock, 140 S. Ct. 1731. Cf. Lawrence, 539 U.S. at 604 (Scalia, J., dissenting) (“The Court today pretends . . . that we need not fear judicial imposition of homosexual marriage. . . . Do not believe it.”).

As Justice Kavanaugh pointed out, Congress would likely have prohibited discrimination on the basis of sexual orientation in the relatively near future. 140 S. Ct. at 1836 (Kavanaugh J., dissenting). Perhaps legislation outlawing discrimination against transgender individuals would also have been enacted fairly soon. But such legislation would almost certainly have included politically popular exceptions that will be difficult, after Bostock, for judges to infer from Title VII as it is now written.


[51]  See 140 S. Ct. at 1739 (“[W]e proceed on the assumption that ‘sex’ [in 1964] signified what the employers suggest, referring only to biological distinctions between male and female.” (emphasis added)).


[52]  See, e.g., McDonald v. Santa Fe Trans. Co., 427 U.S. 273 (1976).


[53]  401 U.S. 424.


[54]  For a short discussion, see Nelson Lund, The Law of Affirmative Action in and after the Civil Rights Act of 1991: Congress Invites Judicial Reform, 6 Geo. Mason L. Rev. 87, 91-101 (1997). For more detail, see Michael Gold, Griggs’ Folly: An Essay on the Theory, Problems, and Origin of the Adverse Impact Definition of Employment Discrimination and a Recommendation for Reform, 7 Indus. Rel. L.J. 429 (1985). For an analysis of the current state of the doctrine that has evolved from Griggs, see Gail L. Heriot, Title VII Disparate Impact Liability Makes Almost Everything Presumptively Illegal . . . ,


[55]  42 U.S.C. § 2000e-2(j) (“Nothing contained in [Title VII] shall be interpreted to require any employer . . . to grant preferential treatment to any individual or to any group because of the race, color, religion, sex, or national origin of such individual or group on account of an imbalance which may exist with respect to the total number or percentage of persons of any race, color, religion, sex, or national origin employed by any employer . . . in comparison with the total number or percentage of persons of such race, color, religion, sex, or national origin in any community, State, section, or other area, or in the available work force in any community, State, section, or other area.”).


[56]  United Steelworkers v. Weber, 443 U.S. 193, 209-16 (1979) (Blackmun, J., concurring).


[57]  Id.


[58]  Id. at 201.


[59]  Id. at 201-02.


[60]  Id. at 202 (quoting Senator Hubert Humphrey). As enacted in 1964, Title VII did not apply to governmental employers.


[61]  480 U.S. 616 (1987).


[62]  Weber, 443 U.S. at 209 (emphasis added).


[63]  480 U.S. at 629 n.7.


[64]  Id. at 640.


[65]  Id. at 671-72 (Scalia, J., dissenting).


[66]  For a more detailed discussion of the 1991 Act and its effect on Weber and Johnson, see Lund, The Law of Affirmative Action, supra note 54.


[67]  42 U.S.C. § 2000e-2m. The proviso at the beginning of this sentence refers to a number of exceptions to the general anti-discrimination rule that are expressly specified in the statutory text.


[68]  42 U.S.C. § 1981 note.


[69]  42 U.S.C. §2000e-5(g). The only other reference to “affirmative action” in the statute concerns the enforcement of affirmative action plans that the government has approved for its contractors. 42 U.S.C. §2000e-17. That provision does not authorize the use of preferences based on race or sex.


[70]  See Lockhart v. United States, 136 S. Ct. 958, 970 (2016) (Kagan, J., dissenting) (“‘When there is a straightforward, parallel construction that involves all nouns or verbs in a series,’ a modifier at the end of the list ‘normally applies to the entire series.’”) (quoting Antonin Scalia & Bryan A. Garner, Reading Law: The Interpretation of Legal Texts (2012)). This canon applies equally to a modifier at the beginning of a series.


[71]  See id. (“‘When the syntax involves something other than [such] a parallel series of nouns or verbs,’ the modifier ‘normally applies only to the nearest reasonable referent.’”) (quoting Scalia & Garner, supra note 70); Barnhart v. Thomas, 540 U.S. 20, 26 (2003) (discussing “the grammatical ‘rule of the last antecedent,’ according to which a limiting clause or phrase . . . should ordinarily be read as modifying only the noun or phrase that it immediately follows . . . .While this rule is not an absolute and can assuredly be overcome by other indicia of meaning, we have said that construing a statute in accord with the rule is ‘quite sensible as a matter of grammar.’”) (citations and case-specific references omitted). As with the series-qualifier canon, this principle applies equally to modifiers that precede a series.


[72]  Note that the savings clause is not a proviso that creates an exception to the general rule against intentional discrimination. It is instead a rule of construction, which warns against possible misinterpretations. To construe a warning against misinterpretations as a warning against the unambiguously plain meaning of a provision would render the statute self-contradictory.


[73]  I am well aware, of course, that the Supreme Court has declared its own opinions to be the supreme law of the land. See, e.g., Cooper v. Aaron, 358 U.S. 1, 18 (1958). I am also aware that this declaration is widely accepted, just as some Roman emperors were once widely worshipped as divinities. Like these ancient rulers, however, Supreme Court Justices are gods only in a metaphorical sense of the term. Similarly, the opinions they issue are law only in a metaphorical sense.


[74]  140 U.S. at 1747 (citations omitted). Gorsuch even goes out of his way to quote a concurring opinion in which Justice Scalia said that “[a]rguments based on subsequent legislative history . . . should not be taken seriously, not even in a footnote.” Id.


[75]  490 U.S. 228 (1989).


[76]  For an overview of the debate, see Roger Clegg, Introduction: A Brief Legislative History of the Civil Rights Act of 1991, 54 La. L. Rev. 1459 (1994).


[77]  137 Cong. Rec. S15, 477-78 (daily ed. Oct. 30, 1991) (memorandum submitted by Sen. Bob Dole).


[78]  For further detail on the Dole memorandum and its relation to similar memoranda submitted by Senators Jack Danforth and Edward Kennedy (the other two leading negotiators), see Lund, The Law of Affirmative Action, supra note 54, at 127 nn.190-91.


[79]  Note that I am not claiming that the inaction by Congress in 1991 itself operated to overrule Weber and Johnson. Rather, my claim is that its deliberate inaction deprived those decisions of their precedential force on Johnson’s own theory of precedent. That means that the Court now has no excuse for refusing to enforce the statute’s completely unambiguous textual commands.


[80]  557 U.S. 557 (2009).


[81]  Bostock, 140 S. Ct. at 1737. Lest one think that “the written word” might include judicial opinions, here is the quotation in context: “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.”