 |
|||||
|
|
|
|
|||
|
|
|||||
|
|
|
|
|||
Petitioner maintains, and a number of courts of appeals have held, that Sullivan v. Louisiana, 508 U. S. 275 (1993) compels the conclusion that the rule of Cage v. Louisiana, 498 U. S. 39 (1990) (per curiam), disapproved in part in Estelle v. McGuire, 502 U. S. 62, 72, n. 4 (1991), qualifies for the second exception to the nonretroactivity rule of Teague v. Lane, 489 U. S. 288 (1989). Brief for Petitioner 24. This argument is a non sequitur. Sullivan addressed whether "Cage error" is susceptible to harmless-error analysis. 508 U. S., at 276. Retroactivity and harmless error are distinct questions and must be addressed separately.
Harmless error analysis and nonretroactivity both have the effect of denying relief to some defendants, despite an aspect of their trials that today would be recognized as constitutional error. The two doctrines therefore have some common aspects, and it is not surprising to see some similar language in the opinions. They also have significant differences, however. The classes of denied defendants are different, and the considerations that go into denying relief in a particular case are different.
A finding that a rule is not retroactive on collateral review cuts off relief from an entire class of defendants, i.e., those whose convictions were final before the rule was made. There is no individual consideration of whether observance of the rule would have made a difference in the particular case. The effect is temporary, however, as nonretroactivity has no effect on future cases. Harmless error is applied case by case, but it applies forever. Arizona v. Fulminante, 499 U. S. 279 (1991) applied harmless error analysis to a rule created half a century earlier in Brown v. Mississippi, 297 U. S. 278 (1936).
Nonretroactivity serves to limit the costs imposed by overturning final convictions based on new rules when those costs outweigh the benefits. See Sawyer v. Smith, 497 U. S. 227, 242-243 (1990). It is the nature of the judicial process that new rules can only be made in cases where they affect the outcome, see Teague, 489 U. S., at 315, and equal treatment demands that those similarly situated be treated the same. See ibid. However, the price of progress would be too great if every new rule required reexamination and possibly retrial of every case when the prior rule was applied. Id., at 310. The focus is the difference in the law between two points in time. Is the cost of retrying cases, which were reasonably deemed properly tried under prior law, outweighed by the value of the additional protection of the new rule, compared to the protections available under prior law?
Harmless error has an altogether different basis. It seeks to achieve the just result in the individual case. While harmless error is also based on "social costs," see Calderon v. Coleman, 525 U. S. 141, 146 (1998) (per curiam), the cost-benefit analysis assesses whether the likelihood that the error made a difference in the particular case outweighs the cost of retrial. See O'Neal v. McAninch, 513 U. S. 432, 443 (1995) (balancing interests and risks). The relevant comparison is between what actually happened and what should have happened. The timing and magnitude of changes in the law are irrelevant. Nonretroactivity on habeas and eligibility for harmless error analysis are both the general rules, and the respective classes of exceptions are both small. They are, however, different classes and small for different reasons.
The jury discrimination cases establish unequivocally that the two classes are not the same. Vasquez v. Hillery, 474 U. S. 254 (1986) rejected the state's argument that racial discrimination in the selection of the grand jury could be harmless error when followed by a conviction by a properly selected trial jury. The main reason is that this "grave constitutional trespass" would have no effective remedy without reversal of the subsequent conviction. See id., at 262. There is no case-by-case assessment to make. The choice is reverse all such cases or reverse none.
Certainly it cannot be said that this error renders the final result unreliable. States do not have to use grand juries at all, Hurtado v. California, 110 U. S. 516, 538 (1884), or even preliminary hearings. See Lem Woon v. Oregon, 229 U. S. 586, 589 (1913). The only argument the Hillery Court could make for any effect on the outcome was the discretion of the grand jury not to indict despite proof of guilt. See 474 U. S., at 263. To say that such an exercise of discretion was highly unlikely in the case of the brutal murder of a 15-year-old girl, see id., at 267 (Powell, J., dissenting), would be an understatement. The "unfairness" that precludes harmless error analysis in racial discrimination cases is not a lack of confidence in the result but rather the denial of equal treatment based on race, contrary to the very core of the Fourteenth Amendment. Statements in other cases regarding structural error and reliability, see, e.g., Neder v. United States, 527 U. S. 1, 8-9 (1999), must be considered with Hillery in mind. Some rules qualify for this small group because of their powerful impact on reliability, but others qualify for other reasons. Another example is the right of self-representation, see id., at 8, the exercise of which rarely improves the defendant's chances of acquittal. See Martinez v. Court of Appeal, 528 U. S. 152, 161 (2000); see also Morris v. Mathews, 475 U. S. 237, 244-245 (1986) (retrial for jeopardy-barred offense "not in any sense harmless," even though conviction reduced to non-barred offense); cf. Caspari v. Bohlen, 510 U. S. 383, 396 (1994) (extension of Double Jeopardy Clause to noncapital sentencing not within second Teague exception).
If structural errors were necessarily Teague exceptions, then all jury discrimination rules would be retroactive on habeas corpus. The definitive counterexample is Teague itself. Teague objected at trial to the prosecutor's use of peremptory challenges, allegedly in a racially discriminatory manner. Teague, 489 U. S., at 293. This claim would later be recognized in Batson v. Kentucky, 476 U. S. 79, 97-98 (1986), but was held nonretroactive on habeas in Allen v. Hardy, 478 U. S. 255 (1986) (per curiam). See Teague, supra, at 294. The Teague Court would not reconsider Allen v. Hardy, id., at 296, and it held that creation of a rule equivalent to Batson under the Sixth Amendment instead of the Equal Protection Clause would be a nonretroactive new rule. Id., at 299 (plurality opinion). The dissent asserted Hillery for the proposition that Teague's proposed rule qualified for the second exception based on its "far graver impact on the fundamental fairness of a petitioner's trial than the discrimination we condemned in Hillery . . . ." Id., at 344 (opinion of Brennan, J.). The Hillery argument is implicitly rejected by the plurality, which held that the error in question "does not undermine the fundamental fairness that must underlie a conviction." Id., at 315; see Clemons v. Mississippi, 494 U. S. 738, 747-748, n. 3 (1990) (argument made in dissent deemed implicitly rejected by majority).
Conversely, coerced confessions are not structural error, see Arizona v. Fulminante, 499 U. S., at 310, even though the rule against them is fundamental for Teague purposes. See Teague, 489 U. S., at 313. To be sure, a few rules are both structural error and Teague exceptions by reason of their powerful and unassessable impact on the reliability of the outcome. The paradigm is, of course, Gideon v. Wainwright, 372 U. S. 335 (1963). See Neder, 527 U. S., at 8 (structural error); Saffle v. Parks, 494 U. S. 484, 495 (1990) (Teague exception). The set of rules that qualify for the Teague exception and the set that are "structural error" are intersecting sets, neither being a subset of the other. Hence, membership in one does not determine membership in the other. The fact that "Cage error" is "structural" does not tell us whether Cage is retroactive on habeas.
Most rules of criminal procedure enhance the accuracy of the proceeding. That is usually why they are made in the first place.(4) The fact that a rule has this effect is not enough to make it retroactive on habeas corpus. The exception is reserved for "watershed rules of fundamental fairness." Sawyer, 497 U. S., at 242. The rule must be "essential to the accuracy and fairness of the criminal process," id., at 243 (emphasis added), rather than merely enhancing accuracy. Cf. id., at 242.
"Watershed" means, in its figurative sense, a "critical point that marks a division or a change of course; a turning point . . . ." American Heritage Dictionary 2019 (3d ed. 1992). Rules that enhance or add an additional layer of protection to existing rules do not qualify. The rule in question must be the bedrock rule, with "the primacy and centrality of the rule adopted in Gideon . . . ." Saffle, 494 U. S., at 495.
Sawyer, supra, has the most detailed discussion of the exception to date, and the case illustrates the incremental versus watershed distinction. Sawyer involved the retroactivity of the rule of Caldwell v. Mississippi, 472 U. S. 320 (1985) regarding prosecutor arguments which diminish the jury's sense of responsibility. See Sawyer, 497 U. S., at 232. Before Caldwell, "the rule of Donnelly [v. De Christoforo, 416 U. S. 637 (1974)] was in place to protect any defendant who could show that a prosecutor's remarks had in fact made a proceeding fundamentally unfair." Id., at 243. Caldwell provided "an additional measure of protection." Id., at 244. This incremental protection had value, but it did not qualify for the exception. Ibid.
Stovall v. Denno, 388 U. S. 293 (1967) provides another illuminating example, even though it comes from the pre-Teague era. Since Teague was a sharp curtailment of habeas retroactivity, see Teague, 489 U. S., at 327 (Brennan, J., dissenting), it follows that rules that were not retroactive under the prior body of law would not qualify for the second Teague exception. Stovall addressed the retroactivity of new rules regarding the right to counsel at pretrial identifications. See 388 U. S., at 294; United States v. Wade, 388 U. S. 218, 226-227 (1967); Gilbert v. California, 388 U. S. 263 (1967). Even though the Wade and Gilbert rules were "justified by the need to assure the integrity and reliability of our system of justice," Stovall, supra, at 299, they were not retroactive, for two reasons. First, although injustices were possible in cases not conforming to these rules, the number of cases where no unfairness or injustice resulted was also large, much more so than in cases where counsel was denied on trial or appeal. See ibid. Second, a defendant whose pretrial confrontation was truly unfair could fall back on pre-Wade due process requirements.
The base right to counsel is, as noted supra, the paradigm of a rule qualifying for the second Teague exception. Yet an extension of that rule to the critical stage of a pretrial identification was not retroactive. Gideon was a watershed rule. Wade's enhancement of Gideon was not.
The watershed rule involved in the present case is In re Winship, 397 U. S. 358 (1970). Winship elevated to "constitutional stature" the requirement of proof beyond a reasonable doubt in criminal cases. Id., at 364. That rule had been universally followed in adult criminal cases before Winship, see id., at 361, but Winship made it a federal constitutional right.
In 1978, before the decision of petitioner's direct appeal, Taylor v. Kentucky, 436 U. S. 478 (1978) established that some level of constitutional protection existed for jury instructions in this area. The actual holding of Taylor involved the presumption of innocence, id., at 490, but the truncated, unclear, and confusing reasonable doubt instruction was a factor in the decision. See id., at 488; see also id., at 486, n. 13 (relation of presumption of innocence to proof beyond a reasonable doubt).
On this foundation, Cage v. Louisiana, 498 U. S. 39 (1990) (per curiam), decided that the instruction used in that case was "contrary to the reasonable doubt requirement articulated in Winship." Id., at 41; see also id., at 41, n. * (citing, inter alia, Taylor). While this holding was not dictated by precedent, and hence "new," it was hardly a "watershed rule." The basic right had been established. Instructions were already subject to scrutiny under this requirement. Cage was a fairly routine case about whether a general principle should be extended to invalidate a particular practice. It was a question of law and not a "mixed question," making Teague analysis of Cage like Butler v. McKellar, 494 U. S. 407, 414-415 (1990), and unlike Wright v. West, 505 U. S. 277, 306 (1992) (Kennedy, J., concurring in the judgment), but there is no "watershed" quality about it.
Before Cage, all defendants were entitled to instructions on the presumption of innocence and the standard of proof beyond a reasonable doubt. After Cage, it was established that the particular instruction in Cage's case was not sufficient. Four years later, it was established that another instruction with only a few differences did meet constitutional minimums. See Victor v. Nebraska, 511 U. S. 1, 19-22 (1994). Under the analysis of Sawyer, the question is whether the incremental change that Cage made was of the fundamental character needed to override the usual rule of nonretroactivity. The answer, amicus submits, is clearly no. This is fine-tuning.
This brings us back to Sullivan v. Louisiana, supra. The fact that Sullivan held that Cage error was not susceptible to harmless error analysis does not, by itself, require the conclusion that it also qualifies for the second Teague exception. Its discussion of Cage does shed some light, though. The reasons for Sullivan's holding do not apply to retroactivity.
Sullivan begins and ends its analysis by citing Duncan v. Louisiana, 391 U. S. 145 (1968) for the importance of having the jury rather than judges make the finding of guilt. Sullivan, 508 U. S., at 277, 281. Yet Duncan was not retroactive. "The values implemented by the right to jury trial would not reasonably be served by requiring retrial of all persons convicted in the past by procedures not consistent with the Sixth Amendment right to jury trial." De Stefano v. Woods, 392 U. S. 631, 634 (1968) (per curiam). Having the judge rather than the jury decide is not inherently unfair in the sense of creating a danger of wholesale convictions of innocent persons, see id., at 633-634, but rather is inconsistent with " 'a profound judgment about the way in which law should be enforced and justice administered.' " Sullivan, supra, at 281 (quoting Duncan).
Rose v. Clark, 478 U. S. 570, 577 (1986) described some structural errors as those without which "a criminal trial cannot reliably serve its function as a vehicle for determination of guilt or innocence . . . ." Sullivan quoted this description. See 508 U. S., at 281. But this passage of Rose does not deal with jury trial. That issue is addressed in the next paragraph. There, Rose says that jury trial is structural because, regardless of how reliable the result may be, "the error in such a case is that the wrong entity judged the defendant guilty." 478 U. S., at 578. Sullivan also quotes and relies on this statement. 508 U. S., at 281. See also Neder, 527 U. S., at 38 (Scalia, J., dissenting) (Sullivan based on allocation, not accuracy, of decision).
Sullivan follows the same line as Vasquez v. Hillery and the rule that denial of self-representation cannot be harmless. See McKaskle v. Wiggins, 465 U. S. 168, 177, n. 8 (1984). No matter how strong our confidence in the accuracy of the trial's result may be, the error is not harmless because the rule in question has a purpose independent of insuring accuracy. For appellate judges to make their own finding about the evidence in the individual case under harmless error analysis would compound rather than cure the error. See Sullivan, supra, at 280.
Conspicuous by its absence from Sullivan is any discussion of how or how much Cage changed the law from what it had been before Cage. Such an inquiry is central to retroactivity and irrelevant to harmless error. Indeed, the only discussion in Sullivan of the specific rule in Cage, as opposed to the general reasonable doubt and jury trial requirements, is in the footnote. See 508 U. S., at 278, n. *. That footnote reserves the question of whether Cage is even correct after Boyde v. California, 494 U. S. 370, 380 (1990) and Estelle v. McGuire, 502 U. S. 62, 72-73, n. 4 (1991). It would be more than a little strange to say that a case establishes an "absolute prerequisite to fundamental fairness," see Teague, 489 U. S., at 314, but then again it might be wrong.
Tyler, Cage, Victor, and Sandoval were all convicted by juries who were instructed that they had to find guilt proven beyond a reasonable doubt. See Victor, 511 U. S., at 7 (Sandoval, California instruction); id., at 18 (Victor, Nebraska instruction). All four therefore received the protection of the bedrock procedural element established in Winship. In Cage, the actual implementation of that element fell a little short of the line. In Victor the implementation cleared the line. See id., at 21-22. The instruction in the present case lies in between. It contains the "abiding conviction" language that Victor found helpful, but not the other clarifying elements of the Nebraska instruction. Compare id., at 21, with supra, at 2-3. These line-drawing exercises on the ragged edges of constitutional doctrine are important and need to be done, but they have "none of the primacy and centrality of . . . Gideon." See supra, at 24. What is the probability that an innocent person was convicted because his jury got the Cage instruction, but the jury would have acquitted him if they had received the Victor instruction instead? Amicus submits it is small.
Cage v. Louisiana prompted an improvement in criminal justice, forcing courts and legislatures to revise and improve instructions long in need of revision. That is a highly desirable result. The cost of progress, however, need not include retrial of every case tried under instructions which were considered entirely proper for many years. That cost is only justified for rules such as Gideon, where the improvement of a single step is so large and dramatic that substantial numbers of innocent people may be left in prison if full retroactivity is not extended to them. The rule of Cage v. Louisiana is not such a rule.
The decision of the United States Court of Appeals for the Fifth Circuit should be affirmed.
February, 2001
Respectfully submitted,
Kent S. Scheidegger
Attorney for Amicus Curiae
Criminal Justice Legal Foundation
|
|
||||
|
|
|
|
||
4. The principal contra-accuracy rule has been virtually banned from habeas corpus. See Stone v. Powell, 428 U. S. 465, 490-491 (1976).