The Goose and the Gander: How Conservative Precedents Will Save Campus Affirmative Action

Case Comment - Volume 102 - Issue 1


In Students for Fair Admissions, Inc. v. President & Fellows of Harvard College and its companion case, Students for Fair Admissions, Inc. v. University of North Carolina (collectively, “SFFA”),[1] the Supreme Court held that the university–defendants’ race-conscious admissions policies violated the Constitution’s Equal Protection Clause and Title VI of the Civil Rights Act of 1964.[2] Although the Court stopped short of explicitly overruling the cases that first authorized such policies, scholars, concurring and dissenting Justices, and even the majority opinion itself all strongly suggest that the holding functionally accomplishes exactly that.

Even before the SFFA ruling was released, the cases provoked much consternation among liberal supporters of affirmative action. The widespread assumption from the start had been that the Court’s conservative majority would rule for the petitioners and against the colleges.[3] The result of such a ruling, progressives maintained, would be the end of preferences for racial minorities in college admissions and a major setback for the cause of racial equality.[4]

We contend, on the contrary, that the SFFA decision will have almost no practical effect at all. To be sure, we agree with the Justices and commenters who argue that the Court has functionally outlawed the consideration of race in college admissions for the purpose of promoting diversity. SFFA thus represents a symbolic victory for affirmative action’s conservative opponents. But not much more. Race-based affirmative action in college admissions will probably not end. It may not even slow down. Colleges that wish to continue practicing it do not need to pause to overhaul their admissions procedures. And perhaps ironically, these possible (non)results flow from a set of previous, non-symbolic conservative victories at the Supreme Court in the fight over racial discrimination.

Here, in brief, is why. First, to understand what will or will not change post-SFFA, one needs to understand the previous state of play. As we will explain below, the Supreme Court’s prior campus-affirmative-action decisions made that state of play a strange one. In effect, race-based preferences in admissions were legal. But the Court placed mysterious—arguably incoherent—restrictions on how those preferences could be structured. Thus, for decades, universities understood that, while they could openly advertise that they practiced affirmative action, they had to diligently obscure all actual instances of admission because of race. Doing otherwise risked legal liability. The SFFA opinion, along with undisputed facts presented in Harvard and UNC’s briefs, confirms that admissions departments conducted themselves accordingly.[5]

What will SFFA change? While the Court did not formally overrule its precedents authorizing campus affirmative action, it placed conditions on such policies that appear to make literally all of them illegal. Ostensibly, then, affirmative action is legally dead. But even before SFFA, under the Court’s prior incoherent restrictions, essentially all affirmative action policies would have been struck down as illegal if their inner workings had been plainly revealed. Moreover, the SFFA opinion makes clear that colleges retain the prerogative to admit students according to multifactorial, discretionary, and ultimately obscure criteria. Thus, today, as before SFFA, colleges are functionally able to make admissions decisions on the basis of race. They must only, as before SFFA, make sure they adequately cover their tracks. Thus, we think that SFFA requires only one practical change to college admissions practices: universities will no longer be able to advertise that they practice affirmative action, even in the abstract. Most will probably have to change the marketing copy on their admissions pages. And they will have to be more circumspect when describing admissions officers’ motivations in litigation. But not much more.

Post-SFFA, one might think that even if universities no longer openly admit that they use race as a factor in admissions, they will be discovered and held accountable. After all, the SFFA briefing was full of statistical analysis purporting to show that Harvard and UNC have made many, many individual admissions decisions on the basis of race.[6] But one would be wrong. Those statistical analyses were irrelevant to the legal arguments in SFFA, and they will remain legally irrelevant in challenges to affirmative action going forward. This is because of the Supreme Court’s decades-old rulings in Washington v. Davis[7] and McCleskey v. Kemp.[8] In those cases, the Court held that statistical proof cannot carry a constitutional discrimination claim.[9] Rather, direct proof of intent to discriminate is required.[10] The statutory rules governing colleges operate similarly.[11] Thus, even after SFFA, affirmative action will invite liability, to a first approximation, only if universities admit that they do it. And given SFFA’s holding, universities will stop admitting it. From there, Washington and McCleskey—cases long reviled by progressives and endorsed by conservatives—will give colleges all the cover they need.

I. The Pre-SFFA State of Play

We begin with Regents of the University of California v. Bakke.[12] There, U.C. Davis’s medical school set aside 16 of 100 admissions seats annually to be filled by members of certain “minority group[s].”[13] No white applicant had ever been admitted for one of those seats.[14] The Supreme Court held that this type of affirmative action program violated the strictures of the Equal Protection Clause.[15] In a plurality opinion by Justice Powell, the Court held that diversity in education was an appropriate constitutional purpose.[16] Thus, the use of race in admissions was constitutionally allowed if every applicant’s merits were considered individually.[17] But the use of a quota that excluded certain applicants from competition based on their race was unconstitutional.[18]

So far, so good. Bakke drew the constitutional line at quotas. Colleges could not refuse to consider students for certain seats because of their race and irrespective of their qualifications. But then came the twin cases of Gratz v. Bollinger[19] and Grutter v. Bollinger.[20]

Those cases reviewed the University of Michigan’s affirmative action policies for undergraduate and law school admissions. Gratz dealt with the university’s points-based system for reviewing undergraduate applications.[21] The maximum score was 150.[22] A score of over 100 would likely result in admission, and a score below 75 would likely result in rejection.[23] Points were awarded for certain nonacademic factors, with membership in a racial minority group garnering 20 points.[24] For comparison, being recruited as an athlete was also worth 20 points, and being a Michigan resident was worth 10.[25] The Court held that this system was unconstitutional under Bakke.[26]

The reasons are somewhat mysterious. Michigan’s system was not a quota; every applicant of any race was eligible for any seat.[27] Nevertheless, the Court thought the system did not sufficiently “consider[] each particular applicant as an individual.”[28] This was apparently because the university “automatically distribute[d]” points to all minority applicants, “embody[ing] the . . . notion[] that a particular applicant, by virtue of race or ethnicity alone, is more valued than other applicants.”[29] Such automatic assignment of value based on race meant that, for many applicants, race would be “decisive” in the decision to admit.[30] Such decisiveness, in turn, violated Justice Powell’s command that “each characteristic of a particular applicant was to be considered in assessing the applicant’s entire application.”[31]

What could all of this mean? Michigan’s undergraduate formula did consider many factors aside from race, from academics to residency to extracurriculars to legacy status.[32] What more could the University have done to ensure its process was sufficiently “individualized?” Moreover, how could the assignment of value based on race, qua race, be forbidden under Bakke? Bakke endorsed affirmative action, and what is affirmative action but a preference—even if not a quota—based on race? Finally, what would it mean to consider race as a factor in admissions without race ever being “decisive” in making an admissions decision?

Grutter provides an example, though not necessarily an explanation. Unlike the university’s undergraduate-admissions office, Michigan Law’s admissions office did not utilize an explicit points system.[33] But it did, like its undergraduate counterpart, consider academic credentials and a number of other factors.[34] Among the “soft variables” considered were, for example, the enthusiasm of recommenders, quality of essays, and membership in a minority group—including “African-Americans, Hispanics, and Native Americans.”[35]

The Court held that this affirmative action program was constitutional.[36] Contrasting it with the undergraduate college’s approach, the Court wrote that, while the Law School used race as a “plus factor,” it did not award “mechanical, predetermined diversity ‘bonuses’ based on race.”[37] The difference, apparently, between Michigan’s undergraduate and law school admissions policies was that the latter paid only “‘[s]ome attention to numbers,’ [not] more.”[38] Years later, the Court would similarly endorse another affirmative action program—at the University of Texas—because it used race as “but a factor of a factor of a factor.”[39]

Confronted with these opinions, what should a law-abiding admissions director do? She knows that she may consider race in admissions for the purpose of ensuring diversity on campus. But how might she operationalize that power? She may not decide in advance to assign a particular amount of weight to race, as compared with other “soft” factors. This would violate Gratz’s rejection of a points-based system. She apparently may not use race in the same predetermined way in every case—even as a consistent tiebreaker between otherwise-identical applications. That would violate the Court’s understanding of “individualized” review. Indeed, it is not clear that she can ever use race as a reason to admit an applicant who would otherwise be rejected—even in relatively close cases—when race is but one of multiple jointly sufficient reasons. That might well violate Gratz’s prohibition of the “decisive” use of race, along with the Court’s command that race only ever be a “factor of a factor of a factor.”

Perhaps she could look to Justice O’Connor’s concurrence in Gratz for guidance. In O’Connor’s view, the main problem with Michigan’s undergraduate affirmative action policy was not that it assigned a fixed amount of value ex ante to race.[40] Rather, it was the amount of value assigned.[41] Racial minority status garnered more points than essentially all other soft factors, including leadership, service, and geographic diversity.[42] In doing so, race dominated the process, “automatically determin[ing] the admissions decision for each applicant.”[43] This resonates with a complaint the majority raised—that Michigan’s points system assigned more points for minority status than for “artistic talent rival[ling] that of Monet.”[44]

Maybe, then, our Director of Admissions could institute a new points-based system where race is scrupulously assigned no more weight than exceptional skill in painting. Even if such a system would satisfy Justice O’Connor, it would address only one of the Gratz majority’s many objections. Such a system would still make race decisive in many admissions decisions. It would still assign value to race qua race, and in doing so, it would fail to comport with the majority’s conception of individualized review.

Best, then, for our Director of Admissions to hew closely to the paradigmatic example of a permissible affirmative action policy—Grutter. What was special about Michigan Law’s approach? Surely, its admissions department did do most of the very things the Court found objectionable about the undergraduate approach. For example, it is almost inconceivable that the law school never treated race as a decisive factor in any particular admissions decision. Likewise, the law school surely valued membership in an underrepresented minority group for its own sake. Treating minority status as a desirable trait and a decisive reason to admit students, at least in close cases, is simply what it means to practice affirmative action.[45]

Instead, Michigan Law’s key strategy—the thing that distinguished it from the undergraduate college—was obscurity. Unlike Michigan’s undergraduate admissions, the law school’s approach to affirmative action ensured that it was not obvious how decisions were being made, at least to the public. Their holistic process—free from points, rubrics, or apples-to-apples comparisons—left no record of why, precisely, any particular applicant was admitted. And if outsiders (or courts) cannot determine why any particular person was admitted, they cannot tell whether race was assigned some weight ex ante. They cannot say whether it was given too much consideration or used too often. They cannot even determine whether it was ever the decisive factor—even conjointly—in admitting any particular applicant. A blunter way of putting this is that, under Grutter and Gratz, college admissions can use race decisively, so long as they leave no record of precisely when and how they have done so.[46]

SFFA reveals that admissions departments got the message. Consistent with the prerogative granted in Bakke, both Harvard and UNC stated that diversity was a critical part of their educational and social mission.[47] Consequently, and again consistent with Bakke, both publicized that they used race as a factor in admissions decisions.[48]

But, following Grutter’s model, Harvard and UNC both repeatedly asserted that their admissions processes were holistic.[49] And indeed, both processes largely obscured the precise reasons for admissions decisions in exactly the same way as Michigan Law’s did. Both universities’ applications requested a wide array of information about the candidate, including, for example, transcripts; standardized test scores; information about extracurricular activities; honors and prizes; essays; family and parental information; financial information; and demographic information, including race.[50]

Based on this information, applications were given an initial holistic read. At Harvard, a “first reader” exercised personal judgment to synthesize the entire application into tentative ratings across four categories: “academic, extracurricular, athletic, and personal.”[51] The first reader then assigned an “overall” rating.[52] The overall rating was, again, judgment-based and holistic, reflecting the reader’s personal “impression of the strength of the application, taking account of all information available at the time.”[53] “All information” here included race, which readers could and did take into account.[54] Readers could also give “tips” for qualities that weren’t easily quantifiable.[55] These included, among other things, diversity in background “that expand[ed] the socioeconomic, geographic, racial, or ethnic diversity of the class.”[56]

After review by a subcommittee, recommended candidates were then referred to the full Admissions Committee.[57] The Committee had several kinds of information it could consider. Although the preliminary rankings “fade[d] to the background” at this stage, the Committee still had the full application file.[58] Beyond those, members of the Committee sometimes received “one-pagers” containing “characteristics of the applicant pool and tentatively admitted class, including . . . race,” among other qualities.[59] The Committee reviewed and discussed all of these documents.[60] It then made tentative application decisions by a vote of the governing Committee.[61] Following those votes, the Committee received further information about the pool of tentative admits, including its racial composition.[62]

Harvard maintained that the demographic information presented to the Committee before its tentative decisions was not used to pursue racial quotas.[63] Instead, only if the data revealed “anomalies in the representation of students with certain characteristics, including race,” might the Committee have given certain candidates further consideration—including on the basis of race.[64] This is certainly compatible with race-based decision making to promote racial diversity. But it is also compatible with the use of race to identify and correct potential racial discrimination in the initial votes, and the SFFA Court did not find otherwise.

Finally, Harvard reduced its pool of tentative admissions to the final class in a process called the “lop.”[65] When deciding whom to cut from the pool of tentative admits, admissions officers were provided four pieces of information about each candidate: legacy status, recruited athlete status, financial aid eligibility, and race.[66] Here again, Harvard agreed that the racial diversity of the class was a consideration in making the lop.[67] But Harvard neither assigned race any set weight nor recorded it as being determinative in any particular decision.[68]

Harvard thus openly admitted that race was used to promote diversity in admissions in three ways: as a non-quantitative, holistic input into the initial candidate rating; as the basis of a tip, which was assigned some ambiguous weight for an unspecified set of admissions decisions; and at the lop, where it was assigned ambiguous weight for an unspecified set of cuts. Harvard thus agreed that race was “determinative” for some portion of the admitted class.[69] But it studiously avoided leaving written records of exactly the cases in which race mattered, the way in which it mattered, or how much it mattered.[70]

UNC’s process was similar. It involved a first reader who rendered a provisional, judgment-based, non-quantitative decision.[71] There was higher review, including by a committee.[72] As with Harvard, UNC was circumspect about how exactly race mattered.[73] It made special note that race was only considered non-mechanically alongside other factors, such that it would only tip the scales in some non-specified set of cases.[74] Notably, UNC contended that this almost never happened,[75] and it kept no records of any individual cases in which race was decisive to an admissions decision.[76]

Both schools, then, pursued Grutter’s opacity strategy at every turn, advertising their use of race in admissions but hiding every instance where it was used. Indeed, even UNC’s illustration in litigation of when race could, in principle, matter was equivocal. UNC proffered the example of an applicant moving from Vietnam to North Carolina and thriving in an unfamiliar setting, which could impress admissions officers.[77] Is race even a factor here? Or would the applicant be treated the same if she happened to be a white immigrant from Vietnam? UNC scrupulously declined to say.[78]

The net result, Justice Sotomayor’s dissent pointed out, is that even after years of discovery and litigation, Harvard and UNC avoided revealing even “a single example of an underrepresented racial minority who was admitted to Harvard or UNC on the basis of ‘race alone.’”[79] Even before the SFFA decision, then, obscurity was the coin of the realm.

II. What Will (Not) Change After SFFA

What will SFFA change, then, about college admissions? The majority opinion is not a model of clarity. On the one hand, the decision does not explicitly overrule Grutter, Gratz, or their brethren. On the other hand, the Court held that educational goals associated with diversity, like “enhancing . . . cross-racial understanding,” would no longer be considered sufficiently “measurable” or “coherent for purposes of strict scrutiny.”[80] The Court also resolved the incoherence inherent in Grutter and Gratz: How can a college possibly practice affirmative action in admissions without ever allowing race to be “decisive” in any decision? It can’t, the SFFA Court candidly said, since “[c]ollege admissions are zero-sum.”[81] We therefore agree with Justice Sotomayor, who in turn agrees with Justice Thomas, that “Grutter is, for all intents and purposes, overruled.”[82]

What, then, will change in this new world of college admissions? Most commentators think the answer is “a great deal.”[83]

We think, on the contrary, that almost nothing will have to change. Here is the one thing that cannot stay the same: how universities talk about admissions. As noted above, before SFFA, Harvard and UNC, like many other universities, maintained explicit language on their websites and elsewhere asserting that they use race in assessing applicants.[84] In a world where the use of race in admissions is verboten, such statements have to go. Universities will now need to update their mission statements, websites, advertisements, and the like. People in leadership positions, too, will have to change how they speak in public about their institutions’ admissions procedures. In particular, they will have to change what they say about those procedures in litigation. Rather than admitting that race matters, while obscuring when and how, they will have to stop admitting it matters.

But that may be all that has to change. Consider what SFFA explicitly allows to remain the same. Colleges will still be able to consider a variety of data inputs, including essays, in deciding whom to admit.[85] And they will still be allowed to use, as they have since Grutter and Gratz, a holistic process as they assess applicants.[86] Nothing in the Court’s decision requires them, for example, to reinstitute mechanical points-based admissions systems but with no points assigned for race.

And as we have shown, these processes of admission are impenetrable to the outside spectator. They are, following Grutter, impenetrable by design. Their highly pluralist decision criteria, their lack of recorded explanations of individual decisions, and their use of multilevel procedures with multimember votes all make it impossible to know how decisionmakers weigh different candidates against each other. The point is to obscure the reasons that any particular candidate was admitted, especially as it relates to race, ethnicity, and other such diversity criteria.

The original reason for these processes was to avoid being caught running afoul of Gratz’s inscrutable command. But they will work just as well to avoid being caught running afoul of SFFA’s clear one.

Suppose, then, that colleges scrub their mission statements, websites, and speech of any admission that race qua race matters. Will remaining silent, at least in public, about race be enough for colleges to maintain their affirmative action programs? We think so. The proof is in the history. After Gratz, colleges understood that they could not generate actionable evidence of any particular race-based admission or else risk liability. They succeeded in covering their tracks. Of the many hundreds of thousands of students who applied in those decades to colleges that practiced affirmative action, vanishingly few have brought claims based on Gratz’s restrictions.[87] Even fewer have won.[88] Absent proof that any admissions committee ever used race in a “decisive,” non-“individualized,” or other forbidden way, such claims were not viable.

These same factors will make post-SFFA claims based on allegations that race mattered to admissions likewise unlikely to succeed. The SFFA litigation itself was viable only because Harvard and UNC, following the Gratz strategy, freely admitted that race factored in.[89] Their admissions were the case’s lynchpin, since mountains of discovery on admissions procedures failed to produce even “a single example of an underrepresented racial minority who was admitted to Harvard or UNC on the basis of ‘race alone.’”[90] After SFFA, such admissions will dry up.

The majority opinion in SFFA declares confidently that evasion by obscurantism will not succeed. “[U]niversities may not simply establish through application essays or other means the regime we hold unlawful today,” the Court wrote.[91] “What cannot be done directly cannot be done indirectly.”[92]

If only it were so. In the majority’s imagined world, every violation of a constitutional right has a remedy. And procedural hurdles—like who bears the burden of proof and what that burden consists of—never defeat an otherwise meritorious claim. But that is not the world in which we live. In our world, what “can be done,” whether “directly” or “indirectly,” depends on how legal rules translate into legal liability. This, in turn, depends on the elements of a claim, the rules of pleading and discovery and evidence, and much more. If the years of litigation in SFFA failed to reveal any individual race-based decisions, it is far from obvious that future claims will succeed.

What about statistics? Perhaps, going forward, they will fill the evidentiary void opened by universities’ new refusals to admit to considering race. The SFFA briefing was chock-full of statistical analyses,[93] some of which the majority opinion reproduced.[94] Even if colleges are excellent at obscuring individual uses of race in admissions decisions, won’t statistical evidence of its aggregate use carry a claim after SFFA?

No. To be sure, the plaintiffs in SFFA did collect and introduce lots of statistics into their briefing.[95] Their analysis purported to show, among other things, that Harvard implemented decisive racial preferences in part by gerrymandering candidates’ personality scores along racial lines.[96] Harvard strenuously denied the charge.[97] The Court likewise took note of gross statistical regularities in the minority share of Harvard’s admitted classes.[98]

From a legal standpoint, however, such statistics were a sideshow. And going forward, they will likely be irrelevant. Here is why: for certain kinds of discrimination claims, statistical proof can be a vital tool of litigation. Under Title VII of the Civil Rights Act, for example, plaintiffs can bring a claim by showing that some policy created a disparate impact on members of different racial (or other protected) groups.[99] Demonstrating aggregate statistical differences in outcomes is sufficient to carry the plaintiff’s initial burden.[100] Then, the burden shifts to the defendant to show that the challenged policy and its disparate impact served some legitimate purpose and was not bare discrimination.[101] If they can, they escape liability, but if not, the plaintiffs win.[102]

But the disparate impact approach and its attendant statistical proof are not always allowed. Notably, it is unavailable in Equal Protection cases. In Washington v. Davis, two Black applicants to the District of Columbia’s Police Department complained that the Department’s recruiting procedures, particularly the written exam, were racially discriminatory.[103] They argued that a disproportionate number of Black applicants failed the verbal exam and that the exam had not been shown to reliably indicate better job performance.[104] The statistics showed that four times as many Black applicants failed compared to white applicants.[105] The Court of Appeals held that the disproportionate impact of the verbal test was enough to constitute a constitutional violation.[106]

The Supreme Court reversed. The Court held that racially disparate impact is not enough to show a constitutional violation.[107] Rather, direct evidence of intent to discriminate was required.[108] Since the exam was neutral on its face and it was appropriate for the government to seek a “modest[] upgrade [in] the communicative abilities of its employees,” the administration of the verbal exam was not unconstitutionally discriminatory.[109]

McCleskey v. Kemp,[110] decided a decade later, reinforced Davis’s holding. There, a Black man convicted of murder challenged his capital sentence on Equal Protection grounds.[111] He relied on “sophisticated statistical studies” showing that Black defendants were sentenced to death much more often than white ones.[112]

The Court rejected the proof as irrelevant to the requisite legal test.[113] Instead, the Court held that a party alleging an equal protection violation must “prov[e] ‘the existence of purposeful discrimination.’”[114] That means showing that “the decisionmakers in [his] case acted with discriminatory purpose.”[115] Statistical evidence, which characterizes the whole class of cases, “is clearly insufficient to support an inference that any of the decisionmakers in [an individual case] acted with discriminatory purpose.”[116]

The same goes for Title VI of the Civil Rights Act, which, unlike the Equal Protection Clause, binds private universities.[117] Title VI does not allow private claimants to prove their discrimination claims using a statistical disparate impact approach.[118] The Supreme Court has yet to decide whether agencies may do so, but it has suggested that interpreting the statute to allow it would be “strange.”[119] Even if agencies could use statistical proof, the statute makes agency enforcement highly cumbersome—requiring, for example, written reports to Congress regarding any penalties imposed.[120] And the remedies are limited: damages are not available.[121] Moreover, even where disparate impact proof is allowed, liability can be avoided by identifying a legitimate, non-racial selection criterion that happens to produce the disparity.[122]

Thus, statistics will not carry post-SFFA claims against universities that continue to use race as a factor in admissions. Under both operative laws—the Equal Protection Clause and Title VI—only direct proof of discriminatory intent in individual admissions decisions will do. And as we have just seen, universities are already adept at ensuring no such proof can be found.

What about pleading standards? Even if statistics alone cannot win a case, perhaps they could state a claim and trigger discovery. That might be enough to deter universities from continuing to use race in admissions, either for fear of litigation costs or for fear that discovery might reveal internal communications containing loose talk about decision criteria. But even here, we think the risks are small. Consider Ashcroft v. Iqbal,[123] which—with Bell Atlantic v. Twombly[124]—gives us the modern federal pleading standard. Iqbal was an Equal Protection case alleging a “disparate, incidental impact” on Muslims of Arab descent from certain post-9/11 national security policies.[125] The Supreme Court held that this was not sufficient to state a claim and obtain discovery. It reasoned that, even at the pleading stage, Courts could not “infer” a conclusion of “invidious discrimination” from statistical imbalances when there were “obvious alternative explanation[s].”[126] Universities will similarly have “obvious alternative explanations” available for statistical evidence of racial disparities in admissions. As already discussed, they will still be allowed to use multifactorial, holistic decision procedures. And as discussed further below, SFFA explicitly affirmed their ability to consider factors that correlate strongly with race—like having overcome discrimination.[127]

What about the practicalities of affirmative action after SFFA? How, going forward, will colleges learn applicants’ race/ethnicity? There are at least two ways in which they currently source the data: (1) a checkbox that asks applicants to volunteer their racial/ethnic identity; or (2) college essays that allow applicants to discuss their racial/ethnic identity if they wish to.[128] Will colleges’ continued collection of racial data via these means invite liability? There are good reasons to think that it will not.

Consider first the checkbox. On one hand, the presence of a “race/ethnicity” box on an application may look like evidence of the very intent that colleges will now have to disavow. Why ask about race if one doesn’t intend to use it in making decisions? A college could argue it is hoping to ensure that its admissions officers are not discriminating on the basis of race. The SFFA did not disturb a state actor’s ability to “remedy a race-based injury that it has inflicted.”[129] Thus, identifying illegal discrimination remains a permissible—and perhaps obligatory—goal post-SFFA. As one of us has written elsewhere, SFFA notwithstanding, colleges can and should implement race-based adjustments to admissions decisions if they are statistically calibrated to remedy what would otherwise be invidious race-based discrimination.[130] Such uses do not necessarily require that an applicant’s racial identification be revealed individually to admissions officers. But one can easily imagine plausible antidiscrimination policies under which they would be.

Essays are an even easier case. As the SFFA Court wrote, “nothing in this opinion should be construed as prohibiting universities from considering an applicant’s discussion of how race affected his or her life, be it through discrimination, inspiration, or otherwise.”[131] This appears to leave the door wide open for individual admissions officers to learn an applicant’s race. Indeed, it seems that the university could explicitly solicit such information by directing each applicant to write about how, if at all, “race affected his or her life.” Harvard University has already announced its intention to do exactly that.[132]

The SFFA majority admonishes that such essays must be used only to make decisions “based on [the applicant’s] experiences as an individual—not on the basis of race.”[133] But the whole point of holistic admissions post-Gratz was to make it impossible to say how any piece of information was used in any given case. And all indicators suggest that colleges succeeded.

All of these factors, we think, cut strongly against the SFFA majority’s assertion that universities “may not simply establish” affirmative action “through . . . other means.”[134] The Court has held that there are legitimate reasons to collect race data on individual applicants and distribute it to admissions officers. From there, universities have decades of practice concealing how, when, and why race is used. They will no longer admit that they consider race, qua race, in their decisions. Cases like Davis and McCleskey suggest that that will be enough to avoid liability, even when statistics suggest illegal conduct. Cases like Iqbal further hold that statistics will not even be sufficient to plead a claim since race-correlated factors may still be considered in admissions. Thus, we tend to agree with Justice Sotomayor’s rejoinder to the majority’s assertion: “No one is fooled.”[135]

Of course, no post-SFFA affirmative action case has yet been litigated, so the doctrinal arguments above remain theoretical. But the experiences of Michigan and California provide some limited empirical evidence. Between the mid-1990s and mid-2000s, both states passed amendments to their constitutions explicitly forbidding affirmative action in college admissions.[136] Both states’ flagship university systems filed amicus briefs in SFFA asserting that: (1) after the state bans, they did, in fact, stop considering race in admissions, and (2) this devastated their ability to enroll a diverse student body.[137] Do these briefs constitute strong evidence that SFFA will do the same?

We think not. To begin, it is far from clear that the state universities’ data shows what they claim—either as to their admissions practices or their downstream effects. Consider first the University of California system (UC). California’s affirmative action ban first took effect for the freshman class of 1998.[138] The demographic reports cited in UC’s brief show that, in 1997, the non-white share of admitted students was about 50%.[139] In 2021, that share was 61%.[140] This represents a steady and ultimately substantial increase, not a decrease, in diversity over time.

Even in 1998—UC’s first year under the new regime—representation among non-white students did not change much. Compared with 1997, the “Hispanic/Latinx” share of admitted students fell by about one percentage point; the same change occurred in the “African American” category.[141] The former fully recovered to 1997 levels in about three years, and the latter in about six.[142] The biggest reduction in representation for the 1998 year was, in fact, for the “White” category, which fell by about seven percentage points.[143] That group has never matched its 1997 share.[144] Rather, it has trended steadily downward over time, while “Hispanic/Latinx” and “International” admissions have trended steadily upward.[145] Other demographic categories have remained remarkably steady—almost exactly at 1997 levels.[146]

UC argued that these system-wide figures obscure inter-campus variation in the system’s ability to recruit—in particular—Black students.[147] It contended, specifically, that African-American representation fell durably at its two most elite schools—UC Berkeley and UCLA.[148] Even if true, this fact would run contrary to the narrative that affirmative action bans uniformly impede the recruitment of Black students. The net system-wide effect, as already noted, has been zero. So any setbacks at Berkeley and UCLA are necessarily offset by leaps ahead at other UC campuses.

But it is not even clear that Berkeley and UCLA have both seen durable setbacks in enrolling Black students. UC cites 2019 figures in support of the claim.[149] But in 2021, a larger share of students admitted to UCLA were African-American than in 1997, and the proportion of those students who enrolled was about the same.[150]

Berkeley alone, then, has apparently seen a durable change. Here, it is worth noting that, in 1997, Berkeley enrolled a substantially larger share of African-American students than the other UCs.[151] Thus, its present (lower) African-American enrollment is in line with the pre-ban figures from both UCLA and the UC system as a whole.[152] It also matches today’s figures from the system as a whole.[153] UC’s brief did not suggest that Berkeley has a special need to be more diverse than its other campuses. It did not even claim that the system tried and failed to achieve such Golden Bear exceptionalism.

The University of Michigan (UM) likewise became somewhat more diverse overall, not less, after the state banned affirmative action. As its brief stated, in 2006—the last year before the ban took effect—“underrepresented minorities made up 12.9% of U-M undergraduates,” and in 2021, the figure was 13.46%.[154] Complicating this top-line picture somewhat, the data does at first appear to show that Black and Native American enrollment, specifically, decreased during the same period.[155]

But there are caveats. First, in 2008—just a year into the ban—UM’s freshman class had more Black representation than it did in either 2006 or 2004, before the ban was in place.[156] Second, as UM’s brief mentioned in passing, in 2010, the federal government changed the racial categories on which UM’s figures are based, so comparing figures before and after that point is difficult.[157] What UM’s brief does not mention is that this change included the addition of “Two or More” as a new category.[158] The new classification may have siphoned off at least some students who would have previously identified as either “Black” or “Native American.” Indeed, after 2010, “Two or More” became a major category. It represented 4.28% of the freshman class as of 2015, compared with 5.76% who identified as Black.[159] Thus, the main change may have been in data coding, not actual student demographics.

Finally, insofar as Black representation fell at UM, that trend appears to have long predated the state’s ban on affirmative action. In 2000, over 9% of UM’s freshman class was Black.[160] By 2006—the final pre-ban year—the figure was down to 7.06%.[161] The state’s ban on affirmative action could not have caused this pre-ban phenomenon. In fact, a naïve projection of the pre-ban trendline would have predicted that, by 2021, there would be no Black freshman enrolled at UM.[162] Consequently, it’s unclear whether UM’s post-ban admissions policies succeeded or failed in promoting diversity. It might be that they did much better than pre-ban expected trends would otherwise have dictated.[163]

Given all of this, UC and UM might well be cited as case studies favoring our view, rather than refuting it. Neither university suffered a collapse in diversity after their states banned affirmative action in college admissions. On the contrary, both have become more diverse, with increased Latino representation leading the way.[164] And even as to the most disadvantaged groups—like Black students—the picture is ambiguous at best. Therefore, the most epistemically conservative conclusion is probably that the bans had little effect on Black representation in two of America’s best public university systems.[165]

This, of course, is our prediction as to SFFA. And in both contexts, the mechanism could easily be the same. Both UC and UM practice “holistic”—that is, discretionary and inherently obscure—admissions.[166] Thus, both could potentially have pursued the path we lay out above: delete race as a consideration on official documents, change nothing else, and rely on Grutter-style obfuscation to protect against liability.

We do not mean to imply that UC and UM did do this. Just that they could have, and that the public data appears to roughly fit that narrative. Other stories are also plausible. First, it is possible that UC and UM did comply because the liability risks under their states’ rules are much higher than those we describe above. Equal Protection and Title VI claims do not allow statistical proof. But state constitutional claims might.[167] Relatedly, it is possible that the race-neutral strategies for increasing diversity that UC and UM implemented turned out to be, in the universities’ view, good substitutes for affirmative action.[168] If so, there would be little reason not to comply with bans. Finally, UC and UM officials might simply have decided that there are independent normative reasons to follow a clear legal directive, irrespective of the threat of liability. With the exception of the differences in state and federal law, these same factors could influence all universities post-SFFA. We therefore make the same claim about them as about UC and UM: not that all colleges will certainly continue to practice affirmative action without much fear of liability, just that they could.

III. The Goose and The Gander

Perhaps we have now aggravated all of our readers. Left-leaning supporters of affirmative action may be offended at our suggestions that: before SFFA, most colleges’ affirmative action programs were quasi-legal at best; that colleges knew this; and that they compensated by intentionally obscuring the programs’ inner workings. Doubly so by our prediction that some colleges will violate the law all the more knowingly now that SFFA has functionally banned it. Right-leaning opponents of affirmative action may be outraged at our suggestion that, going forward, the law will let colleges get away with it.

This is a near exact inversion of the political dynamic that has attended debates over the Equal Protection Clause for the last half century. At least since the Supreme Court’s 1976 decision in Washington v. Davis, it has been conservatives who scoffed at the suggestion that government officials made pervasive, objectionable, and intentionally obscure race-based decisions.[169] And it has been liberals who were outraged that holdings like Davis and McCleskey let them get away with it.[170]

Consider, for example, the 1980 case of City of Mobile v. Bolden.[171] There, Black plaintiffs challenged Mobile’s system for electing its municipal government. The city, like many in the South, eschewed traditional geographically defined legislative districts in favor of election of the entire legislature by the entire voting populace of the city.[172] This ensured that, even in a city with a large Black population, “no Negro had ever been elected to the City Commission.”[173]

Despite this, and despite evidence that white Commissioners had previously discriminated against Black citizens, a plurality of the Court—including its conservatives—found no equal protection liability.[174] Under Davis, they wrote, direct proof of discriminatory intent in the challenged action was required.[175] And irrespective of the Commissioners’ prior discrimination, there was no direct evidence—say, a damning statement—that the city’s lawmakers adopted its electoral system, specifically, for racist reasons.[176] Absent such evidence, the plurality declined to infer such reasons.[177]

Justice Marshall was outraged. First, he thought discriminatory purpose irrelevant.[178] It was reason enough, he argued, to invalidate the electoral scheme that had the “discriminatory impact” of locking Black citizens out of elected office.[179] Second, he saw the plurality’s refusal to infer discriminatory intent as a farce.[180] He invoked, perhaps sardonically, the “useful evidentiary tool, long recognized by the common law . . . that ‘every man must be taken to contemplate the probable consequences of the act he does.’”[181] The voting system’s racially differential effect of at-large voting, Marshall explained, was so obvious as to raise a “strong inference that the adverse effects were desired.”[182]

Post SFFA, the tables have turned. In at least this one area—college affirmative action—cases like Davis suddenly protect the policy interests of liberals at the expense of conservatives. What is there to make of this?

We do not endeavor here to determine whether rules like Davis and McCleskey’s are, in the end, good ones. We do wonder, however, whether the post-SFFA world will present an unusually good opportunity for productive discourse between usual opponents. Perhaps college admissions practices after SFFA will allow each side to see the other’s point of view more clearly. In an ideal world, the result might be legal progress—a new rule that better responds to both camps’ concerns.[183]

Conservatives who think diversity-promoting affirmative action constitutes invidious discrimination may wish to reconsider perennial liberal arguments against Davis and McCleskey. They could begin with those articulated by Justice Marshall in Bolden. First, a process that generates large racial disparities might be normatively objectionable even when no conscious discrimination can be proved—indeed, even where it does not exist. The world is complex, and overlapping institutions—housing, education, government, and more—are marred by histories of conscious racism. The result is that, even where consciously bigoted attitudes have largely fallen away, old processes that seem neutral to those using them can produce disparate results. Maybe the law should try to fix those broken processes in addition to prohibiting intentional discrimination. The SFFA plaintiffs’ emphasis on how affirmative action harms Asian applicants suggests that they, likewise, thought that outcomes mattered independently of intent.[184]

Second, conservatives in the post-SFFA world may feel a new attraction to Justice Marshall’s insistence that intent can be inferred from history and context even when direct proof is lacking. Colleges may or may not outwardly appear to change their admissions processes now that SFFA has been decided. But whatever they do, if the racial mixes of their admitted classes remain relatively steady, surely it is not outré to suggest that the processes were selected with the results in mind.

Liberals, in turn, may find new plausibility in the perennial conservative argument that raw racial disparities do not always indicate racial discrimination. Like it or not, in the modern world, many attributes correlate with race. Some of them are legitimate grounds for decision. Even after SFFA, the majority opinion confirms, colleges may constitutionally assign value, for educational reasons, to diversity of experience in the student body.[185] Such diversity need not be race-based per se. UNC, for example, may continue to value representation of the immigrant experience on its campus. Over two-thirds of immigrants to the United States are non-white.[186] Selection for immigrants will entail, incidentally, selection for non-white race and ethnicity. The same goes, and doubly so, if colleges accept the SFFA Court’s invitation to select for students who have overcome the adversity of racial discrimination.

Relatedly—and contra Marshall’s approach to Mobile’s racist Commissioners—supporters of diversity on campus might come to see some sense in Justice Scalia’s perennial refusal to impute the intent of one member of a body to the body itself.[187] Suppose that, after SFFA, UNC continues to (legally) seek immigrants among its student body. Should its admissions practices be held unconstitutional if, say, one misguided admissions officer, in an informal instant message, suggests selection based on race?[188]

The inversion of political fortunes here, however, is not perfectly symmetrical. Consider that, irrespective of their post-SFFA effect on campus affirmative action, Davis and McCleskey will remain serious impediments to other progressive causes. In areas like voting rights and criminal justice, they will continue to foreclose Equal Protection and other similar lawsuits designed to close persistent racial gaps. As high as the stakes are in college admissions, the stakes of incarceration and representative government are even higher. This might be reason enough for those on the left to maintain their opposition to those cases’ holdings.

But in any case, holdings like Davis and McCleskey—and Iqbal’s application of their principle to pleading—now cut both ways. Thus, any changes the Supreme Court makes in this area will have cross-cutting effects. If it weakens either the substantive liability or pleading standard in Equal Protection and related antidiscrimination cases, it will make all antidiscrimination claims easier to bring. That includes the ones conservatives favor—challenging affirmative action—and the ones liberals favor—challenging criminal prosecutions, voting rules, and more. And vice-versa. What’s sauce for the goose will be sauce for the gander.

IV. The Law and Norms of Disobedience

We have argued that SFFA’s prohibition on the use of race in admissions will, in fact, not dent colleges’ ability to keep employing race-based affirmative action. They simply need to do what they’ve already been doing: maintain a modicum of discretion. This, however, would directly contravene the Supreme Court’s interpretation of the Constitution. So are we suggesting that colleges will disobey the law? That they should? Or that if they did, it wouldn’t be wrongful in some legal or broader normative sense?

The answers to these questions, respectively, are “no,” “no,” and “it depends.” First, we do not know what colleges will actually do. Perhaps colleges will comply with the Supreme Court’s admonishment not to use race in admissions, even if—as we show—they would risk little by ignoring it. Perhaps they value compliance with the law highly and understand the Supreme Court’s interpretations as the final word. More on this latter point in a moment. We do note, however, that colleges also seem to value affirmative action highly. Before SFFA, no one forced them to practice affirmative action in admissions. Yet many said that they considered race as a factor. And if that is true, many must also have been going through the somewhat onerous steps to obscure their decision processes that we described above. Moreover, when their ability to practice affirmative action was challenged, colleges were willing to spend considerable resources in litigation to defend it. We therefore think there is at least some reason to expect that, given the opportunity, colleges will wish to continue their current practices.

We are also expressly not taking a position on whether colleges should continue practicing race-based affirmative action. Our principal point is descriptive—that they could do it if they wanted to without much fear of legal liability. But “can” does not imply “ought.”

Nonetheless, we sketch below some legal and normative arguments that colleges could use to justify—even if only to themselves—defiance of the holding in SFFA. Each seems at least somewhat plausible to us, but we take no position on their ultimate validity. However, if colleges considered any valid, they could continue practicing affirmative action after SFFA with a clear conscience. Some justifications, we note, are advocated by a surprisingly diverse array of legal scholars. But they are also hotly contested.

First, one could think of noncompliance as a kind of civil disobedience. Under this frame, universities accept that the Supreme Court’s interpretation of Title VI and the Constitution is coextensive with the law. But they believe that the law itself is immoral, sufficiently so to justify breaking it. We will not dwell on this possibility except to point out the differences between notable historical examples of civil disobedience and this one. Leaders like Martin Luther King, Jr. and Mohandas Karamchand Gandhi premised their civil disobedience on the idea that defiance of unjust laws—and the ensuing punishment—would expose their injustice and build a mass movement to overturn them. Here, as we argued above, colleges will likely not be punished for disobeying the law. Moreover, it is difficult to argue that the legal rule being resisted here would be as outrageously immoral as those Gandhi and King resisted. For those reasons, and others, it seems unlikely that defiance of SFFA’s holding would, like celebrated examples of civil disobedience, spur a popular movement to change the law.

Second, and alternatively, colleges might not need to think of their post-SFFA decisions to continue practicing affirmative action as disobedience at all. Under this frame, colleges would regard SFFA’s holding neither as coextensive with the law nor as binding on them. Such a position, paired with the belief that SFFA was wrongly decided, leads to the conclusion that there is nothing legally or normatively wrong with ignoring it.

This kind of anti-judicial-supremacist thinking has lately become fashionable among a surprisingly wide variety of legal thinkers. There are many names for it—and many flavors. Progressive law professors Ryan Doerfler and Samuel Moyn call their approach “democratization.”[189] They advocate, among other things, stripping the Supreme Court’s jurisdiction to protect democratically enacted laws from review.[190] Nikolas Bowie and Daphna Renan (incidentally, faculty members at one of the SFFA defendant universities) question whether Congress has any constitutional duty to listen to the courts.[191] On the other side of the aisle, conservative thinkers like Michael Stokes Paulsen argue that coordinate departments of government—especially the executive—may disregard Supreme Court orders they believe are wrong.[192]

Adherence to even a modest anti-judicial-supremacist theory could give colleges all the self-justification they need to continue affirmative action. Consider William Baude’s version. Baude argues that courts’ interpretations of the law are final but only for the purpose of the judgments they issue.[193] Judgments, in turn, bind only the parties rightly before the court.[194] That is, the parties have a duty to do what the court says with respect to one another, but courts do not have the power to impose new legal rules binding the whole world. Outside the boundary of the court’s judgment, everyone may—and perhaps should—continue to act according to their best understanding of the law.

Baude’s is an especially weak anti-judicial-supremacist theory because of the possibility of repeat litigation. The losing party in the original litigation may—as a formal legal matter—act however it likes as to parties who do not yet have a judgment against it. But if it does, and those new parties sue, the results are highly predictable: stare decisis, vertical precedent, and preclusion rules all but guarantee that the original loser will lose again. Thus, in almost all cases, a court’s legal interpretation—especially the Supreme Court’s—is as good as a universally binding rule. But, as we will discuss, not always.

Suppose, then, that Harvard and UNC think the legal analysis in SFFA was wrong. Suppose furthermore that they acknowledge the judiciary’s power to issue judgments binding the parties to the litigation. But they deny that the judgment power goes any further. That is, the SFFA ruling does not change the fact that, in their view, affirmative action is legal under the Constitution and Title VI. What, then, are their legal obligations? Certainly, they must refrain from using race to decide whether to admit any of the applicants who sued in SFFA.[195] Easy. By now, those students will likely have graduated college or at least be well beyond applying to UNC or Harvard. Even if they applied, the universities could comply with the judgment by treating just those applicants as if they were members of whatever group their admissions practices most favored.

As to all other applicants, present and future, Harvard, UNC, and all other universities could continue practicing affirmative action with a clear conscience. By their lights, they are acting in accordance with their best understanding of the law. And, per Baude-ian anti-judicial-supremacist thinking, no judgment compels them to act according to a worse one.

But what about repeat litigation? Won’t future college applicants bring exactly the same claims as the SFFA plaintiffs and, because of SFFA, win? For all the reasons we laid out above, we think not. If universities simply continue on with their Grutter-derived opaque admissions processes, antidiscrimination claims will be hard to bring at all. The factfinding hurdles required to clear even the pleading stage will usually be insurmountable. Here, unlike in other contexts, the single judgment’s effects will be just that—singular.

Third and finally, colleges might justify post-SFFA affirmative action to themselves on grounds of legal incoherence. This is probably the justification they used for their post-Gratz, pre-SFFA, admissions practices. Gratz said that affirmative action was nominally legal.[196] But it then laid down an impossible and arguably incoherent set of conditions under which it could legally be practiced.[197] Given this, colleges may have decided to simply implement good (by their lights) policies compliant (in their view) with the spirit of the law. And if they also had to obscure those policies for fear of capricious liability, then the fault was with the Court for writing such a bad opinion.

The same logic might well apply after SFFA. As already noted, the Court explicitly blessed the use of admissions criteria highly correlated—and possibly coextensive—with racial or ethnic identity. Again, the Court held that colleges may continue to favor students who, among other things, have overcome racial discrimination. Such features of an applicant are, at a minimum, highly correlated with racial/ethnic background. Indeed, they might be so highly correlated as to be coextensive. There simply might not be any Black applicants to Harvard who, by Harvard’s lights, have not faced and overcome anti-Black racism.

How, then, should the diligent judicial-supremacist Director of Admissions ensure compliance with SFFA among her staff? How could she be sure that they were acting on desiderata merely coextensive with race, rather than on race itself? Perhaps more updates to admissions websites will be in order. Or maybe new language in department handbooks will be required. Should she police admissions staff who slip and use the old, functionally identical criteria? Interrogate them to identify their true internal conceptual schema? Fire them if she suspects they have the wrong one? Or should she instead perhaps conclude that drawing these distinctions is a bit like counting angels on pinheads and ignore them? As with Gratz, she might feel justified in moving forward without much change, declining to record the details of admissions decisions, and placing any applicable blame on the Court for issuing yet another mysterious holding on affirmative action.


Many believe that SFFA spells doom for higher education as we know it and constitutes a major setback in the quest for racial justice. To the contrary, we have argued here that the result in SFFA need have no impact at all: colleges will still be able to operate affirmative action programs as they have been with only very minor changes. The reasons trace back to two separate lines of existing Supreme Court precedent. First, Grutter and Gratz have already taught colleges to leave no trace of actual instances in which race mattered in an admissions decision. Thus, direct evidence of what constitutes discriminatory intent after SFFA is and will remain scarce. Second, holdings like Washington v. Davis and McCleskey v. Kemp foreclose Equal Protection claims based solely on statistical evidence. These doctrines have been a thorn in the side of progressive impact litigators for decades. Ironically, they will now be a curse on conservative opponents of affirmative action. The Court could of course abandon those cases, but doing so would again have cross-cutting consequences. Legal challenges to both campus affirmative action and, for example, racially unequal punishment would become easier to mount.

Our analysis has been descriptive, focusing on the operation of legal doctrine. We have tried to describe what colleges could do after SFFA while holding aside what they should do. Perhaps it would be wrong, even if easy, for colleges to continue using race as a factor in admissions now that the Supreme Court has decided doing so is illegal. But that, too, is uncertain. Both the legal and normative status of Supreme Court holdings has lately been questioned by academics spanning a wide range of interpretive and political ideologies. Thus, as we have described, university leaders holding even modest anti-judicial-supremacist views might find legal or normative justifications for defiance. The question then will be whether such leaders can accept the implications of such views in other contexts where the results cut against their interests.

  1. . Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll. (SFFA), 143 S. Ct. 2141 (2023).
  2. . Id. at 2156 n.2, 2166 (2023).
  3. . See, e.g., Adam Liptak, Supreme Court Seems Ready to Throw Out Race-Based College Admissions, N.Y. Times (Nov. 1, 2022), [] (discussing the conservative justices’ evident skepticism of pro-affirmative action arguments presented in oral argument); Ariane de Vogue & Tierney Sneed, Takeaways from SCOTUS Affirmative Action Cases: Conservatives May Overturn Precedent Allowing Race as a Factor in Admissions, CNN (Oct. 31, 2022, 6:51 PM), [] (“With a 6–3 conservative-liberal majority, the question may be not whether the court will strike down affirmative action, but how far it will rule.”).
  4. . Stephanie Saul, If Affirmative Action Ends, College Admissions May Be Changed Forever, N.Y. Times (Jan. 26, 2023), []; see also, e.g., Aziz Huq, Ending Affirmative Action May Be Just the Beginning, Politico (June 29, 2023, 11:50 AM), [] (speculating that conservative justices might seek to build upon SFFA’s ruling by dismantling disparate-impact laws targeting racial discrimination).
  5. . SFFA, 143 S. Ct. at 2166; Brief for Respondent at 46–51, SFFA, 143 S. Ct. 2141 (2023) (No. 20-1199) [hereinafter Harvard Brief]; Brief by Univ. Respondents at 47, Students for Fair Admissions, Inc. v. Univ. of N.C. No. 21-707 (U.S. July 25, 2022) [hereinafter UNC Brief].
  6. . Brief for Petitioner at 23–24, SFFA, 143 S. Ct. 2141 (2023) (No. 20-1199) [hereinafter SFFA Brief]. The majority opinion recapitulates some of the numbers. See SFFA, 143 S. Ct. at 2171 (charting “Share of Students Admitted to Harvard by Race” in the decade prior to the lawsuit).
  7. . 426 U.S. 229 (1976).
  8. . 481 U.S. 279 (1987).
  9. . Davis, 426 U.S. at 244–45; McCleskey, 481 U.S. at 308–09.
  10. . Davis, 426 U.S. at 239–40; see McCleskey, 481 U.S. at 308–09 (declining to accept statistical evidence of racial prejudice influencing capital sentencing decisions).
  11. . See Alexander v. Sandoval, 532 U.S. 275, 280–82 (2001) (acknowledging that “§ 601 prohibits only intentional discrimination” and that its protections are tied to an equal protection analysis).
  12. . 438 U.S. 265 (1978).
  13. . Id. at 274–75.
  14. . Id. at 276.
  15. . Id. at 298–99, 320.
  16. . Id. at 314–15.
  17. . Id. at 317–18.
  18. . Id. at 319–20.
  19. . 539 U.S. 244 (2003).
  20. . 539 U.S. 306 (2003).
  21. . See Gratz, 539 U.S. at 253–55 (discussing the university’s admissions procedure).
  22. . Id. at 255.
  23. . Id.
  24. . Id.
  25. . Id. at 278.
  26. . Id. at 275 (citing Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 317 (1978)).
  27. . Id. at 293–94 (Souter, J., dissenting).
  28. . Id. at 271 (majority opinion).
  29. . Id. (quoting Metro Broadcasting Inc. v. FCC, 497 U.S. 547, 618 (1990) (O’Connor, J., dissenting)).
  30. . Id. at 272 (citing Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 317 (1978)).
  31. . Id. at 271 (paraphrasing Bakke); accord Bakke, 438 U.S. at 317 (discussing pertinent elements of diversity in light of the particular characteristics of the potential student).
  32. . Id. at 253–54.
  33. . Grutter v. Bollinger, 539 U.S. 306, 337 (2003).
  34. . Id. at 315.
  35. . Id. at 315–16.
  36. . Id. at 343.
  37. . Id. at 336–37.
  38. . Id. at 336 (quoting Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 323 (1978)).
  39. . Fisher v. Univ. of Tex. at Austin (Fisher II), 579 U.S. 365, 375 (2016).
  40. . Gratz, 539 U.S. at 279 (O’Connor, J., concurring).
  41. . Id.
  42. . Id.
  43. . Id. at 277.
  44. . Id. at 273 (majority opinion).
  45. . See Students for Fair Admissions, Inc. v. President & Fellows of Harv. Coll. (SFFA), 143 S. Ct. 2141, 2233, 2242 (Sotomayor, J., dissenting) (citing Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 316–18 (1978) and Fisher v. Univ. of Tex. at Austin (Fisher II), 579 U.S. 365, 385 (2016)) (noting Bakke’s and Fisher II’s recognition of the “constitutionality of limited race-conscious college admissions” when exercised as “one factor of many” in an applicant’s file). One could argue that valuing diversity, rather than race qua race, means assigning diminishing marginal value to membership in a minority group as that group becomes more represented on campus. But this is a fine distinction, indeed. And at default margins, minority status gets high value.
  46. . This sounds harsh. And indeed, we think that Grutter and Gratz are badly written opinions. That does not necessarily mean that they produced bad policy. For example, obscuring exactly when and how race is used in admissions might blunt the perennial critique of affirmative action: that it causes students to treat non-white peers as if they did not “deserve” admission. See, e.g., Ariane de Vogue, ‘Silent’ Justice Outspoken on Affirmative Action, ABC News (Sept. 28, 2007, 3:41 PM), [] (discussing Justice Thomas’s experiences with affirmative action and its effect on his credibility as a lawyer). If no one knows how or when race is used in admissions, no one can credibly claim who would not have been admitted but for their race. On the other hand, one can imagine this obscurity fostering the opposite inference—a default presumption that race mattered to all non-white students’ admissions decisions. The net effect is an empirical question on which we take no stance.
  47. . Harvard Brief, supra note 5, at 5–6; UNC Brief, supra note 5, at 4–7.
  48. . See Litigation History, Harvard Admissions Lawsuit,
    admissionscase/lawsuit/ [] (describing Harvard’s use of race in its admissions process); James W. Dean, Jr., The Educational Benefits of Diversity and Inclusion for Undergraduate Students at the University of North Carolina at Chapel Hill 9 (2017), [] (same vis-à-vis UNC).
  49. . UNC Brief, supra note 5, at 8–9; see Harvard Brief, supra note 5, at 7 (asserting that applicants receive ratings “in four areas: academic, extracurricular, athletic, and personal”).
  50. . Harvard Brief, supra note 5, at 6–10; UNC Brief, supra note 5 at 8–10.
  51. . Harvard Brief, supra note 5, at 7.
  52. . Id. at 8.
  53. . Id.
  54. . Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll. (SFFA), 143 S. Ct. 2141, 2154 (2023).
  55. . Harvard Brief, supra note 5, at 8–9.
  56. . Id.
  57. . Id. at 9.
  58. . Id.
  59. . Id.
  60. . Id.
  61. . Id.
  62. . Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll. (SFFA), 143 S. Ct. 2141, 2155 (2023).
  63. . Harvard Brief, supra note 5, at 9.
  64. . Id. at 9–10.
  65. . SFFA, 143 S. Ct. at 2155.
  66. . Id.
  67. . Harvard Brief, supra note 5, at 10.
  68. . SFFA, 143 S. Ct. at 2242, 2252 (Sotomayor, J., dissenting).
  69. . Id. at 2169 (majority opinion).
  70. . See id. at 2242, 2252 (Sotomayor, J., dissenting) (noting the lack of evidence—even after “extensive discovery and [a] lengthy trial[]”—of race’s precise effect on a Harvard applicant’s outcome).
  71. . UNC Brief, supra note 5, at 9.
  72. . Id.
  73. . See id. at 9–11 (taking great pains to paint its use of race as merely a component of its holistic review process).
  74. . Id. at 10–11.
  75. . Id.
  76. . SFFA, 143 S. Ct. at 2252 (Sotomayor, J., dissenting).
  77. . UNC Brief, supra note 5, at 11.
  78. . See id. (“Her story, the former head of admissions testified, ‘reveals sometimes how hard it is to separate race out from other things that [the office] know[s] about a student.’”) (alterations in original).
  79. . SFFA, 143 S. Ct. at 2252 (Sotomayor, J., dissenting).
  80. . Id. at 2166 (majority opinion).
  81. . Id. at 2169.
  82. . Id. at 2239 (Sotomayor, J., dissenting) (quoting id. at 2207 (Thomas, J., concurring)).
  83. . See sources cited supra note 4.
  84. . Litigation History, Harvard Admissions Lawsuit,
    admissionscase/lawsuit/ [] (describing Harvard’s use of race in its admissions process); Common Data Set 2021–22, Univ. of N.C. 8 (last revised August 25, 2023), [https://] (indicating race as a considered factor in admissions decisions).
  85. . SFFA, 143 S. Ct. at 2176.
  86. . See id. (“[N]othing in this opinion should be construed as prohibiting universities from considering an applicant’s discussion of how race affected his or her life, be it through discrimination, inspiration, or otherwise.”); Harvard Brief, supra note 5, at 6 (describing pre-SFFA admissions practices); UNC Brief, supra note 5, at 2 (same).
  87. . We were able to find just three such cases, aside from the SFFA litigations discussed extensively herein. They are: Students for Fair Admissions, Inc. v. Univ. of Tex. at Austin, 37 F.4th 1078 (5th Cir. 2022); Fisher v. Univ. of Tex. at Austin, 579 U.S. 365 (2016); and Smith v. Univ. of Wash., 392 F.3d 367 (9th Cir. 2004).
  88. . The Supreme Court’s SFFA opinions are the only ones representing merits wins for plaintiffs. The plaintiffs lost in Fisher and Smith. Fisher, 579 U.S. at 388 (upholding the validity of the University of Texas at Austin’s admissions policy); Smith 392 F.3d at 382 (upholding the university’s consideration of race). And in Students for Fair Admissions, Inc. v. Univ. of Texas, the plaintiffs won a procedural victory just before the Supreme Court’s decision in the other SFFA cases. Students for Fair Admissions, Inc. v. Univ. of Tex. at Austin, 37 F.4th at 1089 (holding that the lower court erred by deciding that the lawsuit was barred by claim preclusion). The Texas case has presumably now been superseded by the Court’s new rules.
  89. . See Harvard Brief, supra note 5, at 5 (discussing Harvard’s consideration of race); UNC Brief, supra note 5, at 2 (mentioning the same practice at UNC).
  90. . SFFA, 143 S. Ct. 2141, 2252 (2023) (Sotomayor, J., dissenting).
  91. . Id. at 2176 (majority opinion).
  92. . Id. (quoting Cummings v. Mo., 71 U.S. (4 Wall.) 277, 325 (1866)).
  93. . See, e.g., SFFA Brief, supra note 6, at 23–24 (charting both a portion of the racial makeup of Harvard’s recent entering classes and the proportion of various racial groups who were admitted by “academic decile”).
  94. . See SFFA, 143 S. Ct. at 2171 (reproducing one of the charts described above).
  95. . See supra note 93 and accompanying text.
  96. . SFFA Brief, supra note 6, at 16, 31.
  97. . Id.
  98. . See SFFA, 143 S. Ct. at 2171 (noting Harvard’s “numerical commitment” to allocating a given proportion of an entering class to various racial groups).
  99. . 42 U.S.C. § 2000e-2(k)(1). See Griggs v. Duke Power Co., 401 U.S. 424, 432 (1971) (“Congress directed the thrust of the [Civil Rights] Act to the consequences of employment practices, not simply the motivation.”).
  100. . Int’l Bhd. of Teamsters v. U.S., 431 U.S. 324, 337–38 (1977).
  101. . McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802 (1973).
  102. . Griggs, 401 U.S. at 431, 436.
  103. . 426 U.S. 229, 235 (1976).
  104. . Id. at 233, 235.
  105. . Id. at 237.
  106. . Id.
  107. . Id. at 238–39.
  108. . See id. at 241 (discussing the required showing of evidence to demonstrate a constitutional violation).
  109. . Id. at 245–46.
  110. . 481 U.S. 279 (1987).
  111. . Id. at 283, 291.
  112. . Id. at 286; see id. at 287 (“According to [petitioner’s evidence], black defendants were 1.1 times as likely to receive a death sentence as other defendants.”). But see id. at 286 (“The raw numbers . . . indicate a reverse racial disparity according to the race of the defendant: 4% of the black defendants received the death penalty, as opposed to 7% of the white defendants.”).
  113. . See id. at 293–95, 297 (distinguishing the petitioner’s claims from the Title VII and venire-selection contexts, where the Court has “accepted statistical disparities as proof of an equal protection violation,” and finding his evidence “clearly insufficient”).
  114. . Id. at 292 (quoting Whitus v. Ga., 385 U.S. 545, 550 (1967)).
  115. . Id. at 297.
  116. . Id.
  117. . Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll. (SFFA), 143 S. Ct. 2141, 2156 n.2 (2023).
  118. . See Alexander v. Sandoval, 532 U.S. 275, 280–81 (2001) (“Title VI itself directly reaches only instances of intentional discrimination” (quoting Alexander v. Choate, 469 U.S. 287, 293 (1985))).
  119. . Id. at 286 n.6.
  120. . Id. at 290.
  121. . See id. at 289 (reasoning that § 602 does not reflect “an intent to create a private remedy”).
  122. . Consider, for example, the selection criterion of “immigrant,” discussed below at notes 184–187.
  123. . 556 U.S. 662 (2009).
  124. . 550 U.S. 544 (2007).
  125. . Iqbal, 556 U.S. at 682.
  126. . Id.
  127. . Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll. (SFFA), 143 S. Ct. 2141, 2176; see infra Part III.
  128. . Anemona Hartocollis, Colleges Will Be Able to Hide a Student’s Race on Admissions Applications, N.Y. Times (May 26, 2023), [].
  129. . SFFA, 143 S. Ct. at 2186 (Thomas, J., concurring) (citing City of Richmond v. J.A. Croson Co., 488 U.S. 469, 505 (1989)).
  130. . Peter N. Salib, Big Data Affirmative Action, 117 Nw. U. L. Rev. 821, 827 (2022).
  131. . SFFA, 143 S. Ct. at 2176.
  132. . Supreme Court Decision, Harvard Admissions Lawsuit,
    admissionscase/2023/06/29/supreme-court-decision/ [].
  133. . SFFA, 143 S. Ct. at 2176.
  134. . Id.
  135. . Id. at 2251 (Sotomayor, J., dissenting).
  136. . Mich. Const., art. I, § 26; Cal. Const. art. I, § 31(a).
  137. . Brief for the President and Chancellors of the University of California as Amici Curiae Supporting Respondents at 4, Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll. (SFFA), 143 S. Ct. 2141 (2023) (No. 20-1199) [hereinafter UC Brief]; Brief for the University of Michigan as Amicus Curiae in Support of Respondents at 3, Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll. (SFFA), 143 S. Ct. 2141 (2023) (No. 20-1199) [hereinafter UM Brief].
  138. . UC Brief, supra note 137, at 11. The timing of California’s ban on affirmative action may occlude its instructiveness on the effects of SFFA. California’s ban took effect before Grutter and Gratz, and so it may be the case that California’s ban essentially imposed rules against affirmative action akin to those from Grutter and Gratz. Whether such rules had significant impact on the ability to operate an affirmative action program is, as we discuss below, under question—but regardless it is a distinct inquiry from the impact of SFFA on the backdrop of a post-Grutter world.
  139. . Freshman Fall Admissions Summary, Univ. of Cal. (February 27, 2023) [hereinafter UC Admissions Summary],
    freshman-admissions-summary []. We calculated these figures by subtracting the number of students identified as “White,” “International,” and “Other/Unknown” from the total. But this is a conservative estimate, as some share of “International” and “Other/Unknown” students are also non-white. This is the most complete source of data that UC cites. Note, however, that demographic shares are reported to the nearest percentage point. Thus, there is some minimal room for error in characterizing them. However, there is no reason to think that rounding up was any more prevalent than rounding down, so any errors seem likely to cancel one another out, on average.
  140. . Id.
  141. . Id.
  142. . Id.
  143. . Id.
  144. . Id.
  145. . Id.
  146. . Id.
  147. . See UC Brief, supra note 137, at 18–19 (noting that “freshman admissions rates for African American applicants remain well below 1995 levels” even though “admission of students from underrepresented minority groups as a whole [has] increased”).
  148. . Id. at 12.
  149. . Id. at 23.
  150. . UC Admissions Summary, supra note 139.
  151. . Id.
  152. . Id.
  153. . Id.
  154. . UM Brief, supra note 137, at 21.
  155. . Id. at 22.
  156. . Ethnicity Reports, Univ. of Mich., [] (choose “Fall” in “Terms” dropdown, then choose “Freshman” in “Academic Level” dropdown, then choose “Black” in “Ethnicity URM” dropdown). It is worth noting, though, that the 2005 numbers were about six-tenths of a point higher than the 2008 numbers. Id.
  157. . UM Brief, supra note 137, at 21 n.39.
  158. . Id. See also Ethnicity Reports, supra note 156 (the “Two or More” racial category can be found under the “Ethnicity URM” dropdown). The new classification also added the category of “Hawaiian”; reviewing the reports shows that category had a small numerical effect in the University of Michigan’s admissions. See id. (choose “Fall” in “Terms” dropdown, then choose “Freshman” in “Academic Level” dropdown, then choose “Hawaiian” in “Ethnicity URM” dropdown).
  159. . Id. (choose “Fall” in “Terms” dropdown, then choose “Freshman” in Academic Level dropdown, then first choose “Black” in Ethnicity URM dropdown, and second choose “Two or More” and “Two or More URM” in “Ethnicity URM” dropdown).
  160. . Id. (choose “Fall” in “Terms” dropdown, then choose “Freshman” in “Academic Level” dropdown, then choose “Black” in “Ethnicity URM” dropdown).
  161. . Id.
  162. . The pre-ban slope was −0.563 percentage points/year (this regression mean was calculated with the admissions numbers pulled from Michigan’s admissions website). Over 15 years, that would amount to an 8.44-point reduction in Black representation. In 2006, the Black share of freshmen was 7.06%. Id. Thus, the naïve expected Black share of the 2021 freshman class would have been −1.38%.
  163. . Cf. UM Brief, supra note 137, at 21–22 (decrying that UM’s campus-wide enrollment of underrepresented minorities has remained “essentially flat” since 2006).
  164. . See Ethnicity Reports, supra note 156 (choose “Fall” in Terms dropdown; then choose “Freshman” in “Academic Level” dropdown; then choose “Hispanic” in “Ethnicity URM” dropdown—showing that in 2006, the last pre-ban year, Hispanic students represented 5.38% of UM’s admitted class, but that in 2022, almost two decades after UM stopped affirmative action policies, Hispanic students represented 13.4% of UM’s admitted class); UC Admissions Summary, supra note 139 (representing that in 1995, the last year affirmative action was allowed in California, Hispanic students represented 15% of UC’s admitted class, but in 2022, almost three decades after UC stopped affirmative action policies, Hispanic students represented 27% of UC’s admitted class).
  165. . Note that at least one statistical analysis has suggested a larger effect of affirmative action on minority representation in California colleges. See Zachary Bleemer, Affirmative Action and Its Race-Neutral Alternatives, J. Pub. Econ., Apr. 2023, at 1, 6 (discussing the increase in admissions advantages of underrepresented minority applicants at Berkeley and UCLA). But that study limits its analysis of affirmative action to the years 1995–2000, ending well before the recovery in Black representation at the UC schools described above. Id. at 4.
  166. . UC Brief, supra note 137, at 18; UM Brief, supra note 137, at 3.
  167. . Mich. Const., art. I, § 26(6); (“The remedies available for violations of this section shall be the same . . . as are otherwise available for violations of [the state]’s anti-discrimination law.”); Cal. Const. art. I, § 31(g) (same). Note that the availability of the disparate-impact approach is, strictly speaking, a question of proof, not of remedy. So even if Michigan and California generally allow disparate impact proof for discrimination claims, it is still not clear whether their anti-affirmative-action provisions would likewise allow it.
  168. . See UC Brief, supra note 137, at 13–20 (describing race-neutral alternatives); UM Brief, supra note 137, at 12–21 (same).
  169. . See infra notes 175–177 and accompanying text. See also Frank J. Scaturro, The Unjustly Forgotten Legacy of Byron White, Nat’l Rev.: L. and the Courts (March 30, 2017, 5:00 PM), [] (praising Justice Byron White’s decision in Davis on the grounds that “[u]nequal racial outcomes alone did not amount to intentional discrimination prohibited by the Constitution”).
  170. . See, e.g., Khiara M. Bridges, Race in the Roberts Court, 136 Harv. L. Rev. 23, 29–30 (2022) (lamenting that, after Davis, “the judiciary approaches laws that are intended to dismantle the nation’s racial hierarchy with the same degree of constitutional skepticism as it does laws that are designed to reproduce and protect it”). See also infra notes 189–191 and accompanying text; David A. Strauss, Discriminatory Intent and the Taming of Brown, 56 U. Chi. L. Rev. 935, 954 (1989) (discussing Washington v. Davis and its impact on the applicability of Brown v. Board of Education); Mario L. Barnes, “The More Things Change . . .”: New Moves for Legitimizing Racial Discrimination in a “Post-Race” World, 100 Minn. L. Rev. 2043, 2077 (2016) (discussing Davis and McCleskey and how the Court’s decisions in those cases refused to address discrimination).
  171. . 446 U.S. 55 (1980).
  172. . Id. at 58–60 (plurality opinion).
  173. . Id. at 71.
  174. . Id. at 72–74.
  175. . Id. at 72–73.
  176. . See id. at 73–74 (asserting that “no official obstacles” prevented Black citizens from running for office and that white officials’ discrimination provided only “tenuous and circumstantial evidence of the constitutional invalidity of the electoral system”).
  177. . See id. at 73–74, 76 (rejecting four categories of disparate-impact evidence identified by the lower courts as insufficient).
  178. . Id. at 104–05 (Marshall, J., dissenting).
  179. . Id.
  180. . Id. at 106–08.
  181. . Id. at 137 (quoting Townsend v. Wathen, 103 Eng. Rep. 579, 580–81 (K.B. 1808)).
  182. . Id.
  183. . We recognize that this is an optimistic view, to say the least.
  184. . Complaint at 3, Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll., 397 F. Supp. 3d 126 (D. Mass. 2019) (No. 1:14-cv-14176-DJC).
  185. . Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll. (SFFA), 143 S. Ct. 2141, 2176 (2023).
  186. . Abby Budiman, Key Findings About U.S. Immigrants, Pew Rsch. Ctr. (Aug. 20, 2020), [].
  187. . James R. Rogers, Why Scalia’s Originalism Trumps “Original Intent, L. & Liberty (Jan. 10, 2018), [https://]
  188. . See, e.g., UNC Brief, supra note 5, at 14–15 (describing SFFA’s “attempt to disparage UNC” based on a similar scenario).
  189. . See Ryan D. Doerfler & Samuel Moyn, Democratizing the Supreme Court, 109 Cal. L. Rev. 1703, 1726 (2021) (arguing that at least some forms of jurisdiction stripping would be “democratizing”).
  190. . Id. at 1736.
  191. . Nikolas Bowie & Daphna Renan, The Supreme Court Is Not Supposed to Have This Much Power, The Atlantic (June 8, 2022),
    supreme-court-power-overrule-congress/661212/ [].
  192. . See Michael Stokes Paulsen, The Merryman Power and the Dilemma of Autonomous Executive Branch Interpretation, 15 Cardozo L. Rev. 81, 83–84 (1993) (describing tension between federal government branches).
  193. . William Baude, The Judgment Power, 96 Geo. L.J. 1807, 1809, 1811 (2008).
  194. . Id. at 1810.
  195. . We speak here of students as the parties who could claim the protection of the judgment. The plaintiff, Students for Fair Admissions, is an organization. But in this case, the organization claims to assert only the derivative rights of certain current student members. Andrew Hessick, Associational Standing in the Affirmative Action Cases, Yale J. on Regul.: Notice & Comment (Aug. 26, 2022), [].
  196. . See Gratz v. Bollinger, 539 U.S. 244, 270–71 (2003) (opining that though race could be used as a plus factor, it would be important to consider “each characteristic of a particular applicant”).
  197. . See id. at 247 (recognizing that “the implementation of a [constitutionally permissible] program . . . might present administrative challenges”).