Edwards v. Vannoy

593 U.S. 255

• Court: Supreme Court of the United States
• Decided: May 17, 2021
• Docket: 19-5807
• Status: Published

Opinion

(Slip Opinion) OCTOBER TERM, 2020 1

Syllabus

NOTE: Where it is feasible, a syllabus (headnote) will be released, as is being done in connection with this case, at the time the opinion is issued. The syllabus constitutes no part of the opinion of the Court but has been prepared by the Reporter of Decisions for the convenience of the reader. See United States v. Detroit Timber & Lumber Co., 200 U. S. 321, 337.

SUPREME COURT OF THE UNITED STATES

EDWARDS v. VANNOY, WARDEN

CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT

No. 19–5807. Argued December 2, 2020—Decided May 17, 2021 In 2007, a Louisiana jury found petitioner Thedrick Edwards guilty of armed robbery, rape, and kidnapping. At the time, Louisiana law per- mitted non-unanimous jury verdicts if at least 10 of the 12 jurors found the defendant guilty. In Edwards’s case, 11 of 12 jurors returned a guilty verdict as to some crimes, and 10 of 12 jurors returned a guilty verdict as to others. After Edwards’s conviction became final on direct review, Edwards filed a federal habeas corpus petition, arguing that the non-unanimous jury verdict violated his constitutional right to a unanimous jury. The District Court rejected Edwards’s claim as fore- closed by Apodaca v. Oregon, 406 U. S. 404, and the Fifth Circuit de- nied a certificate of appealability. While Edwards’s petition for a writ of certiorari was pending, the Court repudiated Apodoca and held that a state jury must be unanimous to convict a criminal defendant of a serious offense. Ramos v. Louisiana, 590 U. S. ___. Edwards now ar- gues that the Ramos jury-unanimity rule applies retroactively on fed- eral collateral review. Held: The Ramos jury-unanimity rule does not apply retroactively on federal collateral review. Pp. 5–20. (a) A new rule of criminal procedure applies to cases on direct re- view, even if the defendant’s trial has already concluded. But the Court has stated that new rules of criminal procedure ordinarily do not apply retroactively on federal collateral review. The Court has stated that a new procedural rule will apply retroactively on federal collateral review only if the new rule constitutes a “watershed” rule of criminal procedure. Teague v. Lane, 489 U. S. 288, 311 (plurality opinion). When the Teague Court first articulated that “watershed” exception, however, the Court stated that it was “unlikely” that such watershed “components of basic due process have yet to emerge.” Id., at 313. And 2 EDWARDS v. VANNOY

in the 32 years since Teague, the Court has never found that any new procedural rule actually satisfies the purported exception. Pp. 5–7. (b) To determine whether Ramos applies retroactively on federal col- lateral review, the Court must first ask whether Ramos announced a new rule of criminal procedure and, if so, whether that rule falls within an exception for watershed rules of criminal procedure that apply ret- roactively on federal collateral review. The Court concludes that Ra- mos announced a new rule and that the jury-unanimity rule an- nounced by Ramos does not apply retroactively on federal collateral review. Pp. 8–14. (1) The Ramos jury-unanimity rule is new because it was not “dic- tated by precedent existing at the time the defendant’s conviction be- came final,” Teague, 489 U. S., at 301, or “apparent to all reasonable jurists” at that time, Lambrix v. Singletary, 520 U. S. 518, 528. On the contrary, before Ramos, many courts interpreted Apodaca to allow for non-unanimous jury verdicts in state criminal trials. And the Ramos Court expressly repudiated Apodaca. Pp. 8–10. (2) The new rule announced in Ramos does not qualify as a “wa- tershed” procedural rule that applies retroactively on federal collateral review. In an attempt to distinguish Ramos from the long line of cases where the Court has declined to retroactively apply new procedural rules, Edwards emphasizes three aspects of Ramos: (i) the significance of the jury-unanimity right; (ii) Ramos’s reliance on the original mean- ing of the Constitution; and (iii) the effect of Ramos in preventing ra- cial discrimination in the jury process. But the Court has refused to retroactively apply other momentous cases with similar attributes. In DeStefano v. Woods, 392 U. S. 631, the Court declined to retroactively apply Duncan v. Louisiana, 395 U. S. 145, even though Duncan estab- lished the jury right itself. In Whorton v. Bockting, 549 U. S. 406, the Court declined to retroactively apply Crawford v. Washington, 541 U. S. 36, even though Crawford relied on the original meaning of the Sixth Amendment to restrict the use of hearsay evidence against crim- inal defendants. And in Allen v. Hardy, 478 U. S. 255 (per curiam), the Court declined to retroactively apply Batson v. Kentucky, 476 U. S. 79, even though Batson held that state prosecutors may not discrimi- nate on the basis of race when exercising individual peremptory chal- lenges. There is no good rationale for treating Ramos differently from Duncan, Crawford, and Batson. Pp. 10–14. (3) Given the Court’s numerous precedents holding that landmark and historic decisions announcing new rules of criminal procedure do not apply retroactively on federal collateral review, the Court acknowl- edges that the watershed exception is moribund and that no new rules of criminal procedure can satisfy the purported exception for water- shed rules. Continuing to articulate a theoretical exception that never Cite as: 593 U. S. ____ (2021) 3

actually applies in practice offers false hope to defendants, distorts the law, misleads judges, and wastes the resources of defense counsel, prosecutors, and courts. Moreover, no one can reasonably rely on an exception that is non-existent in practice, so no reliance interests can be affected by forthrightly acknowledging reality. The watershed ex- ception must “be regarded as retaining no vitality.” Herrera v. Wyo- ming, 587 U. S. ___, ___ (slip op., at 11) (internal quotation marks omit- ted). Pp. 14–15. Affirmed.

KAVANAUGH, J., delivered the opinion of the Court, in which ROBERTS, C. J., THOMAS, ALITO, GORSUCH, and BARRETT, JJ., joined. THOMAS, J., filed a concurring opinion, in which GORSUCH, J., joined. GORSUCH, J., filed a concurring opinion, in which THOMAS, J., joined. KAGAN, J., filed a dissenting opinion, in which BREYER and SOTOMAYOR, JJ., joined. Cite as: 593 U. S. ____ (2021) 1

Opinion of the Court

NOTICE: This opinion is subject to formal revision before publication in the preliminary print of the United States Reports. Readers are requested to notify the Reporter of Decisions, Supreme Court of the United States, Wash- ington, D. C. 20543, of any typographical or other formal errors, in order that corrections may be made before the preliminary print goes to press.

SUPREME COURT OF THE UNITED STATES _________________

No. 19–5807 _________________

THEDRICK EDWARDS, PETITIONER v. DARREL VANNOY, WARDEN ON WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT [May 17, 2021]

JUSTICE KAVANAUGH delivered the opinion of the Court. Last Term in Ramos v. Louisiana, 590 U. S. ___ (2020), this Court held that a state jury must be unanimous to con- vict a criminal defendant of a serious offense. Ramos repu- diated this Court’s 1972 decision in Apodaca v. Oregon, 406 U. S. 404, which had allowed non-unanimous juries in state criminal trials. The question in this case is whether the new rule of criminal procedure announced in Ramos applies retroactively to overturn final convictions on federal collat- eral review. Under this Court’s retroactivity precedents, the answer is no. This Court has repeatedly stated that a decision an- nouncing a new rule of criminal procedure ordinarily does not apply retroactively on federal collateral review. See Teague v. Lane, 489 U. S. 288, 310 (1989) (plurality opin- ion); see also Linkletter v. Walker, 381 U. S. 618, 639–640, and n. 20 (1965). Indeed, in the 32 years since Teague un- derscored that principle, this Court has announced many important new rules of criminal procedure. But the Court has not applied any of those new rules retroactively on fed- eral collateral review. See, e.g., Whorton v. Bockting, 549 2 EDWARDS v. VANNOY

U. S. 406, 421 (2007) (Confrontation Clause rule recognized in Crawford v. Washington, 541 U. S. 36 (2004), does not apply retroactively). And for decades before Teague, the Court also regularly declined to apply new rules retroac- tively, including on federal collateral review. See, e.g., DeStefano v. Woods, 392 U. S. 631, 635 (1968) (per curiam) (jury-trial rule recognized in Duncan v. Louisiana, 391 U. S. 145 (1968), does not apply retroactively). In light of the Court’s well-settled retroactivity doctrine, we conclude that the Ramos jury-unanimity rule likewise does not apply retroactively on federal collateral review. We therefore affirm the judgment of the U. S. Court of Ap- peals for the Fifth Circuit. I On the night of May 13, 2006, in Baton Rouge, Louisiana, Thedrick Edwards and an accomplice kidnapped Ryan Eaton, a student at LSU. As Eaton was getting out of his car, Edwards and his accomplice confronted Eaton at gun- point and forced him back into the car. Edwards and his accomplice then jumped into the car with Eaton. They drove with Eaton to an ATM where they hoped to withdraw money using Eaton’s card. When they discovered that Eaton did not have any money in his account, they drove to Eaton’s apartment. Once there, they bound and blindfolded Eaton, rummaged through his apartment, and took some of his belongings to Eaton’s car. After they were back in the car, Edwards and his accom- plice coerced Eaton into arranging a meeting with Eaton’s girlfriend. They then drove to the girlfriend’s apartment and, at gunpoint, forced Eaton to knock on the door. When Eaton’s girlfriend opened the door, Edwards and his accom- plice rushed inside. Both Edwards and his accomplice were armed, and Edwards’s accomplice had his gun drawn. Ed- wards and his accomplice instructed Eaton, Eaton’s girl- friend, and two other women in the apartment to lie on the Cite as: 593 U. S. ____ (2021) 3

floor. Edwards then raped one of the women. His accom- plice raped another woman. As they left, they grabbed some personal property from the apartment. Edwards and his accomplice hurried back into Eaton’s car and drove around the corner. They then abandoned the car and fled. Two days later, Edwards and his accomplice confronted another man at gunpoint and forced him to withdraw money from an ATM. Within a day of the second incident, the police collected substantial evidence implicating Edwards in both episodes. The police obtained warrants to search his residence and to arrest him. The day after the police executed the search warrant but before an arrest, Edwards turned himself in to the police and confessed to his crimes. The police vide- otaped Edwards’s confession. (The video is part of the joint appendix. See supremecourt.gov/media/media.aspx.) Edwards was indicted in Louisiana state court for armed robbery, kidnapping, and rape. Edwards pled not guilty and went to trial. Before trial, Edwards moved to suppress the videotaped confession on the ground that the confession was involuntary. The trial court denied the suppression motion. At trial, the jury heard Edwards’s confession and other evidence against him, including the testimony of eyewit- nesses. The jury convicted Edwards of five counts of armed robbery, two counts of kidnapping, and one count of rape. At the time, Louisiana law permitted guilty verdicts if at least 10 of the 12 jurors found the defendant guilty. The jury convicted Edwards by an 11-to-1 vote on one of the armed robbery counts, the two kidnapping counts, and the rape count. The jury convicted Edwards by a 10-to-2 vote on the four remaining armed robbery counts. At sentencing, the trial judge stated: “I can say without hesitation that this is the most egregious case that I’ve had before me.” Record 1113. The judge sentenced Edwards to 4 EDWARDS v. VANNOY

life imprisonment without parole. The Louisiana First Cir- cuit Court of Appeal affirmed the conviction and sentence. In March 2011, Edwards’s conviction became final on direct review. After his conviction became final, Edwards applied for state post-conviction relief in the Louisiana courts. The Louisiana courts denied relief. In 2015, Edwards filed a petition for a writ of habeas cor- pus in the U. S. District Court for the Middle District of Louisiana. He argued that the non-unanimous jury verdict violated his constitutional right to a unanimous jury. The District Court rejected that claim as foreclosed by this Court’s 1972 decision in Apodaca v. Oregon, 406 U. S. 404. In Apodaca, this Court ruled that the Constitution does not require unanimous jury verdicts in state criminal trials. The Apodaca majority consisted of a plurality opinion by four Justices and an opinion concurring in the judgment by Justice Powell. In his opinion, Justice Powell acknowl- edged that the Sixth Amendment requires a unanimous jury in federal criminal trials. 406 U. S., at 371. But in his view, the Fourteenth Amendment did not incorporate that right against the States, meaning that a unanimous jury was not constitutionally required in state criminal trials. Id., at 373, 376–377. In subsequent years, many federal and state courts viewed Justice Powell’s opinion as the con- trolling opinion from Apodaca. See, e.g., Timbs v. Indiana, 586 U. S. ___, ___, n. 1 (2019) (slip op., at 3, n. 1); McDonald v. Chicago, 561 U. S. 742, 766, n. 14 (2010). In Edwards’s case, the District Court likewise followed Justice Powell’s opinion from Apodaca and concluded that a unanimous jury is not constitutionally required in state criminal trials. The U. S. Court of Appeals for the Fifth Cir- cuit denied a certificate of appealability. 2019 WL 8643258 (May 20, 2019). Edwards then petitioned for a writ of cer- tiorari in this Court, arguing that the Constitution requires a unanimous jury in state criminal trials. Cite as: 593 U. S. ____ (2021) 5

II While Edwards’s petition for certiorari was pending, this Court decided Ramos and rejected Justice Powell’s opinion in Apodaca. See Ramos v. Louisiana, 590 U. S. ___ (2020); Apodaca v. Oregon, 406 U. S. 404 (1972). The Court held that the Fourteenth Amendment incorporates the Sixth Amendment right to a unanimous jury against the States. Therefore, in state court as well as federal court, a jury must be unanimous to convict a defendant of a serious of- fense.1 The Court’s decision in Ramos directly affected Louisiana and Oregon, which were the only two States that still al- lowed non-unanimous juries. For those States, this Court’s decision in Ramos immediately triggered a pressing ques- tion: Does Ramos apply retroactively to overturn final con- victions on federal collateral review? We granted certiorari in Edwards’s case to decide that question. 590 U. S. ___ (2020). We conclude that Ramos does not apply retroac- tively on federal collateral review. A A new rule of criminal procedure applies to cases on di- rect review, even if the defendant’s trial has already con- cluded. See Griffith v. Kentucky, 479 U. S. 314, 328 (1987). But under the habeas corpus statute as interpreted by this Court, a new rule of criminal procedure ordinarily does not apply retroactively to overturn final convictions on federal collateral review. See Teague v. Lane, 489 U. S. 288, 310 (1989) (plurality opinion); Penry v. Lynaugh, 492 U. S. 302,

—————— 1 Ramos does not apply to defendants charged with petty offenses,

which typically are offenses that carry a maximum prison term of six months or less. 590 U. S., at ___, n. 7. (slip op., at 3, n. 7); see also Blan- ton v. North Las Vegas, 489 U. S. 538, 543 (1989) (defining petty offense). 6 EDWARDS v. VANNOY

313–314 (1989).2 In stating that new procedural rules ordinarily do not ap- ply retroactively on federal collateral review, Teague rein- forced what had already been the Court’s regular practice for several decades under the retroactivity standard articu- lated in Linkletter v. Walker, 381 U. S. 618 (1965). Linklet- ter set forth a balancing test for determining retroactivity. But even under Linkletter, “new rules that constituted clear breaks with the past generally were not given retroactive effect,” including on federal collateral review. Teague, 489 U. S., at 304 (plurality opinion). As the Court has explained, applying “constitutional rules not in existence at the time a conviction became final seriously undermines the principle of finality which is es- sential to the operation of our criminal justice system.” Id., at 309. Here, for example, applying Ramos retroactively would potentially overturn decades of convictions obtained in reliance on Apodaca. Moreover, conducting scores of re- trials years after the crimes occurred would require signifi- cant state resources. See Teague, 489 U. S., at 310 (plural- ity opinion). And a State may not be able to retry some defendants at all because of “lost evidence, faulty memory, and missing witnesses.” Allen v. Hardy, 478 U. S. 255, 260 (1986) (per curiam) (internal quotation marks omitted). When previously convicted perpetrators of violent crimes go free merely because the evidence needed to conduct a retrial has become stale or is no longer available, the public suf- fers, as do the victims. See United States v. Mechanik, 475

—————— 2 Before Griffith v. Kentucky, 479 U. S. 314 (1987), the Court some-

times would decline to apply new procedural rules even to cases on direct review. See, e.g., Johnson v. New Jersey, 384 U. S. 719, 721 (1966) (rule announced in Miranda v. Arizona, 384 U. S. 436 (1966), applies only to cases in which the trial began after the date of the Miranda decision). Griffith ended that practice and declared that new rules apply to all cases on direct review. Cite as: 593 U. S. ____ (2021) 7

U. S. 66, 72 (1986). Even when the evidence can be reas- sembled, conducting retrials years later inflicts substantial pain on crime victims who must testify again and endure new trials. In this case, the victims of the robberies, kid- nappings, and rapes would have to relive their trauma and testify again, 15 years after the crimes occurred. Put simply, the “costs imposed upon the States by retro- active application of new rules of constitutional law on ha- beas corpus thus generally far outweigh the benefits of this application.” Sawyer v. Smith, 497 U. S. 227, 242 (1990) (internal quotation marks and alteration omitted). For that reason, the Court has repeatedly stated that new rules of criminal procedure ordinarily do not apply retroactively on federal collateral review. The Court has identified only one possible exception to that principle. The Court has stated that a new procedural rule will apply retroactively on federal collateral review only if it constitutes a “watershed” rule of criminal proce- dure. Teague, 489 U. S., at 311 (plurality opinion). But the Teague Court stated that it was “unlikely” that such water- shed “components of basic due process have yet to emerge.” Id., at 313; see also Whorton v. Bockting, 549 U. S. 406, 417 (2007); Schriro v. Summerlin, 542 U. S. 348, 352 (2004); Ty- ler v. Cain, 533 U. S. 656, 667, n. 7 (2001). And in the 32 years since Teague, as we will explain, the Court has never found that any new procedural rule actually satisfies that purported exception.3

—————— 3 By contrast, a new substantive rule—for example, a rule that partic-

ular conduct cannot constitutionally be criminalized—usually applies retroactively on federal collateral review. See Welch v. United States, 578 U. S. 120, 128–129 (2016). The parties here agree, as do we, that the rule announced in Ramos is procedural. The Ramos rule affects “only the manner of determining the defendant’s culpability,” not the “range of conduct or the class of persons that the law punishes.” Schriro v. Sum- merlin, 542 U. S. 348, 353 (2004) (emphasis deleted). 8 EDWARDS v. VANNOY

B To determine whether Ramos applies retroactively on federal collateral review, we must answer two questions. First, did Ramos announce a new rule of criminal proce- dure, as opposed to applying a settled rule? A new rule or- dinarily does not apply retroactively on federal collateral review. Second, if Ramos announced a new rule, does it fall within an exception for watershed rules of criminal proce- dure that apply retroactively on federal collateral review? 1 Ramos held that a state jury must be unanimous to con- vict a defendant of a serious offense. In so holding, Ramos announced a new rule. A rule is new unless it was “dictated by precedent exist- ing at the time the defendant’s conviction became final.” Teague, 489 U. S., at 301 (plurality opinion). In other words, a rule is new unless, at the time the conviction be- came final, the rule was already “apparent to all reasonable jurists.” Lambrix v. Singletary, 520 U. S. 518, 528 (1997). The starkest example of a decision announcing a new rule is a decision that overrules an earlier case. See Whorton, 549 U. S., at 416. The jury-unanimity requirement announced in Ramos was not dictated by precedent or apparent to all reasonable jurists when Edwards’s conviction became final in 2011. On the contrary, before Ramos, many courts interpreted Apo- daca to allow for non-unanimous jury verdicts in state crim- inal trials.4 In addition, in Ramos itself, six Members of the

—————— 4 See, e.g., Timbs v. Indiana, 586 U. S. ___, ___, n. 1 (2019) (slip op., at

3, n. 1) (the “Sixth Amendment requires jury unanimity in federal, but not state, criminal proceedings”); McDonald v. Chicago, 561 U. S. 742, 766, n. 14 (2010) (“The Court has held that although the Sixth Amend- ment right to trial by jury requires a unanimous jury verdict in federal Cite as: 593 U. S. ____ (2021) 9

Court acknowledged that Apodaca allowed non-unanimous jury verdicts in state criminal trials. See 590 U. S., at ___ (SOTOMAYOR, J., concurring in part) (slip op., at 2); id., at ___ (KAVANAUGH, J., concurring in part) (slip op., at 1); id., at ___–___ (THOMAS, J., concurring in judgment) (slip op., at 7–8); id., at ___ (ALITO, J., joined by ROBERTS, C. J., and KAGAN, J., dissenting) (slip op., at 1). And other Members of the Court recognized that Apodaca at least muddied the waters of the Court’s Sixth Amendment jurisprudence. Id., at ___, and n. 36 (plurality opinion) (slip op., at 10, and n. 36). In short, even in Ramos itself, the Court indicated that the decision was not dictated by precedent or apparent to all reasonable jurists. Edwards responds that the Court’s decision in Ramos must have applied a settled rule, not a new rule, because the decision adhered to the original meaning of the Sixth Amendment’s right to a jury trial and the Fourteenth Amendment’s incorporation of that right (and others) against the States. That argument conflates the merits question presented in Ramos with the retroactivity ques- tion presented here. On the merits question, the critical point, as the Court thoroughly explained in Ramos, is that the Constitution’s text and history require a unanimous jury in state criminal trials. On the retroactivity question, —————— criminal trials, it does not require a unanimous jury verdict in state crim- inal trials”); Schad v. Arizona, 501 U. S. 624, 634, n. 5 (1991) (plurality opinion) (a “state criminal defendant, at least in noncapital cases, has no federal right to a unanimous jury verdict”); Burch v. Louisiana, 441 U. S. 130, 137 (1979) (the Court has “approved the use of certain nonunani- mous verdicts in cases involving 12-person juries”); Ludwig v. Massachu- setts, 427 U. S. 618, 625 (1976) (the “holding in Apodaca” was that “the jury’s verdict need not be unanimous”); Smith v. Swarthout, 742 F. 3d 885, 895, n. 4 (CA9 2014) (“The Supreme Court has instructed that the Sixth and Fourteenth Amendments do not require a unanimous verdict in state criminal prosecutions”); see also 6 W. LaFave, J. Israel, N. King, & O. Kerr, Criminal Procedure §22.1(e), p. 23 (4th ed. 2015) (the “Sixth Amendment does not require jury unanimity in state criminal trials”). 10 EDWARDS v. VANNOY

the critical point is that reasonable jurists who considered the question before Ramos interpreted Apodaca to allow non-unanimous jury verdicts in state criminal trials. By renouncing Apodaca and expressly requiring unani- mous jury verdicts in state criminal trials, Ramos plainly announced a new rule for purposes of this Court’s retroac- tivity doctrine. And new rules of criminal procedure ordi- narily do not apply retroactively on federal collateral review. 2 Having determined that Ramos announced a new rule re- quiring jury unanimity, we must consider whether that new rule falls within an exception for watershed rules of crimi- nal procedure that apply retroactively on federal collateral review. This Court has stated that the watershed exception is “extremely narrow” and applies only when, among other things, the new rule alters “our understanding of the bed- rock procedural elements essential to the fairness of a proceeding.” Whorton, 549 U. S., at 417–418 (internal quotation marks omitted). In the abstract, those various adjectives—watershed, narrow, bedrock, essential—do not tell us much about whether a particular decision of this Court qualifies for the watershed exception. In practice, the exception has been theoretical, not real. The Court has identified only one pre- Teague procedural rule as watershed: the right to counsel recognized in the Court’s landmark decision in Gideon v. Wainwright, 372 U. S. 335, 344–345 (1963). See Whorton, 549 U. S., at 419, 421. The Court has never identified any other pre-Teague or post-Teague rule as watershed. None. Moreover, the Court has flatly proclaimed on multiple oc- casions that the watershed exception is unlikely to cover any more new rules. Even 32 years ago in Teague itself, the Cite as: 593 U. S. ____ (2021) 11

Court stated that it was “unlikely” that additional water- shed rules would “emerge.” 489 U. S., at 313 (plurality opinion). And since Teague, the Court has often reiterated that “it is unlikely that any such rules have yet to emerge.” Whorton, 549 U. S., at 417 (internal quotation marks and alteration omitted); see also Beard v. Banks, 542 U. S. 406, 417 (2004); Summerlin, 542 U. S., at 352; Tyler, 533 U. S., at 667, n. 7; Graham v. Collins, 506 U. S. 461, 478 (1993); Sawyer, 497 U. S., at 243; Butler v. McKellar, 494 U. S. 407, 416 (1990). Consistent with those many emphatic pronouncements, the Court since Teague has rejected every claim that a new procedural rule qualifies as a watershed rule. For example, in Beard v. Banks, 542 U. S., at 408, the Court declined to retroactively apply the rule announced in Mills v. Mary- land, 486 U. S. 367, 384 (1988), that capital juries may not be required to disregard certain mitigating factors. In O’Dell v. Netherland, 521 U. S. 151, 153 (1997), the Court refused to retroactively apply the rule announced in Sim- mons v. South Carolina, 512 U. S. 154, 156 (1994), that a capital defendant must be able, in certain circumstances, to inform the sentencing jury that he is parole ineligible. In Lambrix v. Singletary, 520 U. S., at 539–540, the Court de- clined to retroactively apply the rule announced in Espi- nosa v. Florida, 505 U. S. 1079, 1082 (1992) (per curiam), that sentencers may not weigh invalid aggravating circum- stances before recommending or imposing the death pen- alty. In Sawyer v. Smith, 497 U. S., at 229, the Court re- fused to retroactively apply the rule announced in Caldwell v. Mississippi, 472 U. S. 320, 323 (1985), which prohibited a death sentence by a jury led to the false belief that respon- sibility for the sentence rested elsewhere. The list of cases declining to retroactively apply a new rule of criminal procedure extends back long before Teague to some of this Court’s most historic criminal procedure de- cisions. For example, in Johnson v. New Jersey, 384 U. S. 12 EDWARDS v. VANNOY

719, 721 (1966), the Court declined to retroactively apply Miranda v. Arizona, 384 U. S. 436, 444–445 (1966), which required that police inform individuals in custody of certain constitutional rights before questioning them. And in Lin- kletter v. Walker, 381 U. S., at 639–640, the Court refused to retroactively apply Mapp v. Ohio, 367 U. S. 643, 655 (1961), which incorporated the Fourth Amendment exclu- sionary rule against the States. Edwards seeks to distinguish Ramos from the long line of cases where the Court has declined to retroactively apply new procedural rules. Edwards emphasizes three aspects of Ramos: (i) the significance of the jury-unanimity right; (ii) Ramos’s reliance on the original meaning of the Consti- tution; and (iii) the effect of Ramos in preventing racial dis- crimination in the jury process. But Edwards’s attempts to distinguish Ramos are una- vailing because the Court has already considered and re- jected those kinds of arguments in prior retroactivity cases. First, Edwards emphasizes the significance of the jury- unanimity right for criminal defendants. But that argu- ment for retroactivity cannot be squared with the Court’s decisions in Duncan v. Louisiana, 391 U. S. 145 (1968), and DeStefano v. Woods, 392 U. S. 631 (1968) (per curiam). In Duncan, the Court repudiated several precedents and ruled that a defendant has a constitutional right to a jury trial in a state criminal case. 391 U. S., at 149–150, 154–155. Not- withstanding the extraordinary significance of Duncan in guaranteeing a jury trial and expanding the rights of crim- inal defendants, the Court in DeStefano declined to retroac- tively apply the jury right. 392 U. S., at 633; see also Sum- merlin, 542 U. S., at 356–358 (relying on DeStefano and rejecting retroactivity of jury right recognized in Ring v. Arizona, 536 U. S. 584, 589 (2002)). We cannot discern a principled basis for retroactively applying the subsidiary Ramos jury-unanimity right when the Court in DeStefano Cite as: 593 U. S. ____ (2021) 13

declined to retroactively apply the broader jury right itself.5 Second, Edwards stresses that Ramos relied on the orig- inal meaning of the Sixth Amendment. But that argument for retroactivity is inconsistent with Crawford v. Washing- ton, 541 U. S. 36 (2004), and Whorton v. Bockting, 549 U. S. 406 (2007). In Crawford, the Court relied on the original meaning of the Sixth Amendment’s Confrontation Clause to overrule precedent and restrict the use of hearsay evi- dence against criminal defendants. 541 U. S., at 60–69. Notwithstanding Crawford’s reliance on the original mean- ing of the Sixth Amendment, the Court in Whorton declined to retroactively apply Crawford. 549 U. S., at 421. Third, Edwards says that Ramos prevents racial discrim- ination by ensuring that the votes of all jurors, regardless —————— 5 Edwards argues that the Ramos rule mirrors the rule announced in

Burch v. Louisiana, 441 U. S. 130, 134 (1979). In Burch, the Court held that six-person jury verdicts must be unanimous. According to Edwards, the Court retroactively applied Burch in Brown v. Louisiana, 447 U. S. 323 (1980). But the Justices who concurred in the judgment and supplied the decisive opinion in Brown said only that the Burch rule should apply to all cases on direct review. 447 U. S., at 337 (opinion of Powell, J., joined by Stevens, J.). They did not say that the rule should apply retro- actively on federal collateral review. So Brown does not help Edwards here. The Court’s decision in Ivan V. v. City of New York, 407 U. S. 203 (1972) (per curiam), is no more helpful to Edwards. In In re Winship, the Court held that a jury must find guilt “beyond a reasonable doubt.” 397 U. S. 358, 364 (1970). And in Ivan V., the Court held that the rule an- nounced in Winship applied in a case on direct review. 407 U. S., at 205. But in its numerous retroactivity cases, this Court has never identified the Winship rule as a watershed rule of criminal procedure that applies retroactively on federal collateral review. That no doubt explains why, in his submissions to this Court, Edwards himself did not cite Ivan V. in support of his retroactivity argument. In any event, Brown and Ivan V. were pre-Teague decisions. See Teague v. Lane, 489 U. S. 288 (1989). Because Teague tightened the pre- vious standard set forth in Linkletter v. Walker, 381 U. S. 618 (1965), for applying a decision retroactively on federal collateral review, pre-Teague decisions holding that a rule is retroactive are not as relevant as pre- Teague decisions holding that a rule is not retroactive, such as DeStefano. 14 EDWARDS v. VANNOY

of race, matter in the jury room. But that argument for ret- roactivity cannot prevail in light of Batson v. Kentucky, 476 U. S. 79 (1986), and Allen v. Hardy, 478 U. S. 255 (1986) (per curiam). In Batson, the Court overruled precedent and revolutionized day-to-day jury selection by holding that state prosecutors may not discriminate on the basis of race when exercising individual peremptory challenges. 476 U. S., at 92–93, 96–98. Nonetheless, the Court in Allen de- clined to retroactively apply Batson. 478 U. S., at 261; see also Teague, 489 U. S., at 295–296 (reaffirming Allen). The Court’s decisions in Duncan, Crawford, and Batson were momentous and consequential. All three decisions fundamentally reshaped criminal procedure throughout the United States and significantly expanded the constitu- tional rights of criminal defendants. One involved the jury- trial right, one involved the original meaning of the Sixth Amendment’s Confrontation Clause, and one involved ra- cial discrimination in jury selection. Yet the Court did not apply any of those decisions retroactively on federal collat- eral review. Ramos is likewise momentous and consequen- tial. But we see no good rationale for treating Ramos dif- ferently from Duncan, Crawford, and Batson. Consistent with the Court’s long line of retroactivity precedents, we hold that the Ramos jury-unanimity rule does not apply ret- roactively on federal collateral review.6 In so concluding, we recognize that the Court’s many ret- roactivity precedents taken together raise a legitimate question: If landmark and historic criminal procedure deci- sions—including Mapp, Miranda, Duncan, Crawford, Bat- son, and now Ramos—do not apply retroactively on federal

—————— 6 The Ramos rule does not apply retroactively on federal collateral re-

view. States remain free, if they choose, to retroactively apply the jury- unanimity rule as a matter of state law in state post-conviction proceed- ings. See Danforth v. Minnesota, 552 U. S. 264, 282 (2008). Cite as: 593 U. S. ____ (2021) 15

collateral review, how can any additional new rules of crim- inal procedure apply retroactively on federal collateral re- view? At this point, some 32 years after Teague, we think the only candid answer is that none can—that is, no new rules of criminal procedure can satisfy the watershed ex- ception. We cannot responsibly continue to suggest other- wise to litigants and courts. In Teague itself, the Court rec- ognized that the purported exception was unlikely to apply in practice, because it was “unlikely” that such watershed “components of basic due process have yet to emerge.” 489 U. S., at 313 (plurality opinion). The Court has often re- peated that “it is unlikely that any of these watershed rules has yet to emerge.” Tyler, 533 U. S., at 667, n. 7 (alteration and internal quotation marks omitted); see also, e.g., Whor- ton, 549 U. S., at 417; Summerlin, 542 U. S., at 352. And for decades, the Court has rejected watershed status for new procedural rule after new procedural rule, amply demonstrating that the purported exception has become an empty promise. Continuing to articulate a theoretical exception that never actually applies in practice offers false hope to de- fendants, distorts the law, misleads judges, and wastes the resources of defense counsel, prosecutors, and courts. Moreover, no one can reasonably rely on an exception that is non-existent in practice, so no reliance interests can be affected by forthrightly acknowledging reality. It is time— probably long past time—to make explicit what has become increasingly apparent to bench and bar over the last 32 years: New procedural rules do not apply retroactively on federal collateral review. The watershed exception is mor- ibund. It must “be regarded as retaining no vitality.” Her- rera v. Wyoming, 587 U. S. ___, ___ (2019) (slip op., at 11) (internal quotation marks omitted). 16 EDWARDS v. VANNOY

3 We respectfully offer four responses to the dissent. First, in the dissent’s view, if a right is important enough to justify overruling or repudiating precedent (as in Ra- mos), then it often is important enough to apply retroac- tively as a watershed rule of criminal procedure. But the Court’s precedents say the opposite and demonstrate that the dissent’s position erroneously inverts stare decisis and Teague. Teague recognized that the Court would occasion- ally announce new rules of criminal procedure by overrul- ing or repudiating existing precedents. Teague further ex- plained, however, that it was “unlikely” that such new procedural rules would apply retroactively on federal collat- eral review. 489 U. S., at 313 (plurality opinion). In other words, under this Court’s longstanding case law, it is easier to overrule or repudiate a precedent—as the Court did in Mapp, Miranda, Duncan, Batson, and Crawford, for exam- ple—than it is to apply the new procedural rule retroac- tively on federal collateral review—as demonstrated by the Court’s corresponding non-retroactivity decisions in Lin- kletter, Johnson, DeStefano, Allen, and Whorton. The Ramos Court fully understood all of this. Although Ramos stopped short of expressly deciding this retroactivity question (because it was not squarely presented), Ramos discussed retroactivity and plainly foreshadowed today’s decision. The lead opinion in Ramos—which was joined in relevant part by two of today’s dissenters, JUSTICE BREYER and JUSTICE SOTOMAYOR—explained that overruling or re- pudiating Apodaca was not likely to significantly affect Louisiana’s and Oregon’s reliance interests in preserving fi- nal convictions because Ramos was not likely to apply ret- roactively on federal collateral review. In particular, the lead opinion said that the States’ “worries” about Ramos applying retroactively and overturning hundreds of final convictions outstripped “the facts” because “Teague’s test is Cite as: 593 U. S. ____ (2021) 17

a demanding one, so much so that this Court has yet to an- nounce a new rule of criminal procedure capable of meeting it.” Ramos, 590 U. S., at ___ (opinion of GORSUCH, J.) (slip op., at 24); see also id., at ___–___ (KAVANAUGH, J., concur- ring in part) (slip op., at 16–17). The lead opinion added that Teague is “demanding by design, expressly calibrated to address the reliance interests States have in the finality of their criminal judgments.” Id., at ___ (opinion of GORSUCH, J.) (slip op., at 24). In light of that explicit lan- guage in Ramos, the Court’s decision today can hardly come as a surprise. In short, the Court’s holding today—namely, that Ramos does not apply retroactively on federal collateral review— carefully adheres to Ramos and tracks the Court’s many longstanding precedents on retroactivity. Second, the dissent suggests that the Court knows that Ramos should apply retroactively under the watershed ex- ception, but wants to avoid applying Ramos retroactively, and for that reason has decided to just eliminate the water- shed exception altogether. That suggestion is unfounded. Ramos was a momentous decision, and those of us who joined it continue to agree with it. But as we have ex- plained, Ramos itself analyzed the Court’s retroactivity precedents and foretold today’s decision on retroactivity. We are simply following through on what Ramos (as well as the Court’s many other precedents) already said about ret- roactivity to now squarely hold that Ramos does not apply retroactively on federal collateral review. If we thought otherwise and believed that Ramos qualified under the Court’s precedents as a rule that applies retroactively, we would certainly say so. But applying our retroactivity prec- edents, we have concluded that Ramos does not apply ret- roactively—just as the Court has previously held that other historic cases like Mapp, Miranda, Duncan, Batson, and Crawford did not apply retroactively. After reaching that 18 EDWARDS v. VANNOY

conclusion, we then took account of the overall jurispruden- tial landscape of the last several decades in Teague cases and acknowledged what has become unmistakably clear: The purported watershed exception is moribund. Third, on that last point, the dissent responds that Teague nominally identified a retroactivity exception for watershed procedural rules and that we should do so as well. But the problem, as we see it, is that Teague simulta- neously said that it was “unlikely” that new procedural rules would qualify as watershed. 489 U. S., at 313 (plural- ity opinion). So Teague took with one hand what it seem- ingly gave with the other. And in the 32 years since Teague, this Court has never once held that a new procedural rule qualifies for the purported watershed exception. What is more, the Court has regularly repeated that Teague’s wa- tershed exception would likely never be satisfied. The Court today need not and does not overrule any post-Teague cases that held the watershed exception satisfied because there are no post-Teague cases that held the watershed ex- ception satisfied. As noted above, no stare decisis values would be served by continuing to indulge the fiction that Teague’s purported watershed exception endures. No one can reasonably rely on a supposed exception that has never operated in prac- tice. And perpetuating what has become an illusory excep- tion misleads litigants and judges, and needlessly expends the scarce resources of defense counsel, prosecutors, and courts. At this point, given that landmark cases like Mapp, Miranda, Duncan, Batson, Crawford, and now Ramos have not applied retroactively, we are simply acknowledging re- ality and stating the obvious: The purported watershed ex- ception retains no vitality. Fourth, the dissent asserts that the Court is not living up to the promise of Ramos for criminal defendants. To begin with, the dissent cannot reasonably charge the Court with failing to live up to Ramos given that Ramos itself explicitly Cite as: 593 U. S. ____ (2021) 19

forecast today’s decision on retroactivity. Moreover, with respect, JUSTICE KAGAN dissented in Ramos. To be sure, the dissent’s position on the jury-unanimity rule in Ramos was perfectly legitimate, as is the dissent’s position on ret- roactivity in today’s case. And it is of course fair for a dis- sent to vigorously critique the Court’s analysis. But it is another thing altogether to dissent in Ramos and then to turn around and impugn today’s majority for supposedly shortchanging criminal defendants. To properly assess the implications for criminal defendants, one should assess the implications of Ramos and today’s ruling together. And criminal defendants as a group are better off under Ramos and today’s decision, taken together, than they would have been if JUSTICE KAGAN’s dissenting view had prevailed in Ramos. If the dissent’s view had prevailed in Ramos, no defendant would ever be entitled to the jury-unanimity right—not on collateral review, not on direct review, and not in the future. By contrast, under the Court’s holdings in Ramos and this case, criminal defendants whose cases are still on direct review or whose cases arise in the future will have the benefit of the jury-unanimity right announced in Ramos. The rhetoric in today’s dissent is misdirected. Different Members of the Court have reached different con- clusions in Ramos and in this case, but each Member of the Court has acted in good faith in deciding the difficult ques- tions before us. * * * To summarize the Court’s retroactivity principles: New substantive rules alter “the range of conduct or the class of persons that the law punishes.” Summerlin, 542 U. S., at 353. Those new substantive rules apply to cases pending in trial courts and on direct review, and they also apply retro- actively on federal collateral review. New procedural rules alter “only the manner of determining the defendant’s cul- pability.” Ibid. (emphasis deleted). Those new procedural 20 EDWARDS v. VANNOY

rules apply to cases pending in trial courts and on direct review. But new procedural rules do not apply retroactively on federal collateral review. Ramos announced a new rule of criminal procedure. It does not apply retroactively on federal collateral review. We affirm the judgment of the U. S. Court of Appeals for the Fifth Circuit. It is so ordered. Cite as: 593 U. S. ____ (2021) 1

THOMAS, J., concurring

THEDRICK EDWARDS, PETITIONER v. DARREL VANNOY, WARDEN ON WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT [May 17, 2021]

JUSTICE THOMAS, with whom JUSTICE GORSUCH joins, concurring. I join the majority in full because it correctly charts its way through precedent to hold expressly what we have long implied: “New procedural rules do not apply retroactively on federal collateral review.” Ante, at 15. I write separately to highlight that we could also have resolved this case by applying the statutory text of the Antiterrorism and Effec- tive Death Penalty Act of 1996 (AEDPA). AEDPA directs federal courts to deny “any claim that was adjudicated on the merits in State court” unless the state court’s decision “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court.” 28 U. S. C. §2254(d)(1). In 2011, peti- tioner urged a Louisiana court to hold that the Federal Con- stitution requires jury unanimity, and the court rejected that claim on the merits. That conclusion was consistent with Apodaca v. Oregon, 406 U. S. 404 (1972), in which this Court determined that the Constitution does not require unanimous jury verdicts for state criminal convictions. AEDPA thus leaves no room for this Court—or any federal court—to grant relief. 2 EDWARDS v. VANNOY

I A Congress first prescribed federal habeas jurisdiction in the Judiciary Act of 1789. That statute did not clearly de- fine the scope of relief, but “the black-letter principle of the common law [was] that the writ was simply not available at all to one convicted of crime by a court of competent juris- diction.”1 Bator, Finality in Criminal Law and Federal Ha- beas Corpus for State Prisoners, 76 Harv. L. Rev. 441, 465– 466 (1963) (Bator); see Felker v. Turpin, 518 U. S. 651, 663 (1996) (citing Ex parte Watkins, 3 Pet. 193 (1830)). And the writ did not extend at all to prisoners confined under state authority. Bator 465. Congress expanded the writ in the Habeas Corpus Act of 1867. Ch. 28, 14 Stat. 385. This Act extended the writ to prisoners in state custody but again provided only “bare guidelines” about the scope of the writ.2 Wright v. West, 505 U. S. 277, 285 (1992) (plurality opinion). At first, this Court continued to apply the common-law rule that allowed a state petitioner to challenge only “the jurisdiction of the court that had rendered the judgment under which he was in custody.” Ibid. But the Court later “expand[ed] the cat-

—————— 1 Section 14 of the Judiciary Act provided that “courts of the United

States . . . shall have power to issue writs of . . . habeas corpus” and that “justices of the supreme court, as well as judges of the district courts, shall have power to grant writs of habeas corpus for the purpose of an inquiry into the cause of commitment.—Provided, That writs of habeas corpus shall in no cases extend to prisoners in gaol, unless where they are in custody, under or by colour of the authority of the United States, or are committed for trial before some court of the same, or are necessary to be brought into court to testify.” 1 Stat. 81–82. 2 The relevant language of the Act stated that courts, “in addition to

the authority already conferred by [the Judiciary Act of 1789], shall have power to grant writs of habeas corpus in all cases where any person may be restrained of his or her liberty in violation of the constitution, or of any treaty or law of the United States.” 14 Stat. 385. Cite as: 593 U. S. ____ (2021) 3

egory of claims deemed to be jurisdictional for habeas pur- poses.” Ibid. In 1874, for example, this Court found juris- dictional defects whenever state courts imposed sentences under unconstitutional statutes or imposed sentences not authorized by a statute. Ibid. (collecting cases). And a few decades later, the Court expanded relief to situations where, in the eyes of the federal court, “no state court had provided a full and fair opportunity to litigate” a prisoner’s federal claims. Ibid. But absent a “jurisdictional” defect, a state court judgment was entitled to “ ‘absolute respect’ ” as long as the prisoner “ ‘had been given an adequate oppor- tunity to obtain full and fair consideration of his federal claim in the state courts.’ ” Ibid. That rule left no room to grant relief simply because a state court made an error of law. In 1953, this Court abruptly changed course and decided that federal courts could grant a writ of habeas corpus simply because they disagreed with a state court’s judg- ment. See Brown v. Allen, 344 U. S. 443, 463. Around the same time, this Court declared that many constitutional rights of criminal procedure—some old, and some new—ap- plied against the States. See, e.g., Mapp v. Ohio, 367 U. S. 643 (1961) (exclusionary rule); Gideon v. Wainwright, 372 U. S. 335 (1963) (right to a court-appointed attorney); Mi- randa v. Arizona, 384 U. S. 436 (1966) (right to be informed of right against self-incrimination). That combination pre- dictably raised tough questions: Should new rules apply ret- roactively to final state-court judgments, allowing federal courts to grant habeas relief even if state courts did not err when issuing their decisions? And if so, by what authority could federal courts grant that relief? Admitting that “the Constitution neither prohibits nor re- quires retrospective effect,” Linkletter v. Walker, 381 U. S. 618, 629 (1965), the Court took an atextual and ad hoc ap- proach, presumably based on its interpretation of the 1867 Act. The Court declared that some federal decisions apply 4 EDWARDS v. VANNOY

retroactively to final state convictions, thus allowing fed- eral courts to grant habeas relief depending on the “merits and demerits in each case.” Ibid. To guide the analysis, Linkletter announced several factors for federal courts to consider: “the prior history of the rule in question, its pur- pose and effect, and whether retrospective operation will further or retard its operation.” Ibid. This rule did “not le[ad] to consistent results,” so two dec- ades later the Court tried a new interpretation of the 1867 Act. See Teague v. Lane, 489 U. S. 288, 302 (1989) (plural- ity opinion); Danforth v. Minnesota, 552 U. S. 264, 278 (2008) (“Teague’s general rule of nonretroactivity was an ex- ercise of this Court’s power to interpret the federal habeas statute”). Relevant here, Teague allowed federal courts to give new constitutional rules of criminal procedure retroac- tive effect on habeas review only if the new rule was “wa- tershed,” “ ‘bedrock,’ ” or “ ‘essential.’ ” 489 U. S., at 311 (em- phasis deleted).3 B Teague, however, was not the final word on how federal courts should review the decisions of state courts. In 1996, Congress enacted AEDPA, the most significant change to the habeas corpus statute since 1867. AEDPA filled in the “bare guidelines” of the 1867 Act by creating a comprehen- sive system for addressing federal habeas claims brought by state prisoners. See Wright, 505 U. S., at 285. Directly relevant here are two provisions that ensure that —————— 3 Two years before Teague, this Court overruled cases that had allowed

courts to apply the balancing test in Linkletter v. Walker, 381 U. S. 618 (1965), to cases on direct review, holding instead that the Constitution required courts to apply new procedural rules retroactively to cases on direct appellate review. Griffith v. Kentucky, 479 U. S. 314, 328 (1987). Teague also explained that new rules that “accord constitutional protec- tion to . . . primary activity” retroactively apply to cases on federal col- lateral review. 489 U. S., at 307, 310–311. These are considered sub- stantive rules. See Welch v. United States, 578 U. S. 120, 128 (2016). Cite as: 593 U. S. ____ (2021) 5

state courts have the primary role in adjudicating these claims. First, a prisoner must exhaust his claims in state court before he can seek relief in federal court. If “any avail- able [state-law] procedure” remains open, a federal “writ of habeas corpus . . . shall not be granted.” §§2254(b)–(c). Sec- ond, once a state court has had the opportunity to decide that claim, AEDPA demands that federal courts respect that judgment. The law precludes relief “with respect to any claim that was adjudicated on the merits in State court proceedings unless the adjudication of the claim” either (1) “resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States,” or (2) “resulted in a decision that was based on an unreasonable determination of the facts in light of the evi- dence presented in the State court proceeding.” §2254(d). It is not enough for a federal court to disagree with the state court—much less disagree on a point of law that this Court had not yet settled when the state court issued its judg- ment. Rather, the state court’s decision must conflict with clearly established law and be obviously wrong “beyond any possibility for fairminded disagreement.” Harrington v. Richter, 562 U. S. 86, 103 (2011); Greene v. Fisher, 565 U. S. 34, 38–40 (2011). II A Here, the system worked as designed. Edwards pre- sented his unanimous jury claim to a Louisiana court. And the state court reasonably relied on Apodaca in rejecting that claim. AEDPA is clear about what happens next—re- lief “shall not be granted.” §2254(d). Our analysis could have begun and ended there—with §2254(d)(1)’s plain text. Congress, through AEDPA, has made clear that federal courts cannot provide relief in this case. See Montgomery v. Louisiana, 577 U. S. 190, 221 6 EDWARDS v. VANNOY

(2016) (Scalia, J., dissenting); see also Ex parte Bollman, 4 Cranch 75, 94 (1807) (Marshall, C. J.) (“[T]he power to award the writ by any of the courts of the United States, must be given by written law”). B The Court, instead, relies on Teague. I join the Court’s opinion because it correctly applies precedent and leads to the same judgment, but I would be remiss if I did not point out two other problems with Teague. First, it has never been clear what gave this Court au- thority to grant habeas relief to state prisoners based on “new” constitutional rules of criminal procedure. Teague did not explain why the 1867 Act gave federal courts this power. Moreover, Teague primarily focused on moving the law in the opposite direction of Linkletter’s permissive ap- proach to collaterally reviewing final state convictions. See Danforth, 552 U. S., at 278 (“Teague . . . situated the rule it announced in th[e] line of cases adjusting the scope of fed- eral habeas relief in accordance with equitable and pruden- tial considerations”). Even if federal courts had this power, we never decided whether Congress’ most recent version of the habeas statute—AEDPA—continued to allow such re- lief. Given all that, the majority wisely closes a door to ret- roactive relief that likely never existed in the first place. Second, the Court’s reliance on Teague today and in the past should not be construed to signal that AEDPA is an afterthought in analyzing a claim like petitioner’s or that Teague could justify relief where AEDPA forecloses it. AEDPA does contemplate that some new constitutional rules might be retroactive in narrow circumstances. See §§2254(e)(2)(A)(i) (evidentiary hearings), 2244(b)(2)(A) (sec- ond-or-successive bar), 2244(d)(1)(C) (statute of limita- tions). But it does not contemplate retroactive rules upset- ting a state court’s adjudication of an issue that reasonably applied the law at the time. Section 2254(d)—the absolute Cite as: 593 U. S. ____ (2021) 7

bar on claims that state courts reasonably denied—has no exception for retroactive rights. Congress’ decision to cre- ate retroactivity exceptions to the statute of limitations and to the bar on second-or-successive petitions but not for §2254(d) is strong evidence that Teague could never have led to relief here. Russello v. United States, 464 U. S. 16, 23 (1983).4 The plain text applies regardless of what a pre- vious interpretation of a previous statute says.5 * * * A state court rejected petitioner’s claim that he was enti- tled to a unanimous jury verdict. That adjudication was not unreasonable or contrary to clearly established federal law. AEDPA’s explicit directive thus independently resolves this case: “a writ of habeas corpus . . . shall not be granted.” §2254(d).

—————— 4 As JUSTICE GORSUCH correctly points out, federal courts have “equi-

table discretion to decide whether to issue the writ or to provide a rem- edy,” which includes the powers to create doctrines such as harmless er- ror. Post, at 8 (concurring opinion). And federal courts can rely on those doctrines as well as statutory bars to deny relief. That is why, as JUSTICE GORSUCH explains, an equitable retroactivity bar with no watershed ex- ception can independently justify denying relief. Post, at 9, n. 5. 5 The Constitution does not require that habeas relief be available for

new “watershed” rules of criminal procedure. See, e.g., Brown v. Allen, 344 U. S. 443, 532–533 (1953) (Jackson, J., concurring in result); Linklet- ter, 381 U. S., at 629. Teague also acknowledged that a later change in law does not require invalidating a final judgment. See 489 U. S., at 308–309 (plurality opinion). And again, habeas corpus traditionally did not apply at all to prisoners sentenced by a court with valid jurisdiction. See Bator 465–466. Cite as: 593 U. S. ____ (2021) 1

GORSUCH, J., concurring

THEDRICK EDWARDS, PETITIONER v. DARREL VANNOY, WARDEN ON WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT [May 17, 2021]

JUSTICE GORSUCH, with whom JUSTICE THOMAS joins, concurring. Sometimes this Court leaves a door ajar and holds out the possibility that someone, someday might walk through it— though no one ever has or, in truth, ever will. In Teague v. Lane, 489 U. S. 288 (1989), the Court suggested that one day it might apply a new “watershed” rule of criminal pro- cedure retroactively to undo a final state court conviction. But that day never came to pass. Instead, over the follow- ing three decades this Court denied “watershed” status to one rule after another. Rules guaranteeing individuals the right to confront their accusers. Rules ensuring that only a jury may decide a defendant’s fate in a death penalty case. Rules preventing racially motivated jury selection. All failed to win retroactive application. Today, the Court can- didly admits what has been long apparent: Teague held out a “false hope” and the time has come to close its door. Ante, at 15. We take this step not because this Court’s criminal procedure rulings are somehow unimportant. Any decision seeking to enforce liberties enshrined in the Constitution has a claim to “watershed” importance. Instead, we aban- don Teague’s test because it poses a question this Court has no business asking. 2 EDWARDS v. VANNOY

I Though we often refer to the writ of habeas corpus, the common law knew several. See Ex parte Bollman, 8 U. S. (4 Cranch) 75, 97–98 (1807). All had one thing in common: Each required a custodian to produce (habeas) a prisoner’s person (corpus). But they served different ends. Some writs were tools for moving a prisoner from one court to an- other—whether for a new prosecution (ad respondendum) or to execute a prior judgment (ad satisfaciendum). Others functioned more like a subpoena to procure a prisoner’s presence to testify in court (ad testificandum). Others still served to remove a case from an inferior court to a superior one (cum causa). 3 W. Blackstone, Commentaries on the Laws of England 129–131 (1768). Among them all, however, only one came to be known as “The Great Writ.” The writ of habeas corpus ad subjicien- dum was a mechanism for asking “why the liberty of [a] subject[ ] is restrained.” Id., at 131 (emphasis added). Leading up to the English Civil War, monarchs sometimes jailed their subjects summarily and indefinitely, with little explanation and even less process. E.g., Darnel’s Case, 3 How. St. Tr. 1–59 (K. B. 1627). In response, common law courts developed the ad subjiciendum writ to force the Crown to provide reasons for its actions and, if necessary, to ensure adequate process—like a criminal trial—to justify any further detention. See Petition of Right, 3 Car. 1, ch. 1, ¶¶ 5, 8 (1628). In other words, “habeas corpus [w]as the instrument by which due process could be insisted upon.” Hamdi v. Rumsfeld, 542 U. S. 507, 555 (2004) (Scalia, J., dissenting). Great though it was, the writ’s power was never limitless. A prisoner confined under a final judgment of conviction by a court of competent jurisdiction stood on different footing than one confined by the King without trial. A court might issue the writ asking, “What is the reason for confinement?” Cite as: 593 U. S. ____ (2021) 3

But if the return came back: “Because he’s serving a custo- dial sentence after being convicted of a crime,” the inquiry was usually at an end. See Opinion on the Writ of Habeas Corpus, Wilm. 77, 88, 97 Eng. Rep. 29, 36 (K. B. 1758); cf. Anonymus, Cart. 221, 124 Eng. Rep. 928 (C. P. 1671); Ha- beas Corpus Act of 1679, 31 Car. 2, ch. 2, ¶¶ 2, 20. Custody pursuant to a final judgment was proof that a defendant had received the process due to him. See, e.g., Bushell’s Case, Vaugh. 135, 142–143, 124 Eng. Rep. 1006, 1009–1010 (C. P. 1670). In 1789, Congress authorized federal courts to issue the habeas writ. 1 Cong. ch. 20, § 14, 1 Stat. 73, 81–82; Ex parte Bollman, 8 U. S. (4 Cranch), at 93–94. When called upon to interpret that statute, this Court defined the scope of ha- beas review by looking “to the common law.” Id., at 93–94. Unsurprisingly, it proceeded to restate the longstanding rule associated with criminal judgments: Ad subjiciendum provided no recourse for a prisoner confined pursuant to a final judgment of conviction. Ex parte Watkins, 28 U. S. (3 Pet.) 193, 209 (1830). As Chief Justice Marshall rhetori- cally asked, “is not that judgment in itself sufficient cause?” Id., at 202 (emphasis added). If the answer was nearly always yes, one important ex- ception existed both here and in England. A habeas court could grant relief if the court of conviction lacked jurisdic- tion over the defendant or his offense. Id., at 202–203. Still, the exception was “confined” to that “limited class of cases.” Ex parte Parks, 93 U. S. (3 Otto) 18, 21 (1876). One judge could not grant relief just because he might have decided the merits of the case differently than another had. As this Court put it, a perceived “error in the judgment or proceed- ings, under and by virtue of which the party is imprisoned, constitute[d] no ground for” relief. Ex parte Siebold, 100 U. S. (10 Otto) 371, 375 (1880). Any other approach, the Court explained, risked converting the habeas writ into “a mere writ of error,” little more than a chance to redo a trial 4 EDWARDS v. VANNOY

or its appeal. Ibid. Originally, Congress allowed federal courts to issue ha- beas writs only to federal custodians. Reconstruction changed that. After the Civil War, Congress granted fed- eral courts the power to issue habeas writs to state author- ities as well. See Act of Feb. 5, 1867, 39 Cong. ch. 28, § 1, 14 Stat. 385, 385.1 Even then, however, this Court contin- ued to interpret the habeas statute consistent with histori- cal practice. If a prisoner was in custody pursuant to a final state court judgment, a federal court was powerless to re- visit those proceedings unless the state court had acted without jurisdiction. E.g., In re Graham, 138 U. S. 461, 462 (1891); Tinsley v. Anderson, 171 U. S. 101, 104–106 (1898); Markuson v. Boucher, 175 U. S. 184, 185–186 (1899); Med- craf v. Hodge, 245 U. S. 630, 630 (1917) (per curiam). Under the view that prevailed in this country for most of our history, and in England for even longer, Teague’s ques- tion about the “retroactive” application of “watershed” rules of criminal procedure to undo final criminal judgments would have made no sense. Because a final judgment of conviction, pursuant to a full-fledged criminal trial, was the process due to a criminal serving a custodial sentence, the habeas writ had served its purpose. A final judgment evi- denced a lawful basis for confinement and was “binding on all the world.” Ex parte Watkins, 28 U. S. (3 Pet.), at 207. —————— 1 That Act conferred on federal courts the “power to grant writs of ha-

beas corpus in all cases where any person may be restrained of his or her liberty in violation of” federal law. Ibid. (emphasis added). Two earlier statutes extended federal habeas jurisdiction to state custodians, but only for a much narrower class of cases. In 1833, Congress authorized federal courts to issue habeas process to state custodians detaining fed- eral officers for acts taken to implement federal law. Act of Mar. 2, 1833, 22 Cong. ch. 57, § 7, 4 Stat. 632, 634–635. This provision lives on in 28 U. S. C. § 2241(c)(2). And in 1842, Congress permitted federal courts to issue habeas process to state custodians detaining foreign officials whose acts implicated the law of nations. Act of Aug. 29, 1842, 27 Cong. ch. 257, 5 Stat. 539, 539. It, too, lives on in 28 U. S. C. § 2241(c)(4). Cite as: 593 U. S. ____ (2021) 5

II Only in the middle of the twentieth century did things really begin to change. In 1915, this Court suggested that a state court’s extreme departure from “established modes” of criminal trial practice, such as proceeding under the specter of mob violence, might be akin to the loss of “juris- diction,” at least if no corrective mechanism like an appeal existed. Frank v. Mangum, 237 U. S. 309, 326, 335–336 (1915). But if that represented an innovation, it was a mod- est one. The same cannot be said for Brown v. Allen, 344 U. S. 443, 464, 478 (1953). There, this Court effectively recast habeas as another way for federal courts to redress practi- cally any error of federal law they might find in state court proceedings. Never mind that state courts are obligated to follow federal law under the Supremacy Clause. Never mind that those courts may have already passed on a de- fendant’s argument about his federal rights. See id., at 487; id., at 497–501 (Frankfurter, J.). Never mind, too, that the defendant may have lost on appeal within the state court system, and even petitioned this Court for direct review. See id., at 456–457; id., at 489–497 (Frankfurter, J.). Eve- ryone accepts that, in our criminal justice system today, a judgment becomes final only after the completion of a trial and the appellate process, including the opportunity to seek certiorari from this Court on questions of federal law. See Clay v. United States, 537 U. S. 522, 527 (2003).2 Yet, even

—————— 2 When a sovereign furnishes an opportunity to appeal (as state and

federal governments now do), it necessarily invites an appellate court to revisit an initial merits determination. See J. Baker, An Introduction to English Legal History 148–153 (5th ed. 2019) (describing development of the appeal at common law); G. Jacob, A New Law-Dictionary (1729) (de- fining “appeal”). Under an appellate system, then, “[n]one of the[ ] [pre- ceding decisions] are final” in an ultimate sense until any appeals are concluded. 3 W. Blackstone, Commentaries on the Laws of England 411 (1768). 6 EDWARDS v. VANNOY

after all that, Brown held, a federal district court could still vacate a final state court judgment based on any perceived error of federal law it might detect—and do so though the entire state judicial system and this Court had seen nothing amiss. 344 U. S., at 465–487; id., at 506–507 (Frankfurter, J.). The result? As Justice Jackson warned, habeas became little more than an ordinary appeal with an extraordinary Latin name. The Court “so departed from [the finality] principle that the profession now believes that the issues [federal courts] actually consider [in] habeas corpus are substantially the same as would be considered on appeal.” Id., at 540 (Jackson, J., concurring in result). “The fatal sentence that in real life writes finis to many causes”— Judgment affirmed. or Certiorari denied.—became “in legal theory . . . a complete blank.” Id., at 543. Justice Jackson feared that this result not only “trivializ[ed] . . . the writ,” but promised practical problems too. Id., at 536. A large new “haystack” of frivolous habeas petitions was sure to fol- low, making it that much harder for courts to identify the meritorious “needle.” Id., at 537. The only solution Justice Jackson could see was to hold fast to the traditional rule: A final judgment, after completion of trial and the exhaustion of any direct appellate review, was res judicata, and the sole exception was a lack of jurisdiction. Id., at 543–544. Brown not only upended centuries of settled precedent and invited practical problems; it produced anomalies as well. The very same term it decided Brown, this Court re- jected Brown’s fix-any-error approach for final judgments issued by military courts. Burns v. Wilson, 346 U. S. 137, 142 (1953) (plurality opinion); id., at 147 (Minton, J., con- curring in judgment). So only state convicts—not United States service members—were afforded an additional ave- nue for appellate relief in the garb of habeas corpus pro- ceedings. It turned out, too, that only state courts—not ex- ecutive tribunals—were forced to suffer the indignity of Cite as: 593 U. S. ____ (2021) 7

having their final judgments reopened. So federal courts wound up with more power to reopen the judgments of a different sovereign’s courts than the administrative pro- ceedings of the federal government itself. See Rushing v. Wilkinson, 272 F. 2d 633, 641 (CA5 1959); M. Howe, Fore- word: The Supreme Court, 1952 Term, 67 Harv. L. Rev. 91, 160–162 (1953). With time, these implications became clear and, as Jus- tice Jackson predicted, Brown’s innovation proved unsus- tainable. The haystack just grew too large. During the 1960s, this Court incorporated the exclusionary rule against the States. Mapp v. Ohio, 367 U. S. 643, 655 (1961). It announced a Sixth Amendment right to a public de- fender. Gideon v. Wainwright, 372 U. S. 335, 339–340, 342 (1963). It barred the government from speaking to the de- fendant outside defense counsel’s presence. Massiah v. United States, 377 U. S. 201, 204–206 (1964). And it an- nounced a new script governing police interrogations. Mi- randa v. Arizona, 384 U. S. 436, 467–477 (1966). This pro- liferation of new federal procedural rights, combined with a federal post-conviction mechanism that functioned like an ordinary appeal, soon yielded a giant haystack of habeas petitions. For years, this Court struggled to devise rules for sorting the hay from the needles. Its approach varied wildly and inconsistently over time. In a few cases, the Court held a new rule of criminal procedure should not apply retroac- tively to settled convictions. Tehan v. United States ex. rel. Shott, 382 U. S. 406, 409, n. 3, 419 (1966); Linkletter v. Walker, 381 U. S. 618, 622, 639–640 (1965). On occasion, though, it extended the benefit of a new rule to litigants with final criminal judgments—sometimes only to the named petitioner in this Court, Jackson v. Denno, 378 U. S. 368, 377 (1964), other times to everyone laboring under a final judgment, McNerlin v. Denno, 378 U. S. 575, 575 (1964) (per curiam). Justice Harlan called these divergent 8 EDWARDS v. VANNOY

results “an extraordinary collection of rules.” Desist v. United States, 394 U. S. 244, 256–257 (1969) (Harlan, J., dissenting). He even wondered whether they could “properly be considered the legitimate products of a court of law.” Id., at 259. III It was only in this world that Teague’s question about the retroactive application of new “watershed” rules of criminal procedure could even begin to make sense. In an effort to bring some coherence to the area, the Court refocused its attention on the terms of the federal habeas statute. The statute provides that “writs of habeas corpus may be granted”—not that they must be granted. 28 U. S. C. § 2241(a) (emphasis added); see also id. § 2243. The law thus invests federal courts with equitable discretion to de- cide whether to issue the writ or to provide a remedy. Withrow v. Williams, 507 U. S. 680, 716 (1993) (Scalia, J., concurring in part and dissenting in part).3 Exercising this remedial discretion, the Court began to develop doctrines aimed at returning the Great Writ closer to its historic office. It decided that some claims are not cognizable on federal habeas review if state courts provide a mechanism for review. Stone v. Powell, 428 U. S. 465, 494–495 (1976). It established procedural default rules to prevent habeas petitioners from evading independent and adequate state law grounds for sustaining their convictions. Wainwright v. Sykes, 433 U. S. 72, 86–87 (1977). It crafted a heightened harmless error standard, calibrated to reflect the finality interests at stake in the post-conviction context. Brecht v. Abrahamson, 507 U. S. 619, 633–638 (1993). And —————— 3 That is how this Court reads nearly identical text in the Declaratory

Judgment Act (DJA). Because the DJA says federal courts “ ‘may declare the rights and other legal relations of any interested party,’ ” district courts “possess discretion” to award declaratory relief. Wilton v. Seven Falls Co., 515 U. S. 277, 282, 286 (1995). Cite as: 593 U. S. ____ (2021) 9

it applied abuse-of-the-writ rules to prevent an endless cy- cle of petition and re-petition by prisoners with nothing but time on their hands. McCleskey v. Zant, 499 U. S. 467, 489– 493 (1991). Chief among these new-but-old developments was Teague. Drawing on the historic role of habeas, the Court held that newly recognized rules of criminal procedure should not normally apply to cases “which have become fi- nal.” 489 U. S., at 304–310.4 Hard experience since Brown had reminded the Court that finality, “the idea that at some point a criminal conviction reaches an end, a conclusion, a termination, ‘is essential to the operation of our criminal justice system.’ ” Prost v. Anderson, 636 F. 3d 578, 582 (CA10 2011). One might “always continue to ask” whether a particular judgment was “correct.” P. Bator, Finality in Criminal Law and Federal Habeas Corpus for State Prison- ers, 76 Harv. L. Rev. 441, 446–448 (1963). But if the rule of law means anything, it means the final result of proceed- ings in courts of competent jurisdiction establishes what is correct “in the eyes of the law.” Herrera v. Collins, 506 U. S. 390, 399–400 (1993).5 —————— 4 Although the lead opinion in Teague garnered only a plurality, a ma-

jority of the Court adopted the plurality’s rule later that same year in Penry v. Lynaugh, 492 U. S. 302, 313 (1989). 5 Apparently believing our judge-made doctrines did not go far enough,

Congress added further “new restrictions” of its own in the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA). Felker v. Turpin, 518 U. S. 651, 664 (1996). But AEDPA creates only additional conditions to relief; it did not do away with the discretion afforded courts in the habeas statute, or the various rules this Court has formulated in the exercise of that discretion. As this Court has (unanimously) explained, “AEDPA did not codify Teague” and the one “neither abrogates [n]or qualifies the other.” Greene v. Fisher, 565 U. S. 34, 39 (2011). That is why Teague’s retroactivity bar operates “in addition” to AEDPA’s relitigation bar in 28 U. S. C. § 2254(d). Horn v. Banks, 536 U. S. 266, 272 (2002) (per curiam). Because a retroactivity bar and a relitigation bar both pose threshold barriers, JUSTICE THOMAS’s concurrence highlights how AEDPA provides 10 EDWARDS v. VANNOY

IV While Teague did much to return the writ to its original station, it didn’t quite complete the journey. After insisting that final judgments cannot be reopened as a “general rule,” Teague left some wiggle room. It added that some new rules of criminal procedure might yet apply retroactively if they had “watershed” significance. 489 U. S., at 311. Why? Be- cause “ ‘time and growth in social capacity, as well as judi- cial perceptions,’ ” might “ ‘alter our understanding of the bedrock procedural elements’ ” necessary to satisfy notions of “ ‘fairness.’ ” Ibid. To help the world know what a water- shed rule might look like, the Court described it in various ways—“bedrock,” “fundamental,” “central,” an “absolute prerequisite.” Id., at 311–314. A rule fitting that bill, Teague said, would do two things: (1) “ ‘significantly im- prove’ ” existing procedures for determining factual guilt or innocence and (2) “implicate the fundamental fairness of the trial.” Id., at 312–313.6 But all these words have yielded nothing. In more than three decades since Teague, not a single new rule of crimi- nal procedure has satisfied its “watershed” test. Ante, at 10–15. Nor is it as if we have lacked promising candidates.

—————— an additional reason why the Fifth Circuit correctly denied a certificate of appealability here. Ante, at 5 (concurring opinion); see Miller-El v. Cockrell, 537 U. S. 322, 350 (2003) (Scalia, J., concurring). 6 Teague also discussed an exception to the finality rule for certain

“substantive rules.” Because the parties agree that Ramos involved only a new rule of criminal procedure, little need be said about Teague’s “sub- stantive rule” exception. But it is worth noting that substantive rules, which place certain conduct “ ‘beyond the power of the criminal law-mak- ing authority to prescribe,’ ” id., at 307, bear at least some resemblance to this Court’s early cases finding a lack of jurisdiction over a defendant or an offense. See Ex parte Siebold, 100 U. S. (10 Otto), at 376; Ex parte Parks, 93 U. S. (3 Otto), at 20–21. Perhaps this aspect of Teague can be understood as accurately invoking the jurisdictional exception to the fi- nality rule; perhaps not. See Brown, 344 U. S., at 533, n. 4 (Jackson, J., concurring in result). But that question is for another day. Cite as: 593 U. S. ____ (2021) 11

This Court has refused “watershed” status to new rules that seek to ensure death penalty decisions are made by jurors rather than judges. See Schriro v. Summerlin, 542 U. S. 348 (2004) (denying watershed status to Ring v. Arizona, 536 U. S. 584 (2002)). New rules designed to ensure race discrimination plays no role in jury selection met the same fate even before Teague. See Allen v. Hardy, 478 U. S. 255 (1986) (per curiam) (denying retroactive application to Bat- son v. Kentucky, 476 U. S. 79 (1986)). Rules that sought to realign our jurisprudence with the original meaning of the Sixth Amendment’s Confrontation Clause have failed to qualify too. See Whorton v. Bockting, 549 U. S. 406 (2007) (denying watershed status to Crawford v. Washington, 541 U. S. 36 (2004)). The Court’s decision today retraces this familiar path. It denies “watershed” status to Ramos v. Louisiana, 590 U. S. ___ (2020), a decision that (like Crawford) returned us to the original meaning of the Sixth Amendment—and one that (like Ring and Batson) concerns a vital aspect of the jury trial right. The Court explains why this result neces- sarily follows from our post-Teague precedents: If so many other highly consequential rulings have failed to clear Teague’s bar, it’s hard to see how Ramos might. One could even say that any other result would defy this Court’s re- cent precedents. Ante, at 14. At the same time, though, one might also say these prec- edents illustrate how mystifying the whole Teague project has been from its inception. If Teague only prohibits the retroactive application of new rules of criminal procedure, after all, it’s not exactly obvious why that prohibition ap- plies to cases like Crawford or Ramos. Both decisions sought to realign this Court’s decisions with the original meaning of the Sixth Amendment; in that sense, the rights they recognized were anything but new. And to the extent Teague asks whether a new rule is “fundamental” or “bed- rock,” it’s hard to see how rights originally memorialized in 12 EDWARDS v. VANNOY

the Constitution could fail to qualify. Certainly, this Court is in no position to second-guess the judgment of those who wrote and ratified the Constitution. Surely, too, many of the other rules of criminal procedure this Court has found less than “fundamental” since Teague seem anything but that to those whose lives they affect. Nor is it only Teague’s results that mystify. The test itself has been fraught with contradictions from the start. It asks litigants to be on the lookout for new procedural protections “ ‘implicit in the con- cept of ordered liberty.’ ” Beard v. Banks, 542 U. S. 406, 417 (2004). At the same time, we have been told, the fact that “a new procedural rule is ‘fundamental’ in some abstract sense is not enough.” Summerlin, 542 U. S., at 352. For me, it’s here where the history canvassed above mat- ters. This Court’s (in)activity since Teague only begins to make sense when viewed against the backdrop of the tradi- tional rule that old judgments are impervious to new chal- lenges. Yes, this Court’s decisions should apply to all cases pending in trial courts and on direct appeal. But they should not apply retroactively in habeas. The reason has nothing to do with whether Members of this Court happen to think the rules they announce are “new” in some sense or insufficiently “fundamental” in another. It’s simpler than that: The writ of habeas corpus does not authorize federal courts to reopen a judgment issued by a court of competent of jurisdiction once it has become final. Supra, at 2–4. It’s here, too, where today’s decision makes its real con- tribution. If Teague pointed us back in the direction of the traditional rule, each of the cases that has followed in its wake has edged us, step-by-step, closer still. Today’s deci- sion advances the progress by making express what has long been barely implicit: The “watershed” exception for new rules of criminal procedure is no exception at all. Ante, at 15. Not only does this development do much to honor the traditional understanding of habeas review and the great Cite as: 593 U. S. ____ (2021) 13

weight of this Court’s precedents throughout its history. It also allows us to retire a test that was unknown in law until 1989 and whose contours remain unknowable decades later. It frees this Court from the dreary task of needing to concoct reasons to denigrate the importance of obviously important rules like those discussed in Ramos, Ring, Bat- son, and Crawford, which affect the lives and liberty of countless individuals. It does away with the strange busi- ness of having to repackage old rules as new ones. And it eliminates the need for litigants and lower courts to endure years of protracted litigation—tangling with a contradic- tory test and seemingly inexplicable precedents—all sure to achieve nothing. The Court’s candor today is admirable— and correct.7 With these observations, I am pleased to join the Court’s opinion. My vote in similar cases to come will, I hope, “be guided as nearly as [possible] by the principles set forth herein.” Brown, 344 U. S., at 548 (Jackson, J., concurring in result).

—————— 7 The dissent criticizes today’s decision as a departure from modern ha-

beas precedent. Post, at 2; post, at 3, n. 2; post, at 12–13. But the dis- sent’s history is selective. The dissent champions decisions from the 1950s, ’60s, and ’70s. But it disregards how those decisions departed from a century of this Court’s precedents and the common law before that. Supra, at 5–8. At the same time, the dissent’s account overlooks this Court’s precedents refusing to afford retroactive application in every case since the 1980s. Post, at 10–11; post, at 12, n. 7. The dissent may prefer decisions within a particular 30-year window. But it is too much to say this preference is required to “[r]espect[ ] stare decisis.” Post, at 1, n. 1. Cite as: 593 U. S. ____ (2021) 1

KAGAN, J., dissenting

THEDRICK EDWARDS, PETITIONER v. DARREL VANNOY, WARDEN ON WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT [May 17, 2021]

JUSTICE KAGAN, with whom JUSTICE BREYER and JUSTICE SOTOMAYOR join, dissenting. “A verdict, taken from eleven, [i]s no verdict at all,” this Court proclaimed just last Term. Ramos v. Louisiana, 590 U. S. ___, ___−___ (2020) (slip op., at 4–5) (internal quota- tion marks omitted). Citing centuries of history, the Court in Ramos termed the Sixth Amendment right to a unani- mous jury “vital,” “essential,” “indispensable,” and “funda- mental” to the American legal system. Id., at ___, ___, ___ (slip op., at 4, 6, 7). The Court therefore saw fit to disregard stare decisis and overturn a 50-year-old precedent enabling States to convict criminal defendants based on non-unani- mous verdicts.1 And in taking that weighty step, the Court also vindicated core principles of racial justice. For in the Court’s view, the state laws countenancing non-unanimous verdicts originated in white supremacism and continued in our own time to have racially discriminatory effects. See id., at ___−___ (slip op., at 2–3); id., at ___ (SOTOMAYOR, J., concurring in part) (slip op., at 4); id., at ___−___ —————— 1 I dissented in Ramos precisely because of its abandonment of stare

decisis. See 590 U. S., at ___−___ (slip op., at 5–9) (ALITO, J., dissenting); see also Kimble v. Marvel Entertainment, LLC, 576 U. S. 446, 455 (2015) (“Respecting stare decisis means sticking to some wrong decisions”). Now that Ramos is the law, stare decisis is on its side. I take the decision on its own terms, and give it all the consequence it deserves. 2 EDWARDS v. VANNOY

(KAVANAUGH, J., concurring in part) (slip op., at 12–15). Put all that together, and it is easy to see why the opinions in Ramos read as historic. Rarely does this Court make such a fundamental change in the rules thought necessary to ensure fair criminal process. If you were scanning a the- saurus for a single word to describe the decision, you would stop when you came to “watershed.” Yet the Court insists that Ramos’s holding does not count as a “watershed” procedural rule under Teague v. Lane, 489 U. S. 288, 311 (1989) (plurality opinion). The result of to- day’s ruling is easily stated. Ramos will not apply retroac- tively, meaning that a prisoner whose appeals ran out be- fore the decision can receive no aid from the change in law it made. So Thedrick Edwards, unlike Evangelisto Ramos, will serve the rest of his life in prison based on a 10-to-2 jury verdict. Only the reasoning of today’s holding resists explanation. The majority cannot (and indeed does not) deny, given all Ramos said, that the jury unanimity re- quirement fits to a tee Teague’s description of a watershed procedural rule. Nor can the majority explain its result by relying on precedent. Although flaunting decisions since Teague that held rules non-retroactive, the majority comes up with none comparable to this case. Search high and low the settled law of retroactivity, and the majority still has no reason to deny Ramos watershed status. So everything rests on the majority’s last move—the over- turning of Teague’s watershed exception. If there can never be any watershed rules—as the majority here asserts out of the blue—then, yes, jury unanimity cannot be one. The re- sult follows trippingly from the premise. But adopting the premise requires departing from judicial practice and prin- ciple. In overruling a critical aspect of Teague, the majority follows none of the usual rules of stare decisis. It discards precedent without a party requesting that action. And it does so with barely a reason given, much less the “special justification” our law demands. Halliburton Co. v. Erica P. Cite as: 593 U. S. ____ (2021) 3

John Fund, Inc., 573 U. S. 258, 266 (2014). The majority in that way compounds its initial error: Not content to misap- ply Teague’s watershed provision here, see ante, at 10–14, the majority forecloses any future application, see ante, at 14–15. It prevents any procedural rule ever—no matter how integral to adjudicative fairness—from benefiting a de- fendant on habeas review. Thus does a settled principle of retroactivity law die, in an effort to support an insupporta- ble ruling. I Start with what Teague and its progeny repeatedly said about what makes a new rule of criminal procedure “water- shed” (so that, before today, the rule applied retroactively).2 A watershed rule, we held, is “implicit in the concept of or- dered liberty.” Teague, 489 U. S., at 311 (plurality opinion) (internal quotation marks omitted). Such a rule addresses one of “the bedrock procedural elements” of the criminal process. Ibid. (emphasis deleted). Or similarly stated, it plays a “fundamental” and “central[ ]” role in a trial. Beard v. Banks, 542 U. S. 406, 418, 420 (2004). More specifically, a new rule, to qualify as watershed, must be “essential to [the trial’s] fairness.” Whorton v. Bockting, 549 U. S. 406, 418 (2007); see Teague, 489 U. S., at 312 (plurality opinion). And it must go to the defendant’s guilt or innocence, “pre- vent[ing] an impermissibly large risk of an inaccurate conviction.” Whorton, 549 U. S., at 418 (internal quotation marks omitted); see Teague, 489 U. S., at 312 (plurality opinion). Those requirements set a high bar. But they capture—or anyway, were once meant to—a “small core of

—————— 2 Prior to Teague, the Court gave retroactive effect to a somewhat wider

range of new procedural rules. See ante, at 13, n. 5; Danforth v. Minne- sota, 552 U. S. 264, 271–273 (2008). To find the no-retroactivity-ever rule that the majority announces today, a time traveler would have to go back to around 1950—when the Bill of Rights’ protections for criminal defendants did not even apply to the States. See ibid. 4 EDWARDS v. VANNOY

rules” needed to fairly adjudicate a defendant’s guilt. Beard, 542 U. S., at 417. The first clue that the unanimity rule falls within Teague’s small core is that the Court thought its adoption justified overturning precedent. Ramos didn’t just an- nounce a new rule. It reversed a prior, well-settled one. As the majority recounts, “Ramos repudiated this Court’s 1972 decision in Apodaca v. Oregon, 406 U. S. 404, which had allowed non-unanimous juries in state criminal trials.” Ante, at 1. Such a toppling of precedent needs a special justification—more than a run-of-the-mill claim of error. To meet that demand, the Ramos majority described Apodaca as flouting the essential “meaning of the Sixth Amend- ment’s jury trial right,” as revealed in both historical prac- tice and judicial decisions. 590 U. S., at ___ (slip op., at 21). Two concurring Justices added, to support discarding this “egregiously wrong” precedent, that the unanimity rule pre- vents improper verdicts: Apodaca “sanctions the conviction” of some defendants who would otherwise defeat the State’s efforts “to [meet] its burden” of proving guilt. 590 U. S., at ___ (KAVANAUGH, J.) (slip op., at 12); id., at ___ (SOTOMAYOR, J.) (slip op., at 2). And the majority and con- currences alike invoked racial justice to support abandon- ing stare decisis, explaining how a non-unanimity rule has posed a special danger of canceling Black jurors’ votes. See id., at ___–___, ___ (slip op., at 1–2, 21); id., at ___ (SOTOMAYOR, J.) (slip op., at 4); id., at ___–___ (KAVANAUGH, J.) (slip op., at 12–15); infra, at 8. At bottom, then, the Court took the unusual step of overruling prece- dent for the most fundamental of reasons: the need to en- sure, in keeping with the Nation’s oldest traditions, fair and dependable adjudications of a defendant’s guilt. In this much alone, Ramos’s reasoning evokes this Court’s descrip- tions of watershed rules.3 —————— 3 The majority misunderstands my point about the interaction between Cite as: 593 U. S. ____ (2021) 5

And putting talk of stare decisis aside, there remains much more in Ramos to echo Teague. If, as today’s majority says, Teague is full of “adjectives,” ante, at 10, so too is Ra- mos—and mostly the same ones. Jury unanimity, the Court pronounced, is an “essential element[ ]” of the jury trial right, and thus is “fundamental to the American scheme of justice.” 590 U. S., at ___–___ (slip op., at 6–7). The Court discussed the rule’s “ancient” history—“400 years of English and American cases requiring unanimity” leading up to the Sixth Amendment. Id., at ___, ___ (slip op., at 15, 11). As early as the 14th century, English com- mon law recognized jury unanimity as a “vital right.” Id., at ___ (slip op., at 4). Adopting that view, the early Ameri- can States likewise treated unanimity as an “essential fea- ture of the jury trial.” Id., at ___ (slip op., at 5). So by the time the Framers drafted the Sixth Amendment, “the right to a jury trial meant a trial in which the jury renders a unanimous verdict.” Id., at ___ (slip op., at 12) (emphasis in original). Because that was so, no jury verdict could stand (or in some metaphysical sense, even exist) absent full agreement: “A verdict, taken from eleven, was no ver- dict at all.” Id., at ___–___ (slip op., at 4–5) (internal quota- tion marks omitted). Unanimity served as a critical safe- guard, needed to protect against wrongful deprivations of citizens’ “hard-won liberty.” Id., at ___ (slip op., at 15). Or as Justice Story summarized the law a few decades after

—————— stare decisis and Teague. I am not saying that if a “right is important enough to justify overruling” precedent, then it is “important enough to apply retroactively.” Ante, at 16. (If that were my claim, this dissent would be far shorter.) Rather, the overruling of precedent—and more, the justifications given to support that overruling—are elements to con- sider when deciding on a rule’s watershed status. Or, as I say above, “a first clue.” Here, that clue cuts against the majority: Ramos overturned precedent (rather than just announcing a new rule) on grounds strikingly reminiscent of Teague’s criteria for watershed status. Still more clues, pointing in the same direction, appear in the coming pages . . . . 6 EDWARDS v. VANNOY

the Founding: To obtain a conviction, “unanimity in the ver- dict of the jury is indispensable.” Id., at ___ (slip op., at 6). If a rule so understood isn’t a watershed one, then noth- ing is. (And that is, of course, what the majority eventually says.) Once more, from the quotations just above: “funda- mental,” “essential,” “vital,” “indispensable.” No wonder to- day’s majority declares a new-found aversion to “adjec- tives”—or, as a concurring opinion says, “all these words.” Ante, at 10; ante, at 10 (GORSUCH, J., concurring). The una- nimity rule, as Ramos described it, is as “bedrock” as bed- rock comes. Teague, 489 U. S., at 315 (plurality opinion). It is as grounded in the Nation’s constitutional traditions— with centuries-old practice becoming part of the Sixth Amendment’s original meaning. And it is as central to the Nation’s idea of a fair and reliable guilty verdict. When can the State punish a defendant for committing a crime? Re- turn again to Ramos, this time going back to Blackstone: Only when “the truth of [an] accusation” is “confirmed by the unanimous suffrage” of a jury “of his equals and neigh- bours.” 590 U. S., at ___ (slip op., at 4) (quoting 4 Commen- taries on the Laws of England 343 (1769)). For only then is the jury’s finding of guilt certain enough—secure enough, mistake-proof enough—to take away the person’s freedom. Twice before, this Court retroactively applied rules that are similarly integral to jury verdicts. First, in Ivan V. v. City of New York, 407 U. S. 203, 204 (1972) (per curiam), we gave “complete retroactive effect” to the rule of In re Win- ship, 397 U. S. 358 (1970), that a jury must find guilt “be- yond a reasonable doubt.” Like Ramos, Winship rested on an “ancient” legal tradition incorporated into the Constitu- tion. 397 U. S., at 361. As in Ramos, that tradition served to “safeguard men” from “unjust convictions, with resulting forfeitures” of freedom. 397 U. S., at 362. And as in Ramos, that protection plays a “vital” part in “the American scheme of criminal procedure.” 397 U. S., at 363–364. With all that established, the Ivan V. Court needed just two pages to hold Cite as: 593 U. S. ____ (2021) 7

Winship retroactive, highlighting the reasonable-doubt standard’s “indispensable” role in “reducing the risk” of wrongful convictions. 407 U. S., at 204–205. Second, in Brown v. Louisiana, 447 U. S. 323 (1980), we retroactively applied the rule of Burch v. Louisiana, 441 U. S. 130 (1979), that a six-person guilty verdict must be unanimous. Think about that for a moment: We held retroactive a unanimity requirement, no different from the one here save that it ap- plied to a smaller jury. The reasoning should by now sound familiar. Allowing conviction by a non-unanimous jury “im- pair[s]” the “purpose and functioning of the jury,” under- mining the Sixth Amendment’s very “essence.” Brown, 447 U. S., at 331 (plurality opinion). It “raises serious doubts about the fairness of [a] trial.” Id., at 335, n. 13. And it fails to “assure the reliability of [a guilty] verdict.” Id., at 334. So when a jury has divided, as when it has failed to apply the reasonable-doubt standard, “there has been no jury verdict within the meaning of the Sixth Amendment.” Sullivan v. Louisiana, 508 U. S. 275, 280 (1993).4 —————— 4 The majority argues that Ivan V. and Brown applied these new rules

only to cases on direct appeal. See ante, at 13, n. 5. But that isn’t right. Although Ivan V. itself involved a direct appeal, the Court has made clear that the “complete retroactive effect” Ivan V. gave Winship included cases in habeas. See, e.g., United States v. Johnson, 457 U. S. 537, 562– 563, n. 21 (1982). And similarly, lower courts uniformly understood Brown to govern habeas cases, even though a concurring opinion (which supplied the ruling’s fifth and sixth votes) addressed only cases on direct appeal. See, e.g., Atkins v. Listi, 625 F. 2d 525, 526 (CA5 1980); see also Brown, 447 U. S., at 337 (opinion of Powell, J., joined by Stevens, J.). Those applications to habeas cases make sense because the Court of that time did not often distinguish in its retroactivity rulings between direct and collateral review. See Stovall v. Denno, 388 U. S. 293, 300–301 (1967). For that reason, the majority must fall back on the argument that “Brown and Ivan V. were pre-Teague decisions” and “Teague tight- ened the previous standard” for retroactivity. Ante, at 13, n. 5. That is true enough, see supra, at 3, n. 2, but irrelevant here given Brown and Ivan V’s reasoning. As just noted, each of those decisions said everything a court would say today in designating a new rule “watershed”—in essence, that the rule is central to the process of fairly deciding on a 8 EDWARDS v. VANNOY

And something still more supports retroactivity here, for the opinions in Ramos (unlike in Winship or Burch) relied on a strong claim about racial injustice. The Court detailed the origins of Louisiana’s and Oregon’s non-unanimity rules, locating them (respectively) in a convention to “estab- lish the supremacy of the white race” and “the rise of the Ku Klux Klan.” 590 U. S., at ___ (slip op., at 2) (internal quotation marks omitted). Those rules, the Court ex- plained, were meant “to dilute the influence [on juries] of racial, ethnic, and religious minorities”—and particularly, “to ensure that African-American juror service would be meaningless.” Ibid. (internal quotation marks omitted). Two concurring opinions linked that history to current practice. “In light of the[ir] racist origins,” JUSTICE KAVANAUGH stated, “it is no surprise that non-unanimous juries can make a difference”—that “[t]hen and now” they can “negate the votes of black jurors, especially in cases with black defendants.” Id., at ___ (slip op., at 13); see id., at ___ (SOTOMAYOR, J.) (slip op., at 4). But that statement precludes today’s result. If the old rule functioned “as an engine of discrimination against black defendants,” id., at ___ (KAVANAUGH, J.) (slip op., at 13), its replacement must “implicat[e]” (as watershed rules do) “the fundamental fair- ness and accuracy of the criminal proceeding,” Beard, 542 U. S., at 417 (internal quotation marks omitted). Or as JUSTICE KAVANAUGH put the point more concretely, the unanimity rule then helps prevent “racial prejudice” from resulting in wrongful convictions. Ramos, 590 U. S., at ___ (slip op., at 15). The rule should therefore apply not just forward but back, to all convictions rendered absent its pro- tection. II The majority argues in reply that the jury unanimity rule

—————— defendant’s guilt. Cite as: 593 U. S. ____ (2021) 9

is not so fundamental because . . . . Well, no, scratch that. Actually, the majority doesn’t contest anything I’ve said about the foundations and functions of the unanimity re- quirement. Nor could the majority reasonably do so. For everything I’ve said about the unanimity rule comes straight out of Ramos’s majority and concurring opinions. Just check the citations: I’ve added barely a word to what those opinions (often with soaring rhetoric) proclaim. Start with history. The ancient foundations of the unanimous jury rule? Check. The inclusion of that rule in the Sixth Amendment’s original meaning? Check. Now go to func- tion. The fundamental (or bedrock or central) role of the unanimous jury in the American system of criminal justice? Check. The way unanimity figures in ensuring fairness in criminal trials and protecting against wrongful guilty ver- dicts? Check. The link between those purposes and safe- guarding the jury system from (past and present) racial prejudice? Check. In sum: As to every feature of the una- nimity rule conceivably relevant to watershed status, Ra- mos has already given the answer—check, check, check— and today’s majority can say nothing to the contrary.5

—————— 5 The majority does try to say that the plurality opinion in Ramos none-

theless “plainly foreshadowed today’s decision” by noting that Teague’s watershed test was “demanding by design,” in recognition of the States’ reliance interests. Ante, at 16–17 (internal quotation marks omitted); see also ante, at 17–19 (repeating the assertion twice more). But the Ramos plurality’s description of the watershed test was nothing more than objective fact: Yes, the watershed test was purposefully demanding. As to whether the watershed test was so demanding as to exclude the jury unanimity rule, here is what the plurality had to say: “Whether the right to jury unanimity applies to cases on collateral review is a question for a future case where the parties will have a chance to brief the issue and we will benefit from their adversarial presentation.” Ramos, 590 U. S., at ___ (slip op., at 24). Not a lot of “plain[ ] foreshadow[ing]” there. Only a fair bit of wisdom about how to resolve legal issues—which, as I’ll later discuss, the majority could usefully have considered before overrul- ing the watershed exception. See infra, at 13. 10 EDWARDS v. VANNOY

Instead, the majority relies on decisions holding non-ret- roactive various other—even though dissimilar—proce- dural rules. In making that argument from past practice, the majority adopts two discrete tactics. Call the first “throw everything against the wall.” Call the second “slice and dice.” Neither can avail to render the jury unanimity rule anything less than what Ramos thought it—as the ma- jority concedes, “momentous.” Ante, at 14. As its first move, the majority lists as many decisions holding rules non-retroactive as it can muster. See ante, at 11–12 (reviewing a “long line of cases”). The premise here is that sheer volume matters: The majority presents the catalog as if every rule is as important as every other and as if comparing any to the unanimity requirement is beside the point. But that idea founders on this Court’s constant refrain that watershed rules are only a small subset of pro- cedural rules. See, e.g., Graham v. Collins, 506 U. S. 461, 478 (1993). For under that view (as under the very meaning of “watershed”), nothing could be less surprising than that non-watershed rules greatly outnumber watershed ones. That inexorable fact cannot refute the designation of any given rule—much less of jury unanimity—as watershed. And the majority’s kitchen-sink list becomes yet less proba- tive of the issue here because most of its bulk comes from decisions on sentencing. See ante, at 11 (citing Beard, 542 U. S., at 408; O’Dell v. Netherland, 521 U. S. 151, 153 (1997); Lambrix v. Singletary, 520 U. S. 518, 540 (1997); Sawyer v. Smith, 497 U. S. 227, 229 (1990); Schriro v. Sum- merlin, 542 U. S. 348, 352 (2004)). But Teague itself ex- plains why sentencing procedures are not watershed: A wa- tershed rule, the Court said there, must go to the jury’s “determination of innocence or guilt.” 489 U. S., at 313 (plu- rality opinion); see Beard, 542 U. S., at 417. So the major- ity’s indiscriminate inventory of non-retroactive rules can- not get it home. Cite as: 593 U. S. ____ (2021) 11

Enter the majority’s second stratagem, which tries to con- quer by dividing. Here, the majority picks out “three as- pects of Ramos” pointing toward watershed status, and names one prior decision to match each of the three. Ante, at 12. So in addressing the unanimity rule’s “significance,” the majority notes that the Court once held the jury-trial right non-retroactive. Ante, at 12–13 (citing DeStefano v. Woods, 392 U. S. 631, 633 (1968) (per curiam) and Duncan v. Louisiana, 391 U. S. 145 (1968)). In tackling Ramos’s re- turn to “original meaning,” the majority points to our deci- sion that an originalist rule about hearsay evidence should not apply backward. Ante, at 13 (citing Whorton, 549 U. S., at 421 and Crawford v. Washington, 541 U. S. 36 (2004)). And in discussing Ramos’s role in “prevent[ing] racial dis- crimination,” the majority invokes our denial of retroactiv- ity to a rule making it easier to prove race-based peremp- tory strikes. Ante, at 13–14 (citing Allen v. Hardy, 478 U. S. 255, 261 (1986) (per curiam) and Batson v. Kentucky, 476 U. S. 79 (1986)). What the majority doesn’t find—or even pretend to—is any decision corresponding to Ramos on all of those dimen- sions. Take just a pair of examples. The Court has never suggested that requiring a bench trial has race-based pur- poses or effects. See DeStefano, 392 U. S., at 633–635.6 —————— 6 Even on the metric of significance alone, the Court has not previously

ranked the jury-unanimity and jury-trial rights as today’s majority does. As earlier noted, the Court in Brown found a unanimity rule retroactive despite its earlier holding that the jury-trial right was not. See supra, at 6–7. The Court explained that the accuracy and fairness concerns raised by divided juries—where, by definition, at least one person retains rea- sonable doubt—exceed those arising from judicial verdicts. See Brown, 447 U. S., at 334–335, n. 13 (plurality opinion). In insisting otherwise, the majority falls prey to a common greater-includes-the-lesser fallacy— akin to the view that if a State can eliminate a jury, it can impose jury rules of whatever kind it likes. See ante, at 12–13 (reasoning that the treatment of a “broader” right controls that of a “subsidiary” right); Brown, 447 U. S., at 334–335, n. 13 (rejecting precisely that view). 12 EDWARDS v. VANNOY

Similarly, the Court thought its new rule on hearsay evi- dence of less than towering import: Calling the rule “limited in scope,” we doubted that it would have any effect on the “accuracy of factfinding.” Whorton, 549 U. S., at 419. I’ll resist making like points about other permutations because the main point here is a general one. The majority must slice and dice in this way—dividing Ramos into three for purposes of comparison—because it cannot find any rule analogous to jury unanimity on all relevant fronts. And that is for a simple reason: No procedure adopted since Teague so comprehensively fulfills that decision’s criteria for retroactivity. If any rule is watershed, it is jury una- nimity. See supra, at 6. So the majority is left to overrule Teague’s holding on wa- tershed rules.7 On the last page or so of its merits discus- sion (before it turns to pre-butting this dissent), the major- ity eliminates the watershed exception, declaring it “long past time” to do so. Ante, at 15. Teague had said there would not be “many” (retroactive) watershed rules. 489 U. S., at 313 (plurality opinion). The majority now says there will be none at all. If that is so, of course, jury una- nimity cannot be watershed. Finally, the majority offers an —————— 7 In describing the majority as overruling Teague, I do not mean it over-

rules only Teague. That decision doesn’t stand alone in stating the wa- tershed exception as governing law. As I count, the Court has recited the exception 17 more times before today. See Montgomery v. Louisiana, 577 U. S. 190, 198 (2016); Welch v. United States, 578 U. S. 120, 128 (2016); Chaidez v. United States, 568 U. S. 342, 347, n. 3 (2013); Danforth, 552 U. S., at 266, 274–275; Whorton v. Bockting, 549 U. S. 406, 416 (2007); Beard v. Banks, 542 U. S. 406, 416–417 (2004); Horn v. Banks, 536 U. S. 266, 271, n. 5 (2002) (per curiam); Tyler v. Cain, 533 U. S. 656, 665 (2001); Bousley v. United States, 523 U. S. 614, 619–620 (1998); O’Dell v. Netherland, 521 U. S. 151, 156–157 (1997); Lambrix v. Singletary, 520 U. S. 518, 539 (1997); Gray v. Netherland, 518 U. S. 152, 170 (1996); Cas- pari v. Bohlen, 510 U. S. 383, 396 (1994); Graham v. Collins, 506 U. S., 461, 477–478 (1993); Gilmore v. Taylor, 508 U. S. 333, 345 (1993); Saffle v. Parks, 494 U. S. 484, 494–495 (1990); Sawyer v. Smith, 497 U. S. 227, 241–242 (1990). Cite as: 593 U. S. ____ (2021) 13

intelligible reason for declining to apply Ramos retroac- tively. But in taking that road, the majority breaks a core judi- cial rule: respect for precedent. Stare decisis is a foundation stone of the rule of law, “promot[ing] the evenhanded, pre- dictable, and consistent development of legal principles, fos- ter[ing] reliance on judicial decisions, and contribut[ing] to the actual and perceived integrity of the judicial process.” Payne v. Tennessee, 501 U. S. 808, 827 (1991). Adherence to precedent is, of course, “not an inexorable command.” Id., at 828. Ramos itself teaches that much. But Ramos also shows how high stare decisis sets the bar for overruling a prior decision. To reverse course, we insist on compelling reasons, thorough explanation, and careful attention to competing interests. But not here. The majority crawls un- der, rather than leaps over, the stare decisis bar. To begin with, no one here asked us to overrule Teague. This Court usually confines itself to the issues raised and briefed by the parties. See, e.g., United States v. Sineneng- Smith, 590 U. S. ___, ___ (2020) (slip op., at 3) (discussing “the principle of party presentation”). There may be rea- sons to ignore that rule in one or another everyday case. But to do so in pursuit of overturning precedent is nothing short of extraordinary. Cf. Ramos, 590 U. S., at ___, n. 4 (KAVANAUGH, J.) (slip op., at 9, n. 4) (An “important factor” protecting stare decisis “is that the Court typically does not overrule a precedent unless a party requests overruling”). We are supposed to (fairly) apply the prevailing law until a party asks us to change it. And when a party does make that request, we are supposed to attend to countervailing arguments—which no one here had a chance to make. That orderly process, skipped today, is what enables a court to arrive at a considered decision about whether to overthrow precedent. Equally striking, the majority gives only the sketchiest of 14 EDWARDS v. VANNOY

reasons for reversing Teague’s watershed exception. In de- ciding whether to depart from precedent, the Court usually considers—and usually at length—a familiar set of factors capable of providing the needed special justification. See, e.g., Knick v. Township of Scott, 588 U. S. ___, ___ (2019) (slip op., at 20) (listing such considerations). The majority can’t be bothered with that customary, and disciplining, practice; it barely goes through the motions. Seldom has this Court so casually, so off-handedly, tossed aside prece- dent. In its page of analysis, the majority offers just one ground for its decision—that since Teague, the Court has not identified a new rule as watershed, and so “the pur- ported exception has become an empty promise.” Ante, at 15. But even viewed in the abstract, that argument does not fly. That the Court has not found a watershed rule since Teague does not mean it could or would not in the future. Teague itself understood that point: It saw value in the wa- tershed exception even while recognizing that watershed rules would be few and far between. 489 U. S., at 313 (plu- rality opinion). And viewed in the context of this case, the majority’s argument positively craters. For the majority to- day comes face-to-face with a rule that perfectly fits each of Teague’s criteria: Jury unanimity, as described in Ramos, is watershed—even though no prior rule was. See supra, at 4–8. That airtight match between Ramos and Teague re- futes the majority’s one stated reason for overruling the lat- ter decision. The majority could not rely on the absence of watershed rules to topple Teague if it had just faithfully ap- plied that decision to this case. In choosing otherwise, the majority imposes a steep price for overruling Apodaca in Ramos. Taking with one hand what it gave with the other, the Court curtails Ramos’s ef- fects by expunging Teague’s provision for watershed rules. And so too the Court limits the consequences of any simi- larly fundamental change in criminal procedure that may emerge in the future. For the first time in many decades Cite as: 593 U. S. ____ (2021) 15

(since long before Teague, see supra, at 3, n. 2), those con- victed under rules found not to produce fair and reliable verdicts will be left without recourse in federal courts.8 I would not discard Teague’s watershed exception and so keep those unfairly convicted people from getting new tri- als. Instead, I would accept the consequences of last Term’s holding in Ramos. A decision like that comes with a prom- ise, or at any rate should. If the right to a unanimous jury is so fundamental—if a verdict rendered by a divided jury is “no verdict at all”—then Thedrick Edwards should not spend his life behind bars over two jurors’ opposition. I re- spectfully dissent.

—————— 8 The majority’s final claim is that it is properly immune from this crit-

icism—that I cannot “turn around and impugn” its ruling—because “criminal defendants as a group are better off under Ramos and today’s decision, taken together, than they would have been if [my] dissenting view had prevailed in Ramos.” Ante, at 19. The suggestion is surprising. It treats judging as scorekeeping—and more, as scorekeeping about how much our decisions, or the aggregate of them, benefit a particular kind of party. I see the matter differently. Judges should take cases one at a time, and do their best in each to apply the relevant legal rules. And when judges err, others should point out where they went astray. No one gets to bank capital for future cases; no one’s past decisions insulate them from criticism. The focus always is, or should be, getting the case before us right.