Go to:	PDF Version	Briefs Menu

INTEREST OF AMICUS CURIAE

The Criminal Justice Legal Foundation (CJLF) ⁽¹⁾ is a non-profit California corporation organized to participate in litigation relating to the criminal justice system as it affects the public interest. CJLF seeks to bring the due process protection of the accused into balance with the rights of the victim and of society to rapid, efficient, and reliable determination of guilt and swift execution of punishment.

The rules of Teague v. Lane and 28 U. S. C. §2254(d) are important and independent limitations on collateral review. The Court of Appeals' erroneously narrow interpretation of these rules will needlessly delay enforcement of the death penalty, reducing its deterrent effect. These delays are contrary to the rights of victims and society which CJLF was formed to advance.

SUMMARY OF FACTS AND CASE

Over twenty years ago, George Banks "shot fourteen people . . . killing thirteen and wounding one." Commonwealth v. Banks, 521 A. 2d 1, 3 (Pa. 1987), cert. denied, Banks v. Pennsylvania, 484 U. S. 873 (1987). The dead included seven children. See id., at 5. He was sentenced to death in 1983. Id., at 3. Despite the absence of any doubt regarding his identity as the perpetrator, see id., at 8, the execution of the judgment has been delayed for two decades of repetitive review, including a direct appeal, two state collateral reviews, two federal habeas petitions, and associated appeals and petitions for certiorari. See Banks v. Horn, 63 F. Supp. 2d 525, 530-531 (MD Pa. 1999).

In the current round, the United States Court of Appeals for the Third Circuit denied most of his claims, but it vacated the death sentence based on its disagreement with the Pennsylvania Supreme Court regarding the application of Mills v. Maryland, 486 U. S. 367 (1988), a precedent established after Banks' sentence became final on direct review. See Banks v. Horn, 271 F. 3d 527, 540-541 (CA3 2001).

This Court granted certiorari and summarily reversed in Horn v. Banks, 536 U. S. 266 (2002) (per curiam). The per curiam opinion confirmed that the nonretroactivity rule of Teague v. Lane, 489 U. S. 288 (1989) remains an independent limitation on federal habeas corpus and remanded for determination of whether Mills was a new rule. 536 U. S., at 272.

On remand, the Third Circuit panel divided. The two-judge majority held that Mills was not a new rule within the meaning of theTeagueline of cases. Banks v. Horn, 316 F. 3d 228, 229-230 (CA3 2003). The third judge believed Mills was a new rule, but also believed that when a state court waives its procedural default rule and considers on collateral review a claim that could have been raised on direct appeal, the case is not "final" within the meaning of Teague until the completion of the collateral review. Id., at 255 (Sloviter, J., concurring). ⁽²⁾ This Court granted certiorari on September 30, 2003.

SUMMARY OF ARGUMENT

Mills v. Maryland clearly qualifies as a "new rule" under the nonretroactivity jurisprudence of Teague v. Lane and its progeny. The fact that more than one court has come to the opposite conclusion indicates that further clarification of Teague is in order.

A more objective, and less easily evaded, definition can be distilled from Justice O'Connor's opinion in Wright v. West, with a few caveats. Generally, the rule of a case is new if it can be distinguished factually from the precedents that support it, and the distinction is arguably material,i.e., a difference sufficient to support an argument that the earlier cases should not govern. In addition, a rule is new if other precedents could, at the time, reasonably be interpreted as contrary to the rule.

Mills extended federal regulation of state capital sentencing procedure further and more intrusively into an area previously thought to be the province of the states. It is distinguishable from Lockett v. Ohio and Eddings v. Oklahoma, which only governed the substantive factors for decision and not the procedure by which the sentencer reached its decision. This case is therefore like the many prior cases where this Court has held similar expansions (or proposed expansions) to be new rules, including Saffle v. Parks, Sawyer v. Smith, andO'Dell v. Netherland.

The second exception to the rule of Teague v. Lane has been the source of considerable confusion and much unnecessary litigation. In 1971, its purpose was to explain Justice Harlan's concurrence in the retroactive application of Gideon v. Wainwright, which was then only eight years old. Today, it is a relic. No rules of the "primacy and centrality" of Gideon remain to be made, and none have been made for many years.

The unnecessary, unproductive litigation of this obsolete exception can be greatly reduced, in capital cases, by declaring it per seinapplicable to new rules governing the discretionary sentencing decision. Sawyer v. Whitley held that discretionary sentencing issuesnever amount to a "fundamental miscarriage of justice" for the purpose of the exception to the procedural default rule, and the same considerations justify a per serule here.

The present case is distinguishable from Mills. Teague precludes the new rule that the Third Circuit created to apply to Pennsylvania, and the Antiterrorism and Effective Death Penalty Act of 1996 ("AEDPA") precludes collateral attack on the Pennsylvania Supreme Court's reasonable (and correct) application of Mills.

ARGUMENT

I. Mills was a new rule.

A. New Rules and Factual Distinctions.

The Court of Appeals' holding in the present case, that Mills v. Maryland, 486 U. S. 367 (1989) is not a new rule for the purpose ofTeague v.Lane, 489 U. S. 288 (1989), is clearly erroneous for the reasons stated in our brief at the petition stage. Indeed, as we noted there, the opinion "comes close to a catalog of all the ways this Court has held that federal court should not conduct a Teague inquiry." Brief for Criminal Justice Legal Foundation as Amicus Curiae in Support of the Petition for Writ of Certiorari in Beard v. Banks, No. 02-1603, p. 5. This case could readily be decided on the points stated there. However, this case presents an opportunity to clarify theTeague body of jurisprudence, make it more coherent, and reduce unproductive litigation with its attendant expense and delay.

Since it was announced 14 years ago, the rule of Teague has been subject to two main objections. The first was that this Court had improperly cut off habeas relief in cases where Congress had intended that such relief be granted. See Teague, 489 U. S., at 327-328 (Brennan, J., dissenting). Whatever the merits of this position may have been in 1989, AEDPA eliminated any basis for it. There can be no doubt that Congress enacted this law with awareness of Teague and with the intent that it continue. See 28 U. S. C. §2244(b)(2)(A);Tyler v. Cain, 533 U. S. 656, 665 (2001).

The second principal criticism is that the Teague rule is so confusing and generates so much litigation that its cost exceeds its value. Specifically, the claim has been made " 'that Teague has spawned far more confusion than it has eliminated,' . . . [and] [t]he only certain result is additional delay." In re Gallego, 18 Cal. 4th 825, 848, 959 P. 2d 290, 306 (1998) (Brown, J., concurring and dissenting) (quoting Friedman, Failed Enterprise: The Supreme Court's Habeas Reform, 83 Cal. L. Rev. 485, 519 (1995)).

There has indeed been much confusion and unnecessary litigation over Teague issues. The source, however, is not the Teague rule itself but rather the persistent attempts to evade it. The decision of the Court of Appeals in the present case is a prime example. The status ofMills as a new rule is simple and obvious, and the years of additional litigation in this case appear to be the result of that court's determination to overturn this and numerous other Pennsylvania death sentences, rather than any genuine difficulty in the question.

One of the purposes of the Teague rule is to avoid continually forcing "the States to marshal resources in order to keep in prison defendants whose trials and appeals conformed to then-existing constitutional standards," Teague, 489 U. S., at 310, resources which could be better spent in improved trials and initial appeals. See Mackey v. United States, 401 U. S. 667, 691 (1971) (Harlan, J., concurring in part and dissenting in part). This purpose is defeated if "new rule" is not defined clearly enough that easy cases like this one can be quickly resolved. In most cases, the states should have to marshal only minimal resources to have a Teague-barred claim dismissed on that ground in the District Court, and such claims should generally not qualify for a certificate of appealability. See 28 U. S. C. §2253.

In search of a definition of "new rule" that is more objective and less subject to evasion, it is useful to return to the basics of what is a precedent and what is a rule. The essence of precedent is that a new case which is like a previously decided case should be decided the same way. "A solemn decision upon a point of law, arising in any given case, becomes an authority in a like case . . . ." 1 J. Kent, Commentaries on American Law *475 (8th ed. 1854). That is the easy part. The hard part is deciding which cases are alike and which are different in important aspects, given the infinite variations in the complete facts of real cases. See D. Chamberlain, The Doctrine of Stare Decisis: Its Reasons and Extent 11-15 (N. Y. St. Bar Assn. 1885); Schauer, Precedent, 39 Stan. L. Rev. 571, 577 (1987). The most useful discussion for this purpose can be found in a classic law review article by Cambridge Professor Arthur L. Goodhart, Determining the Ratio Decidendi of a Case, 40 Yale L. J. 161 (1930). Goodhart examines and rejects the propositions that the ratio decidendi can be found in either the abstract statement of the rule in the opinion or in the unexplained facts of the case itself. See Brief for Criminal Justice Legal Foundation as Amicus Curiae in Grutter v. Bollinger, No. 02-241, pp. 6-7. Instead, the precedent consists of the material facts of the case and the result based on them. The rule is that any case with the same material facts must yield the same result. Cases which differ in their material facts are not controlled by the precedent. The critical question is which facts are material. "It is by his choice of the material facts that the judge creates law." Goodhart, supra, at 169.

This view of what is a "rule" is consistent with Justice O'Connor's explanation of Teague in Wright v. West, 505 U. S. 277, 304 (1992) (opinion concurring in the judgment).

"To determine what counts as a new rule, Teague requires courts to ask whether the rule a habeas petitioner seeks can be meaningfully distinguished from that established by binding precedent at the time his state court conviction became final. . . . If a proffered factual distinction between the case under consideration and pre-existing precedent does not change the force with which the precedent's underlying principle applies, the distinction is not meaningful, and any deviation from precedent is not reasonable."

A meaningful distinction is not necessarily one that the court will ultimately hold is decisive, so as to require a different result. If it were, there would be no difference between the Teague inquiry and the merits. The "meaningful distinction" question must be considered together with the "inquiry as [to] whether 'reasonable jurists' could disagree as to whether a result is dictated by precedent." Ibid. (quoting Sawyer v. Smith, 497 U. S. 227, 234 (1990)). A meaningful distinction is one that reasonable jurists could conclude requires a different result, not necessarily one that the court will conclude requires a different result.

The evolution of constitutional rules of criminal procedure is typically a series of decisions beginning with a requirement of the Bill of Rights as originally understood and extending its rule further from the core, until a point is reached where countervailing policies permit no further extension. For example, the Self-Incrimination Clause begins with the common-law rule that the defendant in a criminal case cannot be called to the stand and required to testify. See United States v. Balsys, 524 U. S. 666, 674, n. 5 (1998); Miranda v. Arizona, 384 U. S. 436, 510-515 (1966) (Harlan, J., dissenting). From there, Bram v. United States, 168 U. S. 532, 557-558 (1897) held that an out-of-court statement coerced by law enforcement could not be introduced in the trial. This extension was necessary to prevent the core purpose of the rule from being defeated, and it only had a minimal impact on the countervailing policy of bringing to the trier of fact all relevant, reliable evidence, because coerced confessions are so notoriously unreliable. Miranda was a very long extension, excluding from the prosecution's case in chief any in-custody statement taken without warnings and an express waiver. AlthoughMirandafurthered the policies of the earlier cases, it did so at a huge cost to the countervailing policy, excluding confessions which are, in fact, voluntary. See Oregon v. Elstad, 470 U. S. 298, 307 (1985). Edwards v. Arizona, 451 U. S. 477, 484-485 (1981) excluded confessions taken when police ask again for a waiver after the suspect has once invoked his rights. Arizona v. Roberson, 486 U. S. 675, 682-683 (1988) further extended this rule to cases where the second request was made by a different law enforcement agency, investigating a different crime. However, Michigan v. Tucker, 417 U. S. 433, 451 (1974) declined to extend the rule of exclusion to the "fruit" of the interrogation, and New York v. Quarles, 467 U. S. 649, 657 (1984) made an exception for questioning necessary to deal with an imminent threat to public safety.

Each of these cases can be viewed as involving a meaningful factual distinction from the earlier cases in the series. The voluntary but unwarned confessions in Miranda are certainly distinguishable from the coerced confession in Bram, and reasonable jurists could and did disagree that exclusion was required. See Miranda, 384 U. S., at 525 (Harlan, J., dissenting). Miranda was therefore a new rule. SeeJohnson v. New Jersey, 384 U. S. 719, 731 (1966) ("in conformity with previously announced constitutional standards"). Robersonis meaningfully distinguishable fromEdwards, in that a repeated request for a waiver which the suspect has already refused is more threatening to the right against self-incrimination than a new request regarding a different offense, and a rule of exclusion covering multiple offenses keeps out more valid evidence than a rule which only covers one. Reasonable jurists could and did believe that this difference was material and called for a different result. See Roberson, 486 U. S., at 693 (Kennedy, J., dissenting). ThereforeRobersonwas a new rule, even though the question in the case could be characterized as whether to make an exception to the existing rule. SeeButler v. McKellar, 494 U. S. 407, 414-415 (1990). The factual distinction in Quarles was sufficient to convince a majority of the Court that the cost to society of exclusion outweighed the benefit of further protecting the privilege against self-incrimination, even though the facts of the case fit within the abstract statement of the rule in Miranda. See 467 U. S., at 657. The "public safety exception" question was undecided until Quarles resolved it, and Quarles did so without overruling any precedents. Had the decision gone the other way, it would have been a new rule.

These cases demonstrate that it is not sufficient, for the purpose of a Teague analysis, to lift an abstract statement of a principle from a case and find that a subsequent case fits within the terms of that statement. Future cases will present meaningful factual distinctions that were not anticipated in the earlier opinion, and the question of whether the distinction makes a difference remains open until that question is decided in a case which presents it. That is how the law develops gradually, but "[t]he principle announced in Teague serves to ensure that gradual developments in the law over which reasonable jurists may disagree are not later used to upset the finality of state convictions valid when entered." Sawyer v. Smith, 497 U. S. 227, 234 (1990).

We can see this principle in operation throughout this Court's Teague jurisprudence. In Gilmore v. Taylor, 508 U. S. 333, 341 (1993), the fact that the instruction in question involved an affirmative defense rather than the elements of an offense was a meaningful distinction from the In re Winship, 397 U. S. 358 (1970) line of cases. "Because our cases do not resolve conclusively the question . . . , resolution of the issue on habeas would require us to promulgate a new rule." Gilmore, supra, at 351 (O'Connor, J., concurring in the judgment).

In O'Dell v. Netherland, 521 U. S. 151, 165 (1997) instruction about the defendant's legal eligibility for parole, see Simmons v. South Carolina, 512 U. S. 154 (1994) was distinguishable from the jury's consideration of facts regarding a defendant's good behavior in prison in Skipper v.South Carolina, 476 U. S. 1, (1986), even though both bear on the question of future dangerousness. "It is a step from a ruling that a defendant must be permitted to present evidence of that sort to a requirement that he be afforded an opportunity to describe the extant legal regime." O'Dell,supra, at 162. The O'Dell Court did not say how large a step, and it did not need to. Simmons was rationally distinguishable from Skipper on that basis, and therefore it created a new rule.

In Caspari v. Bohlen, 510 U. S. 383, 388 (1994), the Eighth Circuit had rejected a Teague objection, saying " 'it is a short step to apply the same double jeopardy protection to a non-capital sentencing hearing as the Supreme Court applied to a capital sentenc[ing] . . . hearing.' " This Court reversed. Ibid. Whether short or long, it was still a step, and the fact that Bohlen's case was not capital provided a rational basis for distinguishing Bullington v. Missouri, 451 U. S. 430 (1981). See Caspari, supra, at 392.

Three caveats apply to the factual distinction approach to Teague analysis. First, for some rules the operative "fact" is itself a mixed question of law and fact. "If the rule in question is one which of necessity requires a case-by-case examination of the evidence, then we can tolerate a number of specific applications without saying that those applications themselves create a new rule." Wright v. West, 505 U. S., at 308 (Kennedy, J., concurring in the judgment). See Brief for Criminal Justice Legal Foundation as Amicus CuriaeinYarborough v. Alvarado, No. 02-1684, Part II-B.

Second, it is important to keep in the foreground the requirement that the factual distinction need only be one that "[a] reasonable jurist [at the time] could have drawn . . . ." O'Dell, 521 U. S., at 165 (emphasis added). To this end, it is preferable to drop the term "meaningful" used in Justice O'Connor's Wright opinion, because that term is so easily susceptible to evasion. An arguably materialfactual distinction is one that "change[s] the force with which the precedent's underlying principle applies," 505 U. S., at 304, because it either makes the policy behind the precedent less weighty or it makes the countervailing considerations more so. The change in force may not be sufficient to change the outcome when the merits are finally decided, but any change in the right direction is sufficient to make the question an open one and its resolution a new rule.

Third, it must be recognized that the Supreme Court precedents in an area are not always "a model of clarity." Lockyer v. Andrade, 538 U. S. 63, 155 L. Ed. 2d 144, 155, 123 S. Ct. 1166, 1173 (2003). Even when one precedent would appear to dictate a particular result if considered in isolation, another precedent may be reasonably interpreted to point in the other direction. Sawyer, 497 U. S., at 236-237, noted contrary indications from prior cases in concluding that Caldwell v. Mississippi, 472 U. S. 320 (1985) was a new rule. O'Dell, 521 U. S., at 162-166, relied on indications in Caldwell and California v. Ramos, 463 U. S. 992 (1983), that instruction of juries regarding postsentencing review is a matter for the states, in holding that Simmons was a new rule. These contrary indications were sufficient to make Simmons a new rule, "[w]hatever support Gardner and Skipper, standing alone, might lend to petitioner's claim that Simmons was a foregone conclusion . . . ." O'Dell,supra, at 162 (emphasis added). Because "Teague asks state-court judges to judge reasonably, not presciently," id., at 166, the resolution of contrary indications in this Court's precedents is a new rule.

A focus on arguably material factual distinctions should curb the most egregious failures to apply the Teague rule. In Bohlen, there can be no doubt that the capital/noncapital distinction is one that reasonable people might find material. In Saffle v. Parks, 494 U. S. 484, 490 (1990), there is no doubt that an instruction that tells the jury not to consider mitigating evidence at all is different from one that tells them not to be swayed by sympathy, and the argument against the latter is weaker than the argument against the former.

B. The "Old Rule" Cases.

From any survey of this Court's Teague cases, it is apparent that most of this Court's decisions in favor of defendants on full briefing and argument are ultimately held to establish new rules. That is not surprising, since this Court's discretionary certiorari jurisdiction is exercised primarily to resolve disagreements among the federal courts of appeals and state supreme courts, see Supreme Court Rule 10, and most disagreements among these high tribunals are reasonable. Even so, there are a few cases that resolved disputed questions on habeas in the petitioner's favor, and these must be reconciled with the other Teague cases if possible.

In Penry v. Lynaugh, 492 U. S. 302 (1989), the defendant was a three-time rapist who murdered his third victim to keep her quiet. Seeid., at 322. He claimed to be retarded. ⁽³⁾ In accordance with Texas law, the jury was instructed to answer the three Texas "special issues," and the sentence depended on the answer. See id., at 310-311. In effect, the only mitigation a jury can consider under this system is evidence relating to lack of deliberation or intent, nondangerousness in the future, or response to provocation by the victim.

If Lockett v. Ohio, 438 U. S. 586 (1978) (plurality opinion) ⁽⁴⁾ and Eddings v. Oklahoma, 455 U. S. 104 (1982) were the only precedents involved,Penry would have been an easy case. In Lockett, the Ohio statute only allowed three circumstances to avoid a death sentence: victim facilitation of the offense, duress, or a mental defense. See 438 U. S., at 593-594. The plurality announced that states must allow consideration in mitigation of "any aspect of a defendant's character or record and any of the circumstance of the offense . . . ." Id., at 604 (emphasis added).

After reviewing the 1976 decisions, including Jurek v. Texas, 428 U. S. 262 (1976), the Lockett plurality said, "None of the statutes we sustained in Gregg and the companion cases clearly operated at that time to prevent the sentencer from considering any aspect of the defendant's character and record or any circumstances of his offense as an independently mitigating factor." 438 U. S., at 607 (emphasis added). That statement was demonstrably false. Regardless of how broadly the Texas courts allowed the defendant to state his case to the jury, cf. ibid., the law still required the jury to consider only the special issues, and it plainly did not allow other circumstances as "independently mitigating factor[s]."

Eddings confirmed that Lockett meant what it said and elevated its rule to majority opinion status. In a bench trial, the judge believed he could not consider Eddings' bad-childhood evidence in mitigation, and this was a violation of the Lockett rule. See 455 U. S., at 112-115.

Penry is not distinguishable from Eddings. Penry's retardation evidence, if believed, was no less an argument against the death penalty than Eddings' mitigation evidence. See 492 U. S., at 318-319. The bench-trial judge's understanding of the law is not distinguishable in any relevant way from jury instructions. See Eddings, 455 U. S., at 114.

The best argument of the state and the dissent for denying Penry's collateral attack on his sentence was that the Texas system used in his trial had been upheld in Jurek. See 492 U. S., at 355 (Scalia, J., dissenting). To make a petitioner's claim Teague-barred, the interpretation of precedent as contrary to the claim need only be reasonable. See supra, at 11. The Penry dissent's interpretation ofJurekwas certainly reasonable. The problem is that this interpretation had already been nullified by the disingenuous mischaracterization ofJurek in Lockett.

Penry was a hard case because decisions of this Court conflicted with earlier decisions while trying to paper over the conflict, rather than expressly acknowledging it. Under Teague, as under AEDPA, a state court should be able to rely on any precedent of this Court that has not been overruled, if that precedent reasonably appears to be the closest to the case before it. See Lockyer v. Andrade, 538 U. S. 63, 155 L. Ed. 2d 144, 156-157, 123 S. Ct. 1166, 1173-1174 (2003); see also Williams v. Taylor, 529 U. S. 362, 412 (2000) (equating AEDPA "clearly established" law with Teague "old rule"). Penry can be maintained within a coherent body of Teague/AEDPA jurisprudence only on the basis that Penry was "materially indistinguishable" from Eddings ⁽⁵⁾ and that the Penry dissent's interpretation of Jurek, however reasonable on the face of Jurek, did not survive Lockett. The simpler path, and the one that would form a more coherent body of jurisprudence, would be to simply acknowledge thatPenry's "new rule" holding is inconsistent with later developments in this area and is no longer good law. Under Sawyer, O'Dell, and Lambrix v.Singletary, 520 U. S. 518, 536 (1997), the reasonable interpretation ofJurek as contrary to Penry's claim should have been sufficient to makePenry a new rule.

Stringer v. Black, 503 U. S. 222 (1992) involved the question of invalidly vague "aggravating" circumstances. The jury found true the problematic "especially heinous, atrocious or cruel" circumstance, along with two crisply defined circumstances, and weighed all three against the mitigating factors in its final selection decision. Id., at 225-226. Any one of the three would have made Stringer "death-eligible," i.e., bringing him into the range where the jury had the discretion to impose a death sentence on the basis of "weighing."

The first issue was whether the holding of Maynard v. Cartwright, 486 U. S. 356 (1988), striking down identical language, was a new rule or merely an application of Godfrey v. Georgia, 446 U. S. 420 (1980). Mississippi did not contest that applying Godfrey to this language was not a new rule. "Especially heinous" is not distinguishable from "outrageously . . . vile," since both phrases merely take an adjective that applies to all murders and add an adverb of degree. See 486 U. S., at 364. The differences between the Georgia and Mississippi systems also do not distinguish the case in the direction against applying the rule. In Georgia, the aggravating circumstances as such have no further application after the defendant has been found death-eligible, while in Mississippi those same circumstances go on to be the exclusive factors weighed on the prosecution side in the penalty phase. See Stringer, 503 U. S., at 229-230. This is a difference that, if it has any effect, strengthens rather than weakens the argument for applying Godfrey.

The second issue was whether the Godfrey error was harmless under Zant v. Stephens, 462 U. S. 862 (1983) and Barclay v. Florida, 463 U. S. 939 (1983). On this point, it becomes extremely difficult to reconcile Stringer with a "reasonable jurists could disagree" standard. See Stringer, 503 U. S., at 247-248 (Souter, J., dissenting). In terms of our proposed analysis in Part I-A, Stringer is more than arguably distinguishable fromGodfrey in that the invalid eligibility factor was redundant. Both Stevens and Barclay had already distinguishedGodfrey on that ground. See id., at 245-247.

If Stringer can be reconciled with this Court's past definitions of "new rule," see, e.g., Lambrix, 520 U. S., at 538 ("whether no otherinterpretation was reasonable"), or with our proposed formulation, it can only be because there is something different about harmless error analysis as opposed to basic rules of procedure. Uncertainty in whether a state can salvage a judgment with an admitted error arguably does not impact the state as severely as overturning judgments that were error-free under the law at the time they were rendered. Cf. Sawyer v. Smith, 497 U. S., at 234. If Stringer is not to be overruled, it should at least be limited to its harmless-error context.

In summary, then, a case decided after petitioner's conviction became final or a proposal made by petitioner in the present case is a new rule if either: (1) the case is distinguishable in an arguably material way from the precedents that petitioner claims dictate it; or (2) there were precedents that, at the time, could reasonably be interpreted as contrary to the rule. Observance of these limitations will assist in reducing litigation and confining the habeas writ to its underlying purpose, see Parks, 494 U. S., at 488, which is to " 'guard against extreme malfunctions in the state criminal justice systems.' " Brecht v. Abrahamson, 507 U. S. 619, 634 (1993) (emphasis added) (quoting Jackson v. Virginia, 443 U. S. 307, 332, n. 5 (1979) (Stevens, J. concurring in the judgment)). When the state courts decide the claim without violating the precedents in effect at the time, there has been no extreme malfunction, and the habeas inquiry properly ends there.

C. Mills as a New Rule.

Mills v. Maryland presents none of the problems that made Stringer and Penry difficult. None of this Court's pre-Mills capital cases resembledMills anywhere near as closely as Godfrey resembled Stringer or Eddings resembled Penry. Instead, Mills follows the same pattern of extensions or proposed extensions of existing rules into new areas that we saw in Parks, Sawyer, Taylor, Bohlen, O'Dell, andTeague itself.

For the proposition that the jury instructions in Mills violated the Eighth Amendment, the Mills Court cited only three precedents:Lockett,Eddings, and Skipper v. South Carolina, supra. None of these cases involved an issue regarding how the sentencer went about answering the questions presented. Lockett and Eddings both involved constraints placed by law, or believed to be placed, on what the sentencing judge could consider. Skipper, 476 U. S., at 3, involved evidence withheld from the sentencing jury and therefore not considered by it at all.

In Mills, the jury was instructed to consider defendant's evidence of each mitigating factor and determine if he had proven that factor. See 486 U. S., at 387-388. In a broad sense, all the mitigating factors were "considered" by the sentencer. Whether federal constitutional constraints should go beyond what the sentencer as a whole must consider into the inner workings of the jury was very much an open question in 1988.

Whenever a defendant proposes extending federal control deeper into the workings of the state criminal justice system, the case is distinguishable from cases establishing more general requirements. The Eighth Amendment, in particular, does not "place within constitutional ambit micromanagement tasks that properly rest within the State's discretion to administer its criminal justice system."Harris v. Alabama, 513 U. S. 504, 512 (1995).

In Teague, the petitioner proposed extending the Sixth Amendment cross-section requirement past the drawing of the venire into the selection of the trial jury. See 489 U. S., at 299. The cross-section requirement itself was well-established law. The extension would have furthered the requirement's purposes, and four Justices thought it should be so extended. See id., at 322 (Steven, J., concurring in the judgment); id., at 341 (Brennan, J., dissenting). Federal control under the cross-section requirement had not extended to the petit jury before, see id., at 301, so this further micromanagement would have been a new rule.

Parks involved a reliance on Lockett and Eddings in the same general sense that the Court of Appeals relied on those cases in the present case. The Court of Appeals in the present case believed that those cases established a "constitutional rule that the Eighth Amendment prohibits allbarriers to the sentencer's consideration of any and all mitigating evidence . . . ." Banks v. Horn, 316 F. 3d 228, 235 (CA3 2003) (emphasis added). If they really had, then they would also, as Parks argued, have "required . . . that jurors be allowed to base the sentencing decision upon the sympathy they feel for the defendant after hearing his mitigating evidence." 494 U. S., at 489. But the Court rejected Parks' argument. "Parks asks us to create a rule relating, not to what mitigating evidence the jury must be permitted to consider in making its sentencing decision, but to how it must consider the mitigating evidence." Id., at 490 (emphasis in original). This distinction is precisely on point in the present case.

When the evidence of a mitigating circumstance goes in to the jury room and is debated and voted on by the jury, that circumstance has been considered. Parks and Mills are therefore distinguishable from all of the cases before them where a mitigating circumstance had not been considered because the law did not allow it (Lockett), the sentencing judge thought the law did not allow it (Eddings), or the judge instructed the jury not to consider it. See Hitchcock v. Dugger, 481 U. S. 393, 398-399 (1987). Parks and Mills involve a finer level of detail in the state's capital sentencing structure, where federal regulation had not reached up to that point. The extension would have been new in Parks, and it was new in Mills.

Sawyer rejected an argument that the general principles of Lockett and Eddings dictated a rule on prosecutorial argument, when neither they nor any other Eighth Amendment case had addressed the prosecutor's arguments. See 497 U. S., at 236. The Caldwell rule pushed deeper into the traditional province of the states. Gray v. Netherland, 518 U. S. 152 (1996) similarly rejected, as a proposed new rule, an attempt to extend federal regulation to advance notice of the prosecution's evidence. See also O'Dell v. Netherland, 521 U. S., at 165-166 (informing jury about parole was previously considered policy choice for state).

Until Mills v. Maryland, this Court had interpreted the Eighth Amendment to require only that the sentencer not be precluded from considering the defendant's mitigating evidence. The internal workings of this jury as sentencer had not been subject to any Eighth Amendment regulation. Mills is unlike Penry, which involved an incompletely resolved conflict in this Court's precedents, and unlikeStringer, which involved the harm caused by an undisputed error. Mills is like the Teague cases where an existing rule was extended or proposed for extension into a new area. All of these were new rules, and Mills was a new rule.

II. Rules affecting only the discretionary sentencing decision should be categorically excluded from the
second Teague exception.

"Mills changed nothing. In addition, Mills changed everything." That, in essence, is the holding of the Sixth Circuit in Gall v. Parker, 231 F. 3d 265, 322-324 (2000), declaring that Mills v. Maryland, 486 U. S. 367 (1988) was both an "old rule" and a rule qualifying for the second exception to Teague v. Lane, 489 U. S. 288 (1989).

To qualify for "old rule" status under Teague, a rule must be "dictated by precedent." 489 U. S., at 301 (emphasis in original). To qualify for the second exception, a rule must be a "watershed rule[] of criminal procedure . . . [that] 'alter[s] our understanding of the bedrock procedural elements' . . . ." Id., at 311 (emphasis omitted). These two categories are beyond distinct. They are separated by the Grand Canyon. A watershed rule is a rule of the revolutionary character of Gideon v. Wainwright, 372 U. S. 335 (1963). See Safflev.Parks, 494 U. S. 484, 495 (1990). It is not just a change in the law, but a dramatic change.

In his concurring and dissenting opinion in Mackey v. United States, 401 U. S. 667 (1971), Justice Harlan suggested an exception "for claims of nonobservance of those procedures that . . . are 'implicit in the concept of ordered liberty.' " Id., at 693 (quoting Palkov.Connecticut, 302 U. S. 319, 325 (1937)). These are the very basic rules of fundamental fairness that were considered part of the Fourteenth Amendment's Due Process Clause even before the incorporation doctrine. ⁽⁶⁾ The purpose of this exception appears to be to explain Justice Harlan's continued concurrence in granting habeas relief to prisoners with claims under Gideon v. Wainwright, 372 U. S. 335 (1963), which was then only eight years old.

While Teague adopted Justice Harlan's proposal in most other respects, it raised the bar for this exception even higher. To qualify for the second exception, a new rule would have to meet both the Palko test of fundamentalness and Justice Harlan's original proposal of " 'rules which significantly improve the pre-existing factfinding procedures.' " Teague, 489 U. S., at 312 (quoting Desist v. United States, 394 U. S. 244, 262 (1969)). Such a high standard raised the obvious question of whether any such rules remained to be made over a quarter century after Gideon. The Teague plurality answered cautiously, "we believe it unlikely that many such components of basic due process have yet to emerge." Id., at 313 (emphasis added).

In the fourteen years since Teague, this Court has not found a single new rule to qualify for this exception. See O'Dell v. Netherland, 521 U. S. 151, 171 (1997) (Stevens, J., dissenting). It has found a great many rules and proposed rules that do not. A handful of circuit cases have found rules to qualify, but Congress had so little confidence in the Court of Appeals' decisions on retroactivity that it limited the retroactive rule exception for successive petitions to rules "made retroactive . . . by the Supreme Court." See 28 U. S. C. §2244(b)(2)(A).

In the post-Teague cases, this Court has only marginally clarified the definition of the second exception. There are repeated statements along the lines that a qualifying rule would have to be of the "primacy and centrality" of Gideon. See Parks, 494 U. S., at 495.

The farther we go down the path of the evolution of criminal procedure, the less likely it becomes that any "absolute prerequisite to fundamental fairness," Teague, 489 U. S., at 315, remains unrecognized. Justice Stevens recognized the connection between the fundamental nature of a rule and the time of its recognition in his dissent in Rose v. Lundy, 455 U. S. 509, 544, and nn. 9-11 (1982), where he illustrated "fundamental" by reference to the "classic grounds" for habeas corpus --all rules which predate Gideon.

Tyler v. Cain, 533 U. S. 656 (2001) contains an important shift in language which recognizes this further diminution in the probability of the existence of any new, fundamental rule. There the Court said, "it is unlikely that any of these watershed rules 'ha[s] yet to emerge.' "Id., at 667, n. 7 (emphasis added). The shift from "many" in 1989 to "any" in 2001 recognizes that the probability is asymptotically approaching zero.

Because the second exception exists in theory, it must be litigated for every new rule in every circuit, except for the rules where this Court decides the question. Yet this Court cannot decide the question on every rule. The fact that the retroactivity of Mills remains undecided 15 years after that decision, despite long-standing circuit splits and numerous petitions for certiorari, vividly illustrates the problem. See, e.g., Petition for Writ of Certiorari in Dixon v. Williams, No. 92-111 (requesting resolution of the circuit split eleven years ago). The Teague opinion denounced the "unfortunate disparity in the treatment of similarly situated defendants on collateral review" that resulted from varying decisions in the lower courts on retroactivity under the prior doctrine. 489 U. S., at 305. Yet the same result has followed under Teague from confusion over the second exception.

In the not-too-distant future, amicus CJLF believes it will be appropriate for the Court to accept that there are no rules of the "primacy and centrality of . . . Gideon" remaining to be made. The second exception can then be formally retired as unnecessary and the litigation it creates ended. For now, we propose a more limited curtailment. We propose a bright-line rule that new rules which govern the process of choosing a sentence within the range for which the defendant is eligible are per se outside the second exception.

Both this Court and the Congress have altered the law of habeas corpus over the course of the last thirty years with two goals in mind: to keep the writ available to correct fundamental miscarriages of justice and to curtail its misuse for endless relitigation of questions already reasonably resolved once. To reconcile these competing goals, new limitations on the writ have typically come with "actual innocence" exceptions. SeeMurray v. Carrier, 477 U. S. 478, 495-496 (1986); 28 U. S. C. §2244(b)(2)(B)(ii). The unifying theme is to tighten restrictions in the interest of finality and federalism but relax them to avoid a fundamental miscarriage of justice. The Teague rule is an important part of the tightening, and the second exception is based on the actual innocence consideration. See Withrow v. Williams, 507 U. S. 680, 699-700 (1993) (O'Connor, J., dissenting).

The second exception to Teague is closely related to the actual innocence exception of Carrier, as Teague itself makes clear. See 489 U. S., at 313. Application of this exception to capital cases is therefore illuminated by the two cases defining the Carrier exception:Sawyer v. Whitley, 505 U. S. 333 (1992) and Schlup v. Delo, 513 U. S. 298 (1995).

Sawyer noted that in an ordinary criminal case, the "fundamental miscarriage of justice" exception meant that the person was "actually innocent," which in turn simply meant that "the State has convicted the wrong person of the crime." See 505 U. S., at 340. In the context of capital sentencing, Sawyer rejected the argument that the concept extended beyond eligibility for the death penalty (i.e., truth of at least one aggravating factor) to "the ultimate discretionary decision between the death penalty and life imprisonment." Id., at 343. A "fundamental miscarriage of justice," for this purpose, means only a death sentence imposed on a person legally ineligible for it, not a sentence within the discretionary range. See id., at 347. Schlup further distinguished actual innocence of the crime from ineligibility for the penalty. The greater importance of realinnocence warranted a more relaxed standard for the exception to finality. See 513 U. S., at 324.

If we take the Sawyer/Schlup implementation of Carrier along with Teague's reliance on Carrier for the second exception, a simple, clear, bright line becomes visible. Just as issues relating only to the discretionary sentencing decision never qualify for the "fundamental miscarriage of justice" exception of Carrier, so new rules relating only to the operation of that discretionary step never qualify for the second exception toTeague.

The bar for death-eligibility is very high. First, the defendant must be guilty of murder, not some lesser offense. See Coker v. Georgia, 433 U. S. 584, 592 (1977) (plurality opinion). The class of murderers must be further narrowed by objective factors in the form of aggravating circumstances or a higher degree of murder. See Tuilaepa v. California, 512 U. S. 967, 971-972 (1994). The defendant must not qualify for a long and growing list of exclusion criteria. See Thompson v. Oklahoma, 487 U. S. 815, 838 (1988) (plurality opinion) (age); Enmund v. Florida, 458 U. S. 782, 797 (1982) (minor accomplice without intent to kill); Atkins v. Virginia, 536 U. S. 304, 318 (2002) (mentally retarded). Once the case has cleared all those hurdles, it is difficult to imagine a case where the justice of the death penalty is not a matter of opinion on which reasonable people can and will differ.

A bright-line rule "is designed to avoid the costs of excessive inquiry where [it] will achieve the correct result in almost all cases."Coleman v.Thompson, 501 U. S. 722, 736 (1991). Categorically excluding the regulation of discretionary sentencing from the secondTeague exception satisfies that criterion. The only rules in this area that are even arguably of the "primacy and centrality of . . .Gideon" are those of Gregg v.Georgia, 428 U. S. 153 (1976) and its companion cases. Since then, every defendant sentenced to death has been entitled to a narrowed definition of eligibility and an individualized consideration of mitigating circumstances. By comparison, all the rules created since then have amounted to fine tuning within these broad outlines.

The continued existence in theory of an exception which never applies in fact serves only to delay justice in cases where it is already long overdue and to squander resources better spent elsewhere. We have devoted far too large a share of our case review resources on the claims of unquestionably guilty murderers, claims which have nothing to do with guilt or innocence.

The only consequence of a Teague bar is that the habeas petitioner's sentence is judged by the standards in effect at the time of his direct appeal. See Sawyer v. Smith, 497 U. S. 227, 243 (1990). That includes all the requirements of Gregg and its companion cases. For issues having no relation whatever to actual innocence of the murder, that is sufficient to avoid a fundamental miscarriage of justice.

The rule of Mills v. Maryland has nothing to do with guilt of the offense and nothing to do with eligibility for the death penalty. Even its holding that the instruction risked the imposition of a death penalty despite compelling mitigating circumstances is far-fetched, for the reasons stated in Part III, below. The second-exception argument could be summarily dismissed, as this Court has done in many cases. See, e.g., Gray v.Netherland, 518 U. S. 152, 170 (1996). However, a substantial step toward achieving the goals set out in Teague could be achieved by simply declaring the second exception to be per se inapplicable to rules governing the discretionary sentencing decision.

III. Even if Mills applies, both Teague and AEDPA
require that the state court judgment stand.

A. Teague and AEDPA.

On state collateral review, the Pennsylvania Supreme Court chose not to rely on the nonretroactivity of Mills v. Maryland, 486 U. S. 367 (1988), but instead to rely on its earlier precedents holding that the Pennsylvania instruction in question does not violate Mills. SeeCommonwealth v.Banks, 656 A. 2d 467, 470 (Pa. 1995). ⁽⁷⁾ The state court's decision on the merits is immune from collateral attack unless it is "contrary to, or involved an unreasonable application of," this Court's precedents. See 28 U. S. C. §2254(d)(1).

The instruction and verdict form in this case were different from the ones in Mills, and the differences raise another but very similar issue. As is typical in Teague cases, there is a question whether the precedent the petitioner relies on is a new rule and a distinct question of whether extension of that precedent to cover this case would be another new rule. See Stringer v. Black, 503 U. S. 222, 228 (1992). However, when the state court has ruled on the merits and §2254(d)(1) applies, this second-step Teague inquiry is subsumed in the AEDPA inquiry. Any extension of the precedent that would be "new" for Teague is not "clearly established" for AEDPA. SeeWilliamsv. Taylor, 529 U. S. 362, 412 (2000). Any rule which is not "clearly established" cannot be used to avoid §2254(d)'s limitation on granting relief. See Lockyer v. Andrade, 538 U. S. 63, 155 L. Ed. 2d 144, 155-156, 123 S. Ct. 1166, 1173-1174 (2003).

B. Two Kinds of Unanimity.

A requirement that the jury reject and not consider any mitigating circumstance on which it is not unanimous is a violation of theMillsrule. On the other hand, a requirement that the jury continue deliberating until it is unanimous one way or the other is perfectly valid. See McKoy v. North Carolina, 494 U. S. 433, 455, n. * (1990) (Kennedy, J., concurring in the judgment). The Court of Appeals in this case completely failed to recognize the distinction. This argument is presented in our brief at the petition stage, and space does not permit its repetition here. See Brief for Criminal Justice Legal Foundation as Amicus Curiae in Support of the Petition for Writ of Certiorari in Beard v. Banks, No. 02-1603, pp. 11-16.

C. The Boyde Criterion.

There is another point which apparently needs to be reiterated. The Court of Appeals in this case relied on Mills v. Maryland for the proposition that "the critical question is . . . whether a reasonable jury could have interpreted the instructions in an unconstitutional manner . . . ." Banks v.Horn, 271 F. 3d 527, 544 (CA3 2001) (emphasis added). Mills did indeed say that, as the dissenters so vigorously asserted in Boyde v. California, 494 U. S. 370, 390 (1990) (Marshall, J., dissenting) ("unequivocally confirmed"). However, the Boyde majority found the precedents as a whole unclear and adopted the "reasonable likelihood" standard. Id., at 380. Mills is simply no longer good law on this point.

The Court of Appeals in this case was oblivious to the distinction between the Mills standard and the Boyde standard, and it cited them interchangeably as if they were the same. Indeed, in the very same paragraph in which the court cites the discredited Mills standard, it also citesBoyde. See 271 F. 3d, at 544.

The "reasonable likelihood" inquiry of Boyde allows the reviewing court to give the jurors some credit for common sense. TheMillsCourt agonized over the nightmare scenario of a jury that felt compelled to return a verdict of death even though eleven agreed that life was the proper verdict, or even if all twelve agreed but for different reasons. See 486 U. S., at 373-374. Applying Boyde to an arguably ambiguous instruction, we must ask how realistic it is that the twelve jurors would adopt the interpretation that forces them to a result that most or all of them believe to be unjust, when there is another interpretation available that would produce the result they believe is right. We should not analyze a jury instruction like a digital electronic circuit, tracing down every conceivable path. See Cuppv.Naughten, 414 U. S. 141, 149 (1973) ("abstract and conjectural emanations"). Under Boyde, we ask what is realistic.

Jurors have enough sense to know that the final verdict is what is really important. Even when they are unambiguously instructed to reach an agreement on intermediate findings, they will typically put aside disagreements and mark whatever steps are needed to reach the agreed-upon conclusion. See Ginsburg, Special Findings and Jury Unanimity in the Federal Courts, 65 Colum. L. Rev. 256, 268 (1965).

The verdict form in Mills unambiguously required unanimity to mark a mitigating circumstance "yes" and omitted the word "unanimously" for marking a circumstance "no." See 486 U. S., at 387. The verdict form in the present case is substantially different. It refers to unanimity only with regard to the ultimate issue. "We the jury have found unanimously . . . One or more aggravating circumstances which outweigh any mitigating circumstance or circumstances." Banks v. Horn, 271 F. 3d, at 549-550 (emphasis omitted). Applying the clearly established law ofBoyde, a court could reasonably conclude there is no reasonable likelihood that the jurors interpreted the form in the unjust way, contrary to common sense, feared by the Mills Court. Indeed, that conclusion is not only reasonable, it is correct.

In this case, the highest court of the state and the federal court of appeals simply disagree. The state court has not acted in a renegade manner. It recognized the controlling precedents of this Court, applied them to the facts, and reached a result well within the bounds of reasonable disagreement among competent, conscientious jurists. Congress has decided that in such circumstances the state court judgment should stand, unless reviewed by this Court on certiorari.

CONCLUSION

The decision of the Court of Appeals for the Third Circuit should be reversed.

November, 2003

Respectfully submitted,

Kent S. Scheidegger

1. This brief was written entirely by counsel for amicus, as listed on the cover, and not by counsel for any party. No outside contributions were made to the preparation or submission of this brief.

Both parties have given written consent to the filing of this brief.

2. Judge Sloviter made no attempt to reconcile her unique definition of finality with the fact that this Court applied the usual definition in identical circumstances in Sawyer v. Smith, 497 U. S. 227 (1990). See Brief for Criminal Justice Legal Foundation as Amicus CuriaeinHorn v. Banks, No. 01-1385, pp. 4-5.

3. A jury has since found that Penry is not, in fact, retarded. Babineck, Texas Jury Rejects Rapist's Retarded Claim, Associated Press, July 3, 2002.

4. Although Justice Blackmun's concurrence is the opinion on the "narrowest grounds," cf. Marks v. United States, 430 U. S. 188, 193 (1977), we refer only to the plurality opinion because of its later acceptance by the Court in Eddings.

5. Cf. Andrade, 155 L. Ed. 2d, at 157, 123 S. Ct., at 1174 ("Andrade's sentence also was not materially indistinguishable from the facts inSolem").

6. The holding of Palko was that the Fifth Amendment's Double Jeopardy Clause was not such a rule. By the time of Mackey, the Double Jeopardy Clause was considered "incorporated," and Palko had been overruled. See Benton v. Maryland, 395 U. S. 784, 796 (1969).

7. The mandate of Caspari v. Bohlen, 510 U. S. 383, 389 (1994), that Teague be addressed as a threshold issue, is limited to the federal courts. Cf.Johnson v. Fankell, 520 U. S. 911, 919-920 (1997) (federal rule on interlocutory appeal in §1983 case, created for the benefit of states and their officers, not binding on state court).