Harvard Law Review Harvard Law Review Harvard Law Review

Discrimination

Discriminatory Taint

The truism that history matters can hide complexities. Consider the idea of problematic policy lineages. When may we call a policy the progeny of an earlier, discriminatory policy, especially if the policies diverge in design and designer? Does such a relationship condemn the later policy for all times and purposes, or can a later decisionmaker escape the past? It is an old problem, but its resolution hardly seems impending. Just recently, Supreme Court cases have confronted this fact pattern across subject matters as diverse as entry restrictions, nonunanimous juries, and redistricting, among others. Majority opinions seem unsure whether or why “discriminatory predecessors” matter, and individual Justices who agree that they do squabble over methodology.
One could answer these questions by banishing them. Thus, some would simply treat any nonidentical policy predecessor as minimally relevant, and only relevant insofar as it suggests present-day bad intent. Anything else, they suggest, risks an unmoored original sin jurisprudence, with courts claiming to know guilt when they see it. Simple is not always better, however, especially if it risks eliding information material to a policy’s validity. But again: how do we divine materiality?

Better approaches are possible. While our law broadly appreciates that continuity matters to legal meaning and responsibility, constitutional law has undertheorized it. Deploying continuity here helps conceptualize, and craft guideposts for, “discriminatory taint”: an objectively ascertainable relationship between an earlier policy and a later, similar policy. Thus defined, taint can impugn some policies that might otherwise have passed constitutional muster. Yet it also facilitates realistic approaches — judicial and nonjudicial — to distinguishing genuine purging of taint from its laundering. And it supplements debates on the nature of wrongful discrimination by underscoring how continuity can help identify persistent constitutional problems even absent subjective bad intent.

Introduction

If the federal government implements a facially neutral, country-targeted travel ban, does it matter that it closely follows, and invokes the same justifications, as two similar bans, both enjoined on Establishment Clause grounds?1×1. See, e.g., Joseph Landau, Process Scrutiny: Motivational Inquiry and Constitutional Rights, 119 Colum. L. Rev. 2147, 2173–75, 2177 n.164 (2019) (discussing litigation leading up to Trump v. Hawaii, 138 S. Ct. 2392 (2018)). What if Louisiana invokes neutral reasons for reaffirming a law permitting nonunanimous jury verdicts, seventy-six years after adopting it for racist reasons?2×2. See Ramos v. Louisiana, 140 S. Ct. 1390, 1394, 1401 n.44 (2020); id. at 1426 (Alito, J., dissenting). Or if Texas’s redistricting maps draw partly on older maps that a court concluded were unconstitutional?3×3. See Abbott v. Perez, 138 S. Ct. 2305, 2313, 2316–18, 2325–26 (2018). Or if a city closes pools, ostensibly for safety and cost reasons, but only after a court enjoined their segregated operation?4×4. See Palmer v. Thompson, 403 U.S. 217, 218–19 (1971). Assuming it does matter, how much, and for how long?5×5. See, e.g., id. at 230 (Blackmun, J., concurring) (raising fear that impeaching contemporary facially neutral action with past actions could unjustly “lock[] in” that policy choice indefinitely, notwithstanding later-manifesting legitimate reasons).

Judging from the majority and separate opinions in these (and other6×6. See infra section I.C, pp. 1203–11 (discussing, inter alia, Espinoza v. Mont. Dep’t of Revenue, 140 S. Ct. 2246 (2020); and Flowers v. Mississippi, 139 S. Ct. 2228 (2019)). ) cases, the Supreme Court’s answer resembles a shrug. Of course, as conventional antidiscrimination doctrine recognizes, the past can matter.7×7. See Village of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 267 (1977) (directing equal protection attention to “[t]he historical background of the decision” and “[t]he specific sequence of events leading up to the challenged decision”); see also McCreary County v. ACLU of Ky., 545 U.S. 844, 862 (2005) (same principle for the Establishment Clause); Church of the Lukumi Babalu Aye, Inc. v. City of Hialeah, 508 U.S. 520, 540 (1993) (citing Arlington Heights, 429 U.S. at 266) (same for the Free Exercise Clause). Throughout the federal judiciary, however, that general recognition has not generated consistency in cases posing what we might call the “discriminatory predecessor” problem, i.e., where an older policy can credibly be called the progenitor of a contemporary policy actually under review.8×8. See infra section I.C, pp. 1203–11. This is distinct from whether a past violation’s lingering effects preclude the termination of continuing injunctive relief connected to that violation. See Wendy Parker, The Future of School Desegregation, 94 Nw. U. L. Rev. 1157, 1163–64, 1164 n.61 (2000) (discussing vestiges).
These diverse recurrences reveal practical importance. The varied approaches9×9. See infra section I.C, pp. 1203–11. reveal theoretical importance. Despite deploying the mens rea–inflected language of “intent,”10×10. See, e.g., Aziz Z. Huq, What Is Discriminatory Intent?, 103 Cornell L. Rev. 1211, 1214–24 (2018). courts in constitutional antidiscrimination cases necessarily infer wrongful discrimination from the circumstances. We seem to intuit that discriminatory predecessors might be unusually relevant — but past that intuition, we splinter.
The theoretical muddle may stem from the tricky questions relating to just how powerful evidence of discriminatory predecessors should be. If the key is the permissibility of contemporary action at Time 2 (T2), surely some earlier (Time 1/T1) action “cannot, in the manner of original sin, condemn government action that is not itself unlawful.”11×11. Abbott v. Perez, 138 S. Ct. 2305, 2324 (2018) (emphasis added) (quoting City of Mobile v. Bolden, 446 U.S. 55, 74 (1980) (plurality opinion)). Or take the very idea of a “predecessor.” Naturally, “the world is not made brand new every morning,”12×12. McCreary County, 545 U.S. at 866. and some past events connect causally to some present events. But, formally, new policies are just that: new. As such, they may have new facially legitimate justifications, or deviate in form from their ostensible predecessors. Indeed, at least sometimes defendants might challenge the predecessors’ wrongfulness and insist that this contestability renders their character uninformative to the present substantive question.13×13. See infra section V.A.2, pp. 1261–64. Safer, perhaps, to say only that history broadly matters as context that might aid an all-things-considered analysis of T2 policies, which must stand and fall on their own merits.14×14. See, e.g., Rogers v. Lodge, 458 U.S. 613, 625 (1982); Eric Schnapper, Perpetuation of Past Discrimination, 96 Harv. L. Rev. 828, 829 (1983).
Greater illumination is possible. Start with pretext. Those who disagree on the frequency of wrongful discrimination should nevertheless agree that wrongdoers should not be able to continue acting with unreconstructed unconstitutional aims. But policy change over time can cloak precisely that behavior, especially when we often must infer a policy’s illegitimacy from a context that bad faith actors can manipulate. We can thus gain much from better understanding temporal pretext. This Article accordingly describes a type of relationship distinguished by the formal and functional continuity of temporally separated policies. At minimum, this contribution richens our account of how change over time affects pretext-based claims.15×15. Thus, the “genetic fallacy” — the “alleged mistake of arguing that something is to be rejected because of its suspicious origins” — is inapposite here. Genetic Fallacy, The Oxford Dictionary of Phil. (Simon Blackburn ed., 2d rev. ed. 2008) (emphasis added), https://www.oxfordreference.com/view/10.1093/acref/9780199541430.001.0001/acref-9780199541430-e-1360 [https://perma.cc/2297-CPR4]. After all, “[f]requently such reasoning is, actually, quite appropriate,” id., and often especially in the legal sphere, see Charles L. Barzun, The Genetic Fallacy and a Living Constitution, 34 Const. Comment. 429, 433 (2019).

But taking constitutional continuity seriously can tell us more. Once comprehendible, these relationships do more than just suggest pretext. The persistence of an older policy’s operative core can manifest a “discriminatory taint” that alone should impugn an otherwise facially legitimate policy. Understanding what taint is and how to find it advances antidiscrimination discourse in multiple areas. It can help courts navigate what taint means for adjudication (for example, whether and when they should deem taint purged and how to proceed when it is not), guide nonjudicial decisionmaking (for example, choosing how to act and justifying said choices), and inform scholarly understanding of wrongful discrimination (both over time and as a generalizable phenomenon).

In pursuit of the foregoing, this Article makes three contributions.

First, this Article opened with diverse examples to emphasize the wide-ranging need for careful thought about the relevance of time and change in this context. To be sure, the broad idea that problematic history could affect present-day analysis is not new.16×16. See, e.g., Charles L. Black, Jr., The Lawfulness of the Segregation Decisions, 69 Yale L.J. 421, 424 (1960) (describing segregation’s “apostolic succession from slavery”); Elise C. Boddie, Adaptive Discrimination, 94 N.C. L. Rev. 1235, 1285 (2016); Owen M. Fiss, Groups and the Equal Protection Clause, 5 Phil. & Pub. Affs. 107, 144–45 (1976) (discussing the possibility of heightened scrutiny for facially neutral state action that perpetuates the effects of past wrongful discrimination); Daryl J. Levinson, Framing Transactions in Constitutional Law, 111 Yale L.J. 1311, 1313–14 (2002); Melissa Murray, Race-ing Roe: Reproductive Justice, Racial Justice, and the Battle for Roe v. Wade, 134 Harv. L. Rev. 2025, 2062, 2086 (2021); Schnapper, supra note 14, at 829–31 (discussing a variety of types of “perpetuation of racial discrimination,” id. at 829); Reva B. Siegel, “The Rule of Love”: Wife Beating as Prerogative and Privacy, 105 Yale L.J. 2117, 2178–80 (1996) (discussing “continuity” between legal regimes that “enforce[] social stratification,” id. at 2178); David A. Super, Temporal Equal Protection, 98 N.C. L. Rev. 59, 61–64 (2019). Nor are more specific considerations of a discrete past policy’s possible relevance to an arguable descendant’s constitutional meaning.17×17. See, e.g., Jessica A. Clarke, Explicit Bias, 113 Nw. U. L. Rev. 505, 560–71 (2018) (discussing taint as a remedial complication of recognizing explicit bias); Justin Driver, Constitutional Outliers, 81 U. Chi. L. Rev. 929, 943 (2014) (describing “backup[s],” “replacement state [policies] designed to preserve as much as possible of a legal model that either has recently been invalidated or seems certain to be invalidated in the near future”). In the October 2019 Term alone, dueling opinions in a redistricting case invoked “taint” explicitly,18×18. Compare Abbott v. Perez, 138 S. Ct. 2305, 2324–25 (2018), with id. at 2346–47, 2352 (Sotomayor, J., dissenting). and multiple other cases grappled with the concept under other names.19×19. See generally Ramos v. Louisiana, 140 S. Ct. 1390 (2020). Compare Espinoza v. Mont. Dep’t of Revenue, 140 S. Ct. 2246, 2267–74 (2020) (Alito, J., concurring), with id. at 2293 n.2 (Sotomayor, J., dissenting). Commentators, too, have noted the potential thorny issues the T1/T2 pattern can raise, but almost invariably in passing.20×20. See, e.g., John Hart Ely, Democracy and Distrust 141 & n.26 (1980); Micah Schwartzman, Official Intentions and Political Legitimacy: The Case of the Travel Ban, in NOMOS LXI: Political Legitimacy 201, 219–23 (Jack Knight & Melissa Schwartzberg eds., 2019); Gabriel J. Chin, Rehabilitating Unconstitutional Statutes: An Analysis of Cotton v. Fordice, 71 U. Cin. L. Rev. 421 (2002); Clarke, supra note 17, at 560–71; Kim Forde-Mazrui, The Constitutional Implications of Race-Neutral Affirmative Action, 88 Geo. L.J. 2331, 2389–95 (2000) (raising the specter of readoption of “suspect policies for nonsuspect reasons,” id. at 2389); Brandon L. Garrett, Unconstitutionally Illegitimate Discrimination, 104 Va. L. Rev. 1471, 1473–74 (2018) (noting possible relevance of taint to policy reenactment); Paul Gowder, Equal Law in an Unequal World, 99 Iowa L. Rev. 1021, 1045 n.82 (2014); Richard L. Hasen, The Supreme Court’s Pro-partisanship Turn, 109 Geo. L.J. Online 50, 67 & n.95 (2020); Leslie Kendrick & Micah Schwartzman, The Supreme Court, 2017 Term — Comment: The Etiquette of Animus, 132 Harv. L. Rev. 133, 149–50 (2018) (briefly discussing “[f]utility and [t]aint,” id. at 149); Richard C. Schragger, Of Crosses and Confederate Monuments: A Theory of Unconstitutional Government Speech, 63 Ariz. L. Rev. 45, 49 n.21, 64 (2021); Toby J. Heytens, Note, School Choice and State Constitutions, 86 Va. L. Rev. 117, 147, 149 (2000) (noting the possibility of “taint of original invidious intent” but finding it unclear whether and how it might be “purge[d]” from a policy reenactment, id. at 149); see also sources cited supra note 17. Professor Gabriel Chin’s project is closest to mine, but he deals with a single case where the temporal relationship could not seriously be disputed, which obviated the need to delve into harder cases of relation. See Chin, supra, at 423. Our proposed solutions also diverge. A forthcoming project from Rebecca Aviel will also grapple with many of the questions of intertemporal discrimination this Article raises. See Rebecca Aviel, Second-Bite Lawmaking, 100 N.C. L. Rev. (forthcoming 2022) (on file with the Harvard Law School Library).

This Article probes deeper. The passing commentary has spent little time on details like whether taint requires perfect identity, and, if not, how to determine which differences preclude it. Courts forced to address the issue without reliable theoretical frameworks have done so in inconsistent and undeveloped ways.21×21. See infra section I.C, pp. 1203–11. This Article aims to flesh out the taint concept as a detectable type of relationship between an earlier policy and a later policy. I accordingly offer a framework that looks first to the earlier policy’s operation and next to markers of material continuity in the subsequent policy. In so doing, I draw on and build upon diverse areas in which continuity over time shapes assignments of legal responsibility, including mootness law and criminal law,22×22. See infra Part II, pp. 1212–27. constructing my proposal for constitutional responsibility as a species of institutional temporal realism.23×23. See, e.g., Richard H. Pildes, Institutional Formalism and Realism in Constitutional and Public Law, 2013 Sup. Ct. Rev. 1, 2. A tainted relationship embodies a historical fact that is uniquely relevant to any T2 inquiry — so relevant, in fact, that it justifies treating the T2 policy with increased skepticism.

Delineating taint’s contours generates the Article’s second contribution. Taint can aid searches for present-day bad intent, but its implications for longstanding debates over how we detect wrongful discrimination24×24. See, e.g., Benjamin Eidelson, Discrimination and Disrespect 1–7, 74 (2015); Deborah Hellman, When Is Discrimination Wrong? (2008); Richard H. Fallon, Jr., Constitutionally Forbidden Legislative Intent, 130 Harv. L. Rev. 523, 525–34 (2016); Schwartzman, supra note 20, at 203–20. go further. The Supreme Court’s antidiscrimination doctrine is widely understood as requiring specific, subjective intent to harm because of a protected trait.25×25. See, e.g., Paul Brest et al., Processes of Constitutional Decisionmaking: Cases and Materials 1097 (7th ed. 2018); Jamal Greene, The Supreme Court, 2017 Term — Foreword: Rights as Trumps?, 132 Harv. L. Rev. 28, 43 (2018); Ian Haney-López, Intentional Blindness, 87 N.Y.U. L. Rev. 1779, 1833, 1837–38 (2012); Charles R. Lawrence III, The Id, the Ego, and Equal Protection: Reckoning with Unconscious Racism, 39 Stan. L. Rev. 317, 318, 324 (1987); Reva B. Siegel, Blind Justice: Why the Court Refused to Accept Statistical Evidence of Discriminatory Purpose in McCleskey v. Kemp — And Some Pathways for Change, 112 Nw. U. L. Rev. 1269, 1279 (2018). But see, e.g., William D. Araiza, Animus 91 & n.2 (2017) (contesting this doctrinal view and noting academic disagreement); infra section III.D, pp. 1234–35. Whatever might be said about that approach generally, taint accentuates its weaknesses. As described in this Article, taint can coexist with genuinely pure-hearted T2 decisionmakers.26×26. See, e.g., David E. Pozen, Constitutional Bad Faith, 129 Harv. L. Rev. 885, 934–36, 945–47 (2016); David A. Strauss, Discriminatory Intent and the Taming of Brown, 56 U. Chi. L. Rev. 935, 960–61 (1989). The same facts that concretize the taint concept, however, also indicate that more matters than pure-heartedness when evaluating a tainted T2. That is, properly understood, taint describes a scenario where a decision rule focused on specific intent is uniquely inapt. In making this point, the Article offers a novel, targeted critique of antidiscrimination decision rules focused on decisionmaker “intent” and engages recent literature on the necessary prerequisites to wrongful discrimination.27×27. See, e.g., Huq, supra note 10, at 1218, 1223–24; sources cited supra note 24.

Finally, the Article addresses implementation. Taint is a fact about a T2 policy that triggers a taint-sensitive way of evaluating validity, not an independently sufficient demonstration of invalidity. That understanding of taint as a trigger facilitates a nuanced approach that both prevents the perpetuation of wrongful discrimination and enables taint’s genuine purging, rather than its laundering. I argue that this can aid both judicial and nonjudicial actors.

Courts that find taint should focus on whether the state can re-earn whatever clean-slate treatment the T2 policy would otherwise have received. Key here is a targeted disinterring of constitutional disparate impact, notwithstanding usual judicial skepticism.28×28. See, e.g., Washington v. Davis, 426 U.S. 229, 239–48 (1976); Lawrence, supra note 25, at 318–21. On this two-pronged approach, courts first ask whether the state can show that the contemporary policy has eliminated any meaningful disparate impact. Second — if the state cannot so show — it must make a heightened showing of why it cannot eliminate the disparate impact and why the legitimate need for this means of pursuing a non-discriminatory government interest outweighs the harm of shielding the disparate impact of a tainted rule. Although I offer a unique approach to that heightened showing, I draw on analogies to other examples of heightened scrutiny to show that courts are well equipped to conduct such analyses.

Moreover, taint offers utility beyond adjudication. Many limits on judicial action (for example, institutional role, judicial procedure, and the countermajoritarian difficulty)29×29. See Alexander M. Bickel, The Least Dangerous Branch 16, 25–26 (1962); see also Michael W. McConnell, The Supreme Court, 1996 TermComment: Institutions and Interpretation: A Critique of City of Boerne v. Flores, 111 Harv. L. Rev. 153, 186–90 (1997) (emphasizing the relevance of institutional differences to permissible government action). do not constrain nonjudicial actors.30×30. Cf. Jack M. Balkin, The Reconstruction Power, 85 N.Y.U. L. Rev. 1801, 1807–08 (2010) (considering these role differences vis-à-vis congressional enforcement power under the Reconstruction Amendments). Moreover, nonjudicial status does not terminate an actor’s duty to consider independently how constitutional norms ought to shape its self-conception, obligations to its polity, and actions. For those actors, this Article’s framework can be a tool of democratic empowerment. For example, in some cases where a court might not find taint judicially actionable, political actors can deploy it — if they choose — to locate taint and justify ostensibly purgative policy.

Two prefatory limitations warrant mention. To facilitate depth over breadth, I treat taint as a constitutional concept and do not attempt full-bore engagement with the vagaries of statutory antidiscrimination law. Furthermore, I elaborate the concept in the “status-based discrimination” context, that is, discrimination relating to “the cultural markers . . . that distinguish groups.”31×31. J.M. Balkin, The Constitution of Status, 106 Yale L.J. 2313, 2325 (1997). Accordingly, my paradigm cases embrace the equal-treatment problems familiar to the Equal Protection Clause and constitutional contexts with some convergent doctrinal evolution, such as the Religion Clauses.32×32. See id. at 2346–49; see also infra section V.B.3, pp. 1267–68. With the concept in hand, I plan in a future work to offer a more comprehensive taxonomy of the different settings in which taint might manifest and explore whether those different settings require treating taint differently.

This Article proceeds in five Parts. Part I starts by elaborating the problem from earlier core manifestations through to contemporary, murkier examples. Part II steps back to consider the relevance of intertemporal continuity to legal responsibility and how such continuity can shape a functional, rigorous, and transsubstantive concept of discriminatory taint. Part III types the descriptive phenomenon as a constitutional concept that ought to trigger a taint-specific anti-discrimination decision rule that, importantly, rejects the centrality of specific intent. Part IV offers prescriptions for various actors in implementing the concept, and Part V concludes by engaging the most pressing critiques and noting potential implications of the analysis for other legal questions.

I. An Unappreciated Problem

The current problem does not stem from disagreement about history’s relevance, understood abstractly. Most everyone agrees that context matters, and that history is context. As it turns out, however, courts are distinctively at sea when it comes to the detection and import of discriminatory policy predecessors.

A. Roots

A modern framework can benefit from examining the approach to continuity and change over time reflected in older, polar cases. Two excellent examples are the Grandfather Clause cases and the White Primary Cases.

The former set began with Guinn v. United States,33×33. 238 U.S. 347 (1915). which involved an Oklahoma constitutional amendment that imposed a literacy requirement for voting, but then exempted any person who was either entitled to vote on January 1, 1866, or descended from such a person.34×34. Id. at 357. Oklahoma, notably, secured its 1907 admission to the Union with a suffrage provision that would pass muster today (but for its reservation of the franchise to male citizens). See id. at 355, 357. “Shortly after” admission, it effected this “radical change” in the provision. Id. at 355. This facially race-neutral provision’s aim was clear. No one could imagine any basis for pegging the exemption to 1866 other than an attempt to perpetuate sub silentio what the Fifteenth Amendment prohibited, and the Court invalidated it as such.35×35. See id. at 364–65, 367. Because the Fifteenth Amendment passed in 1870, Oklahoma’s provision was in a sense overinclusive by four years, but that hardly disguised its discriminatory descent.

Undeterred, Oklahoma immediately enacted a new scheme.36×36. Samuel Issacharoff et al., The Law of Democracy 53 (5th ed. 2016). It automatically qualified anyone who voted in 1914 (under the suffrage provision Guinn invalidated), but barred everyone else from registering unless they applied during two weeks at the start of May 1916.37×37. See id. But the changed form couldn’t insulate the new provision: “The Amendment nullifies sophisticated as well as simple-minded modes of discrimination. . . . [The new scheme] partakes too much of the infirmity of [the old].”38×38. Lane v. Wilson, 307 U.S. 268, 275 (1939). The two-week new-registration period “operated unfairly against the very class” for whom the Amendment was passed.39×39. Id. at 277; see also Driver, supra note 17, at 945–46 (classifying the new provision as a “backup”).

Similarly, Texas worked overtime in the White Primary Cases40×40. Terry v. Adams, 345 U.S. 461 (1953); Smith v. Allwright, 321 U.S. 649 (1944) (overruling Grovey v. Townsend, 295 U.S. 45 (1935)); Grovey, 295 U.S. 45; Nixon v. Condon (Nixon II), 286 U.S. 73 (1932); Nixon v. Herndon (Nixon I), 273 U.S. 536 (1927). to repackage an invalid policy as constitutional. Post-Reconstruction, Texas was a one-party Democratic state.41×41. Issacharoff et al., supra note 36, at 266. Over roughly thirty years, Texas Democrats deployed: statutory bars on any Black person’s participation in Democratic primaries (Nixon I42×42. 273 U.S. 536. ); express statutory direction to state party executive committees to determine their members, with predictable results (Nixon II43×43. 286 U.S. 73. ); an ostensibly independent party resolution barring Black participation (Grovey44×44. 295 U.S. 45. and Smith45×45. 321 U.S. 649. ); and (in at least one county) an ostensibly private, voluntary “club” with de facto control over nominations that limited its membership to White persons (Terry46×46. 345 U.S. 461 (1953). ).47×47. See id. at 266–79. That the policies discriminated was always obvious. After Nixon I, the impediment was the absence of obvious state action.48×48. See N.Y. State Bd. of Elections v. Lopez Torres, 552 U.S. 196, 203 (2008); Smith, 321 U.S. at 661–62. Yet, in asking whether state action indeed produced the discrimination, the Court surely was not ignorant of what a four-Justice concurrence in Terry called “[a]n old pattern in new guise.”49×49. 345 U.S. at 480 & n.7 (Clark, J., concurring) (citing Nixon I, 273 U.S. 536; Nixon II, 286 U.S. 73; and then Smith, 321 U.S. 649).

These are some of the clearest premodern50×50. By premodern I mean before the late 1970s, when equal protection doctrine settled on a focus on government “intent.” See, e.g., Caleb Nelson, Judicial Review of Legislative Purpose, 83 N.Y.U. L. Rev. 1784, 1850–51 (2008); Strauss, supra note 26, at 951–52. cases of discriminatory descent. Also relevant are cases of clear attempts to evade a previously stated universal rule, even if not announced in the first instance against the evader. These include cases growing out of Massive Resistance to Brown v. Board of Education,51×51. 347 U.S. 483 (1954); see, e.g., Brest et al., supra note 25, at 1040; Justin Driver, The Schoolhouse Gate 263 (2018) (discussing Green v. Cnty. Sch. Bd., 391 U.S. 430 (1968)); Alan David Freeman, Legitimizing Racial Discrimination Through Antidiscrimination Law: A Critical Review of Supreme Court Doctrine, 62 Minn. L. Rev. 1049, 1090 (1978) (citing Monroe v. Bd. of Comm’rs, 391 U.S. 450 (1968)); Strauss, supra note 26, at 949 n.46 (citing Griffin v. Cnty. Sch. Bd., 377 U.S. 218 (1964)). or cases dealing with clever denials of the right to jury service,52×52. See, e.g., Batson v. Kentucky, 476 U.S. 79, 100 n.25 (1986) (overruling Swain v. Alabama, 380 U.S. 202 (1965)); Swain, 380 U.S. at 221–26 (declining to hold that an individual peremptory strike of a Black juror could be unconstitutional unless part of a striking and systematic pattern of racially disparate strikes); Akins v. Texas, 325 U.S. 398, 407 (1945) (failing to invalidate a conviction obtained under a facially neutral scheme executed so as to permit just one, but no more, Black person per grand jury); Hill v. Texas, 316 U.S. 400, 404 (1942) (invalidating a conviction obtained under a facially neutral scheme that worked to eliminate any Black person’s participation on juries); Norris v. Alabama, 294 U.S. 587, 598–99 (1935) (same); Williams v. Mississippi, 170 U.S. 213, 221, 225 (1898) (failing to invalidate a facially neutral state scheme intended to bar Black persons from jury service); Strauder v. West Virginia, 100 U.S. 303, 312 (1880) (invalidating a West Virginia statute explicitly banning Black persons from jury service). and the “unremitting and ingenious” voting discrimination that Congress found to justify the Voting Rights Act’s (VRA) preclearance provision.53×53. South Carolina v. Katzenbach, 383 U.S. 301, 309 (1966); see also id. at 309–15. It breaks no new ground to observe that “those . . . who are of a mind to discriminate”54×54. Batson, 476 U.S. at 96 (quoting Avery v. Georgia, 345 U.S. 559, 562 (1953)). may treat a formal legal proscription on discrimination as a prompt to discriminate more efficiently rather than to cease. More important are the relationships these examples demonstrate. Terry taken alone might be one thing; Terry as the culmination of multifarious attempts to exclude Black Texans from exercising political power is another.

B. The Problem’s Persistence

Notwithstanding these older cases, one might wonder whether any conceptual problem remains. Since the 1970s, the Supreme Court’s equal protection cases have instructed55×55. See, e.g., Pers. Adm’r v. Feeney, 442 U.S. 256 (1979); Village of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252 (1977); Washington v. Davis, 426 U.S. 229 (1976). courts to ask whether an ostensibly neutral action was actually motivated at least in part by discriminatory purpose.56×56. See Feeney, 442 U.S. at 279; Arlington Heights, 429 U.S. at 265–68; see also Bertrall L. Ross II, The Representative Equality Principle: Disaggregating the Equal Protection Intent Standard, 81 Fordham L. Rev. 175, 178 (2012). The analysis is contextual and includes attention to the “historical background of the [challenged] decision” as a legitimate “evidentiary source.”57×57. Arlington Heights, 429 U.S. at 267. Free exercise analysis of intentional religious discrimination has incorporated that approach.58×58. See Church of the Lukumi Babalu Aye, Inc. v. City of Hialeah, 508 U.S. 520, 540–41 (1993); see also Huq, supra note 10, at 1238. The deceptively simple conclusion is that discriminatory descent simply folds into the holistic enterprise as part of historical background. But that would leave at least two problems unanswered: how to determine discriminatory descent and what exactly folding it in entails. To sharpen the point, consider McCleskey v. Kemp.59×59. 481 U.S. 279 (1987).

McCleskey is perhaps best known for its skepticism of statistical analysis.60×60. See, e.g., John Charles Boger, McCleskey v. Kemp: Field Notes from 1977–1991, 112 Nw. U. L. Rev. 1637, 1678 & n.184 (2018). But McCleskey also offered what the majority called “historical evidence” of “Georgia laws in force during and just after the Civil War.”61×61. 481 U.S. at 298 n.20. That sterile phrasing understated those laws, which made execution race-dependent; McCleskey argued that Georgia’s contemporary capital laws operated as their de facto descendants.62×62. See id. at 329–30 (Brennan, J., dissenting) (describing Georgia’s “dual system” of capital punishment, id. at 329); Randall L. Kennedy, McCleskey v. Kemp: Race, Capital Punishment, and the Supreme Court, 101 Harv. L. Rev. 1388, 1411–13, 1424 (1988) (elaborating upon that dual system and its “continuity,” id. at 1412, with the capital system challenged in McCleskey). The Court dismissed the argument in a curt footnote: While a decision’s “historical background . . . is one evidentiary source” of proof,63×63. McCleskey, 481 U.S. at 298 n.20 (quoting Village of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 267 (1977)). “unless historical evidence is reasonably contemporaneous with the challenged decision, it has little probative value.”64×64. Id. Notably, the Court cited Hunter v. Underwood, 471 U.S. 222 (1985), which invalidated a 1901 Alabama constitutional provision that disenfranchised anyone convicted of a crime involving “moral turpitude.” Id. at 223. The Court treated Underwood as easy, given the 1901 constitutional convention’s avowedly White supremacist aims and the provision’s present-day disparate impact. Id. at 227, 229. McCleskey could be read to imply, therefore, that this was the sort of historical background that interested the Court, not background based on lines of descent. Thus, “official actions taken long ago” are unacceptable “as evidence of current intent.”65×65. McCleskey, 481 U.S. at 298 n.20.

This begins to bring the problem into focus. Any contemporary government policy exists subsequent to many other policies. The interesting, and difficult, puzzle for this Article is whether a policy might relate to a predecessor such that the fact of that relationship directs a different evaluative outcome. The problem suggests a spectrum; how do we know how related is related enough?66×66. Cf. Levinson, supra note 16, at 1373 n.210 (“[E]verything is related to everything else under . . . some time frame. The meaningful question[ is] . . . which time frame.”). And how should we treat a related policy, once identified?

For constitutional law purposes, it would not do to take McCleskey’s footnote to mean that we just ignore any policy that is not “reasonably contemporaneous” (whatever that means). The footnote is undertheorized, even beyond the vagueness of “reasonably.” Understanding it as denying the relevance of any history beyond a modern policy’s immediately contemporaneous history (for example, legislative history67×67. See, e.g., supra note 64. ) would be inconsistent with the approaches seen in section I.A, Arlington Heights’s attention to “historical background,” and (as we will see) how continuity is treated elsewhere in law.68×68. See infra section II.A, pp. 1212–15. The Court has not yet held that the only history that matters is the history the government endorses.

Observing that McCleskey’s inchoate skepticism does not a theory make is not itself a theory. But it does emphasize the persistence of a problem. Failing to think carefully about why and how past related policies matter risks cursory, inconsistent analyses of wrongful discrimination. The next section observes this risk’s modern manifestation.

C. The Problem in Practice

Before proceeding, a note on what I bracket. We might be interested in how noncontemporaneous history at a high level of abstraction should shape analysis of discrete contemporary action. Perhaps, for example, a jurisdiction’s history of discrimination should always inform the evaluation of its contemporary behavior. That argument would confront judicial hostility to supposedly “amorphous” concepts of “societal discrimination,” a likely target of the McCleskey footnote just discussed.69×69. Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 307 (1978) (opinion of Powell, J.); see also Veasey v. Abbott, 830 F.3d 216, 231–32 (5th Cir. 2016) (discounting, in a voter ID case, the relevance of Texas’s general history of voting discrimination (citing McCleskey)); Hayden v. Paterson, 594 F.3d 150, 166–67 (2d Cir. 2010). Although that hostility can be critiqued, that is not this Article’s goal.

Nor am I examining propositions of the form “because this type of regulation was used, or could have been used, for suspect ends in the past (somewhere), all contemporary such regulation is suspect today.”70×70. Cf. Darrell A.H. Miller, Tainted Precedent, 74 Ark. L. Rev. 291, 295 (2021) (critiquing such arguments); Murray, supra note 16, at 2086 (critiquing anti-abortion argument premised on connecting American history of eugenics with the potential for abortion’s use in eugenicist ways). Defending that claim would require more — and different — emphasis on how one demonstrates a past-present nexus and connects particularized present harms to a more nebulous past.71×71. For similar reasons, I mostly bracket interrogation of private discrimination qua private discrimination, although past private discrimination that generates discrete past public discrimination could be relevant to a taint-based analysis of a successor public act. Cf. Evans v. Abney, 396 U.S. 435, 436–39 (1970) (considering, after judicial invalidation of a public park’s segregated operation under the trust that willed the city the park property, whether a state court could constitutionally terminate the trust and revert the property to decedent’s heirs). Taint might well matter in understanding how a general recognition of a discrimination-riddled past might generate or explain a variety of present-day disparate outcomes. I just believe that such a use is more expansive in a way that would require further theorization.

I focus instead on a narrower, yet still important question. Sometimes, a government act isn’t just earlier in time than another act, but seems related to the later act in a generative way. The challenge is concretizing that intuition. The nexus problem remains (Part II seeks to solve it) but in less amorphous form than more freewheeling inquiries into the past’s relevance to the present.

We can learn much from this narrower focus. As this section will show, substantial judicial disagreement exists on whether we should even care about a supposedly problematic genealogy, on how to identify it if it is a legitimate quarry, or on how its presence ought to affect the validity of the action being reviewed. The first task is grasping the disagreement.

To do so, I have selected a fairly illustrative set of cases presenting the T1/T2 pattern, and attempted a rough categorization. Though the cases sometimes defy easy placement in boxes, the approaches to the problem of policy continuity fall on a spectrum from temporally minimalist to temporally maximalist. In some cases, the court avoids confronting the question, which only emphasizes the need for methodological order.

1. Temporally Minimalist. — Some courts treat the issue in a temporally minimalist way. That is, their decisions exude deference to legislatures and the formal processes of legislation in a way that minimizes the relevance of similar past policies. These courts focus on whether the substance, procedural history, or enactor of a T2 policy is at all different from those aspects of a T1 policy. For them, changes reflected in and resulting from the formal procedures of lawmaking justify applying conventional scrutiny to the policy under review, notwithstanding its past. None of these courts expressly deem the past irrelevant. But what does the work is whether the T2 policy can be seen as different, and if so, the discriminatory predecessor recedes into the background.

An apt recent example, Abbott v. Perez,72×72. 138 S. Ct. 2305 (2018). blasted the idea that “taint” associated with Texas’s 2011 redistricting maps determined the validity of 2013 maps based (in part) on the 2011 maps.73×73. See id. at 2324–25. To be sure, the Court acknowledged that if the 2011 maps were discriminatory, that would be “relevant” to the ultimate question of whether the 2013 districting plans were unconstitutional.74×74. Id.; see also id. at 2327. But it condemned the lower court’s use of taint as having wrongly “reversed the [plaintiff’s] burden of proof” to show that the 2013 legislature acted with discriminatory taint.75×75. Id. at 2325. And it is hard to find in the majority opinion’s lengthy reweighing of the record any reference to, or use of, the district court’s finding that the 2011 plans were discriminatory.76×76. See id. at 2326–30. Perhaps such reweighing was improper, see id. at 2356 (Sotomayor, J., dissenting), but it occurred and is thus worth scrutinizing.

The minimalist approach lurked in Trump v. Hawaii77×77. 138 S. Ct. 2392 (2018). as well. After Donald Trump’s campaign-trail calls for “a total and complete shutdown” of Muslim immigration,78×78. See id. at 2417. his Administration instituted three successive restrictions on foreign-national entry that — to varying degrees — targeted majority-Muslim countries. Ban 3 followed the judicial blocking of Ban 1 and the partial judicial blocking of Ban 2’s temporary restrictions.79×79. See id. at 2403–05 (recounting the development of bans in tandem with litigation challenges and noting expiration of Ban 2’s temporary measures). All bans asserted the same goal, pursued in the same facially neutral manner: combat terrorist threats through limitations on the entry of foreign nationals.80×80. Compare id. at 2403 (first ban titled “Protecting the Nation from Foreign Terrorist Entry into the United States”), with id. at 2404 (third ban titled “Enhancing Vetting Capabilities and Processes for Detecting Attempted Entry into the United States by Terrorists or Other Public-Safety Threats”). All, plaintiffs said, actually violated the Establishment Clause’s prohibition of preferring one religion over another.81×81. See id. at 2406.

Given the Court’s “assum[ption]” that it might consider “extrinsic evidence,”82×82. Id. at 2420; see also id. at 2419–20. It “assumed” this after first flirting with saying that national security precedent barred considering any such evidence. See id. context (such as candidate Trump’s statements and the policy’s disparate religious effect) could cast doubt on a facially neutral action.83×83. See McCreary County v. ACLU of Ky., 545 U.S. 844, 862 (2005). Lower courts that invalidated those bans had followed that course.84×84. See, e.g., Hawaii v. Trump, 241 F. Supp. 3d 1119, 1135–36 (D. Haw. 2017). The third ban, however, presented a potential complication. Unlike its predecessors, it followed a worldwide review undertaken pursuant to Ban 2, included two countries that are not majority-Muslim, and removed two majority-Muslim countries included in Ban 1.85×85. See Trump, 138 S. Ct. at 2404–05 (describing worldwide review and new inclusion of North Korea and Venezuela); id. at 2422 (describing removal of majority-Muslim countries). Chad, a third majority-Muslim country, was removed from Ban 3 postinception pursuant to the order’s review process. See id. at 2406.

Those differences seemed to matter. To be sure, the Court never said that the modifications saved Ban 3. But it did emphasize the “persuasive evidence” that the ban rested on legitimate national security concerns — evidence that appeared primarily to be findings generated by the “worldwide” multiagency review and the revision of Ban 3 compared to Ban 1.86×86. Id. at 2421. Nor did it ever hint that the prior “uninsulated” policies affected the disposition. In short, the Court was able to deploy benign context unique to Ban 3 against the damning context common to all bans.87×87. See Landau, supra note 1, at 2173 (“The more the government could show its policy was thoroughly vetted, the less the President’s disparaging remarks about Islam seemed to undermine the case for deference.”). It could do so only by minimizing the earlier bans’ import, a move that itself was facilitated by the last ban’s distinctions.

Important lower court examples of temporal minimalism also exist.88×88. Reaching back further could produce more cases, but my goal here is illustrative, not exhaustive. Three circuits have considered a felon disenfranchisement law that had been revised and replaced in some form over time.89×89. See Hayden v. Paterson, 594 F.3d 150, 164–65, 167 (2d Cir. 2010); Johnson v. Florida, 405 F.3d 1214, 1223–25 (11th Cir. 2005); Cotton v. Fordice, 157 F.3d 388, 391 (5th Cir. 1998). All acknowledged (or assumed) that an earlier version of the provision reflected racial bias, but nevertheless validated the current provision. The reasoning is notable. One court implied that merely amending the provision to remove burglary from the list of disenfranchising crimes sufficed to “supersede[] the previous provision and remove[] the discriminatory taint associated with the original version.”90×90. Cotton, 157 F.3d at 391. The court also emphasized a later amendment that added murder and rape to the list — “crimes historically excluded from the list because they were not considered ‘black’ crimes.” Id. (The additions, incidentally, seem equally consistent with approval of the original plan and desire to make it work “better.”) In any event, the court was clear that “each amendment superseded the previous provision and removed the discriminatory taint.” Id. (emphasis added). Another placed dispositive purgative weight on an 1894 provision’s removal of a sunset provision that was included in that provision’s three (discriminatory) predecessors.91×91. Hayden, 594 F.3d at 164–67 (finding that plaintiffs sufficiently alleged discriminatory purpose underlying the 1821, 1846, and 1874 constitutional provisions that required the legislature to pass a disenfranchisement law at the next legislative session (but that made that choice permissive thereafter), but concluding that the 1894 provision’s removal of that sunset provision “substantively change[d]” the provision enough to sever any inferential link, id. at 167). A third was satisfied that a century-later revision’s modification of the class disenfranchised under its predecessor was a “substantive[] alter[ation]”92×92. Johnson, 405 F.3d at 1225. and reenactment, to be validated absent independent contemporaneous evidence of bias.93×93. See id. at 1221–25.

Some lower court voter ID cases have also taken this approach. After a federal district court invalidated an earlier voter ID law, Texas passed a new law ostensibly intended to cure any legal defects.94×94. See Veasey v. Abbott, 888 F.3d 792, 795–97 (5th Cir. 2018). After the district court invalidated the new law because it was a “vestige” of the earlier one,95×95. Id. at 798. the Fifth Circuit reversed, reasoning that “unless remedial legislation . . . is itself infected with a discriminatory purpose, federal courts are obliged to defer to the legislative remedy.”96×96. Id. at 800. Critically, in its view, “substantial, race-neutral alterations in an old unconstitutional law may remove the discriminatory taint.”97×97. Id. at 802 (citing Cotton v. Fordice, 157 F.3d 388, 391–92 (5th Cir. 1998)). This phrasing at least seems to concede that a new law could theoretically be tainted. Similarly, a North Carolina voter ID case closely tracked Abbott in casting a lower court’s discussion of taint as an improper shifting of the burden of proof.98×98. See N.C. State Conf. of the NAACP v. Raymond, 981 F.3d 295, 303–05 (4th Cir. 2020) (reversing the district court’s invalidation of a 2018 voter ID law that followed a 2013 law’s invalidation by the Fourth Circuit). Like Abbott, that case paid lip service to Arlington Heights’s historical-background factor, but said almost nothing that took seriously the possible continuity between old and new laws.99×99. See id. at 302–11 (restricting discussion of historical background to a general passing reference to the state’s “historical background favor[ing] finding discriminatory intent,” id. at 305). Section I.C.2, pp. 1207–10, discusses a bit of maximalism that peeked through.

The minimalist approach is thus well established. What makes these cases minimalist, to be clear, is not that they ignore historical context. What places them at this pole is their parsimonious approach to treating policy continuity as meaningful. That approach seems to stem from distaste for the idea of shackling the present with the past. Broadly speaking, that orientation means these courts treat substantive change effected through formal lawmaking procedures as sufficient to discharge any taint of the distasteful past.

2. Temporally Maximalist. — The Grandfather Clause and White Primary Cases offer temporally maximalist attention to the realities of how a particular practice came to be “over[] time.”100×100. Pildes, supra note 23, at 2. More recent examples are also available.

One is Flowers v. Mississippi.101×101. 139 S. Ct. 2228 (2019). The background: In five trials over roughly a decade,102×102. See Flowers v. State, 240 So. 3d 1082, 1093 (Miss. 2017). District Attorney Doug Evans engaged in a pattern of racially disparate strikes in jury selection, with his prosecutorial misconduct generating three reversals, including one on Batson grounds.103×103. Flowers, 139 S. Ct. at 2236–38. At jury selection for his sixth attempt at convicting Curtis Flowers, Evans used five of his six peremptory strikes against Black jurors.104×104. Id. at 2237. Under these “extraordinary facts,” the Court found a Batson violation.105×105. Id. at 2235.

But what was so extraordinary? Evans gave facially coherent reasons for every strike.106×106. See id. at 2251–52 (Alito, J., concurring) (claiming the reasons for the strikes “were not only facially legitimate but . . . would be of concern to a great many attorneys,” id. at 2251, and asserting that in the usual case they would pass muster); id. at 2253 (Thomas, J., dissenting) (listing reasons). Batson aficionados know well that providing such reasons is usually not difficult and usually sufficient to rebuff a challenge.107×107. See, e.g., Thomas Ward Frampton, For Cause: Rethinking Racial Exclusion and the American Jury, 118 Mich. L. Rev. 785, 787 & n.3 (2020) [hereinafter Frampton, For Cause] (collecting studies on the persistence of racially disparate peremptory strikes); Thomas Ward Frampton, The Jim Crow Jury, 71 Vand. L. Rev. 1593, 1626–27 (2018) [hereinafter Frampton, Jim Crow Jury]. What’s more, a Black juror was actually seated in Flowers’s sixth trial.108×108. See Flowers, 139 S. Ct. at 2246. One can easily imagine an opinion nodding to the history but stressing that the task was to determine the constitutionality of these strikes, made under different circumstances for different reasons. Instead, the majority held that the racial disparities of past strikes in past trials coupled with multiple past Batson violations by this same state actor mattered significantly to the question of discriminatory intent in this trial.109×109. See id. at 2245–46, 2250–51; see also id. at 2234, 2236 (emphasizing the identity of the prosecutor); id. at 2251 (Alito, J., concurring) (same); Paul Butler, Mississippi Goddamn: Flowers v. Mississippi’s Cheap Racial Justice, 2019 Sup. Ct. Rev. 73, 80 (emphasizing the decision’s contextual grounding).

Here, too, lower court maximalist exemplars exist. Although (as noted above) the litigation terminated in a largely minimalist decision,110×110. See supra p. 1207. the maximalist approach manifested twice in the recent litigation over North Carolina’s voter ID law. In 2018, North Carolina’s legislature implemented a recent ballot measure requiring voter identification.111×111. See N.C. State Conf. of the NAACP v. Raymond, 981 F.3d 295, 299 (4th Cir. 2020); see also Johnson v. Florida, 405 F.3d 1214, 1225 (11th Cir. 2005). The ballot measure followed the Fourth Circuit’s holding multiple provisions of a 2013 voter ID law to be racially discriminatory.112×112. See Raymond, 981 F.3d at 299 (citing N.C. State Conf. of the NAACP v. McCrory, 831 F.3d 204, 216–18 (4th Cir. 2016)). The legislature submitted the ballot measure.

Initially, a district court invalidated the implementing law’s voter ID provisions.113×113. See N.C. State Conf. of the NAACP v. Cooper, 430 F. Supp. 3d 15, 53 (M.D.N.C. 2019). Its context-laden opinion emphasized pre-2013 history, substantial identity between legislative supporters of the two bills, legislative statements indicating disagreement with the first bill’s invalidation and desire to protect the new bill from constitutional challenges, and the maintenance of elements the Fourth Circuit had found problematic.114×114. See id. at 29–35. On the last point, specifically, the new law addressed the disparate treatment of absentee and in-person voting the Fourth Circuit had noted, but did not alter the rejection of public-assistance IDs as acceptable identification. See id. at 34. Despite new provisions that “could significantly limit” the new law’s disparate impact,115×115. Id. at 35. the court emphasized that the expansion of eligible IDs “continue[d] to primarily include IDs which minority voters disproportionately lack” and exclude those which nonwhite voters were more likely to have.116×116. Id. at 38. So, although the new law changed enough to not be a “barely disguised duplicate” of the old,117×117. Id. at 43 (quoting Amended Memorandum of Law in Support of Plaintiff’s Motion for Preliminary Injunction at 10, Cooper, 430 F. Supp. 3d 15 (No. 18-cv-01034), ECF No. 91). the court concluded that the legislature had declined to “cleanse the discriminatory taint” of the old law and instead attempted to circumvent the Fourth Circuit’s opinion.118×118. Id.

That approach to locating and cognizing problematic continuity epitomizes maximalism. I noted the Fourth Circuit’s largely minimalist reversal above. One piece of its analysis, however, deviated. In the court’s view, the “interject[ion]” of the statewide electorate,119×119. N.C. State Conf. of the NAACP v. Raymond, 981 F.3d 295, 305 (4th Cir. 2020). through the ballot measure postdating the 2013 law, “undermined the . . . link[]” between the old and new.120×120. Id. at 306. In contemplating the possibility that the link might matter, and looking to the specifics of the policies’ relationship to examine whether it did, the court reasoned in a temporally maximalist way.

Similarly, a district court rejected the minimalist approach in evaluating 8 U.S.C. § 1326’s criminalization of the reentry of certain removed aliens.121×121. See United States v. Carrillo-Lopez, No. 20-cr-00026, 2021 WL 3667330, at *1, *9 (D. Nev. Aug. 18, 2021). Both parties agreed that proscribed motives animated the provision’s initial enactment in 1929.122×122. Id. at *7. The provision, however, had been reenacted six times over the next sixty-seven years, with alterations to penalties and scope.123×123. See id. at *4 n.11, *23–24. Drawing on Abbott and the circuit-court felony-disenfranchisement cases, the government argued that any reenactment per se rendered the tainted origins irrelevant.124×124. See id. at *9 & n.19.

The court disagreed. In its view, 1929 mattered both as historical background (in the Arlington Heights sense) and because the original discrimination “infect[ed]” subsequent policies.125×125. Id. at *9–10. For unclear reasons, the court focused primarily on 1929’s relevance to the 1952 reenactment, rather than the subsequent policies. See id. at *4 n.10; see also *9–18. In my view, more front-end energy was warranted to explain why the infection had persisted to the present. It appears that an underelaborated piece of the reasoning was that the later policies also lacked substantial alteration from the 1929 and 1952 versions. See id. at *23–24; see also infra p. 1210. The key symptom, it concluded, was the lack of “substantial[] alter[ation]” over time.126×126. See id. at *10. The policy’s alterations were not to the contrary; no later Congress had “confronted” the problematic roots, and no alterations had “changed [the law’s] function” or “functional operation.”127×127. Id. at *24; see also id. at *23–25. Coupled with additional evidence of proscribed motivations in 1952, this infection invalidated the present-day provision. Other district courts considering this provision and its sibling provision, 8 U.S.C. § 1325, have reached the opposite conclusion, often explicitly embracing the idea that the passage of time rendered the initial law’s wrongfulness minimally relevant.128×128. See, e.g., United States v. Machic-Xiap, No. 19-cr-407, 2021 WL 3362738, at *1–2, *15 (D. Or. Aug. 3, 2021) (considering section 1326); United States v. Rios-Montano, No. 19-CR-2123, 2020 WL 7226441, at *4 (S.D. Cal. Dec. 8, 2020) (considering section 1325).

3. The Muddle. — These admittedly high-level labels aim to outline the landscape, not to imply perfect distinctions in approach. One can always debate categorization. Take Flowers. District Attorney Evans was a constant in every trial. Perhaps Flowers is thus only maximalist in an anodyne way dependent on a willingness to assume considerable individual motivational continuity over time. That willingness, perhaps, might be less forthcoming where that sort of individualized continuity is less likely (such as multimember institutions). That critique cannot be overstated, though; people, like institutions, change over time.129×129. See, e.g., Levinson, supra note 16, at 1373 n.211.

More importantly, my claim does not require two crisply demarcated categorical approaches. The point, in fact, is that the categories are not clean, with the diversity itself indicating need for further analysis. After all, one could frame Trump as involving individualized continuity no less than Flowers — but the government won the former and lost the latter. Perhaps the best demonstration of a lack of consensus lies in recent cases where a standard approach would have helped, if one existed.

For example, when it considered a nonunanimous jury provision in Ramos v. Louisiana,130×130. 140 S. Ct. 1390 (2020). the Court made sure to emphasize that the provision’s forbearer was adopted with racist intent and effect.131×131. See id. at 1394. But the Court’s Sixth Amendment resolution failed to specify why that historical relationship mattered.132×132. Id. at 1426–27 (Alito, J., dissenting) (raising this objection). Two months later, Espinoza v. Montana Department of Revenue133×133. 140 S. Ct. 2246 (2020). reversed an application of a state constitutional provision restricting aid to religious schools; its predecessor provision modeled a failed federal constitutional amendment motivated by anti-Catholic animus.134×134. See id. at 2268–69, 2271 (Alito, J., concurring); see also id. at 2258–59 (majority opinion). Espinoza noted the history, but said nothing about whether it might inform the analysis.135×135. See id. at 2259 (stating only that “the historical record is ‘complex’”).

Further emphasizing the range of thought, two Justices who concurred in both cases to discuss the predecessor found themselves on functionally different sides: Justice Alito stressed the discriminatory predecessor in Espinoza and argued against its relevance in Ramos, while Justice Sotomayor did the opposite.136×136. Compare id. at 2273 (Alito, J., concurring) (emphasizing textual and impact continuity from T1 to T2, disputing that the provision was “cleansed of its bigoted past” and arguing that “it . . . does not matter whether Montana readopted the . . . provision for benign reasons”), and Ramos, 140 S. Ct. at 1426 (Alito, J., dissenting) (treating the jury provision’s origins as irrelevant), with id. at 1410 (Sotomayor, J., concurring) (arguing that the jury provision’s origins are “worthy of . . . attention” because of the provision’s persistent discriminatory effects and the reenacting legislature’s failure to “grapple[] with the laws’ sordid history”), and Espinoza, 140 S. Ct. at 2293 n.2 (Sotomayor, J., dissenting) (arguing that the reenacting legislature sufficiently engaged the past and that some Catholics supported the reenactment). I do not mean to imply disingenuousness. As I will argue, taint is context-sensitive, and one might well conclude (and both Justices attempted to show) that it is more outcome determinative in one of these cases than the other. My point is simply that the lack of consensus on approach generates significant differences in engagement even among those who are willing to acknowledge that a discriminatory predecessor might matter.

* * *

The cases confirm the prevalence of allegedly discriminatory antecedents. They also confirm the law’s undeveloped state, notwithstanding the broad judicial understanding that context matters. Courts never ignore history. But what they fail to do is apply a consistent theoretical framework to that history that permits robust evaluation of policy continuity. One can, perhaps, excuse them for failing to apply what does not exist. History, especially a problematic predecessor, surely matters. But because we do not agree on why, how, or what it takes to cancel it out, courts invariably operate in a critically undertheorized way. This issue is too important — and too certain to persist — for us not to do better. Accordingly, the next Part pulls together the strands of theory embedded in the cases, explains why they matter, and employs them to construct a framework that can help judicial and nonjudicial actors better organize and systematize their analysis of discriminatory predecessors.

II. The Missing Piece: Discriminatory Taint

A crisper concept of “taint” can demystify many of the recurring problems of temporality and change. This Part looks to broader principles of legal continuity to build a better descriptive account that reflects and ties together the key recurring problems in discriminatory predecessor cases.

A. Legal Continuity, Meaning, and Responsibility

A step back from doctrine may help. Professor Richard Pildes has offered a useful public law taxonomy for describing whether and how much contingent context should shape evaluation of government action. An approach based on “institutional realism” focuses on how an institution “actually function[s] in, and over, time.”137×137. Pildes, supra note 23, at 2. Its antonym, “institutional formalism,” treats institutions in a “black box” manner: “intentionally ignor[ing]” contingent institutional features, including the institution’s previous behavior in substantially similar circumstances.138×138. See id. When faced with this tension, Pildes notes, courts are “all over the map.”139×139. Id. at 37. That aptly summarizes section I.C’s portrait of courts addressing allegedly discriminatory predecessors.

This clash is unsurprising. Our constitutional tradition empowers new democratic actors to set new courses.140×140. See, e.g., Greene, supra note 25, at 128; Super, supra note 16, at 77–78. Formalism advances that end when it declines to impugn facially legitimate actions with “more contingent, specific features of institutional behavior.”141×141. Pildes, supra note 23, at 2. Formalism also embraces the rule-of-law benefits of treating (formally) like cases alike.142×142. See id. at 7; Gowder, supra note 20, at 1023–24. These are real benefits. Yet institutions transcend their present personnel.143×143. See, e.g., Anna Stilz, Collective Responsibility and the State, 19 J. Pol. Phil. 190, 193, 196 (2011). The essence of that transcendence is, naturally, the subject of contestation, but I take as a given the broad acceptance that it is true in some sense. See, e.g., Kendy Hess, The Unrecognized Consensus About Firm Moral Responsibility, in The Moral Responsibility of Firms 170–71, 182–84 (Eric W. Orts & N. Craig Smith eds., 2017). Formally “new” institutional actors had predecessors whose actions shaped the institutional world their successors occupy, in ways that empower and constrain.144×144. See Karen Orren & Stephen Skowronek, Institutions and Intercurrence: Theory Building in the Fullness of Time, in Nomos XXXVIII: Political Order 111, 140–41 (Ian Shapiro & Russell Hardin eds., 1996); Daphna Renan, The President’s Two Bodies, 120 Colum. L. Rev. 1119, 1178 (2020); Ernst H. Kantorowicz, The King’s Two Bodies: A Study in Medieval Political Theology 294–95, 299, 301–02, 309–13 (1997) (noting medieval application of the principle of “identity despite changes” to states, id. at 311); Natasha Wheatley, Spectral Legal Personality in Interwar International Law: On New Ways of Not Being a State, 35 Law & Hist. Rev. 753, 774 (2017) (citing Professor Ernst Kantorowicz’s analysis of the concept of “the state as legal corporation”); David Ciepley, Is the U.S. Government a Corporation? The Corporate Origins of Modern Constitutionalism, 111 Am. Pol. Sci. Rev. 418, 423 (2017).

That much is broadly uncontroversial. For example, public law remedies for government wrongdoing often constrain what later officials may do.145×145. See Parker, supra note 8, at 1161–62. Decisionmakers for corporations and market-participant governments can contractually bind their successors.146×146. See Jon D. Michaels, Privatization’s Pretensions, 77 U. Chi. L. Rev. 717, 739–44 (2010); see also Odette Lienau, Rethinking Sovereign Debt: Politics, Reputation, and Legitimacy in Modern Finance 1–4, 10–11 (2014) (discussing “norm of sovereign debt continuity” — “that sovereign states should repay debt even after a major regime change,” id. at 2). And some government decisions (like exploiting nonrenewable resources) are simply irreversible, whatever later actors may prefer. But the common thread in these easier cases is a T1 commitment to courses of action that inherently outrun the decisionmakers’ tenure, like an ongoing contractual relationship or logging of a forest. It is more controversial to suggest that some aspect of a T1 action can outlive that action’s formal demise to generate constraints on nonidentical T2 action.

A formalist response to that suggestion might simply say T1 does not “constrain” T2 at all, or only can if it is probative of T2 decisionmakers having behaved impermissibly. On this view, the time to address T1 was when it was in force. Now that it is not, discussion of wrongfulness should be addressed to its replacement. Surely this response is sometimes appropriate.147×147. See Post Hoc Ergo Propter Hoc, The Oxford Dictionary of Phil. (Simon Blackburn ed., 3d ed. 2016), https://www.oxfordreference.com/view/10.1093/acref/9780198735304.001.0001/acref-9780198735304-e-2448 [https://perma.cc/4S37-9RWF] (noting the “fallacy of arguing that because one event happened after another, it happened because of it”). Institutional realism, though, can justify a richer inquiry: whether T1’s validity and its relationship to T2 independently impeaches T2.

As it turns out, it is unnecessary to rely on institutional realism in the abstract. It points toward an amply precedented approach. Across various areas, the law often recognizes that formal separateness and nonidentity may not bar the intertemporal transmission of blameworthiness. The common element in these disparate approaches is the recognition that substantial continuity matters. Responsibility may depend on what a discrete event relates to over time within its institutional context.

To make this more concrete, consider some institutional examples. In corporate law, “[a] corporation that acquires, merges with, consolidates with, or spins off from a criminal corporation automatically inherits a criminal taint.”148×148. Mihailis E. Diamantis, Successor Identity, 36 Yale J. on Regul. 1, 4 (2019). Continuity can affect corporate civil liability under similar circumstances, too.149×149. 4 James Cox & Thomas Lee Hazen, Treatise on the Law of Corporations § 22:8 (3d ed. 2020) (observing “well recognized exemptions” to the rule that a corporation that purchases another corporation’s assets is not subject to the selling corporation’s liabilities, including where the transaction is a “de facto merger” or “when the surviving company carries forward a common identity with the selling corporation”); 15 Fletcher Cyclopedia of the Law of Corporations § 7123.20 (discussing similar continuity-based “continuity of enterprise” and “continuation of business” doctrines). Similarly, one district attorney’s nonprosecution promises have constrained successors’ ability to decide otherwise.150×150. See, e.g., Commonwealth v. Cosby, 252 A.3d 1092, 1130–31, 1136 & n.24, 1144 (Pa. 2021). Even the innocuous concept of stare decisis counts: courts face otherwise inapplicable constraints on some cases’ resolution merely because of resolutions in a “previous similar case” —here, “the past is supposed to govern the present.”151×151. Austin Sarat & Thomas R. Kearns, Writing History and Registering Memory in Legal Decisions and Legal Practices: An Introduction, in History, Memory, and the Law 1, 4 (Austin Sarat & Thomas R. Kearns eds., 2002); see also Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833, 854 (1992) (joint opinion) (“[T]he very concept of the rule of law underlying our own Constitution requires such continuity over time that a respect for precedent is, by definition, indispensable.” (emphasis added)).

Continuity-related responsibility for natural persons tracks the treatment of artificial persons. Accordingly, many criminal law doctrines transmute otherwise blameworthy acts into blameless acts “because [they are] rooted in or determined by factors that preceded the criminal incident.”152×152. Mark Kelman, Interpretive Construction in the Substantive Criminal Law, 33 Stan. L. Rev. 591, 611 (1981); see also Paul H. Robinson, Imputed Criminal Liability, 93 Yale L.J. 609, 611–12 (1984). Duress and insanity are two well-known examples.153×153. See Kelman, supra note 152, at 611. The opposite is also true; consider conspiracy law. There, B’s prior entry into an enterprise with A may make B culpable for blameworthy acts A takes growing out of that enterprise.154×154. Robinson, supra note 152, at 617, 665–68. I thank Professor Aziz Huq for suggesting this example. Theories of implied-in-law contracts or unjust enrichment can be understood similarly, as rendering otherwise legitimate enrichment impermissible based on preceding conduct.155×155. See, e.g., Scarlett v. Air Methods Corp., 922 F.3d 1053, 1064–65 (10th Cir. 2019) (discussing implied-in-law contracts); Town of New Hartford v. Conn. Res. Recovery Auth., 970 A.2d 592, 611–13, 611 n.25 (Conn. 2009) (same). Indeed, across diverse settings, responsibility determinations are shaped by a treatment of two or more temporally separated events as continuous, or not.156×156. See, e.g., J.M. Balkin, The Rhetoric of Responsibility, 76 Va. L. Rev. 197, 229–33, 241–43 (1990).

In short, while maintaining a commitment to connecting blameworthiness to culpability, the law has long understood culpability flexibly. In doing so, it often recognizes that continuity between past and present discrete acts informs (or, if you like, constrains) the validity or meaning of later-in-time acts. This can occur even when such acts are formally separated and substantively distinct, and it does not depend on a T2 actor’s willing endorsement or embrace of those constraints.

From this perspective, the anti-continuity embraced in many recent constitutional law cases seems more like an outlier.157×157. See, e.g., Abbott v. Perez, 138 S. Ct. 2305, 2324–25 (2018); cf. Shelby County v. Holder, 570 U.S. 529 (2013) (invalidating a VRA provision that restricted voting changes by certain Southern states based on a formula pegged to past state conduct); Cristina M. Rodríguez, The Supreme Court, 2020 Term — Foreword: Regime Change, 135 Harv. L. Rev. 1, 144 (2021) (arguing that Brnovich v. Democratic Nat’l Comm., 141 S. Ct. 2321 (2021), “continue[d] the line of thought [the Court] began in Shelby County, of downplaying the discriminatory effects of voting rules by divorcing the state laws under its consideration from any history of discrimination”); Boddie, supra note 16, at 1289–90; Guy-Uriel E. Charles & Luis E. Fuentes-Rohwer, The Court’s Voting-Rights Decision Was Worse than People Think, The Atlantic (July 8, 2021), https://www.theatlantic.com/ideas/archive/2021/07/brnovich-vra-scotus-decision-arizona-voting-right/619330 [https://perma.cc/4NHE-JJR8] (describing Brnovich as embodying a recent “impulse” in election law cases to “unmoor the present from the country’s history of racial discrimination”). At the very least, continuity’s recurrence in legal responsibility determinations should caution against dismissing it in the constitutional discriminatory predecessor context.158×158. Constitutional theorists have often advocated greater attention to history and continuity, see Siegel, supra note 16, at 2119 (articulating a “preservation through transformation” theory for “regime[s]”), but have not emphasized the uniqueness of T1/T2 policy relationships, see Driver, supra note 17, at 943 (discussing backups, which do pose the T1/T2 fact pattern, but for reasons not focused on their temporal aspects). Opponents of greater attention to history have often emphasized amorphousness problems. See, e.g., Kim Forde-Mazrui, The Canary-Blind Constitution: Must Government Ignore Racial Inequality?, 79 Law & Contemp. Probs. 53, 53, 64–65 (2016); Boddie, supra note 16, at 1279. Any attempt to resolve the uncertainty section I.C describes should recognize as much. The question of how much constitutional significance we should afford policy continuity is, at least, an open one.

The old tension between realism and formalism cannot be resolved jurispathically.159×159. Cf. Robert M. Cover, The Supreme Court, 1982 Term — Foreword: Nomos and Narrative, 97 Harv. L. Rev. 4, 40 (1983) (observing a judicial “jurispathic” role in “suppres[sing]” one view of the law in favor of another); Pildes, supra note 23, at 53–54. We need both, inside and outside constitutional law. Deciding which should drive our approach to constitutional law continuity must draw on some normative priors.160×160. See Levinson, supra note 16, at 1372, 1375–76; cf. Richard Thompson Ford, Bias in the Air: Rethinking Employment Discrimination Law, 66 Stan. L. Rev. 1381, 1388 (2014) (observing the private law truism that “legal ‘causation’ does not refer to an objective phenomenon but is instead a legal conclusion informed by normative judgments”). I turn now to sketching some priors that counsel in favor of temporally maximalist realism in this context.

B. Guiding Principles

Perhaps surprisingly, wide agreement exists regarding the general principles that support a temporally maximalist approach to the taint problem: effectuating constitutional mandates requires, among other things, that they are not evaded through artifice and that past unconstitutional conduct is not extended into the present.

Consider evasion. Here, courts have long aimed to extinguish bad faith attempts to comply with the form but not the substance of constitutional rules. Indeed, so-called “anti-evasion” doctrines occur throughout constitutional law, reflecting a sense that constitutional rules are not made to be broken.161×161. Brannon P. Denning & Michael B. Kent, Jr., Anti-evasion Doctrines in Constitutional Law, 2012 Utah L. Rev. 1773, 1779; see also Elena Kagan, Private Speech, Public Purpose: The Role of Governmental Motive in First Amendment Doctrine, 63 U. Chi. L. Rev. 413, 414 (1996). Most disagreements concern what should be deemed evasion.

The past’s extension into the present poses some thornier questions. Sometimes constitutional law takes this quite seriously, despite the Supreme Court’s recent anti-continuity strain.162×162. See sources cited supra note 157. For example, detecting unconstitutional discrimination authorizes a court not just to enjoin the discrimination but to ensure the eradication of its effects.163×163. See, e.g., United States v. Virginia, 518 U.S. 515, 547 (1996); Douglas Laycock, The Death of the Irreparable Injury Rule, 103 Harv. L. Rev. 687, 718 (1990) (“[G]overnment defendants who are immune from damages are subject to reparative and structural injunctions ordering them to undo the consequences of past wrongs.” (footnote omitted)). Disputes here are over how far courts may go in eradicating.164×164. See, e.g., Freeman v. Pitts, 503 U.S. 467, 506 (1992) (Scalia, J., concurring) (arguing that the passage of time necessarily implies the dissipation of a constitutional violation’s effects); cf. Levinson, supra note 16, at 1354–55. Similarly, the Court has recognized that remedying the effects of past discrimination (public or private) is a compelling interest, at least if identified with specificity.165×165. Parents Involved in Cmty. Schs. v. Seattle Sch. Dist. No. 1, 551 U.S. 701, 720 (2007); Shaw v. Hunt, 517 U.S. 899, 909 (1996). Here, the nuanced cases involve whether express government use of racial categories to craft such remedies survives strict scrutiny.166×166. See, e.g., Brest et al., supra note 25, at 1128–30. But the interest qualifies as compelling irrespective of whether the Court thinks a particular means of pursuing the interest passes constitutional muster.

These principles generate difficult questions at the margins. But incompletely theorized agreements on the general principle still can generate a sophisticated means of evaluating facially neutral governmental acts with discrete, specific past discriminatory predecessors.167×167. Cf. Cass R. Sunstein, Incompletely Theorized Agreements, 108 Harv. L. Rev. 1733, 1739 (1995) (discussing the phenomenon of “comfortable . . . agreement on a general principle” notwithstanding the possibility of “disagreement about [some] particular cases”). We need not specify the precise limits of proper governmental responses to what the Supreme Court calls historical “societal discrimination.”168×168. Shaw, 517 U.S. at 910. Nor is consensus necessary on how often governments may voluntarily use race-conscious curative measures. We can build on the wide agreement that specific past acts of wrongdoing can, but should not, extend themselves into the present and that governments should not be permitted to do indirectly what they cannot do directly.

Any nonarbitrary decision about how the law should treat time and continuity involves value judgments.169×169. See Balkin, supra note 156, at 262–63; Levinson, supra note 16, at 1315–16, 1375–76. See generally Kelman, supra note 152, at 593–95, 638–39, 666–67. Those whose judgments differ from mine may reach different conclusions. Here, though, more consensus exists than one might first think. Enough exists, at least, to recognize that the relationship between some modern government acts and some earlier discriminatory government acts may be uniquely relevant to the modern acts’ validity. This Article’s remainder takes up the task of defining and justifying that relevance as a constitutional antidiscrimination concept.

C. Two Types of (Potential) Taint

Before drawing on the foregoing principles to animate a better descriptive account of taint, it is worth pausing to highlight an important taxonomical point they facilitate. Every case in section I.C at least plausibly implicated legal continuity and the possibility of persistent past wrongful discrimination. Phrased that way, section I.C reveals an unappreciated genus of cases involving ostensibly discriminatory predecessor policies.

Yet section I.C also reveals an important species-level distinction within that genus. On one hand are cases like Ramos and Espinoza, where the T1 is quite far away, with complete decisionmaker turnover within the relevant institution. Call these “distant” cases. On the other hand are cases like Trump and Abbott, where the T2 closely follows the T1, often with substantial or complete decisionmaker identity within the relevant institution. Call these “compressed” cases.

Part of the problem any taint concept must resolve — and probably the source of some of the methodological muddle — lies in uncertainty about how the taint idea would apply to these related but distinguishable species. In distant cases, how powerful should the mere passage of time be? (For example, does a taint claim’s force necessarily dissipate with time?) In compressed cases, what does the compactness tell us — is it relevant only to potential, proscribed T2 intentions, or is there some sense in which the T2 situation can be questioned irrespective of intent?

In describing taint, then, I am sensitive to this distinction. Yet because the types exist within a genus, my proposal simultaneously recognizes that distant and compact cases manifest the same problem, on different scales. A unified genus-level approach is worth pursuing.

D. Describing Taint

I turn now to proposing an approach to discriminatory descent problems, with these principles and taxonomy in hand. The key first-cut need is a better definition of a tainted relationship. We can’t know how to treat taint until we agree on what we mean when we invoke it. I propose a less deferential approach that nevertheless preserves the possibility that a contemporary policy is legitimate despite relevant discriminatory history.

Relevance matters. By analogy, just as we do not treat the big bang (for example) as a legally relevant cause, not every public act that occurred before the putative T2 is relevant to a taint analysis.170×170. See Andrew Verstein, The Failure of Mixed-Motives Jurisprudence, 86 U. Chi. L. Rev. 725, 791 n.207 (2019) (observing that, depending on scope, “every feature of the universe [is] a but-for cause” (emphasis omitted)); Alexandra D. Lahav, Essay, Chancy Causation in Tort Law, J. Tort L. (forthcoming 2022) (manuscript at 3) (on file with the Harvard Law School Library) (observing that beyond causation’s “irreducible factual core: that an event has a tendency to produce an outcome,” legal causation reflects “policy choices used to achieve normative goals”). This distinction is important for nonidentical later-in-time policies. Take the example of the nonunanimous jury provision from Ramos.171×171. 140 S. Ct. 1390, 1394 (2020). It might be academically interesting to show that, given the right policy relationship, this past policy could inform responsibility determinations for the present one. But it would be almost entirely academic if the relationship became irrelevant were Louisiana merely to reenact the provision but change the number of jurors needed for a conviction from nine to ten.172×172. As Louisiana did. See, e.g., id. at 1426 (Alito, J., dissenting); State v. Hankton, 122 So. 3d 1028, 1038 (La. Ct. App. 2013). Such a trivial move should not be dispositive. But saying that much requires some sort of heuristic for ascertaining how much similarity taint requires.

An analogous problem from mootness law can help rationalize an approach. Some cases require deciding whether a supposedly new governmental policy is different enough from a policy it replaced that it moots litigation on the older policy. Thus, in a case challenging a city’s affirmative action ordinance, repealed and replaced with a similar ordinance on the same subject, the Court looked to the “gravamen” of the pending complaint and rejected mootness because the new ordinance “disadvantage[d]” the plaintiffs “in the same fundamental way.”173×173. Ne. Fla. Chapter of Associated Gen. Contractors of Am. v. City of Jacksonville, 508 U.S. 656, 662 (1993); see also id. at 660–63; id. at 662 n.3 (framing the question as whether the new ordinance “is sufficiently similar to the repealed ordinance that it is permissible to say that the challenged conduct continues”). The idea is that mootness doctrine would become an evasion doctrine if insubstantial changes extinguished any challenge to a policy’s substance.174×174. See id. at 662; see also Am. Freedom Def. Initiative v. Wash. Metro. Area Transit Auth., 901 F.3d 356, 361–63 (D.C. Cir. 2018) (finding no mootness where the plaintiffs were disadvantaged in the “same fundamental way,” id. at 362, under a new advertisement policy); 13C Charles Alan Wright et al., Federal Practice and Procedure § 3533.6, at 281 & n.32 (3d ed. 2008). Similarly, in the search for a tainted relationship, we can look to the gravamen of the policies and to the sorts of claims plaintiffs at either time might have raised.175×175. Cf. Levinson, supra note 16, at 1372–73 (arguing that, in the public law context, questions of proper “aggregation [of harms and benefits] over time . . . usually take the form of a nexus or germaneness test” and offering examples).

Edge cases are inevitable. Those are the price of an antidiscrimination approach that responds to the reality that wrongful discrimination is initially undertaken for a reason that may persist through time.176×176. See, e.g., United States v. Fordice, 505 U.S. 717, 747 (1992) (Thomas, J., concurring) (“[D]iscriminatory intent does tend to persist through time . . . .”); Strauss, supra note 26, at 956 (“Understanding what is wrong with discrimination may require some tolerance for relatively vague notions; there is no guarantee that the prohibition against discrimination can be expressed in a way that is not somewhat vague.”). But it is workable to ask whether the T2 policy acts upon, or operates within, the world in the same way as the supposed T1. That is, we should ask whether an intertemporal policy relationship’s characteristics demonstrate the kind of distinctive continuity that the law often treats as relevant to responsibility.177×177. I generally speak in terms of distinctive continuity, but not much would change if this were viewed in causation terms, that is, as an argument about what we ought to recognize as the constitutionally pertinent results of a discriminatory predecessor. See Lahav, supra note 170 (manuscript at 3).

Three factors apparent in section I.C’s cases can guide that inquiry. Their presence distinguishes mere post hoc temporal relationships from those suggesting responsibility-laden policy continuity.

An important threshold note: I postpone to section V.A.2 the possibility that the government might dispute the original policy’s wrongfulness. Assuming (I think reasonably) that some older policies are wrongful, the first question is whether we can detect meaningful continuity between such policies and modern successors. The second question is whether and to what extent such continuity drives constitutional responsibility. To be sure, if we can answer those antecedent questions, the resolution of disputes over the nature of original policies may affect the set of cases when taint is relevant. But the set will not be empty.178×178. As section V.A.1 notes, if the original policy’s wrongfulness is implausible, a court would properly decline to entertain a claim premised on a taint theory.

1. Same Function. — Drawing on the mootness analogy, tainted relationships require the subsequent government act to carry forward the predecessor’s functional operation. The easiest example is the T2 reenactment of an exact duplicate. Such a case is essentially indistinguishable from a provision untouched since its inception, maintaining its effects.179×179. See, e.g., Hunter v. Underwood, 471 U.S. 222, 227–30, 232–33 (1985); see also supra note 64. It is easiest to see here how the technically new law “brings the old soil with it.”180×180. Cf. United States v. Davis, 139 S. Ct. 2319, 2331 (2019) (quoting Sekhar v. United States, 570 U.S. 729, 733 (2013)) (discussing meaning implications of transplanted statutory phrasing). I do not claim these cases are common. Note, though, that this already illuminates the old futility objection from Palmer v. Thompson181×181. 403 U.S. 217, 225 (1971). : a law repassed in identical form for “different reasons” may be distinguished from an identical, clean-slated new law precisely because it was (1) repassed (2) in that form.182×182. See id.

While it surely is not difficult to avoid enacting an exact duplicate, the extreme case is a small step from acknowledging that a nonidentical later policy may functionally replicate a predecessor’s real-world operation. Indeed, we expect legislatures to update and amend laws because the original law’s impetus may, in the main, persist. A realist approach will examine the benefits and burdens the provisions distribute, and the way in which they do so.183×183. See Ne. Fla. Chapter of Associated Gen. Contractors of Am. v. City of Jacksonville, 508 U.S. 656, 662 n.3 (1993) (inquiring whether the “new ordinance is sufficiently similar to the repealed ordinance” that it presents the same controversy). Thus, the nonunanimous jury rule from Ramos functionally duplicated the original nonunanimous rule, notwithstanding minor changes.184×184. The reenactment reduced the number of allowable dissenting jurors from three to two and expanded the carveout for cases requiring unanimity. See, e.g., Ramos v. Louisiana, 140 S. Ct. 1390, 1426 (2020) (Alito, J., dissenting); State v. Hankton, 122 So. 3d 1028, 1038 (La. Ct. App. 2013). So too for the North Carolina voter ID law passed after a previous ID law’s invalidation.185×185. See N.C. State Conf. of the NAACP v. Cooper, 430 F. Supp. 3d 15, 26 (M.D.N.C. 2019) (citing N.C. State Conf. of the NAACP v. McCrory, 831 F.3d 204, 216 (4th Cir. 2016)).

A hypothetical case may elucidate this factor’s limits. Suppose that, a few years after Ramos, Louisiana gives prosecutors double the peremptory strikes defense attorneys have.186×186. This would alter the current Louisiana framework, which allots both sides the same number of peremptory strikes. La. Code Crim. Proc. Ann. art. 799 (2021) (giving both sides same number). Insofar as nonunanimous juries “allow[ed] backdoor and unreviewable peremptory strikes,”187×187. Ramos, 140 S. Ct. at 1418 (Kavanaugh, J., concurring); see also id. at 1417–18. continuity might seem plausible, especially given prosecutorial strikes’ racially disparate effects.188×188. See Frampton, For Cause, supra note 107, at 786–88. Nevertheless, although the setting is suspicious and probably warrants a conventional equal protection challenge, the policies are insufficiently similar for taint. The first diluted minority-juror influence. The second (by hypothesis) concedes a minority juror’s ability to torpedo a guilty verdict but affords prosecutors more front-end power to affect jury composition. Despite potential overlap in ultimate effect, the policies do not operate, and jurors are not “disadvantage[d,] . . . in the same fundamental way.”189×189. Ne. Fla. Chapter of Associated Gen. Contractors of Am. v. City of Jacksonville, 508 U.S. 656, 662 (1993). This is not to make this factor coextensive with mootness, but simply to draw on judicial treatment of similarity in that context to illuminate this one.

In short, continuity in real-world operation and effect confirms subject-matter continuity. To be sure, judgment is required at the margins. But, as the mootness doctrine from which it borrows recognizes, flexible judgment calls are the price of a doctrine that responds to real-world complexity. Understood that way, and once fleshed out, substantial similarity is neither novel nor unduly manipulable. This Article’s concept of taint requires this sort of continuity.

2. Continuity and Institutional Responsibility. — Another key is the actor’s identity. The easiest cases, involving persistent individual executive actors,190×190. See Flowers v. Mississippi, 139 S. Ct. 2228, 2245–46, 2250–51 (2020); see also id. at 2234, 2236 (emphasizing the continuous presence of the lead prosecutor); id. at 2251 (Alito, J., concurring) (same). do not exhaust the set. The problems of evasion and lingering effects of past wrongdoing do not disappear simply because multimember bodies have inconstant personnel.191×191. In any event, the multi-actor composition of many “executives” makes them “theys” rather than “its” on a realist approach. Cf. Kenneth A. Shepsle, Congress Is a “They,” Not an “It”: Legislative Intent as Oxymoron, 12 Int’l Rev. L. & Econ. 239 (1992). And not even individual persons are continuous over time, which could theoretically raise T1/T2 responsibility allocations in the same way that multimember-body discontinuity does. See, e.g., Levinson, supra note 16, at 1373 n.211. An institutional realist approach suggests that institutional continuity helps explain how identity matters, as well as the nature of the responsibility at T2. Taint recognizes responsibility as connected not just to individual decisionmakers but to the institutions of which they are a part.

Here, it helps to disentangle two concepts of responsibility — “blame responsibility” and “task responsibility.”192×192. See Stilz, supra note 143, at 194 & n.11. This approach to responsibility often surfaces in nuanced moral philosophy debates over the concept of “moral taint.” See, e.g., Marina A. L. Oshana, Moral Taint, 37 Metaphilosophy 353, 359 (2006). This piece aligns with defenders of some version of that concept insofar as they reject the claim that no one can be responsible (in any sense) for actions over which they never exercised control. See, e.g., David Silver, Collective Responsibility, Corporate Responsibility and Moral Taint, 30 Midwest Stud. Phil. 269, 271 (2006). That said, this Article is a legal intervention rooted in legal approaches to responsibility and culpability. I draw on blame and task responsibility to provide a vocabulary for that intervention, not to endorse a particular side in that broader philosophical debate. It is beyond this Article’s scope to take sides on, for example, the aptness of “shame” here. See Oshana, supra, at 373–74 (noting, additionally, other disagreements in the philosophy literature). The former recognizes responsibility in a sense that incorporates an actor’s “moral fault or virtue.”193×193. Stilz, supra note 143, at 195. In this classic sense, one is responsible because one’s blameworthy behavior caused harm. Conversely, task responsibility concedes the actor’s moral innocence, yet nevertheless recognizes some responsibility by virtue of some other nonblameworthy behavior they have engaged in. For example, someone who voluntarily joins an organization or group may have task responsibility for harms the organization causes irrespective of individual blame, just as they would share the fruits of organizational success that they did not individually cause.194×194. See id. at 194–95.

This distinction animates a realist approach. Assume arguendo that formal policy separation and nonidentity (characteristic of the discriminatory predecessor fact pattern) preclude imputing identical blame responsibility at T1 and T2. Nevertheless, even when political decisionmakers have changed, new actors have opted into a going concern that has taken action, secured benefits, and incurred liabilities tied to the institution, not its constituent individuals.195×195. A supportive analogy is the classic justification for the potentially infinite transmission of previously incurred debt “to successor governments and states” — “those who follow . . . enjoy . . . the benefit of the credit that [was] extended to their predecessors.” Lee C. Buchheit et al., The Dilemma of Odious Debts, 56 Duke L.J. 1201, 1210 (2007). Importantly, this opting in granted T2 actors control over whether to hew to or break from the T1 policy. That control, considered within the institution’s historical context (including T1), demonstrates institutional continuity.

That continuity, in turn, reveals task responsibility associated with any persistent, T1-rooted harms that T2 causes.196×196. See, e.g., John M. Parrish, Collective Responsibility and the State, 1 Int’l Theory 119, 127–28 (2009) (linking nonblame responsibility to “participation in a chain of social connections that helps to create or sustain [a] negative outcome,” id. at 127). Whatever might have been the case absent reenactment, we can presume that a T2 reenactment was undertaken to carry forward some of T1’s perceived benefits. Institutional embrace of the sweet of the past properly travels with limits on action based on responsibility for the bitter.

The fact of control also supports some assignment of blame based on the T2 actors’ active choice to reaffirm T1’s operative core. Consider, again, an analogy to conspiracy law, which sometimes holds B responsible for blameworthy acts by A that B did not intend.197×197. Robinson, supra note 152, at 665–66. That legal choice can be justified on the ground that B’s less-blameworthy intentional act of conspiring with A may be “added” to B’s recklessness or negligence regarding the possibility that A might engage in more-blameworthy acts in relation to the conspiracy.198×198. Id. at 666. Similarly, here, we may grant that the T2 institutional decisionmakers are less blameworthy than their predecessors. Yet we might still legitimately assign some blame responsibility based on the combination of the less-blameworthy intentional reaffirmation and their negligence as to the nature and continued effect of the T1 policy.

The key question, then, is whether the policy relationship reflects the sort of institutional continuity that justifies recognizing past-rooted responsibility. Taint requires the contemporary actors’ active participation in the institution that created T1, and specifically in the reaffirmation of T1’s function.

This approach recognizes that it may be appropriate to constrain institutions composed of multiple transitory individuals based on what those individuals’ institutional predecessors did in the institution’s name.199×199. See supra section II.A, pp. 1212–15. That analysis captures turnover in multimember bodies. Thus, we can still perceive institutional continuity in Ramos, notwithstanding T1 and T2’s three-quarter-century separation. This approach also recognizes personal continuity. Even though, like institutions, individuals change over time, common sense tells us that a District Attorney’s continuous presence, as in Flowers, is germane. So, too, we can find evidence of continuity where “a majority of the [same party’s] legislators who voted for [the previous law] voted for [the new law.]”200×200. N.C. State Conf. of the NAACP v. Cooper, 430 F. Supp. 3d 15, 31 (M.D.N.C. 2019). But not everything germane is indispensable.

Before proceeding, it is worth emphasizing the harder questions this framing need not answer. T2 decisionmakers’ voluntary entry into, and action within, a particular organizational relation pretermits the possibly harder (yet important) question of when we ought to impute responsibility absent that level of volition.201×201. Cf. Fred O. Smith, Jr., The Other Ordinary Persons, 78 Wash. & Lee L. Rev. 1071, 1078 (2021) (noting the possibility that “[o]ne generation can become complicit” in a past generation’s wrongdoing). And because the T1/T2 pattern includes some functional reenactment at T2, the account pretermits questions of whether some reparative responsibility would attach even absent functional reaffirmation of T1 (and if so, how much). That is to say, while plenty exists to argue about on this account, it nevertheless presents a relatively easier case, especially given the plausible synergy of blame and task responsibility in these fact patterns.

In short, the origin story matters to a claim that the connection between T1 and T2 shapes T2’s validity. Part of that story includes individuals, but when governmental responsibility is involved, the institution is a necessary part of that connective tissue. Continuity-based responsibility — in the sense that transcends individual blameworthiness — thus properly requires a single institutional lineage. That lineage, in turn, helps justify tying current institutional and policy responsibility to discriminatory policy predecessors. The justification embraces some imputation of responsibility to less responsible actors who, for example, reaffirm a past problematic policy. But, importantly, taint also invokes the idea of never-terminated institutional responsibility, revealed in institutional continuity.

3. Related Events. — Finally, given my emphasis on temporal maximalism, it is worth considering how related events may inform the taint determination.

Starting with intervening events, an expansive approach recognizes that an event can reveal something important about the relationship between events that it bisects. Here, specifically, an event that postdates and relates to an earlier event may help show the first event’s connection to an event that postdates them both. Unsurprisingly, temporal closeness often drives public law responsibility determinations.202×202. See, e.g., Rochon v. Gonzales, 438 F.3d 1211, 1220 (D.C. Cir. 2006) (Title VII); Liaquat Ali Khan, Temporality of Law, 40 McGeorge L. Rev. 55, 72–74 (2009) (collecting examples).

A representative example is litigation, or its threat, at “T1.5.” Consider, on that note, a few instances in which the state’s T2 action in relation to an intervening event creates evidence of a T1/T2 connection.

Most straightforward are what one might call postruling changes, where the government shifts to a same-subject T2 after litigation generates an unfavorable ruling on the merits of a putative T1. In Trump, the alterations to travel bans after losses in lower courts would fit this bill. So too if the city in Palmer had lost and then in fact attempted to reenact a pool-closure policy (perhaps closing some other public facilities to create a patina of difference).203×203. Cf. Palmer v. Thompson, 403 U.S. 217, 225 (1971) (speculating about a reenactment for “different reasons”). In both cases, it makes sense to understand the final travel ban, or the new pool closure, differently than if they had emerged absent litigation.

Changes pending a ruling illuminate for different reasons. Governments routinely, while litigation is pending and sometimes before any merits ruling, announce a change in the challenged policy.204×204. See, e.g., Joseph C. Davis & Nicholas R. Reaves, The Point Isn’t Moot: How Lower Courts Have Blessed Government Abuse of the Voluntary-Cessation Doctrine, 129 Yale L.J.F. 325, 329–32 (2019). They generally then assert mootness, and sometimes succeed.205×205. See id. Even if some of these cases are indeed moot, that need not sever continuity for taint purposes. Mootness doctrine’s attention to similarity helps confirm its usefulness for conceptualizing taint, but the two are not coextensive. It may be that changes both moot the case and sever any degree of constitutionally relevant connection, but maybe not. For our different purposes — locating continuity bearing on responsibility — changes pending a ruling are powerful evidence.

Events entirely external to past or pending litigation can also inform a continuity determination. Think of demand letters, negative press framing a practice as illegal, or a court ruling on materially similar facts. Such events raise a concrete, imminent threat of impending litigation on an issue.206×206. As spoliation doctrine recognizes. See, e.g., Kronisch v. United States, 150 F.3d 112, 126–27 (2d Cir. 1998). If the jeopardized policy is amended or “replaced” with a policy that functions similarly, the intervening event of threatened litigation is a key piece of information binding the policies together. This discussion of litigation-related changes is not exhaustive. What illuminates a particular relationship will depend on the policies involved.

To be clear, related events are one factor. Without more, altering policies in response to litigation is not inherently suspect, let alone necessarily fatal to the new policy. This factor must be understood in connection with the focus on function. A new policy that functions in a fundamentally different way is outside the scope of taint, regardless of its relation to litigation. When a relation to litigation does exist, however, we reasonably treat a policy differently if it also preserves the earlier policy’s operative core. And, in close same-function cases, related events can properly tip the scales in taint’s favor. Finally, as Part IV explains, like any tainted policy, a policy tainted under these circumstances can still be justified.

Related events can cut both ways. Expanding the scope of relevance beyond immediate temporal context may sweep in intervening events that generate an anti-taint inference. Thus, it could sever continuity if an independent institution concluded for different, newly arising reasons that the sort of policy at issue is beneficial and nondiscriminatory.

Finally, I do not wish to discount the possibility that a preexisting institutional tradition of pursuing a facially legitimate policy embodied in a T1/T2 relationship could weaken the negative inferences the relationship generates.207×207. This idea draws on Professors Samuel Issacharoff and Trevor Morrison’s defense of increased deference to the government where a challenged policy is “facially within the bounds of long-settled authority,” Samuel Issacharoff & Trevor Morrison, Constitution by Convention, 108 Calif. L. Rev. 1913, 1941 (2020), and reflects “systematic practices that have defined” government operation “over a prolonged period of time,” id. at 1920. Such a tradition could, in a sense, be thought of as “good taint.” Still, when this tradition is genuinely preexisting, such that the problematic T1 policy has the closest nexus to the modern policy, such a tradition should be unlikely to sever otherwise demonstrable continuity. This does not mean a preexisting salutary tradition does not matter — as I discuss later, it bears on whether a government can “purge” taint once detected.208×208. See infra section IV.B, pp. 1238–45.

* * *

Summing up: Discriminatory predecessors raise questions about whether and when responsibility for past government acts constrains modern government behavior. We cannot answer those questions with a view from nowhere. But we can answer them against the backdrop of widespread legal recognition that continuity informs meaning and responsibility, and in light of widely accepted antidiscrimination commitments. Those principles permit distinguishing legally relevant predecessors of Louisiana’s modern nonunanimity rule (for example, its initial 1898 enactment) from the legally irrelevant (the Louisiana Purchase). Continuity in subject, effect, institution, and as inferred from surrounding events intelligibly identifies when the relationship of an earlier, wrongful policy and a successor embodies a taint.

Taint, defined this way, dictates no outcomes. The ability to distinguish tainted progeny from an otherwise identical untainted policy helps justify distinctions in responsibility allocations. But circumstances and people change, governments should correct missteps, and correction may well involve a policy with some resemblance to a discriminatory predecessor. Indeed, the idea of correction as a governmental responsibility undergirds this entire Part. Unfortunately, changes can also represent faux correction that launders or insulates past wrongdoing. Part IV’s elaboration of how to treat taint aims, in part, to help distinguish faux and legitimate correction. The point so far is that an antecedent task is distinguishing the tainted from the untainted.

This Part began by observing that the alternatively minimalist and maximalist judicial approaches to the discriminatory predecessor problem surface an old tension between institutional formalism and realism. This proposal attempts to attend to both perspectives’ valid concerns, settling closer to the realist pole. The standard thus has some play in the joints. Still, contra usual concerns about realism devolving into arbitrary free-for-alls,209×209. Cf. Pildes, supra note 23, at 8–9, 39 (noting “predictable rule-of-law questions,” id. at 39, associated with realism). realist pursuit of legally relevant continuity can be rigorous. We need not have finally resolved whether formalism or realism is better to believe that in particular settings a realist approach best responds to the current problems.210×210. See id. at 53–54.

It is possible to achieve a better concept of constitutional continuity for antidiscrimination law. The taint concept offers part of the solution. And though taint is not the final word, understanding what it is will help guide our thinking about what it looks like to purge it.

III. Taint as Trigger and Wrongful Discrimination

Part I explained that muddled approaches to the discriminatory predecessor problem highlight the need for a framework that advances how we treat these instances. Part II offered the first piece of that solution by sketching a temporally maximalist, institutionally realist approach to detecting meaningful policy similarity across time. When detected, we can call that relationship “taint.” But staging the analysis this way isolates the next questions: What is taint? And what are its implications?

This Part begins to answer those questions by explaining what kind of concept taint is, and what effects that concept entails. Taint is a relational fact that can, all else equal, be evidence of T2 pretext. But it is far more significant in its role as a constitutional concept that triggers a specific method of detecting the violation of constitutional antidiscrimination principles. As such, it intervenes in broader conversations about the meaning, and detection, of wrongful discrimination. Finally, this Part’s conceptual work lays the foundation for Part IV’s prescription for taint’s practical operation.

A. Beyond Pretext

Before defending my more ambitious vision of taint, I want to start with what I hope will be common ground on its utility. Some policy is pretextual.211×211. See Robin Charlow, Tolerating Deception and Discrimination After Batson, 50 Stan. L. Rev. 9, 20 (1997) (defining pretext and connecting it to intentionality). That recognition often animates judicial approaches.212×212. See Denning & Kent, supra note 161, at 1780–84. All else equal, and especially in compressed cases, taint raises the likelihood that a facially neutral T2 is in fact a bad faith attempt to launder an illegitimate T1. The concept is thus a quintessential anti-evasion tool,213×213. See id. at 1776. and it is therefore independently useful to any search for conscious malintent in a world where such intent can be subtle and cloaked behind policy change over time.

Limiting taint’s role to smoking out specific bad intent, however, would raise some reasonable critiques, such as the fear of false positives. Surely some cases of descriptive taint will in fact not stem from specific present-day proscribed intent. And the strength of an inference of bad specific intent likely decays over time. This limitation would also be vulnerable to the long-running ontological and epistemic critiques of talking about bad intent in a world of multimember bodies and nonobservable mental states.214×214. See Schwartzman, supra note 20, at 203–04; Landau, supra note 1, at 2156–57.

To be sure, pessimists regarding our society’s expungement of present-day bad intent may deem a flood of false positives unlikely. And concerns about intent’s artificiality have not deterred courts or the person on the street from invoking intent as a useful concept.215×215. See W. Kerrel Murray, Populist Prosecutorial Nullification, 96 N.Y.U. L. Rev. 173, 212 (2021). But these critiques offer some reason to go beyond pretext. That is, taint justifies a decision rule that does not require specific proscribed intent. This is the more controversial claim, but our descriptive foundation can lend it some support.

B. Constitutional Decision Rules and Wrongful Discrimination

The broader understanding of taint is best characterized as a trigger for a targeted antidiscrimination “decision rule.”216×216. Mitchell N. Berman, Constitutional Decision Rules, 90 Va. L. Rev. 1, 9 (2004). Under this rubric, borrowed from Professor Mitchell Berman, the distinction is between rules that define the “proper meaning” of a constitutional provision and rules that define “how to decide whether” a provision has been complied with.217×217. Id.

Applying this distinction helps show what taint is not. It is not, for example, an independent constitutional violation. Nor is it a conclusive demonstration of such a violation. Nor does it depend on the particular content of the underlying violation that it may help show (which is why it can work as a transsubstantive concept). Rather, taint is shorthand for a type of intertemporal continuity that warrants thinking differently about whether and why we should assign constitutional responsibility at T2. That different way of thinking is a taint-specific method for determining whether a constitutional antidiscrimination provision has been violated.218×218. See id. at 9–10; Kermit Roosevelt III, Constitutional Calcification: How the Law Becomes What the Court Does, 91 Va. L. Rev. 1649, 1657–58 (2005).

The remainder of this Part fills out the decision rule’s content. Specifically, it argues that the rule is best thought of as an objective one that rejects any need, at T2, for specific intent to harm or otherwise act unconstitutionally.

Understanding the scope of this intervention requires situating taint within longer-running debates over how the law detects proscribed discrimination.219×219. See generally Eidelson, supra note 24; Hellman, supra note 24; Lawrence, supra note 25; Strauss, supra note 26. See also Frank I. Michelman, Reasonable Umbrage: Race and Constitutional Antidiscrimination Law in the United States and South Africa, 117 Harv. L. Rev. 1378, 1383 n.30 (2004). Those debates stem from the reality that at least any nonuniversal government action (and some universal ones) “discriminate” in the sense that they draw distinctions or create disparate effects among people.220×220. Hellman, supra note 24, at 13–14, 27; see Gowder, supra note 20, at 1030–31 (suggesting that all legislative acts are nonuniversal in the sense that “they indicate conditions for the application of law,” id. at 1031, that will not always be met). We thus need a way to identify which of those distinctions are “wrongful discrimination” and which are not.221×221. See, e.g., Gowder, supra note 20, at 1049–50. For present purposes, the contestants fall into two camps, although considerable nuance exists within them.

One perspective treats something akin to specific intent (either to harm or to classify in some inherently wrongful way) as a necessary prerequisite to showing wrongful discrimination.222×222. See, e.g., Kennedy, supra note 62, at 1419–21; Siegel, supra note 25, at 1279. But see infra section III.C, pp. 1232–34. A lurking complexity stems from the modern Court’s tendency to treat the mere use of race as a constitutional problem requiring justification, even absent any alleged intent to harm. See, e.g., Haney-López, supra note 25, at 1781–89. That approach does not obviously flow from doctrinal formulations focused on purposeful adverse effects. E.g., Pers. Adm’r v. Feeney, 442 U.S. 256, 279 (1979). Elaborating whether, and how, that colorblindness approach can be reconciled with cases like Feeney is beyond this Article’s scope. For present purposes, it suffices to note that the Court’s approach in such cases evinces an already existing judicial ability to understand flexibly wrongful discrimination’s relation to intent. It is theoretically possible to apply this approach to collective bodies.223×223. See Fallon, supra note 24, at 537–38 (advancing a coherent concept of collective legislative intent, but noting its difficulties in the area of “forbidden” intent). But, because no mental state (individual or collective) is directly observable, this approach can devolve into a search for smoking guns like outlandish avowals of racism or a disparate effect explainable only by specific intent.224×224. Classic examples of such effects, although preceding modern equal protection doctrine, include Gomillion v. Lightfoot, 364 U.S. 339, 341 (1960); and Yick Wo v. Hopkins, 118 U.S. 356, 374 (1886).

The primary alternative contains multitudes, all of which consider the specific-intent vision inadequate, and all of which (generally) look to effects to guide the decision rule.225×225. See, e.g., Strauss, supra note 26, at 941–45. Some alternatives are framed primarily by their advocates as offering an alternative vision of the underlying constitutional meaning. No surprise; the distinction between decision rules and the underlying “operative propositions” can be murky.226×226. Berman, supra note 216, at 79–89; Roosevelt, supra note 218, at 1655; see also Henry P. Monaghan, The Supreme Court, 1974 Term — Foreword: Constitutional Common Law, 89 Harv. L. Rev. 1, 32–33 (1975) (exploring the “differences of degree,” id. at 33, involved in similar attempted distinctions). Even if these alternatives speak more in the register of constitutional meaning, they all need a theory of the appropriate decision rule that excludes reference to specific intent.

So, for example, for some the essential problem is “selective indifference”227×227. See, e.g., Paul Brest, The Supreme Court, 1975 Term — Foreword: In Defense of the Antidiscrimination Principle, 90 Harv. L. Rev. 1, 7–8 (1976) (“racially selective sympathy and indifference,” id. at 7); Michael Klarman, An Interpretive History of Modern Equal Protection, 90 Mich. L. Rev. 213, 263–64 (1991); Lawrence, supra note 25, at 347–49; Strauss, supra note 26, at 1007. to burdens or harms to one group that the state would not permit for a socially preferred group.228×228. See, e.g., Lawrence, supra note 25, at 348–49 (envisioning decisionmakers who, unconsciously, “would have weighed the costs and benefits differently” if the harm had fallen on White people, id. at 348, and decisionmakers who never thought about race one way or the other and nevertheless “inadvertent[ly] devalu[ed]” Black persons’ interests, id. at 349); Strauss, supra note 26, at 1007 (contemplating “legislators . . . free of conscious discrimination” who “may nevertheless be better able to sympathize with the suffering of members of their own race than with the plight of [another racial] group”). Specific intent is unnecessary, although this framing would seem to require some alternative-history speculation to conclude that a harm-generating decision would have been made otherwise if the harm fell on another (in practice, the dominant) group.229×229. See Klarman, supra note 227, at 298–99.

Another view focuses on ascertaining demeaning social meaning.230×230. See Eidelson, supra note 24, at 6–9; see also Hellman, supra note 24, at 29, 33. Here again, the focus is not on intent, but on whether the act “expresses the appropriate valuations of (that is, attitudes toward) persons.”231×231. Elizabeth S. Anderson & Richard H. Pildes, Expressive Theories of Law: A General Restatement, 148 U. Pa. L. Rev. 1503, 1504 (2000). This approach needs a decision rule for determining the reasonable and conventional understanding of the action.232×232. See id. at 1512–13; Eidelson, supra note 24, at 7; Gowder, supra note 20, at 1036, 1043–44. The possibility of contesting an action’s expressive character, rooted in social context, can pose a problem for this view.233×233. See Hellman, supra note 24, at 33, 74; Anderson & Pildes, supra note 231, at 1570.

Another view that rejects intent might be dubbed “unintentional pretext.”234×234. Cf. Charlow, supra note 211, at 20 (noting intentionality of conventional pretext). Consider that some decisionmakers who genuinely believe they are acting for legitimate reasons may have unconsciously engaged in motivated reasoning or self-deception.235×235. See, e.g., Dan M. Kahan, The Supreme Court, 2010 Term — Foreword: Neutral Principles, Motivated Cognition, and Some Problems for Constitutional Law, 125 Harv. L. Rev. 1, 19–26 (2011); Pozen, supra note 26, at 934–36. Such behavior differs from classic pretext insofar as decisionmakers mislead not just others but themselves.

These approaches may overlap. For example, selective-indifference evidence could include what the action and its effects communicate. They are also not coextensive. The selective-indifference and expressive-meaning approaches might reach cases where the actor’s mind is more empty than deluded (contra “unintentional pretext”). And one might imagine an action impugned by its social meaning even without sufficient evidence that decisionmakers would have acted differently had the burden fallen on a dominant group.

More importantly, these approaches unite in rejecting conscious intent as the essential quality of constitutionally proscribed discrimination and in needing a decision rule focused on objective facts about the world.236×236. See, e.g., Burton v. Wilmington Parking Auth., 365 U.S. 715, 725 (1961); Anderson & Pildes, supra note 231, at 1523–24; Pozen, supra note 26, at 934–35; Strauss, supra note 26, at 946, 961–62. They thus track judicial interpretation of the First and Fourth Amendments. See, e.g., Citizens United v. Fed. Election Comm’n, 558 U.S. 310, 340 (2010) (First); Whren v. United States, 517 U.S. 806, 813 (1996) (Fourth).

* * *

To sum up: One could hinge constitutional antidiscrimination decision rules on specific, subjective, discriminatory intent. Even if one acknowledges that such intent must be inferred, when facially neutral policies are at issue, this worldview pushes toward invalidating only those policies impeached by smoking guns.237×237. See Haney-López, supra note 25, at 1790, 1837–39 (discussing inferences of government “innocence” from benign context). Alternately, we might recognize the relevance of subjectivity but emphasize that wrongfulness need not depend on it.238×238. See, e.g., Clarke, supra note 17, at 517.

Taint can intervene in this old debate. Recall that none of Part II’s indicia of legal continuity require specific intent at T2 (although, as section A notes, they may be relevant to pretext). The ability to describe taint this way supports understanding the decision rule it generates as not requiring specific intent.

C. Against Specific Intent

To reiterate, taint can be useful even understood narrowly as a mere auxiliary of an inquiry into specific intent. But the descriptive account can justify a broader view — one that dovetails with the just-discussed alternative approaches to demonstrating wrongful discrimination.

Begin with some truths about any wrongful T1.239×239. I consider the import of the government possibly disputing the T1 policy’s wrongfulness in section V.A.2, pp. 1261–64. Any T1 had some actual effect in the world. Here, I mean any effect, not just disparate effect on the basis of some trait. For example, some would cast Palmer as an example of evil intent without disparate racial effect, because the city’s public-pool closure made such pools unavailable to everyone.240×240. See, e.g., Klarman, supra note 227, at 296 (noting that “[the closure’s] impact seemed nondiscriminatory”). Even if one says the closures had no disparate racial effect (which I do not), denying public pools to everyone is a cognizable real-world “effect.”241×241. I am not sure exactly what a genuine bad-intent-with-zero-effect case looks like, but it is outside this Article’s scope.

Recognizing that T1 did something in the real world supports the next point. The descriptive framework offers a vocabulary for describing when an intertemporal policy connection matters more for meaning and responsibility than for the mine-run relationship between two policies. Thus it looks to continuity in function, real-world effect, and institutional actor, as well as inquiring into whether any related events inform that connection.242×242. See supra section II.D, pp. 1218–27. That approach provides both a factual basis and a reason for concluding that T1 ought to inform T2 responsibility determinations, much the same way that, for example, continuity between corporations is a reason for concluding that the earlier corporation’s criminal liability ought to pass to the later.243×243. See Diamantis, supra note 148, at 4; see also supra section II.A, pp. 1212–15 (outlining examples of continuity-connected responsibility). And it can do so without ever requiring reference to T2 “intent.”

This conclusion is not compelled. Saying that taint “can” support a view is not to say that it must. That said, consider the key facts of the relevant situation. A T1 policy, with real-world effects, constituted wrongful discrimination. Now, at T2, we have a policy connected in a uniquely close way to that earlier policy, including in its real-world function. We must determine whether the second-in-time policy complies with constitutional antidiscrimination mandates. Should the descriptive fact of taint matter to the decision rule, and if so, how much?

It is not incoherent to say that taint matters merely as evidence of specific proscribed intent, such that the mere passage of time would render it progressively irrelevant. In my view, however, the principles shaping the descriptive account reveal why the better decision rule cannot be reduced to motivation. It cannot, for the same reasons that district courts must go beyond enjoining illegal activity to eradicating the past’s discriminatory effects, and that state actors have a compelling interest in remedying the effects of past public or private discrimination.244×244. See supra section II.B, pp. 1216–17. In short, past harm can extend into the present in judicially cognizable ways, and the state can and should address the past’s problematic persistence. Those principles animate taint, and a decision rule focused on T2 bad intent gives them short shrift.

Taint, in other words, cuts through arguments over the general value of a specific-intent decision rule. Whatever might be said for or against such a rule generally, it is particularly inapt here if one buys the descriptive account of taint.245×245. This is not to deny specific bad intent’s relevance. It is simply to insist that tainted situations demonstrate specific intent’s failure, without more, to provide a satisfactory, comprehensive account of wrongful discrimination. See, e.g., Strauss, supra note 26, at 962. To be sure, this proposal steps beyond a judicial decree eradicating the discriminatory effects of the specific activity enjoined. But the step is small. And treating a specifically identified relationship with a discriminatory predecessor as relevant, irrespective of T2 subjective intent, is consistent with the compelling-interest status of remedying the effects of discrimination. Slippery-slope critiques of the supposedly unbounded results of recognizing amorphous “societal discrimination” are thus off point here.246×246. See, e.g., Boddie, supra note 16, at 1279–81 (citing, inter alia, the slippery-slope argument in Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265 (1978)).

The foregoing points reveal taint’s intervention within broader debates over motive in antidiscrimination law. Where taint exists, the specific-intent framework lacks the vocabulary to describe fully what matters. Here, it is unusually easy to see what the broader view’s advocates generally maintain: our evaluative judgments of people and our evaluative judgments of their actions can be treated independently.247×247. Gowder, supra note 20, at 1050. Importantly, in triggering a more expansive decision rule, taint need not embrace conclusively any particular broader locution of the underlying constitutional problem. In every case, taint fits comfortably into a view that focuses on what the government act effects rather than what the government actors intended.

For example, taint is probative of selective indifference because we know that a past, substantially similar constellation of facts did in fact constitute wrongful discrimination. That may not be conclusive, as Part IV emphasizes. But, on the selective-indifference view, we would want all information relevant to indifference, and taint provides important information even absent pretext. Indeed, taint helps regiment the selective-indifference approach by reducing the need for hard-to-know historical counterfactuals.248×248. See Strauss, supra note 26, at 972–73.

Taint also fits well into approaches focused on expressive or social meaning. If the question is how we should interpret a particular set of facts pertaining to a government act, a history of this institution wielding this sort of act wrongfully is important, regardless of modern intent. Again, taint can reduce the need for speculation (here, about the action’s social meaning).249×249. See Gowder, supra note 20, at 1039–40 (noting this common critique).

Finally, recognizing taint helps reveal likely cases of unintentional pretext. After all, in a world without truth serum, speculating about what someone would “really think if they were honest with themselves” can be a dangerous game. But a rigorously identified tainted relationship gives us purchase — an objective reason to suspect that someone might in fact be engaged in genuinely successful self-deception.

D. The Taint Trigger and Doctrine

A few words are warranted on doctrinal fit. To be clear, if necessary, taint should be understood as a uniquely justified exception to the normally obtaining decision rules in antidiscrimination cases. One can make a case for its compatibility, however.

As noted, it is generally said that the Court’s discriminatory purpose doctrine makes conscious intent to harm necessary for an equal protection violation.250×250. See, e.g., Haney-López, supra note 25, at 1783, 1795–96; Klarman, supra note 227, at 298; Siegel, supra note 25, at 1279. The most restrictive language comes from Personnel Administrator v. Feeney251×251. 442 U.S. 256 (1979). : “‘Discriminatory purpose’ . . . implies more than intent as volition or intent as awareness of consequences”; it “implies that the decisionmaker . . . selected or reaffirmed a particular course of action at least in part ‘because of,’ not merely ‘in spite of,’ its adverse effects upon an identifiable group.”252×252. Id. at 279 (citation omitted). One might read that case and its progeny as imposing a conscious-intent requirement inconsistent with taint’s broadest deployment.

On the other hand, Feeney did not present as groundbreaking. It focused primarily on explaining that the then-recently created intent framework distinguished “because of” from “in spite of.”253×253. Id.; see also id. at 278 (“The appellee’s ultimate argument rests upon the presumption . . . that a person intends the natural and foreseeable consequences of his voluntary actions.”); Katie Eyer, The But-For Theory of Anti-discrimination Law, 107 Va. L. Rev. 1621, 1632–33 (2021) (arguing, similarly, that the “only aim” of Washington v. Davis, 426 U.S. 229 (1976), and its progeny “was to reject a pure disparate impact standard,” Eyer, supra, at 1633, and that the cases left inchoate the Court’s conception of “discriminatory purpose”). That is, some harms to identifiable groups that we might colloquially say were “intended” (because foreseeable) are not “intentional” for equal protection purposes because they occurred, in truth, “in spite of” the group’s identity.254×254. See Feeney, 442 U.S. at 279. But see id. at 279 n.25 (recognizing that in some circumstances foreseeability is strong evidence of proscribed purpose). But that oft-criticized distinction does not resolve for all purposes the proper decision rule for divining “because of.”255×255. Cf. John F. Manning, Without the Pretense of Legislative Intent, 130 Harv. L. Rev. 2397, 2431 (2017) (noting that what we deem a government actor’s “intended decision” must proceed from a “value judgment about what should count as that . . . decision”). Feeney’s take on the matter seems entirely in line with its predecessor, Washington v. Davis,256×256. 426 U.S. 229. in recognizing that “[p]roof of discriminatory intent” may “rely on objective factors,” legitimately determined “from the results” achieved or avoided.257×257. Feeney, 442 U.S. at 279 n.24; see also id. at 279 & n.25; Washington v. Davis, 426 U.S. at 242; id. at 253 (Stevens, J., concurring).

Because “discriminatory purpose” must be determined on the basis of objective facts in the world rather than (impossible) direct observation of often-nonexistent mental states (such as of multimember bodies),258×258. See Andrew Verstein, The Jurisprudence of Mixed Motives, 127 Yale L.J. 1106, 1123 (2018) (“[A]ll internal mental states must be proven by recourse to external circumstances . . . .”). it is consistent with reading “because of” in a way that does not demand a specific-intent decision rule. That is, it remains possible to say that some cases that do not reflect heartland specific-intent cases are also not best described as harm merely inflicted “in spite of” group identity.

Taint can fit into this class of cases. Consider the selective-indifference approach. If, due to selective indifference, a decisionmaker adopts a policy it would not have adopted had the burdens fallen on a more-favored group, one can intelligibly say the adoption was “because” the burdens in fact fell on a less favored group. We may understand unpurged taint as at least demonstrating selective indifference to the harms of perpetuating a tainted law and the responsibility generated by discriminatory predecessors. If so, one can, consistently with Feeney, treat such a policy as constituting a violation.259×259. Cf. Strauss, supra note 26, at 962 & n.80 (contending that Feeney is more capacious than often suggested, and that it permits a “reversing the groups” approach to “discriminatory intent” that would embrace selective-indifference principles).

At the very least, advocates and lower courts need not read governing Supreme Court doctrine to preclude a specific intent–exempt decision rule in taint cases. Finally, if this reading does not resonate descriptively, I also mean it normatively. That is, if Supreme Court doctrine really does preclude the decision rule I advocate (and elaborate further in Part IV), taint’s existence is yet another strike against that doctrine’s persuasiveness.

IV. Operationalizing Taint

Part III advocated understanding taint as a trigger for a specialized decision rule that (at minimum) justifies denying a tainted policy “clean-slate” treatment, irrespective of intent or facial validity. This Part drills down further on how that decision rule would work, including what follows from the rule and what it would mean to say taint has been purged. Readers still committed to specific intent should read on; as noted, taint can aid in all cases as a tool for detecting pretext. More broadly, one can adhere to a decision rule that embraces specific intent and believe that taint justifies imputation of whatever degree of intent one thinks necessary. Ample precedent exists for that perspective; criminal law often recognizes the propriety of imputing mental states.260×260. See, e.g., Robinson, supra note 152, at 614–19 (outlining a “diverse group of doctrines,” id. at 615, governing imputation of culpable mental states).

A definitional note: I treat “purged” as shorthand for the conclusion that the relationship described in Part II ought not to count against the modern policy. In other words, the framework outlined in this Part bears on two related but distinct questions: (1) what the government must do to defend a T2 policy on the grounds that taint should not weigh against its validity and (2) what the government might be able to do at “T3” to enact a similar policy if it cannot validate its T2 policy.

This Part begins by advocating a judicial approach focused on disparate impact and reason-giving. First, given a showing of taint, courts should ask whether a T2 policy continues to carry a non–de minimis disparate impact of the sort T1 imposed. A government can purge otherwise applicable taint by showing the absence of such impact. Second, if disparate impact persists, governments must offer a more substantial justification than a facially neutral act normally requires. The taint shapes what that justification looks like; the reasoning must specifically engage with the import of the tainted relationship for the present policy, including an explanation of why disparate impact can’t be eliminated and why this policy’s necessity outweighs its tainted impact.

This approach fills out the decision rule further. A tainted relationship is prima facie evidence that justifies shifting a burden of production to the government of demonstrating the taint’s extirpation. Although the ultimate burden of proof never shifts, it counts in the plaintiff’s favor on the ultimate question if the government cannot purge the taint by showing no disparate impact or a substantial, taint-sensitive showing of necessity.

Taint represents a setting in which receding from judicial disparate impact skepticism is unusually justified. At the same time, this approach aims to show the workability of good faith purging. Eliminating real-world disparate effects suggests dissipation of the primary wrong, which I take to be disparate status–based harm. Absent that, because taint can exist absent present-day bad intentions, the government’s burden is not to show that it really has good intentions, but to show why the policy is nevertheless necessary. After explaining the approach, I test its usefulness on some real-world cases and discuss some potential implementation difficulties.

I close this Part by emphasizing that taint is a constitutional concept that is portable to nonjudicial contexts. Understanding it this way highlights that taint is not just a tool of (sometimes justified) judicial overriding of democratic will. It can also be a mechanism that empowers democratic actors committed to understanding and enforcing their constitutional duties.

A. Reinvigorating Disparate Impact

As a theory, disparate impact has generated endless debates over its meaning, justifications, and implications. One particularly relevant justification is its usefulness in eradicating the continuing effects of past discrimination.261×261. See, e.g., Brest et al., supra note 25, at 1088 & n.73; Nicholas O. Stephanopoulos, Disparate Impact, Unified Law, 128 Yale L.J. 1566, 1604–05 (2019). Although Washington v. Davis refused to give disparate impact constitutional dimension,262×262. 426 U.S. at 248. lower courts had previously permitted disparate impact to establish liability, at least presumptively.263×263. See, e.g., Reva B. Siegel, The Supreme Court, 2012 Term — Foreword: Equality Divided, 127 Harv. L. Rev. 1, 12–14 (2013); Strauss, supra note 26, at 950–51. Taint justifies a recovery of some form of this approach.

Recall first that we can know that a policy is tainted without necessarily knowing whether it has a disparate impact.264×264. To be sure, the benefits and burdens of the T2 policy vis-à-vis the T1 policy are key to determining whether the policies are functional duplicates for purposes of taint. See supra section III.C, pp. 1232–34. But, at that initial taint-detection stage, what matters is whether the T2 law tracks the T1 law by (for example) making it harder to vote by imposing a voter ID requirement. If taint exists, this Article then looks to any disparate burden to inform the possibility of purging taint. Thus, finding taint leaves open the question of whether the T2 rule creates any non–de minimis disparate impact on an identifiable group.265×265. I am not wedded to a single definition of de minimis. One venerable standard asks whether the rate for the identifiable group “is less than four-fifths . . . of the rate for the group with the highest rate.” Greene, supra note 25, at 45; see also Jennifer L. Peresie, Toward a Coherent Test for Disparate Impact Discrimination, 84 Ind. L.J. 773, 773 (2009) (advocating synthesis of four-fifths and “statistical significance” approaches to defining non–de minimis disparate impact). See generally Stephanopoulos, supra note 261. At minimum an actionable impact should be both “statistically” and “practically” significant. See Brnovich v. Democratic Nat’l Comm., 141 S. Ct. 2321, 2358 n.4 (2021) (Kagan, J., dissenting); see also Stephanopoulos, supra note 261, at 1571–72. Governments can purge taint by showing the absence of meaningful disparate impact.266×266. In this way, I deviate from the mine-run statutory disparate impact framework, in which the plaintiff has the burden of showing a disparity. See Stephanopoulos, supra note 261, at 1595. I think this makes better sense of taint’s fit with antidiscrimination law, since the alternative would de facto make the passage of time presumptively purge taint, which would let the state off the hook more easily than I think is warranted. Though this is an escape hatch, it is not easily manipulated.

On one hand, a disparate impact–focused path to purging addresses the fear that taint will forever straitjacket governments, notwithstanding genuine changes in circumstances.267×267. Cf. Palmer v. Thompson, 403 U.S. 217, 230 (1971) (Blackmun, J., concurring) (expressing fear of indefinite “lock[] in”); Clarke, supra note 17, at 561–63. That concern has some validity, regardless of one’s optimism regarding the speed of progress. At least when a policy tainted in form is no longer discriminatory in effect, it is appropriate to consider the policy on its own terms.

At the same time, governments cannot escape by simply manipulating the policy’s form or waiting for the passage of time to dissipate the taint. With respect to pretext, bad faith actors who simply wish to launder malice will have a harder time if they must simultaneously expunge any disparate impact. Focusing on disparate impact also advances approaches that decenter specific intent to harm. Eliminating disparate impact is strong evidence of equal treatment today, notwithstanding the past’s meaning.

B. Reason-Based Purging

1. Why Permit Reason-Giving? — In some cases, disparate impact will persist. In others, such as when the T2 is new, determining whether it will persist may require undue speculation about downstream effects.268×268. Sometimes a new policy’s changes will be sufficiently straightforward to permit informed, reliable speculation about its effects, cf. Stephanopoulos, supra note 261, at 1640–41 & n.421, so I cannot rule out the possibility of drawing predictive disparate impact conclusions. One could simply treat those cases as the end of the line, such that taint plus governmental inability to dispel disparate impact equals conclusive determination of wrongful discrimination. That is a coherent suggestion, but it goes too far.

As an initial matter, permitting the government to show why this disparate impact is justified and unavoidable would be consistent with the general approach to disparate impact in other areas.269×269. See id. at 1595, 1597–98, 1604. Symmetry alone does not resolve matters, since taint could be a unique beast. But the general approach is worth noting, especially if good reasons exist for it. Moreover, and again, taint is a heuristic helpful in determining wrongful discrimination, not the determination itself. Prima facie cases are rebuttable,270×270. Cf. Washington v. Davis, 426 U.S. 229, 241 (1971) (discussing burden shifting in the prima facie context). and I doubt that showing a lack of disparate impact should be the only way to rebut this one. Although taint helps show why a broader decision rule is preferable, it does not follow that governments should lose whenever they cannot disprove persistent disparate impact.

Consider cases where we simply cannot know T2’s impact. Not every case will be like Trump v. Hawaii, where the question reduces to whether permissible or forbidden reasons explain an indisputable disparate impact. Treating such situations as automatic government losses would be simplistic. A tainted policy might well, under present-day circumstances, be both supportable by legitimate nondiscriminatory reasons and also indispensable for tackling a given policy problem. Especially given taint’s decoupling from intent, uncertainty about impact should not work to lock governments into the status quo indefinitely. Perhaps the government can show that what seems to be (for example) selectively indifferent or demeaning action actually is not.

Even where disparate impact can be foreseen or has already occurred, an absolutist approach may be unwarranted. Some necessary policies may have an unavoidable disparate impact that compelling justifications outweigh. For example, some might critique any government benefit not pegged to recipient need as generating a disparate impact because of preexisting entrenched disparate distributions of power and resources.271×271. See, e.g., Kevin E. Jason, Dismantling the Pillars of White Supremacy: Obstacles in Eliminating Disparities and Achieving Racial Justice, 23 CUNY L. Rev. 139, 151 (2020) (describing “colorblind benefits” as, inter alia, “includ[ing policy] solutions that involve the sometimes equal, but always inequitable [i.e., unjust], allocation of resources and opportunities” and noting that “one’s starting position is critical to determining how one will fare”). Sometimes rejecting progressivity may be illegitimate precisely because it will produce such impacts, even if it lifts all boats overall.272×272. See id. at 151–52, 159, 180. But sometimes it may not. Imagine a public good that by its nature “cannot be provided to anybody unless [it is] provided to everybody” on the same terms.273×273. Liam Murphy & Thomas Nagel, The Myth of Ownership: Taxes and Justice 46 (2002). Especially for substantial interests,274×274. Consider, for example, national defense or certain anti–air pollution measures. it is a harder question whether the disparate impact of providing it might be justified because of the difficulty of making it progressive and the substantive importance of the benefits.

2. Reason-Based Purging. — Given those preliminaries, what sort of reasons might purge taint where disparate impact persists or cannot be disproven? Tiers-of-scrutiny talk can sometimes obscure.275×275. See, e.g., Andrew M. Siegel, Equal Protection Unmodified: Justice John Paul Stevens and the Case for Unmediated Constitutional Interpretation, 74 Fordham L. Rev. 2339, 2340 (2006). But the tiers may help illuminate inappropriate approaches.

Consider the application of strict scrutiny to facially neutral policies found to be animated at least in part by discriminatory purpose.276×276. See, e.g., Ralph Richard Banks et al., Racial Justice and Law 410 (1st ed. 2016). Taint can exist absent that showing. Requiring the T2 government to satisfy strict scrutiny based on nothing more than taint would transform it into the underlying violation that it aids in demonstrating. Yet the highly deferential rational basis review usually applicable to facially neutral government acts is too lax. Taint with teeth must excuse plaintiffs from disproving “every conceivable basis which might support [the policy],”277×277. Box v. Planned Parenthood of Ind. & Ky., Inc., 139 S. Ct. 1780, 1782 (2019) (quoting Armour v. City of Indianapolis, 566 U.S. 673, 685 (2012)). and deny governments the boon of having policies upheld on reasons articulated for the first time in litigation or conceived independently by the courts.278×278. Brest et al., supra note 25, at 567.

Something intermediate is needed to fill out the decision rule applicable when governments cannot purge taint via eliminating disparate impact. Helpfully, antidiscrimination law has a conventional framework for dealing with disparate impact.279×279. Stephanopoulos, supra note 261, at 1595. On that approach, given disparate impact, a defendant must show that the challenged “practice is necessary to achieve a substantial interest,” and, if so, the plaintiff “may try to demonstrate that this interest could be attained through different means that yield smaller [disparities].”280×280. Id. The taint approach need not duplicate the usual approach, which itself varies somewhat depending on context.281×281. See, e.g., id. at 1600 n.190 (noting deviations in the Age Discrimination in Employment Act framework). Thus, for example, I reverse the “usual framework” by placing the burden on the government to show no disparate impact.282×282. See supra note 266. Still, reinventing the wheel is unnecessary.

In crafting the best approach for a taint-specific rule, then, I draw on the usual framework’s insight that otherwise-suspect disparities may be justified upon a heightened showing.

Furthermore, although the approach is not importable wholesale, I see much value in administrative law’s reasoned-decisionmaking approach.283×283. Professor David Super has also noted the potential relevance of administrative law principles to evaluating policy change over time, although he deploys them differently. See Super, supra note 16, at 84–87. Formally, it requires that a decision be “based on a consideration of the relevant factors.”284×284. Dep’t of Homeland Sec. v. Regents of the Univ. of Cal., 140 S. Ct. 1891, 1905 (2020) (quoting Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 416 (1971)). Importantly, courts “may not supply a reasoned basis for the agency’s action that the agency itself has not given.”285×285. Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983) (citing SEC v. Chenery Corp., 332 U.S. 194, 196 (1947)). Here, “meaningful judicial review” requires an agency to “disclose the basis” of its action,286×286. Dep’t of Com. v. New York, 139 S. Ct. 2551, 2573 (2019). and actions may be invalid where there is a “significant mismatch between the decision . . . made and the rationale . . . provided.”287×287. Id. at 2575.

Disputes over how rigorous this review is in practice need not be resolved here.288×288. See, e.g., Jacob Gersen & Adrian Vermeule, Thin Rationality Review, 114 Mich. L. Rev. 1355, 1356, 1358–59 (2016) (arguing that administrative lawyers and academics overestimate the toughness of reasoned-decisionmaking review). I invoke the standard to embrace the idea of legitimacy that appears to support it — that democratic legitimacy flows from the capacity and ability to provide “public-regarding reasons that all might accept.”289×289. Jerry L. Mashaw, Reasoned Administration and Democratic Legitimacy 168 (2018) (arguing that administrative law principles implement this vision of legitimacy); see also Thomas P. Schmidt, Judicial Minimalism in the Lower Courts, 108 Va. L. Rev. (forthcoming 2022) (on file with the Harvard Law School Library). If the state is to require citizens to live under a law whose form was tainted and that continues to impose disparate impacts, the state ought to be able to offer robust explanations, applicable to all, for why it is necessary.

3. Filling Out the Approach. — Building on these raw materials, purgative reason-giving must reflect robust engagement with the tainted relationship. Persistent disparate impact does not necessarily doom the policy. But, to purge the taint, governments must offer (1) a legitimate, nondiscriminatory interest that will be substantially impaired absent this policy, (2) direct engagement with the past problematic history, (3) an explanation of why the disparate impact cannot be eliminated, and (4) an explanation of why the legitimate need for this means of pursuing the legitimate interest outweighs the harm of shielding the disparate impact of a tainted rule. These reasons cannot be post hoc, and a court may not conjure them up on the government’s behalf.

Much of this approach will depend on context-specific “qualitative judgments” made “in light of competing values and social facts,”290×290. Greene, supra note 25, at 61. and I will apply it to specific cases in section C. Here, I will simply say a bit about the types of social facts that will matter and why.

For example, courts should care about the passage of time and changed personnel, not because those necessarily purge taint, but because time’s passage generates new problems, and new decisionmakers identify new solutions. These facts matter, in short, because they make it more plausible that a substantial, novel government need has emerged to independently justify the policy. I do not mean to suggest that an unprecedented need is necessary, but its existence would be a strong sign that the animating aims truly have changed.

Balancing the need against the persistent harm will also be multifaceted. Much will turn on whether the legitimate need for this policy in this form outweighs the specific harm of the tainted rule’s preservation. In advocating the disinterring and protection of a right against a tainted rule’s disparate impact, this approach can be seen as a species of the proportionality approach that Professor Jamal Greene advances as a general approach to rights questions.291×291. See id. at 58–60. Again, I will sketch a few examples.

Return to plausible alternatives for pursuing the asserted governmental interest. Could a state purge a felon disenfranchisement law’s taint by merely invoking “deterrence” as a legitimate, nondiscriminatory reason? No, because the burden is not just to show that a legitimate, nondiscriminatory justification can be articulated, but that a legitimate interest has generated a genuine nondiscriminatory need for this policy in this form. Deterrence is too general, because felonies can be deterred without such a law. Deterrence might well be somewhat impaired from the state’s perspective, but taking taint seriously means that policy-choice freedom is somewhat constrained. Practically, then, the legitimate, nondiscriminatory interest will have to be stated with meaningful precision; otherwise, a plausible alternative will be relatively easy to identify and will militate against a claim of necessity.

Continuing in this vein, plausible alternatives to pursuing the asserted interest will also undermine a claim that disparate impact can’t be eliminated. As the felon disenfranchisement example suggests, the purgative analysis incorporates “tailoring” elements. Importantly, though, this is not to say that plausible alternatives are fatal to an attempt to purge taint. This is not a strict scrutiny analysis. A state could conceivably show that a legitimate, nondiscriminatory interest would be substantially impaired without policy X without showing that policy X is the only way of pursuing the interest. But plausible alternatives do suggest that a substantial impairment claim may be unwarranted.

On the other side, the state can bolster a necessity claim by identifying “good taint,” that is, a long tradition of nonsuspect institutional uses of this sort of policy for this interest.292×292. See supra pp. 1224–25. An institutional history of legitimate needs generating functionally similar conduct is itself a reason to believe the government if it claims, today, that legitimate, nondiscriminatory needs exist for the policy.

Regarding the strength of the interest pursued, although governments need not identify an interest that a court would call compelling, it surely will help. On the harm side of the ledger, as the so-called tiers of scrutiny recognize, the nature of the harm affects what it takes to justify it.293×293. See, e.g., United States v. Virginia, 518 U.S. 515, 533–34 (1996) (noting varying permissibility of “inherent differences” justifications for differential treatment). We can acknowledge the potential contestability of the “nature” of harm and still recognize that a persistent disparate impact in the administration of criminal justice (for example) is particularly inexcusable.294×294. Cf. Pena-Rodriguez v. Colorado, 137 S. Ct. 855, 868 (2017) (asserting that racial discrimination is “especially pernicious in the administration of justice” (quoting Rose v. Mitchell, 443 U.S. 545, 555 (1979))).

What of direct engagement with problematic history? In section V.A.2, I discuss further the possibility that a government contests the wrongfulness of T1. For now, though, note that this approach does not require governmental concession on this point. It simply requires explicit engagement with the reasons that could support such a conclusion. That is, the government must treat them as genuine concerns to be addressed to the maximum extent possible consistent with the (by hypothesis) important reasons that justify the policy notwithstanding the persistent disparate impact.295×295. My suggestion thus fleshes out Justice Sotomayor’s interest in examining whether the state has “grappled” with the policy’s history. See, e.g., Ramos v. Louisiana, 140 S. Ct. 1390, 1410 (2020) (Sotomayor, J., concurring). This hearkens back to the need for public-regarding reasons acceptable by all to justify the maintenance of this policy — whether or not “all” agree on taint.296×296. See supra p. 1241. Those reasons should at least acknowledge the relevance of the “uncomfortable past.”297×297. Espinoza v. Mont. Dep’t of Revenue, 140 S. Ct. 2246, 2293 n.2 (2020) (Sotomayor, J., dissenting) (quoting id. at 2273 (Alito, J., concurring)). Engagement engenders productive disruption by forcing reconsideration and justification beyond “this is how we’ve always done it.” Even where the government ultimately can justify the policy, this process may generate a policy whose impacts fall more equally across the population.

To sum up, taint triggers a decision rule that first looks to disparate impact. If necessary, courts then apply heightened scrutiny. Although, in a sense, the plaintiff’s burden of production is lessened vis-à-vis the usual case — that is, she need “only” show taint to shift a burden of production to the government — so too is the government’s responsive burden of production.298×298. I emphasize that shifting the burden of production does not shift the ultimate burden of proof (sometimes called the burden of persuasion). See Dir., Off. of Workers’ Comp. Programs, Dep’t of Lab. v. Greenwich Collieries, 512 U.S. 267, 272–76 (1994) (distinguishing the concepts); cf. Johnson v. California, 545 U.S. 162, 170–73, 171 n.6 (2005) (discussing how Batson’s prima facie model never shifts the ultimate burden of proof, but does shift the burden of production to “determine the persuasiveness of the . . . constitutional claim,” id. at 171). In the normal case where a plaintiff shows the action was motivated in part by proscribed intent, the government must show that the motivation was not a but-for cause, and satisfy strict scrutiny if it cannot.299×299. See Hunt v. Cromartie, 526 U.S. 541, 546 (1999); Hunter v. Underwood, 471 U.S. 222, 228 (1985); Banks et al., supra note 276, at 410; Verstein, supra note 258, at 1125, 1127, 1144, 1162; cf. Grutter v. Bollinger, 539 U.S. 306, 343–44 (2003) (upholding race-conscious affirmative action policy under strict scrutiny); Shaw v. Reno, 509 U.S. 630, 643–44 (1993) (stating that strict scrutiny principles apply to race-neutral policies that are, “on their face, ‘unexplainable on grounds other than race.’” (quoting Village of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 266 (1977))). Conversely, where taint supports a prima facie case, the government need not show narrow tailoring that satisfies a compelling interest. It need only show that legitimate reasons support the policy, that it has accounted for the problematic history, and that any persisting disparate impact is unavoidable and outweighed by the benefits of the legitimate justifications.

This approach avoids strict scrutiny without merely offering a blueprint to unaccountable taint laundering. First, unlike rational basis review, where a government need not have any particularly good reasons for passing a law, a tainted law must be justified by legitimate, nondiscriminatory reasons (beyond inertia) that independently would generate the policy. Second, unlike the mine-run case, disparate impact still has to be justified. And showing that it is unavoidable often will require at least showing that making the policy progressive is impossible due to the policy’s nature. Third, because post hoc reasons will not suffice, taint will sometimes operate to force formal reenactment to place sufficient reasons on the record. That ought to increase confidence that the reasons given actually are weighty.

This also guards against a potential misunderstanding. Abbott raised concerns about improper “shifting” of the burden of proof in cases involving discriminatory predecessors.300×300. 138 S. Ct. 2305, 2324–25 (2018) (“[I]t was the plaintiffs’ burden to overcome the presumption of legislative good faith and show that the [T2] Legislature acted with invidious intent. The [lower] court contravened these basic principles.” Id. at 2325.). The Court thought the lower court erred by forcing Texas to prove a legislative “change of heart” and engagement in a “deliberative process to ensure” that the new policy “cured any taint.”301×301. Id. at 2325 (quoting Perez v. Abbott, 274 F. Supp. 3d 624, 649 (W.D. Tex. 2017)).

However one reads the lower court’s opinion in that case, taint (properly understood) is perfectly consistent with conventional allocations of the burden of proof. Showing taint proves wrongful discrimination no more than a Batson prima facie case does.302×302. Cf. Johnson, 545 U.S. at 169–71, 169 n.5 (noting that “a prima facie case of discrimination can be made out by offering a wide variety of evidence,” id. at 169, and emphasizing that Batson recognized that step one was satisfied where defendant timely objected to the striking of all Black persons on the venire). Taint does generate a burden for the government defendant, however — a burden of production, just as a prima facie case does in Batson and Title VII disparate treatment cases.303×303. See USPS Bd. of Governors v. Aikens, 460 U.S. 711, 714–17 (1983) (describing the prima facie burden-shifting method applicable in Title VII cases); cf. Johnson, 545 U.S. at 171 n.7 (noting similarity between Title VII burden shifting and Batson burden shifting). Indeed, mine-run equal protection doctrine embraces the same approach by shifting the burden to the government to disprove that discriminatory motivation was a but-for cause when a plaintiff shows such motivation played a part in a decision. See Arlington Heights, 429 U.S. at 270 & n.21. Here, as there, failure to discharge that burden counts against the government in the final determination.304×304. See Johnson, 545 U.S. at 171 & n.6. I leave unresolved the possibility that taint cases might warrant making such a failure conclusive.

None of this means that taint changes the ultimate burden of proof, or that past wrongfulness is dispositive of present wrongfulness. Rather, it simply points to one reasonable path for some plaintiffs to satisfy their burden of proof. Perhaps that means that plaintiffs in taint cases have an easier time relative to others, but that seems no more problematic than plaintiffs in cases involving overt discrimination having an easier time than plaintiffs in covert discrimination cases. Differently situated plaintiffs have different advantages. Thus, Abbott’s complaints of original sin “condemn[ing] governmental action that is not itself unlawful” and disrespect for the presumption of good faith305×305. Abbott, 138 S. Ct. at 2324 (quoting City of Mobile v. Bolden, 446 U.S. 55, 74 (1980) (plurality opinion)). are orthogonal. Taint means that, at least sometimes, present good faith is insufficient. Sometimes, the unpurged taint of past discrimination may be a fact about contemporary government action that indicates its unlawfulness, irrespective of its good faith enactment.

C. Exemplars

I turn now to consider whether taint actually can aid judges considering T1/T2 cases. Exemplar cases from the Supreme Court and lower courts demonstrate that it can. As my discussion in section IV.D should make clear, any conclusions that courts should not find taint should not be taken as conclusions that taint “does not exist” in some metaphysical sense, or that nonjudicial actors are barred from concluding otherwise and acting accordingly. Finally, because purging inquiries are context sensitive, different or additional facts could generate different conclusions. This is a necessary and laudable feature of a functional approach that both takes taint seriously and finds value in permitting its genuine purging.

1. Ramos v. Louisiana. — Treated only as a Sixth Amendment case, one could fairly ask (as the Ramos dissent did) why the racist origins of Louisiana’s jury-nonunanimity rule mattered, especially given the adoption and endorsement of similar rules by apparently blameless institutions.306×306. Ramos v. Louisiana, 140 S. Ct. 1390, 1426–27 (2020) (Alito, J., dissenting). The majority opinion noted the origins while criticizing previous Justices’ reasoning, but did not elaborate on the Sixth Amendment relevance (if any).307×307. See id. at 1401 & n.44 (majority opinion) (indicating that even a nonunanimity rule with benign origins would be unconstitutional); see also id. at 1405. One concurrence noted the origins as supporting overruling any precedent that would permit the rule. See id. at 1417 (Kavanaugh, J., concurring). Nor, vis-à-vis those origins, did it offer a detailed answer to the dissent’s emphasis on the readoption of the rules “under different circumstances.”308×308. Id. at 1426 (Alito, J., dissenting). A concurrence assayed an answer: the readoption insufficiently “grappled with the laws’ sordid history.” Id. at 1410 (Sotomayor, J., concurring). But it did not detail what that should look like. See supra note 295.

Of course, the Court could resolve only the claim brought. But one suspects that litigation decisions were informed by concern that courts would treat the readoption as defanging a Fourteenth Amendment claim. Taint could have cleaned up the analysis by showing how a court could be made to see the readoption’s unique constitutional problem. The argument, in short: continuity in institution, subject matter, and effect (permitting nonunanimous convictions) showed patent continuity between old and new, notwithstanding the passage of time and the tweak in the jurors needed for conviction from nine to ten.

Importantly, saying that much does not dictate a result. Taint makes relief possible absent present-day bad motives, but analysis remains.309×309. As noted, taint should still be relevant to even the strictest specific-intent framework as evidence of potential pretext. See supra section III.A, pp. 1227–28. While a bit of speculation is required, the best analysis would be that Louisiana could not defend the rule as it came to the Court. The rule’s disparate impact persisted, and the disparity’s source — the rule’s interaction with Black Americans’ minority status — seemed essentially inherent.310×310. See, e.g., Ramos, 140 S. Ct. at 1418–19 (Kavanaugh, J., concurring); Frampton, Jim Crow Jury, supra note 107, at 1598–99. The disparate impact fell in the criminal justice area. And Louisiana’s refusal to acknowledge the rule’s racist origins311×311. See Ramos, 140 S. Ct. at 1426 & n.2 (Alito, J., dissenting). would pose obstacles to showing that the legitimate need for this rule in this form outweighed the harm of shielding its tainted disparate impact. Although we cannot know what interest Louisiana would have articulated under this Article’s framework, I cannot think of one that is precisely stated enough to make it plausible that nonunanimity is necessary to pursue it. Invalidation on equal protection grounds was thus appropriate.312×312. Interestingly, unlike the actual holding, this holding might have permitted states with different institutional histories to experiment with nonunanimity. See, e.g., Murray, supra note 215, at 225 n.291 (noting support and critique of the states-as-laboratories-of-democracy tradition). The desirability of such experiments is unclear.

What then? If committed to nonunanimity, Louisiana would need to find some version of that policy that eliminated disparate impact, or produce more robust justifications that actually attended to the policy’s history. If such justifications sounded in judicial efficiency,313×313. See Ramos, 140 S. Ct. at 1427 n.3 (Alito, J., dissenting) (offering such justifications). it would count against Louisiana that there are alternative ways to pursue that vaguely stated goal. Absent a more precisely articulated goal that Louisiana genuinely needed nonunanimity to pursue, it might well be that discharging the taint would be impossible given persistent disparate impact. I cannot say for sure without knowing every possible potential interest, but the important takeaway is that there may well be cases where persistent disparate impact is indeed fatal to reenactment.

2. Trump v. Hawaii. — Taint also illuminates the Travel Ban case. The compressed temporality, intervening litigation events, and clear subject-matter continuity strongly support a line of descent from Bans 1 and 2 to Ban 3.314×314. See supra p. 1204. For the majority, all that mattered was whether Ban 3 itself violated the Establishment Clause notwithstanding its facial validity.315×315. See Trump v. Hawaii, 138 S. Ct. 2392, 2418 (2018). As against the extrinsic evidence of candidate and President Trump’s statements and behavior, the majority emphasized the worldwide review process that preceded Ban 3 and the government’s ultimate removal of three majority-Muslim countries from the third list.316×316. Id. at 2421–22. As previously observed, these were two of Ban 3’s most notable differences.317×317. See id. at 2404–05, 2422; Issacharoff & Morrison, supra note 207, at 1943.

An opinion that grappled with taint would look quite different. True, the government never conceded invalidity. But the earlier bans’ wrongfulness was not just plausible, but actually found in reasoned adjudications.318×318. I bracket the context-specific question of whether the national security context could possibly require ignoring the “extrinsic evidence” critical to this assertion. See Trump, 138 S. Ct. at 2419–20 (raising but not deciding this question). And the third ban still created a disparate religious burden. A better analysis would have consequently shifted a burden of production to the government to demonstrate that a legitimate, nondiscriminatory need for this policy (notwithstanding its origins) outweighed the persistent disparate impact.

No doubt, any legitimate, nondiscriminatory reasons that the interagency review generated would be relevant. The government would still need to show that legitimate interests would be substantially impaired without this type of ban. Moreover, in determining ultimate validity, those reasons would have to be reconciled with the damning pre–interagency review evidence, with more skepticism than the Court applied.319×319. See id. (applying rational basis review). Trump appeared to give near-dispositive weight to the interagency review and certain “ex post procedures.”320×320. Cf. Landau, supra note 1, at 2152 (categorizing Trump as an example of how courts “frequently sustain challenged acts . . . when those acts are the result of a thorough process”). Taint renders insufficient the mere existence of facially neutral text or procedures. The government should still have had to engage with the problematic predecessor evidence and articulate expressly how legitimate needs outweighed persistent disparate impact.

Trump thus shows how taint justifiably counterbalances the sort of apparently benign context that can insulate a facially neutral policy from scrutiny.321×321. See Haney-López, supra note 25, at 1839 (observing judicial use of contextual evidence to “build the case for [governmental] innocence”); Landau, supra note 1, at 2173. For example, perhaps a history of unquestionably legitimate orders of this “form and substance” indeed preexisted this Presidency.322×322. See Issacharoff & Morrison, supra note 207, at 1947–48. Something like “good taint” might well matter. But the government should have had to articulate, and the Court should have had to evaluate, whether those more distant policies tipped the purgative scale.323×323. Regarding a statutory interpretation question, the Court parenthetically referenced past entry restrictions issued under the relevant statute, but did not suggest they mattered to the constitutional question. See Trump, 138 S. Ct. at 2409–10, 2413.

Again, some tainted policies can be justified, so proposing a different approach to Trump does not dictate an outcome. The case, regardless, helps show how and why a compressed temporal span matters. Whatever one’s ultimate view of Ban 3, the facts are consistent with an attempt to launder taint through the involvement of subordinates. That risk exists in any compressed-time case, and suggests that taint is hardest to purge where the same individual final decision-maker affirms a functionally identical institutional decision within a tight time span. Here, the blame and task responsibility underpinning taint is at its zenith. Accordingly, absent a final-decisionmaker change, a substantial showing of altered circumstances would have been the most reliable way for the government to earn clean-slate treatment for a follow-on travel ban.

To be sure, it is fair to acknowledge that courts may find it uniquely challenging to weigh the substantiality of national security claims. Taint would not necessarily bar a similar ban if, years later, the specific countries involved posed a specifically identified, novel national security threat. Still, national security is not a Constitution-free zone, and section IV.B’s standard would demand more than a nonspecific invocation of “national security” to defend a tainted policy.

Ultimately, the sensitivity of the analysis that taint calls for complicates any self-assured declaration of who ought to have won, considered post hoc. At minimum, the facts recounted in the opinion itself would not have sufficed to purge the taint of the initial policy. Perhaps more could have been offered. If so, taint would at least have extracted a more robust showing and an unequivocal disavowal of Trump’s damning statements.324×324. See Verstein, supra note 170, at 768, 782 n.168. Regardless, any outcome must stem from more than context laundering — whether pretextual or merely corollary to time’s passage and genuinely well-intentioned behavior.

3. Abbott v. Perez. — Abbott helps show the framework’s useful granularity.

The story began with Texas drawing its post-2010 census redistricting maps, which generated immediate litigation in both Texas federal court and the D.C. federal court.325×325. Abbott v. Perez, 138 S. Ct. 2305, 2315 (2018); Texas v. United States, 887 F. Supp. 2d 133, 138–39 (D.D.C. 2012) (noting Texas’s request for judicial preclearance of its maps for Federal House elections and Texas House and Senate elections). A Texas federal district court devised interim maps pending the litigation’s resolution, but the Supreme Court vacated those interim maps in 2012.326×326. Abbott, 138 S. Ct at 2315–16. The maps were vacated as insufficiently deferential to the originals. In remanding, the Court instructed the federal district court to draw interim maps based on the original maps except insofar as modification was required to avoid “incorporat[ing] . . . any legal defects.”327×327. Id. at 2316 (quoting Perry v. Perez, 565 U.S. 388, 394 (2012) (per curiam)). Thus, the lower court was to modify any district regarding which it thought plaintiffs were likely to succeed on claims under the Constitution or section 2 of the VRA, or which it thought could support any “not insubstantial” claim under section 5 of the VRA.328×328. Id.

After the lower court did so,329×329. Perez v. Abbott, 274 F. Supp. 3d 624, 632–33 (W.D. Tex. 2017), rev’d, 138 S. Ct. 2305 (2018). the D.C. federal court held that section 5 of the VRA barred preclearance of the original maps.330×330. Abbott, 138 S. Ct. at 2316. As relevant here, the court denied preclearance because Texas failed to prove (as section 5 required) that its Federal House of Representatives map was enacted without discriminatory purpose. Texas, 887 F. Supp. 2d at 151–52, 161–62. It also failed to prove that its State House map had no impermissibly “retrogressive effect.” Id. at 177; see also id. at 139. Texas then repealed the original maps, adopted the interim map for Federal House elections without change, and adopted the interim map for State House elections with some modification.331×331. Perez, 274 F. Supp. 3d at 634; see also Abbott, 138 S. Ct. at 2317. Next, the Texas federal court concluded that the first maps were unconstitutional and that “the racially discriminatory intent and effects [of those maps] carr[ied] over into the 2013 [T2] plans where . . . district lines remain unchanged.”332×332. Perez, 274 F. Supp. 3d at 686 (Federal House plan); Perez v. Abbott (Perez II), 267 F. Supp. 3d 750, 757–58, 794–95 (W.D. Tex. 2017) (same finding for State House plan). Abbott examined those T2 maps.

The majority’s core argument was that the lower court erred by reversing the ultimate burden of proof.333×333. Abbott, 138 S. Ct. at 2324–25. One can argue about what the district court actually did. In my view, however, it appeared to have been reaching for something similar to this Article’s proposal to understand taint as shifting a burden of production to the state without changing the ultimate burden of proof. If so, the court (and other similarly situated courts) could have prevented reviewing-court confusion by employing this Article’s vocabulary. Similarly, the taint framework could have strengthened the dissent’s challenge to the burden-of-proof argument,334×334. See id. at 2353–54 (Sotomayor, J., dissenting). and offered guidance for lower courts in need of a framework consistent with Abbott’s concerns.335×335. Cf. N.C. State Conf. of the NAACP v. Raymond, 981 F.3d 295, 303–05 (4th Cir. 2020) (concluding that the district court reversed the burden of proof in contravention of Abbott).

Insofar as Abbott implicitly equates recognizing taint with reversing the burden of proof, this Article shows why that reasoning fails. On these complex facts, however, the majority perhaps had a (different) point.

Treating the initial maps as T1, it was at “T1.5” that the Texas federal court followed the Supreme Court’s instructions to create interim maps that eliminated potential legal defects.336×336. See Abbott, 138 S. Ct. at 2328. Then, at T2, Texas’s new maps essentially (or verbatim) adopted the T1.5 maps. To be sure, those interim maps did take as a baseline the T1 maps, and the interim maps were expressly preliminary.337×337. Id. at 2316. But for purposes of detecting a tainted relationship, it is striking that a different institution, operating under a materially different mandate, intervened at T1.5 to change the T1 policy in a way explicitly pegged to eliminating wrongfulness.

Normally, the institutional and subject matter continuity, close temporal relationship, and relationship to litigation would support a taint finding. But, if the animating idea is continuity, continuity was arguably cut in a clear way here.338×338. This could be countered in the dissent’s register: the T1.5 intervention was too preliminary to support this weight. See, e.g., id. at 2348 (Sotomayor, J., dissenting). The Court came close to recognizing this in its emphasis on that intervening event, although it confused matters by framing its displeasure as rooted in burdens of proof.339×339. See id. at 2325 (majority opinion). The better argument would have been that the lower court wrongly added tainted continuity to the plaintiffs’ side of the scale. Put bluntly, taint should not have aided the plaintiffs here. As all recognized, the 2011 maps were relevant — but in the conventional, holistic way that history matters, not the special-decision-rule way taint justifies.

4. Federal Criminal Unlawful Entry and Reentry Provisions. — As this Article goes to press, lower-court challenges to 8 U.S.C. §§ 1325–1326 are percolating. These immigration law provisions criminalize unlawful entry and unlawful reentry under certain conditions.340×340. See 8 U.S.C. §§ 1325(a), 1326. The first versions of these provisions appeared in the 1929 Undesirable Aliens Act,341×341. Act of Mar. 4, 1929, Pub. L. No. 70-1018, ch. 690, § 2, 45 Stat. 1551. which also criminalized unlawful entry and reentry.342×342. Id. No court considering challenges to the provisions has disputed the substantial evidence that the 1929 Act enacted wrongful discrimination.343×343. See, e.g., United States v. Carrillo-Lopez, No. 20-cr-00026, 2021 WL 3667330, at *7–9, *14 (D. Nev. Aug. 18, 2021) (citing legal historians, legislative history, and congressional testimony to explain how the Act was enacted to keep out “undesirable” groups, such as migrants from Mexico, and “to keep America’s identity white,” id. at *14); see also id. at *7 & n.17 (noting the government’s concession that the 1929 law violated equal protection principles); United States v. Novondo-Ceballos, No. 21-CR-383, 2021 WL 3570229, at *1 (D.N.M. Aug. 12, 2021) (discussing the “[Undesirable Alien’s Act]’s admittedly racist origins,” id. at *6).

Since 1929, Congress has reenacted and amended these criminal provisions multiple times, including as part of major immigration reforms in 1952 and 1990.344×344. Carrillo-Lopez, 2021 WL 3667330, at *4 & n.11, *13, *23–24 (section 1326); United States v. Rios-Montano, No. 19-CR-2123, 2020 WL 7226441, at *3 (S.D. Cal. Dec. 8, 2020) (section 1325). But the operative core — criminalizing unlawful entry and reentry — has persisted unchanged. Changes have worked around the edges, to increase penalties, governmental enforcement authority, or the scope of persons subject to penalties.345×345. Carrillo-Lopez, 2021 WL 3667330, at *13, *23 (section 1326); Rios-Montano, 2020 WL 7226441, at *4 (calling section 1325 “substantially similar” to its original form). And no intervening event identified in the cases thus far rises to the level of cutting the continuity chain — the mere passage of time is insufficient, as is the mere presence of intervening reenactment. Accordingly, we can fairly impute to subsequent policy enactors the task and blame responsibility contemplated in section II.D.

Stepping through the analysis this way reveals the error in validating the present policies on the grounds that we should not “automatically impute . . . past motivations to the current law and forgo analysis of the enacting legislature.”346×346. E.g., United States v. Gallegos-Aparicio, No. 19-CR-2637, 2020 WL 7318124, at *3 (S.D. Cal Dec. 10, 2020); Rios-Montano, 2020 WL 7226441, at *4. Taint is relevant to claims of intentional pretext. But the concept also clarifies and supports claims not premised on proscribed motivation, which thus do not require any automatic motivational imputation.

On this Article’s approach, the current laws are tainted. For present purposes, I will assume that persistent disparate impact is demonstrated by the fact that the overwhelming majority of persons apprehended at the border are from Mexico or Latin America.347×347. See, e.g., Novondo-Ceballos, 2021 WL 3570229, at *1; Carrillo-Lopez, 2021 WL 3667330, at *15; United States v. Machic-Xiap, No. 19-cr-407, 2021 WL 3362738, at *10–11 (D. Or. Aug. 3, 2021). This assumption could be contested on baseline grounds, i.e., disparate relative to what nondisparate baseline? Compare Dep’t of Homeland Sec. v. Regents of the Univ. of Cal., 140 S. Ct. 1891, 1915 (2021) (downplaying an immigration policy’s substantial impact on Latinos “because Latinos make up a large share of the unauthorized alien population”), with Carrillo-Lopez, 2021 WL 3667330, at *6–7 (rejecting the government’s attribution of impact to “geography” and Latinos’ higher share of unauthorized alien population, in part because of “over-policing of . . . the Southern border,” id. at *7). As things stand, the present laws reflect insufficient engagement with past problematic history. We may stipulate that substantial benign context accompanied later reenactments, and that changes elsewhere in the immigration laws embraced nondiscrimination principles.348×348. See, e.g., Rios-Montano, 2020 WL 7226441, at *5–7 (finding, in 1990 legislative history, general “balancing of valid immigration considerations,” support from progressive organizations, and an absence of contemporary racial animus, id. at *7); United States v. Gutierrez-Barba, No. CR-19-01224-001, 2021 WL 2138801, at *4 (D. Ariz. May 25, 2021) (discussing a 1965 amendment to other parts of immigration laws to include a nondiscrimination provision and removal of national-origin quotas). But, as even a court that rejected a challenge acknowledged, Congress’s reenactments have neither specifically referenced the 1929 law nor engaged with the risks of perpetuating a plausibly discriminatory past.349×349. See, e.g., Rios-Montano, 2020 WL 7226441, at *5–6; see also Carrillo-Lopez, 2021 WL 3667330, at *24. Whatever else might be said about other provisions of the original law, the entry and reentry provisions have been carried forward functionally unchanged. At least where, as here, no one seriously disputes the original law’s illegitimate nature, specific engagement demands more of the contemporary decisionmaker. Congress could show the necessary engagement by, for example, acknowledging the reprehensible mentalities animating the original act, investigating the historical and ongoing impact of the criminal provisions, articulating why the criminal provisions remain necessary notwithstanding ongoing impact, and taking feasible steps to minimize that impact. In short, the necessary engagement at least requires appreciating the persistent impact of the past.

While the specific-engagement factor does not require the government to have formally conceded the original law’s illegitimacy, it may often warrant at least initial invalidation of a tainted law. But lest the reader think taint inevitably “fatal in fact,”350×350. Cf. Grutter v. Bollinger, 539 U.S. 306, 326 (2003) (quoting Adarand Constructors, Inc. v. Peña, 515 U.S. 200, 237 (1995)). these laws are otherwise strong candidates for plausible purging if a subsequent reenactment evinces better engagement.

Assuming the legitimacy of controlling entry into the country, preventing unlawful entry is a legitimate, nondiscriminatory interest. I have assumed disparate impact thus far, but the demographics of the population attempting entry at least makes plausible a government claim that any attempt to prevent unlawful entry will generate some such impact. Despite that impact, some policy compelling compliance with this legitimate interest seems indispensable, lest that interest be a mere suggestion. Thus, the government would have strong arguments that the legitimate interest would be substantially impaired absent this policy, in a way that demonstrates that the need for the policy outweighs the persistent taint-rooted harm.

The strongest counterargument would be the plausibility of preventing unlawful entry through civil violations only, rather than criminal violations.351×351. See, e.g., Caitlin Dickerson, Some Democrats Want to Decriminalize Illegal Border Crossings. Would It Work?, N.Y. Times (July 31, 2019), https://www.nytimes.com/2019/07/31/us/border-crossing-decriminalization.html [https://perma.cc/Y3BA-MFE9]. While this counter is substantial, taint does not mandate strict scrutiny–like narrow tailoring. Given a weighty, legitimate interest in preventing unlawful entry, and the magnitude of attempts to accomplish such entry,352×352. See, e.g., Joel Rose, Border Patrol Apprehensions Hit a Record High. But That’s Only Part of the Story, NPR (Oct. 23, 2021, 7:47 AM), https://www.npr.org/2021/10/23/1048522086/border-patrol-apprehensions-hit-a-record-high-but-thats-only-part-of-the-story [https://perma.cc/R3NZ-2ZM5]; Eileen Sullivan & Miriam Jordan, Illegal Border Crossings, Driven by Pandemic and Natural Disasters, Soar to Record High, N.Y. Times (Oct. 22, 2021), https://www.nytimes.com/2021/10/22/us/politics/border-crossings-immigration-record-high.html“>https://www.nytimes.com/2021/10/22/us/politics/border-crossings-immigration-record-high.html”>https://www.nytimes.com/2021/10/22/us/politics/border-crossings-immigration-record-high.html [https://perma.cc/E8NW-P3PM]. a court can reasonably accept the government’s claim that the pursuit of that interest would be substantially impaired if it were precluded from using criminal sanctions. Note, still, that the necessity claim is specific and tied to the policy’s existence in this form; a vague invocation of “deterrence” as a legitimate interest would be insufficient. To pass muster, the claim also has to be tied to the specific, history-conscious engagement sketched above.

To be sure, one might reject the strength of the interest in preventing unlawful entry or conclude that on balance the availability of civil penalties vitiates the necessity claim for criminal penalties. The foregoing analysis simply aims to demonstrate that a robust argument for purging is possible on these facts, given the precision with which the interest can be stated and connected to the policy. In the end, though, these examples are given not because the standard mandates only one possible answer, but to show how the taint standard can regiment and routinize the important questions we should care about.

5. Lower Court Felon Disenfranchisement and Voter ID Cases. — Finally, a conceptual note on the lower court felon disenfranchisement and voter ID cases may illuminate taint’s value. I will limit my analysis to explaining how taint would clarify and streamline the appropriate analysis.

The felon disenfranchisement cases, and one of the voter ID cases, advanced the temporally minimalist proposition that any substantive change severed any potential tainted continuity, with one going so far as to treat the removal of a sunset provision as making all the difference.353×353. See Hayden v. Paterson, 594 F.3d 150, 164–67 (2d Cir. 2010); see also supra section I.C.1, pp. 1204–07. Taint precludes such a cursory analysis. Assuming present-day disparate impact,354×354. See, e.g., Derek T. Muller, The Democracy Ratchet, 94 Ind. L.J. 451, 459 & n.56 (2019) (noting and collecting sources on disparate impact of felony disenfranchisement laws); see also supra note 265. the right question is whether the state can justify restricting access to the vote in this tainted manner. These cases cannot be resolved by simply invoking the Fourteenth Amendment’s apparent general endorsement of some felon disenfranchisement.355×355. See Richardson v. Ramirez, 418 U.S. 24, 54 (1974) (relying on section 2 of the Fourteenth Amendment’s “affirmative sanction” of such laws).

The fact-intensive nature of the analysis cautions against fact-independent speculation regarding the outcomes of these varied cases. But taint does place some governmental defenses out of bounds. For example, because a legitimate, nondiscriminatory interest that would be substantially impaired is necessary, vague invocations of (for example) “fraud” would be insufficient. Of course, if the state could show actual fraud to which the policy in question directly responded, the story might be different. These cases also suggest how other constitutional values would affect the analysis. Weighted against any state justification must be voting’s constitutionally “fundamental” status.356×356. Harper v. Va. Bd. of Elections, 383 U.S. 663, 670 (1966).

Other aspects of the lower court opinions can be clarified along lines similar to those previously discussed in relation to Abbott. These cases cannot be resolved by talismanic invocations of the burden of proof and presumptions of good faith.357×357. See, e.g., N.C. State Conf. of the NAACP v. Raymond, 981 F.3d 295, 303–05 (4th Cir. 2020); Veasey v. Abbott, 888 F.3d 792, 801–02 (5th Cir. 2018). But see Raymond, 981 F.3d at 306–07 (suggesting that a ballot measure was a relevant “intervening event,” id. at 306). Taint’s effect must be understood precisely, as going to burdens of production, not burdens of proof.

D. Taint as a Nonjudicial Tool

To this point, this section has prioritized operationalizing taint in a way that can help judges. But taint is a “constitutional concept” that any political actor can understand, although its decisionmaking implications may differ depending on the actor.358×358. See Lawrence Gene Sager, Fair Measure: The Legal Status of Underenforced Constitutional Norms, 91 Harv. L. Rev. 1212, 1213–14 (1978). I pause here to examine those broader possibilities.

First, taint can shape how nonjudicial actors understand their obligations to their constituents. So, for example, a city council or legislature might gather information on a current policy’s link to a discriminatory predecessor, with an eye for the continuity Part II describes. And, in a more demosprudential or popular constitutionalist vein,359×359. See Lani Guinier & Gerald Torres, Changing the Wind: Notes Toward a Demosprudence of Law and Social Movements, 123 Yale L.J. 2740, 2749 (2014). an engaged citizenry could draw on the language of taint to prompt such information gathering (or to guide independent information gathering).

Second, nonjudicial taint conclusions need not duplicate judicial conclusions. Practically speaking, these actors “do” different things. Courts must resolve concrete cases brought to them based on the information the parties provide in an adversarial setting, under constraints like rules of evidence, justiciability, and others.360×360. See, e.g., Balkin, supra note 30, at 1829–31. The relative freedom of other actors may mean they simply will have more — or better — information bearing on taint.

Conclusions may differ for deeper reasons. Even if a constitutional challenge invoking taint fails in litigation, nonjudicial actors could conclude that it exists and that it warrants remediation. After all, information access aside, there can be room for disagreement about taint at multiple stages of this analysis.361×361. See infra section V.A, pp. 1261–65. If this is not a feature, it is at least not a bug. Constitutional law often generates reasonable disagreement,362×362. See, e.g., Adam M. Samaha, Dead Hand Arguments and Constitutional Interpretation, 108 Colum. L. Rev. 606, 656 (2008). and the entailments of the “taint” concept are hardly less determinate than those of “equal protection.”363×363. See Richard H. Pildes, Forms of Formalism, 66 U. Chi. L. Rev. 607, 613 (1999). Within their spheres, nonjudicial actors can apply, and act upon, their independent constitutional judgment.364×364. Although departmentalism and popular constitutionalism can be controversial, “almost no one — and maybe no one at all — thinks that nonjudicial officials should not make constitutional judgments, and act on them, even in some contexts in which their judgments diverge from those that courts have made or would make.” Richard H. Fallon, Jr., Judicial Supremacy, Departmentalism, and the Rule of Law in a Populist Age, 96 Tex. L. Rev. 487, 499 (2018). Since extended exploration of those arguments is beyond this Article’s scope, I should be taken as advocating whatever level of nonjudicial constitutional judgment the reader finds congenial. So far, I simply mean to observe that different actors might agree on constitutional meaning and the approach to divining compliance with that meaning and still reach good faith, disparate conclusions on outcomes. Examples include simple disagreement about the existence, in a particular case, of a sufficient intertemporal nexus, or disagreement about whether the T2 actors sufficiently purged the taint.

Moreover, nonjudicial actors might simply reject a judicial decision rule. Imagine that a court agrees that the factual basis for “taint” exists in a particular equal protection case, but refuses relief for lack of specific intent at T2.365×365. This is one way to understand a recent case rejecting an equal protection claim for lack of “racial animus” despite simultaneously recognizing that “the historical foundation of the current law can be traced back to the earlier statute.” United States v. Machic-Xiap, No. 19-cr-407, 2021 WL 3362738, at *15 (D. Or. Aug. 3, 2021); see id. at *10. That decision would turn on how that court determines compliance with the constitutional equal protection mandate — that is, by searching for specific intent. Nonjudicial actors need not embrace that decision rule. They could instead apply a test of compliance that does not require specific intent in taint cases. This choice could be informed by disagreements over underlying constitutional meaning (say, a belief that wrongful discrimination simply “is” more capacious than actions taken with specific intent), but it need not be. It could be justified on nothing more than a belief that whatever the content of the underlying constitutional meaning, specific intent is orthogonal to enforcing that meaning.366×366. At the margin, the distinction between decision rules and underlying meaning can get murky. See Berman, supra note 216, at 79–83, 108–13; Roosevelt, supra note 218, at 1655. But, as the text accompanying this footnote should indicate, nothing here turns on a bright-line distinction. If the court refuses relief because it simply thinks that taint is inconsistent with the “real” meaning of the Constitution, a nonjudicial actor could still disagree.

Third, taint can justify added judicial deference to a democratic actor’s independent detection and correction of a problem. Judicial and nonjudicial actors, charged equally with following the Constitution,367×367. See U.S. Const. art. VI, cl. 3. might simply disagree in good faith on the taint concept’s applicability or application. Given taint’s constitutional dimensions, and the institutional advantages of nonjudicial bodies, courts should defer to that divergent nonjudicial judgment. Sometimes, added deference would just overdetermine the outcome. But some decisions are of a sort that courts might otherwise scrutinize skeptically, such as race-conscious decisions. There, courts should look approvingly on a decision credibly invoking a taint finding, even if a taint-based litigation challenge to the status quo ante would have failed.

Perhaps the most apt place for such deference is the congressional power to “enforce” the Reconstruction Amendments “by appropriate legislation.”368×368. Balkin, supra note 30, at 1808 (quoting U.S. Const. amend. XIII, § 2). I would approach similarly the several later amendments that contain similar language. See id. I bracket questions regarding whether and to what extent the power to “enforce” includes congressional power to determine constitutional meaning.369×369. See, e.g., Gillian E. Metzger, Congress, Article IV, and Interstate Relations, 120 Harv. L. Rev. 1468, 1527 & n.246 (2007); McConnell, supra note 29, at 170–71. Irrespective of one’s view on judicial interpretive supremacy, “appropriate” at least should embrace deference to a congressional act premised on a factual finding of taint and otherwise claiming no interpretive authority.370×370. This much seems justifiable given the Reconstruction Framers’ well-established distrust of the judiciary. See, e.g., McConnell, supra note 29, at 181–84; Balkin, supra note 30, at 1850 & n.185. Indeed, for two of the three Reconstruction Amendment enforcement powers, judicial doctrine already countenances that view.

Most clearly, the Court’s extant approach to the Thirteenth Amendment enforcement power defers to congressional power “rationally to determine . . . the badges and the incidents of slavery.”371×371. Jones v. Alfred H. Mayer Co., 392 U.S. 409, 440 (1968). This approach recognizes that meaningful aspects of the “slavery” that section 1 abolishes can persist. Just as important, it recognizes the propriety of deferring to Congress’s judgment that it has persisted.

Taint-based deference also fits the Fourteenth Amendment context. True, the Court here has cut back on more expansive visions, insisting that appropriate legislation is only that which is “remedial” in a way that is “congruent and proportional” to the Supreme Court’s understanding of constitutional meaning.372×372. City of Boerne v. Flores, 521 U.S. 507, 519–20 (1997); see also McConnell, supra note 29, at 170–74 (critiquing City of Boerne’s view of the enforcement power). The Court has not applied the “congruent and proportional” test to the Thirteenth and Fifteenth Amendments. Putting aside the critiques of that approach,373×373. See, e.g., Akhil Reed Amar, Intratextualism, 112 Harv. L. Rev. 747, 818–27 (1999); McConnell, supra note 29, at 170–74. a congressional finding of taint should still generate increased deference. After all, the congruent-and-proportional test still permits prophylactic action to “deter[] or remed[y] constitutional violations,”374×374. City of Boerne, 521 U.S. at 518. so long as Congress does not claim “to define the substance of constitutional guarantees.”375×375. Nev. Dep’t of Hum. Res. v. Hibbs, 538 U.S. 721, 728 (2003). Treating taint as a predicate for the exercise of the Fourteenth Amendment enforcement power requires no substantive redefinition. It need imply only that Congress is persuaded of the need to remedy the persistent taint of a constitutional violation that a court has found or would find.376×376. Cf. City of Boerne, 521 U.S. at 536 (“It is for Congress in the first instance to ‘determin[e] whether and what legislation is needed to secure the guarantees of the Fourteenth Amendment,’ and its conclusions are entitled to much deference.” (quoting Katzenbach v. Morgan, 384 U.S. 641, 651 (1966))). And it should matter to a court when Congress is persuaded.

As for the Fifteenth Amendment, the most notable recent case, Shelby County v. Holder,377×377. 570 U.S. 529 (2013). illuminates precisely because of its insufficient deference to taint-based legislative judgment. At its heart, the case was about the enforcement power implications of Congress independently finding and addressing taint. Specifically, it boiled down to the permissibility of Congress’s conclusion that the discrimination that originally justified the VRA’s preclearance coverage formula persisted in an evolved form that supported carrying the formula forward unchanged.378×378. See id. at 536, 553–57. Compare id. at 554 (asserting that the “fundamental problem” was that “Congress . . . reenacted a formula based on 40-year-old facts having no logical relation to the present day”), with id. at 559 (“The question . . . is who decides whether [the Act] . . . remains justifiable, this Court, or [the] Congress . . . .” (Ginsburg, J., dissenting) (footnote omitted)), and id. at 575–76 (“Congress also found that voting discrimination had evolved into subtler second-generation barriers . . . . The evidence . . . grounded Congress’ conclusion that the remedy should be retained . . . .”). Even granting the norm of “equal state sovereignty” as the Court articulated it,379×379. But see Leah M. Litman, Inventing Equal Sovereignty, 114 Mich. L. Rev. 1207, 1211–16 (2016) (criticizing the concept’s deployment in Shelby County). the decision turned on whether Congress’s taint-based judgment sufficed to justify a deviation from that norm. Taking taint seriously as a constitutional concept within Congress’s concurrent bailiwick would have produced a majority opinion much more like Justice Ginsburg’s dissent and the original validation of the VRA in South Carolina v. Katzenbach.380×380. 383 U.S. 301 (1966).

Beyond the Federal Congress, other nonjudicial actors can also invoke taint to claim deference. I will sketch three quick examples.

Consider, initially, how President Joseph Biden’s revocation of President Donald Trump’s travel bans indicted them as “discriminatory” and inconsistent with “religious freedom and tolerance.”381×381. Biden Administration Reverses Trump Administration Policies on Immigration and Asylum, 115 Am. J. Int’l L. 340, 341–42 (2021) (quoting Ending Discriminatory Bans on Entry to the United States, Proclamation No. 10,141, 86 Fed. Reg. 7,005, 7,005 (Jan. 25, 2021) [hereinafter Biden Proclamation]). Although I would prefer a more involved analysis of the taint-related basis for those assertions, President Biden’s proclamation does provide a template for a taint-conscious, nonjudicial divergent judgment. Ideally, such a judgment would apply the approach to detecting taint outlined in Part II. Note, however, that this Part’s burden-shifting constraints on what courts should do before a taint-based invalidation382×382. See supra sections IV.A–B, pp. 1237–45. are less integral to what nonjudicial actors should do when deciding how to respond to taint.

Additionally, President Biden’s proclamation went beyond mere termination to embrace measures aimed at giving individuals who were denied visas under the previous regime a second chance under an untainted one.383×383. Biden Proclamation, 86 Fed. Reg. at 7,005; see also Biden Administration Reverses Trump Administration Policies on Immigration and Asylum, supra note 381, at 342. Explicitly framing the revocation as a taint-based constitutional decision could have justified even more substantial reparative measures that, for example, expressly targeted race, religion, or nationality.

For another example, consider the facts in Parents Involved in Community Schools v. Seattle School District No. 1.384×384. 551 U.S. 701 (2007). No court ever held Seattle’s schools to have been segregated by law.385×385. See id. at 712; id. at 806–07, 820 (Breyer, J., dissenting). But, during the decades following Brown, multiple lawsuits challenged the city’s monochromatic schools as the unlawful products of school board action.386×386. See id. at 808–10 (Breyer, J., dissenting). The board responded to those lawsuits by adopting integrative plans and — in one instance — by reaching a formal settlement agreement.387×387. See id. at 807.

The Parents Involved Court indicated that the absence of a formal court holding of de jure segregation barred Seattle from invoking past discrimination as an interest justifying its race-conscious school-assignment plans.388×388. See id. at 720–21 (majority opinion). A taint-sensitive analysis would counsel otherwise. Sufficient evidence existed from which a modern board could conclude that past school-board policies in fact constituted wrongful discrimination.389×389. See, e.g., id. at 807–09 (Breyer, J., dissenting) (recounting, inter alia, past school board memoranda acknowledging racially disparate student-transfer policies and lawsuits attributing monochromatic schools to race-conscious school-attendance policies and district boundary lines). Irrespective of whether a court formally declared those policies invalid, taint as a constitutional concept recognizes modern decisionmakers as competent to conclude that contemporary policies with persistent disparate effect are tainted by that past. If they so conclude, and act to purge that taint, this Article’s model calls for judicial deference to that constitutionally based judgment.

Finally, more hypothetically, imagine the Supreme Court had applied a taint analysis in Ramos and upheld the Oregon nonunanimous jury provision that the state had reenacted decades after its original, problematic enactment.390×390. See Ramos v. Louisiana, 140 S. Ct. 1390, 1426 (2020) (Alito, J., dissenting). Actually, Oregon was a nonparty, but I tweak the facts for simplicity since Louisiana had repealed its law by the time of Ramos. Oregon could always have repealed the provision. But nonjudicial actors also could have independently applied the basic idea of taint to argue for repeal. Indeed, my proposed judicial decision rule could be instructive here: they might argue that complete revocation is necessary because the policy is insufficiently important to justify the amount of disparate impact that would persist if retained in any form.

Importantly, once detected, taint can justify going beyond mere repeal. Those steps could be flexible and targeted, untrammeled by doctrines like retroactivity or the permissible scope of judicial relief. In the Ramos-based example, new trials for convicted persons might be one option. Another might be targeted investment in areas whose residents were disproportionately affected by the provision. For an election law example, take Virginia’s recent State Voting Rights Act, passed by a state formerly subject to preclearance under the Federal VRA.391×391. See Reid J. Epstein & Nick Corasaniti, Virginia, the Old Confederacy’s Heart, Becomes a Voting Rights Bastion, N.Y. Times (June 8, 2021), https://www.nytimes.com/2021/04/02/us/politics/virginia-voting-rights-northam.html [https://perma.cc/69PH-AKHL]; see also Va. Code Ann. § 24.2-105 (West 2021), https://lis.virginia.gov/cgi-bin/legp604.exe?212+ful+CHAP0533+pdf [https://perma.cc/B2C9-SSYQ]. Although the legislative text does not explicitly draw this connection, Virginia’s legislature could have invoked taint to justify the new burdens the new law places on state and local actors. In all events, these policies should benefit, if challenged, from the deference that nonjudicial invocation of taint deserves.

V. Moving Forward

I close by exploring some complications and potential areas for future consideration.

A. Complications

Three complications may help illuminate the taint concept further. First, how should we think about showing taint in litigation? Second, what if the government contests T1’s wrongfulness? Third, how might this approach alter incentives?

1. Methods of Proof. — Taint is a relationship, inferred from historical facts, that may aid the proof of an underlying constitutional violation. Thus, plaintiffs should include facts supporting the relationship’s existence in their complaint and deploy them in line with this Article’s framework if a defendant challenges the underlying claim’s viability. Then, the court should determine whether the invocation of taint satisfies the requirements applicable at that stage of litigation; if the case goes to trial, the ultimate factfinder must decide.392×392. Cf. Alexandra D. Lahav, Procedural Design, 71 Vand. L. Rev. 821, 823 (2018) (discussing the staged, escalating nature of “textbook” civil procedure).

Decisionmakers must apply the same judgment used to resolve other contested questions. So, for example, a court might ask whether it is plausible that the ostensible T1 and T2 are functional duplicates, or whether a reasonable jury could find that an intervening event supports treating two events as uniquely connected. Like most sensitive, contextual inquiries, the inquiry will never be fully mechanical. Still, nothing about taint requires courts to treat an implausible invocation as plausible.

2. The Contested T1. — What about disagreement over T1? Even where most would say the “origins are clear,” as with the nonunanimous jury provision adopted by the delegates of the avowedly White supremacist 1898 Louisiana constitutional convention, a state may contest those origins.393×393. Compare Ramos, 140 S. Ct. at 1394 (noting that “courts in both Louisiana and Oregon” had “acknowledged that race was a motivating factor” in those states’ adoption of a nonunanimity rule), with id. at 1426 n.2 (Alito, J., dissenting) (observing that those states’ briefs did in fact contest the rules’ racist origins). Even if Louisiana’s refusal to concede the point was not credible in Ramos, surely good faith disagreement will sometimes be possible. To be clear, taint is conceptually useful regardless of this complication, so long as some wrongful T1 policies exist. And this problem will not affect cases arising after a final adjudication of the T1 policy’s wrongfulness.394×394. This Article contains multiple examples of that fact pattern. See, e.g., Palmer v. Thompson, 403 U.S. 217, 218–19 (1971); N.C. State Conf. of the NAACP v. Raymond, 981 F.3d 295, 299 (4th Cir. 2020). I do think, however, that more can be said about this potential problem.

It is not hard to imagine a government obstructing the efficacious ascertainment of whether T1 is discriminatory by replacing it with a policy that has new factual and procedural context, but operates similarly. That is a problem even if taint is only an anti-evasion device.395×395. See supra section III.A, pp. 1227–28; see generally Denning & Kent, supra note 161. Moreover, uncoupling taint from conscious malintent means we can expect some good faith T2 contestation. But taint would be substantially underinclusive of its purposes if its applicability turned on whether the defendant agreed that T1 was problematic. Indeed, it would be least likely to apply where its usefulness may command the widest agreement: specific-intent T2 pretext. Still, the answer cannot be absolute deference to the plaintiff’s characterization.

The best solution recalls that taint works as a methodological heuristic for understanding when history should deny the government whatever clean-slate treatment facially neutral policies normally receive. Within that framework, we do not examine T1 to determine whether the government “should have been” liable in the past. We examine it because T1/T2 continuity makes T1’s circumstances relevant to the government’s T2 justificatory burden. The purpose is not to issue an advisory opinion on T1, but to ascertain legally relevant aspects of T1 to the T2 analysis.

Therefore, it is unhelpful to talk of contesting whether T1 was “really” discriminatory. The question is whether what we know about T1 should trigger a taint analysis at T2, not whether T1 was independently unconstitutional. Courts are well equipped to characterize the meaning of a past event when relevant to evaluating a subsequent, related event.396×396. Consider the issue preclusion context, where courts must determine (and the parties often contest) whether an issue was in fact previously litigated and determined by a valid final judgment to which the determination was essential. See, e.g., Lucky Brand Dungarees, Inc. v. Marcel Fashions Grp., Inc., 140 S. Ct. 1589, 1594 (2020); Stephen N. Subrin et al., Civil Procedure: Doctrine, Practice, and Context 976–79 (5th ed. 2016). Though different, compare the need for courts, sometimes, to “characteriz[e] the significance or meaning of state law for constitutional law purposes,” Henry Paul Monaghan, Supreme Court Review of State-Court Determinations of State Law in Constitutional Cases, 103 Colum. L. Rev. 1919, 1935 (2003), distinguishable from “redetermination of state law,” id. at 1936; see id. at 1935–37. Analogously, courts can characterize the significance of the law at T1 for T2 constitutional law purposes without purporting to “determine” the T1 law. Given continuity, T1’s wrongfulness becomes a material fact for evaluating T2. Decisionmakers can and should resolve disputes about that fact like any other dispute. Accordingly, nothing here advocates retrospective “prophecies of what the courts w[ould have done] in fact,”397×397. Oliver Wendell Holmes, The Path of the Law, 10 Harv. L. Rev. 457, 461 (1897) (equating “the law” to such prophecies). but only a present-day evaluation of past facts. Taint aims to identify present-day situations requiring remediation because of their relationship to the past, not to speculate about what past courts would have done in their day.

Consider Ramos as an exemplar. Modern courts did not simply defer to Louisiana’s characterization of the original nonunanimity rule, or terminate analysis of the original rule’s character because Redemption-era courts would have validated it. Instead, they investigated its wrongfulness by examining, among other things, its adoption at a racism-suffused convention.398×398. Ramos, 140 S. Ct. at 1394. Similarly, a taint analysis in Trump would find it important that past courts found the predecessor bans constitutionally suspect and that Ban 3 was generated in close proximity to the litigation challenging those bans.399×399. See Trump v. Hawaii, 138 S. Ct. 2392, 2403–04 (2018); id. at 2436–38 (Sotomayor, J., dissenting); see also supra pp. 1205–06. As always, the original law’s disparate impact or enforcement (or lack thereof) can support an inference of wrongful discrimination.400×400. See Pers. Adm’r v. Feeney, 442 U.S. 256, 279 nn.25–26 (1979). Nothing about the taint concept requires a final judicial decision at T1.

Hard cases will surely exist, with easy-to-imagine examples including where substantial passage of time reduces the amount of useful T1 evidence. Sometimes, that may mean that courts get it “wrong,” in the sense that they might have reached a different decision with better information. That means the framework may miss some instances, but so would unqualified deference to either party’s characterization. Better, I think, to understand taint as a sometimes-difficult question whose importance justifies the inquiry.

On a final note, taint is everyone’s business, which means the most complete solution is sometimes political. Perhaps courts sometimes should underenforce taint, and perhaps the hardest cases of contested wrongfulness fit that bill.401×401. Cf. Sager, supra note 358, at 1226–27 (discussing institution-dependent constitutional norm enforcement). But other actors need not duplicate the judiciary.402×402. See supra section IV.D, pp 1255–60. So, for example, perhaps a municipality successfully contests the discriminatory character of its past policy. Within constitutional limits, state or federal governments with the power to overrule that municipality might wield taint to justify action that effects remedies similar to those sought in the litigation. Less affirmatively, a government that might be able to contest successfully T1’s wrongfulness might still modify its behavior to obviate a taint attack, for example, by working to eliminate disparate impacts. If taint generates those results via in terrorem effect,403×403. See Daryl J. Levinson, Rights Essentialism and Remedial Equilibration, 99 Colum. L. Rev. 857, 905–06 (1999); Stephanopoulos, supra note 261, at 1633–34. it matters little whether it does so via court order or cautious government bureaucrats. Broadly, then, the concept does not rise or fall on the resolution of this interesting question.

3. Incentives, and Palmer. — Two related critiques might proceed from a fear of bad incentives.

One is stasis. That is, governments fearing taint will fear to update policies because a new policy may well be continuous with an old policy in form and effect (after all, some governmental problems are constant). This would hinder necessary updates in important areas. This seems more of a feature than a bug. We should want to encourage policy change only if such change will not perpetuate past taint. Otherwise, change may work primarily to insulate the problematic character of the old policy from challenge, which is exactly what the taint concept should prevent. Far from dissuading new policy, taint should encourage governments to work affirmatively to purge taint in ways that ensure new policies are in fact new.

Another concern might be the futility fear that animated Palmer v. Thompson.404×404. 403 U.S. 217, 225 (1971). On this view, taint is a time-wasting paper tiger because governments unable to validate their T2 policy will simply go back and “do it better” at T3 and produce essentially the same program. Again, I do not see this objection as uncovering a major demerit.

First, a government’s ability to justify essentially the same program under a heightened standard keyed to the harms of the past is evidence of legitimate purging. Here, taint works similarly to a Calabresian “constitutional remand.”405×405. See Quill v. Vacco, 80 F.3d 716, 738 (2d Cir. 1996) (Calabresi, J., concurring). One might say, for example, that a law impugned by taint is “neither plainly unconstitutional . . . nor plainly constitutional,” at least in the sense that, in its form, it is not per se invalid.406×406. Id. Under those circumstances, taint forces the government to demonstrate, in a substantial way, that legitimate contemporary grounds justify employing this tainted form.407×407. Cf. id. (explaining that “constitutional remand” is for governmental “current and clearly expressed statements . . . of the state interests involved”).

Second, sometimes the government simply will be unable to do it “better,” perhaps because disparate impact persists without a plausibly legitimate reason. So, imagine the city lost in Palmer. Assuming disparate impact,408×408. See infra p. 1266. the government would need new, legitimate reasons to demonstrate that a legitimate interest would be substantially impaired absent a pool closure. This would not mean the city could never close the pools,409×409. See Palmer, 403 U.S. at 230 (Blackmun, J., concurring). but it would require some substantial showing of changed circumstances to show that pools previously thought worth the cost now were not.410×410. The changed circumstances of private resistance to constitutional mandates would not be a valid basis. See Palmore v. Sidoti, 466 U.S. 429, 433–34 (1984); Randall Kennedy, Reconsidering Palmer v. Thompson, 2018 Sup. Ct. Rev. 179, 203.

Third, sometimes, the government will be unable to do it better because (whether due to unreconstructed animus or pandering to animus) it is unwilling to treat the plausibly problematic past as a genuine concern, as taint requires. That might, for example, explain President Trump’s inability to unequivocally disavow his past statements in Trump.411×411. See Trump v. Hawaii, 138 S. Ct. 2392, 2439 & n.4 (2018) (Sotomayor, J., dissenting).

Fourth, taint focuses on severing links to the discriminatory past, not eliminating all justice-based policy problems. If injustice persists, as it often will, the solution must lie in some other legal doctrine (for example, taint does not preclude a conventional equal protection challenge) or in the political process.

B. Future Lessons

Finally, some downstream implications of taint should be highlighted for potential future inquiry.

1. Leveling Down (and Palmer, Again). — Taint may shed some light on the leveling-down problem. That is, an equal-treatment violation can technically be remedied in two ways: providing the benefit to the aggrieved claimant (leveling up) or denying the benefit to all instead of some (leveling down).412×412. See, e.g., Brest et al., supra note 25, at 1122; Hellman, supra note 24, at 127; Pamela S. Karlan, Race, Rights, and Remedies in Criminal Adjudication, 96 Mich. L. Rev. 2001, 2027 (1998). Usually the aggrieved party wants leveling up,413×413. See Karlan, supra note 412, at 2027. and leveling down sometimes smacks of the monkey’s paw.414×414. That is, a wish granted perversely. See, e.g., Thomas B. Colby, The Sacrifice of the New Originalism, 99 Geo. L.J. 713, 715 & n.6 (2011) (citing W.W. Jacobs, The Monkey’s Paw, in The Lady of the Barge and Others, Entire Collection 14 (The Echo Libr. 2009) (1902)).

Taint can help articulate the intuition that something is wrong here. (Re)consider Palmer, a classic leveling-down case.415×415. Karlan, supra note 412, at 2027. Recall that Jackson closed its pools only after an injunction against their previous segregation.416×416. Palmer v. Thompson, 403 U.S. 217, 218–19 (1971). Understanding T1 as the segregated status quo and T2 as the pool closure, a taint analysis permits us to articulate the substantial similarity between the policies despite the facial difference between open and closed pools. The same institutional actor took both actions, and the litigation and injunction as intervening events help cement the two policies’ nexus.

Any objection based on the formal equality of the pool closure (for example, White people were unburdened before but are burdened now) carries us to the second stage of the analysis. On disparate impact, the reality was that more White people had access to private pool alternatives than Black people.417×417. See, e.g., id. at 252 (White, J., dissenting) (observing that a formerly public pool was now operated privately on a White-only basis). It will often be thus: “Disproportionate effects . . . are not difficult to discern in many leveling down cases, given the greater access of nonstigmatized groups to substitute benefits.”418×418. Deborah L. Brake, When Equality Leaves Everyone Worse Off: The Problem of Leveling Down in Equality Law, 46 Wm. & Mary L. Rev. 513, 528 n.49 (2004). To be sure, one can dispute whether this counts as disparate impact, but my point is that at least one legitimate view of the concept would detect it here. Unable to utilize the impact escape hatch, the city would be faced with the burden of showing substantial justification for closing pools that it was happy to operate until being instructed to operate them constitutionally.

In short, leveling down will generally generate clear continuity, channeling the inquiry into considering whether any persistent real-world disparate impact can be justified. And, as is true generally, the availability of a high-level legitimate reason (like cost) won’t suffice to show that the burden of cost outweighs the harm of persistent disparate impact, especially if the cost is the cost of complying with the Constitution. Because taint forces the state to make a real case that leveling down was necessary, it helps pick out some of the more pernicious cases of leveling down (including those undertaken pretextually) and helps explain why they are problematic: they work to lock in the real-world harms of the original policy. This could be an underelaborated explanation of the constitutional problem with Virginia’s originally proposed remedy for its all-male military school in United States v. Virginia419×419. 518 U.S. 515 (1996). : a new all-female military school.420×420. See, e.g., id. at 526–28, 534, 548–49.

2. Dissipation over Time and the Right-Remedy Relationship. — This Article says little about cases where a wrongful act simply persists without modification or reenactment, on the theory that these are easier cases generally and particularly with respect to the temporality problem.421×421. This is Hunter v. Underwood, which did get easy-case treatment. See supra note 64. Although this is probably a solid rule of thumb, harder variants exist, such as when the purpose and meaning of a formally unchanged “longstanding . . . practice[]” supposedly evolves over time.422×422. Am. Legion v. Am. Humanist Ass’n, 139 S. Ct. 2067, 2083 (2019); see id. at 2082–84 (considering this possibility in the Establishment Clause context and noting the Statue of Liberty’s changing meaning).

Taint can help analyze such cases. For example, if change supposedly saves such an otherwise problematic policy, taint can support an inquiry into whether the policy in this T2 context is sufficiently continuous with the T1 context. In a manner analogous to the heartland taint case, we might profitably ask whether the policy continues to act in a sufficiently similar way in the world and whether any intervening events bear on how it should be understood today (such as, but not limited to, changes in policy needs and relevant governing institutions). If such continuity exists, taint might also prompt us to ask whether disparate impact persists.423×423. The analogy is imperfect, since it seems inappropriate to let the state “justify” a formally unchanged T1 that in fact continues to operate in the same way, including disparate impact. At that point, conventional constitutional analysis seems appropriate.

Taint can also inform inquiries into the propriety of continuing injunctive relief for a T1 violation, such as school desegregation cases.424×424. See, e.g., Parker, supra note 8, at 1163–64. If the question is whether the injunctive relief has “worked,” taint can at least underscore that the mere passage of time is not enough to assume past evils have dissipated.425×425. See Boddie, supra note 16, at 1294–96 (critiquing the Supreme Court’s assumption in desegregation cases “that the passage of time cures the constitutional violation,” id. at 1294). Its utility here makes sense. Taint asks questions in the T1/T2 context that resemble the important questions informing the termination of injunctive relief. Indeed, the definition of the present-day rights violation in a taint scenario includes the existence of a past, not-fully-remedied rights violation. In this way, taint is another example of the instability of the distinction often drawn between “rights” and “remedies.”426×426. See Levinson, supra note 403, at 884–85, 900. I thank David Pozen for prompting my appreciation of this implication.

3. State Action. — In the Equal Protection Clause’s terms, this Article deploys taint to reveal “den[ials]” of “equal protection” rather than whether a “state” has acted.427×427. U.S. Const. amend. XIV. Still, the treatment of similarity over time may also illuminate cases where nonstate actors appear to have taken the state’s discriminating oar.

The White Primary Cases may be emblematic. Every case after the first considered whether concededly exclusionary rules constituted state action.428×428. See Issacharoff et al., supra note 36, at 266–78. In the hardest case (Terry), four Justices’ state-action conclusion emphasized continuity.429×429. See Terry v. Adams, 345 U.S. 461, 480 & n.7 (1953) (Clark, J., concurring). That conclusion was less obvious for the private Jaybird Party’s rules than for the original statute barring Black Texans’ participation.430×430. See supra p. 1200. A taint-like consideration of continuity in real-world effect might be an underelaborated basis for Terry: every rule in fact excluded Black Texans from participating in the only meaningful primary. Coupled with the multiple previous Supreme Court interventions in that aspect of Texan governance, that continuity perhaps revealed the initial state action’s persistent “taint.”

This reasoning would extend the thesis. Rather than treating institutional continuity as a prerequisite to taint, taint would play a role in showing that continuity. I am more tentative on this necessarily less-constraining framework’s desirability. But the potential extension is at least plausible where a public-to-private shift appears to thwart constitutional scrutiny. Even sans institutional continuity, the framework seems apt for determining whether the state “has become entangled with a private entity” or “approved, encouraged, or facilitated private conduct.”431×431. Geoffrey R. Stone et al., Constitutional Law 1536 (8th ed. 2017). Absent institutional continuity, additionally, the framework might be informative for analysis of problematic “borrowing” between jurisdictions. Again, this is a tentative observation warranting future analysis.

4. Other Wrongful Discrimination. — In taking equal-treatment principles as the paradigm, this Article focuses primarily on the Equal Protection Clause and Religion Clause cases that basically raise equal-treatment complaints.432×432. See supra p. 1198. For example, the Establishment Clause claim in Trump v. Hawaii was that the policy “singl[ed] out Muslims for disfavored treatment.” 138 S. Ct. 2392, 2417 (2018). The only apparent reason this sort of claim is not an equal protection claim is that courts prefer religion-specific constitutional provisions. See, e.g., Burlington N. R.R. Co. v. Ford, 504 U.S. 648, 651 (1992) (recognizing, in dicta, religion as a suspect equal protection class); Caroline Mala Corbin, Nonbelievers and Government Speech, 97 Iowa L. Rev. 347, 379 (2012). Putting aside the scholarly value of depth over breadth, I think, in line with Professor Jack Balkin, that status-discrimination cases are uniquely pernicious.433×433. See Balkin, supra note 31, at 2368. Even if that justifies this Article’s scope, discrimination matters in other constitutional areas, such as content-discrimination free speech doctrine,434×434. See, e.g., Leslie Kendrick, Content Discrimination Revisited, 98 Va. L. Rev. 231, 286–92 (2012). Article IV, and the inferred dormant commerce clause.435×435. See, e.g., Metzger, supra note 369, at 1471–72. Moreover, many federal and state statutory provisions aim to address legislatively defined wrongful discrimination.

Although taint might be transplanted to these areas, different considerations might apply. For example, the Court analyzes some Religion Clause questions arguably involving discrimination in ways quite different from equal protection questions.436×436. See, e.g, Tandon v. Newsom, 141 S. Ct. 1294, 1296 (2021) (per curiam) (holding strict scrutiny appropriate “whenever [government] treat[s] any comparable secular activity more favorably than religious exercise,” which might create de facto disparate impact liability (first emphasis added)); cf. Fulton v. Philadelphia, 141 S. Ct. 1868, 1896–97 (2021) (Alito, J., concurring) (asserting that the Free Exercise Clause contains no equal-treatment principle); Caroline Mala Corbin, Intentional Discrimination in Establishment Clause Jurisprudence, 67 Ala. L. Rev. 299, 304 (2015) (sketching the Establishment Clause’s unsettled doctrinal contours). Dormant commerce clause discrimination’s concern with economic protectionism is not necessarily coextensive with the Equal Protection Clause’s concerns.437×437. See, e.g., Huq, supra note 10, at 1219 n.32, 1220 n.33. Something similar can be said about the Free Speech Clause’s core concerns,438×438. See Kendrick, supra note 434, at 296. which are themselves the subject of considerable debate.439×439. See, e.g., Levinson, supra note 403, at 902–03.

In the statutory realm, despite much similarity between Title VII’s race-discrimination prohibition and constitutional equal protection constraints,440×440. See Charles A. Sullivan, Tortifying Employment Discrimination, 92 B.U. L. Rev. 1431, 1454–55, 1454 n.97 (2012). Title VII’s “motivating factor” analysis permits liability under circumstances that would not support constitutional liability.441×441. See, e.g., id. at 1454–55 (observing possible lack of symmetry in Title VII and constitutional “intent” discourse); Verstein, supra note 258, at 1155–59. So too for section 2 of the VRA.442×442. See Brnovich v. Democratic Nat’l Comm., 141 S. Ct. 2321, 2332 (2021).

My sense is that this Article’s view of policy continuity is largely portable, but that how we treat such continuity may vary as substantive contexts change. For example, taint might fit best with statutes that expressly reject the specific-intent view of discrimination (for example, Title VII and the VRA). More broadly, the core concerns animating the constitutional or statutory provision may warrant different treatment of T1 policies within its sphere. Regardless, this Article’s definition of taint permits its location in a way that does not lock us into a single way of dealing with it across subject matters. That is a final benefit of staging the analysis.

Conclusion

It is useful, but sometimes inadequate, to know that history “matters.” When constitutional discrimination is at issue, cases of policy continuity over time show that truism’s limits. Here, at least, we can and should say more about how that history matters. By letting us identify problematic continuity in a regimented way, discriminatory taint offers one novel way to do so.

The concept does not answer every associated question. Indeed, properly understood, knowing that a policy is tainted does not dictate its validity. But taint offers reasoned guideposts that shape an important inquiry in ways consistent with broader commitments and consonant with the need to eradicate discrimination and avoid unwarranted “lock in.” In so doing, it advances antidiscrimination principles in a transsubstantive way and offers insight into what we mean — and should mean — when we talk about wrongful discrimination.


* Associate Professor of Law, Columbia Law School. For helpful conversations, I thank participants in workshops and panels at the Columbia Law School Faculty Workshop Series, the University of North Carolina School of Law, the 2021 John Mercer Langston Workshop at Berkeley Law, the 2021 Minnesota Law School Public Law Workshop, the 2021 Culp Colloquium at Duke Law, and the 2022 National Conference of Constitutional Law Scholars at the University of Arizona James E. Rogers College of Law’s Rehnquist Center. Special thanks to Jessica Bulman-Pozen, John Coyle, Justin Driver, Kevin Douglas, Luis Fuentes-Rohwer, Kellen Funk, Paul Gowder, Jamal Greene, Deborah Hellman, Esther Hong, Eisha Jain, Kevin E. Jason, Craig Konnoth, Daniel Harawa, Vinay Harpalani, Aziz Huq, Tim Lovelace, Bill Marshall, Kim Forde-Mazrui, Henry Monaghan, David E. Pozen, Jed Purdy, Thomas Schmidt, Micah J. Schwartzman, Fred O. Smith, Jr., Susan Sturm, and Chas Tyler. I received excellent research assistance from Michael Boyd, Jake Pagano, and Xin Wang. Thanks are also due to the Columbia Law School librarians, especially Lena Rieke. I also greatly appreciate the careful attention of the team of editors at the Harvard Law Review. Finally, this piece is dedicated to my parents, Walter and Karena, who made it possible. Errors or misunderstandings are my own.