106 F.3d 1381
7th Cir.
1997
EASTERBROOK, Circuit Judge.

In 1976 the Supreme Court of Wisconsin held that under Wis.Stat. § 974.06(4) criminal defendants may present on collateral attack any constitutional contentions they omitted from direct appeal—no matter why these claims were omitted. Bergenthal v. State, 72 Wis.2d 740, 242 N.W.2d 199 (1976). Eighteen years later, on June 22, 1994, that court overruled Bergenthal, concluding that it had misread this statute. State v. Escalona-Naranjo, 185 Wis.2d 168, 517 N.W.2d 157 (1994). Today in Wisconsin a prisoner needs a “sufficient reason to raise [in a collateral attack] a constitutional issue ... that could have been raised on direct appeal”. 517 N.W.2d at 164 (emphasis in original). State courts apply the holding of Escalona-Naran-jo to prisoners whose direct appeals were filed before Bergenthal was overruled.

Jon Liegakos, whose conviction for murder was affirmed in 1987, is such a person. Wisconsin’s judiciary dismissed his collateral attack, because he declined to give any reason for omitting from his direct appeal the arguments he raised in the collateral proceeding. Confined under a sentence of life imprisonment, Liegakos now wants a federal writ of habeas corpus. The district court denied his petition for a combination of procedural and substantive reasons. 928 F.Supp. 799 (1996).

Liegakos’s principal argument is that the application of Escalona-Naranjo to a case in which the direct appeals preceded June 22,1994, violates the due process clause of the fourteenth amendment, and that he is therefore entitled to an adjudication in Wisconsin’s courts (which use doctrines more favorable to prisoners than those in § 2254 litigation) as if Bergenthal remained the law. Anything less, he contends, allows a state to set a trap for unwary litigants.

After Marks v. United States, 430 U.S. 188, 97 S.Ct. 990, 51 L.Ed.2d 260 (1977), and Bouie v. Columbia, 378 U.S. 347, 84 S.Ct. 1697, 12 L.Ed.2d 894 (1964), the due process clause places judges under the same basic constraint as the ex post facto clause does for legislatures: new rules that increase the punishment for crime, or make additional acts criminal, cannot be applied to conduct predating the change. It does not follow, however, that all procedural changes fall under this ban. See California Department of Corrections v. Morales, — U.S. -, 115 S.Ct. 1597, 131 L.Ed.2d 588 (1995); Collins v. Youngblood, 497 U.S. 37, 110 S.Ct. 2715, 111 L.Ed.2d 30 (1990).

Alterations in procedures, including the law of collateral attack, are frequent. For example, in 1977 the Supreme Court of the United States held that prisoners who want to present arguments that have been procedurally defaulted—such as those omitted from the briefs on direct appeal, Murray v. Carrier, 477 U.S. 478, 106 S.Ct. 2639, 91 L.Ed.2d 397 (1986)—may raise them on collateral attack only if they can establish cause and prejudice. Wainwright v. Sykes, 433 U.S. 72, 97 S.Ct. 2497, 53 L.Ed.2d 594 (1977). This departure from the former law, which permitted a prisoner to raise all contentions that had not been deliberately bypassed, see Fay v. Noia, 372 U.S. 391, 83 S.Ct. 822, 9 L.Ed.2d 837 (1963), was applied to Sykes himself, and to all later petitioners no matter when their convictions occurred. Liegakos’s position implies that the Supreme Court violated the due process clause, for the effect of Sykes was more substantial than that of Es-calonar-Naranjo. (The “sufficient reason” standard is more generous to prisoners than the cause-and-prejudice standard.)

Changes in the law of collateral attack constitutionally may be applied to persons who were convicted while greater opportunities for collateral review existed. Felker v. Turpin, — U.S. -, -, 116 S.Ct. 2333, 2340, 135 L.Ed.2d 827 (1996); Lindh v. Murphy, 96 F.3d 856, 867-68, 871-74 (7th Cir.1996) (en banc), cert. granted on a different issue, — U.S. -, 117 S.Ct. 726, 136 L.Ed.2d 643 (1997); United States v. Burnom, 27 F.3d 283 (7th Cir.1994). Liega-kos would have a stronger argument if he had relied on Bergenthal in 1987 when deciding which issues to present. But he does not say that he did-—and we presume that reliance on the former state of the law would be a “sufficient reason” to make the claim belatedly. Cf. Burris v. Parke, 95 F.3d 465 (7th Cir.1996) (en banc). Having declined to advance any reason why the claims he now presents were withheld in 1987, Liegakos cannot successfully maintain that the state must entertain a collateral attack. Application of Wis.Stat. § 974.06(4), as Escalona-Naranjo understands it, to persons convicted before June 22, 1994, does not violate the Constitution; a state may curtail or even abolish collateral review as it pleases.

At this point Wisconsin drops the other shoe: failure to present the arguments on appeal in 1987 is a procedural default, which forecloses review under § 2254 in the absence of cause and prejudice. Lack of a “sufficient reason” for the omission necessarily means a lack of “cause,” the state contends. Maybe so—but the framework of Sykes applies only if the defendant forfeited the claim under a rule that supplies an independent and adequate ground of decision. Whether the ground is independent depends on state law, see Hogan v. McBride, 74 F.3d 144 (7th Cir.1996); whether it is adequate depends on federal law.

NAACP v. Alabama ex rel. Patterson, 357 U.S. 449, 457, 78 S.Ct. 1163, 1169, 2 L.Ed.2d 1488 (1958), holds that a rule of procedure is not adequate to prevent federal collateral review when the defendant could not be “deemed to have been apprised of its existence” at the time he omitted the procedural step in question. See also Barr v. Columbia, 378 U.S. 146, 149, 84 S.Ct. 1734, 1736, 12 L.Ed.2d 766 (1964) (state procedural rules “not strictly or regularly followed” do not bar review). James v. Kentucky, 466 U.S. 341, 348-51, 104 S.Ct. 1830, 1835-37, 80 L.Ed.2d 346 (1984), generalized these holdings when concluding that only a “firmly established and regularly followed state practice” prevents federal review. What this means in practice is that the state rule of practice must have been in place, and enforced, “by the time as of which it is to be applied.” Ford v. Georgia, 498 U.S. 411, 424, 111 S.Ct. 850, 857-58, 112 L.Ed.2d 935 (1991). See also Trevino v. Texas, 503 U.S. 562, 566-68, 112 S.Ct. 1547, 1549-50, 118 L.Ed.2d 193 (1992); Del Vecchio v. Illinois Department of Corrections, 31 F.3d 1363, 1380-81 (7th Cir.1994) (en banc). Ford holds that a rule of state procedure adopted by the state’s highest court two years after the defendant’s trial was not an adequate ground. The Supreme Court did not require Ford to show that he relied on the old rules; all that mattered was what the announced rules were on the date of the act or omission said to work the forfeiture. The inquiry is objective, which greatly simplifies the task of application. Escalonar-Naranjo overrules a case that defined the state’s law of forfeiture at the time Liegakos took his direct appeal. Whether or not Liegakos relied on Bergen-thal in 1987, he could have presented his constitutional claims to the state courts between 1988 and 1993. Under the holdings of Ford and its cousins (including our opinion in Del Vecchio), the doctrine of Escalona-Nar-anjo is not an “adequate” state ground for appeals briefed before its announcement.

What is more, for all but one of Liegakos’s arguments (the one he presented on appeal in 1987), the 1996 amendment to § 2254(d) does not apply: it affects only a “claim that was adjudicated on the merits in State court proceedings”. Although the state’s trial court rejected on the merits the arguments Liegakos presented for the first time on collateral attack, the court of appeals relied entirely on Escalona-Naranjo, and the disposition of the last state court to issue an opinion determines whether the state has invoked a ground of forfeiture. Ylst v. Nunnemaker, 501 U.S. 797, 111 S.Ct. 2590, 115 L.Ed.2d 706 (1991); Prihoda v. McCaughtry, 910 F.2d 1379 (7th Cir.1990). Wisconsin may be surprised that more restrictive standards for collateral attacks in state court (Escalona-Naranjo) and more restrictive standards in federal court (Title I of the Antiterrorism and Effective Death Penalty Act of 1996, discussed in Felker and Lindh) cancel each other out: we must review Liegakos’s claim as if nothing had changed since 1987. But there is no irony. Both Escalona-Naranjo and the new federal standards put a premium on following the rules of procedure. States must do likewise, if they seek to reduce the federal role. An obligation to turn square corners applies across the board.

Liegakos did not testify at his trial. He now contends that he did not recognize the benefits of testifying, and therefore did not intelligently surrender his right to do so. According to Liegakos, his lawyer stressed the disadvantages of testifying but did not discuss with him the corresponding advantages. The judge did not inquire on the record whether Liegakos understood these things and did not elicit a formal waiver of the right to testify. Liegakos insists that because the right to testify is so important, see Rock v. Arkansas, 483 U.S. 44, 52, 107 S.Ct. 2704, 2709-10, 97 L.Ed.2d 37 (1987), a waiver on the record, after advice from the judge, must be required, just as it is for a waiver of the right to counsel. See Johnson v. Zerbst, 304 U.S. 458, 464, 58 S.Ct. 1019, 1023, 82 L.Ed. 1461 (1938). We have held otherwise, United States v. Brimberry, 961 F.2d 1286 (7th Cir.1992), and for a compelling reason: whether to testify is a fundamental element of a strategy that the defendant may want (indeed, may be entitled) to keep in confidence. A judge could probe whether the accused is making an intelligent choice only by exploring all of the options open to the defense—what evidence is available, what witnesses reliable, what could be brought out on cross-examination if the defendant testifies, and so on. Defendant and his lawyer should explore these issues and options carefully, but as a rule the judge need not, and should not, inquire into the choice of defense strategy.

Even if we were inclined to doubt the wisdom of Brimberry, which we are not, Liegakos could not benefit. Only rules established before the conclusion of the direct appeal may be applied on collateral attack. Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989). Liegakos contends that by relying on Escalona-Naranjo the state waived its right to invoke Teague. How could this be? Wisconsin has raised all available procedural objections to Liegakos’s contentions. A state with two grounds (failure to raise this issue on appeal in 1987, and the fact that the law as of 1987 did not require a waiver on the record) need not elect between them. They are not inconsistent, see Lockhart v. Fretwell, 506 U.S. 364, 372-73, 113 S.Ct. 838, 844-45, 122 L.Ed.2d 180 (1993), and anyway a litigant may take inconsistent positions.

On at least five occasions Liegakos and his lawyer discussed whether he would testify, and the lawyer told him that his was the final decision. Although Liegakos asserts that his lawyer did not tell him the potential benefits of testifying, the state judge found otherwise after an evidentiary hearing. This finding cannot be disturbed under the standards of § 2254(e). Counsel’s inability to recall, seven years after the events, the exact words he used when discussing this subject with Liegakos is hardly such a suspicious shortcoming that the district court should have disregarded the state judge’s finding and held his own evidentiary hearing. Memories would be fainter in 1997 than they were in 1994. Liegakos could have made this claim in 1987, and because his is the burden of persuasion, the consequences of the long delay fall on him.

Recasting this argument as an attack on the adequacy of counsel adds nothing. Nor are we impressed by the contention that counsel had a duty to mislead the jury. For this is another strand of Liega-kos’s attack on counsel’s performance. Lie-gakos concedes stabbing and killing Richard Lundgren during a fight. At trial Liegakos attempted to reduce his culpability to manslaughter by arguing “imperfect self-defense”—that he believed his conduct justified by' Lundgren’s aggression, even though the fatal attack was not a reasonable response to Lundgren’s threat. The effort to reduce the degree of responsibility in this fashion was undercut by the testimony of Sonya Roggan-buck that Liegakos told her after the fight: “This guy was hassling me, and I had to stab him.... This guy gave me this big shiner, so I had to stab him twice more.” Counsel did not challenge Rogganbuck’s testimony at trial. Liegakos now argues that, had counsel conducted a good investigation, he would have discovered that Liegakos did not make this statement to Rogganbuck, and therefore would have been able to discredit her testimony. Counsel concedes that he did not investigate this subject and that he elected not to cross-examine Rogganbuck about it. He also gave a reason: that Liegakos admitted that he had said those words to Roggan-buck. At the evidentiary hearing in state court, Liegakos denied telling counsel this; the lawyer testified that Liegakos had done so; the judge believed the lawyer. Again this finding is conclusive under § 2254(e).

A lawyer’s failure to discredit truthful testimony does not make the resulting verdict a violation of the Constitution. Some advocates can make black seem white, or get an honest witness to appear deceitful. These skills fetch high prices in the market for legal services, but the Constitution does not entitle a defendant to a trial at which the truth will be undermined. Perhaps counsel could have shown that Rogganbuck has a bad memory, or that someone in the room with Liegakos when he spoke on the phone to Rogganbuck does not remember the words she (and Liegakos) both recall. Facts that could have been brought out on eross-exami-nation, such as Rogganbuck’s friendship with Lundgren, might have led the jury to doubt her testimony. Yet whether or not counsel’s decision flunked the “performance” test of Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), it did not cause “prejudice”—for prejudice means an unreliable verdict. See Nix v. Whiteside, 475 U.S. 157, 175-76, 106 S.Ct. 988, 998-99, 89 L.Ed.2d 123 (1986); id. at 184-85, 106 S.Ct. at 1003-04 (Blackmun, J., concurring); Lockhart v. Fretwell, 506 U.S. at 369-70, 113 S.Ct. at 842-43. After Nix and Lockhart a decision that makes the verdict more reliable cannot establish prejudice. In Nix the lawyer refused to present testimony he believed to be perjured; in Lockhart counsel failed to take advantage of a case that was later overruled. See also Holman v. Page, 95 F.3d 481, 491-92 (7th Cir.1996) (failure to have probative evidence excluded cannot establish prejudice); John C. Jeffries, Jr. & William J. Stuntz, Ineffective Assistance and Procedural Default in Federal Habeas Corpus, 57 U.Chi.L.Rev. 679, 691-93, 712-13 (1990). Confinement following a trial at which truth-speakers are allowed to testify unchallenged is hardly the sort of problem that a federal court should root out by issuing a writ of habeas corpus.

One final argument rests on the compulsory process clause of the sixth amendment. Liegakos called Boyd Smith as a defense witness at trial. Smith was an eyewitness to the stabbing; at one time the prosecutor believed that Smith was himself culpable and had charged Smith with being a party to the crime of first degree murder. Although the prosecutor dismissed that charge, he informed Smith that he would reinstate it if more evidence came to light. Understandably concerned, Smith invoked his privilege against self-incrimination. So far, no problem. Many cases hold.that the sixth amendment does not entitle a defendant to testimony that the witness has a fifth amendment privilege not to give. E.g., Gleason v. Welborn, 42 F.3d 1107, 1109 (7th Cir.1994). The novelty here is that the prosecutor was willing to offer Smith assurances that his testimony would not be used against him. Such assurances replace the privilege, and the witness then may be compelled to testify. Kastigar v. United States, 406 U.S. 441, 92 S.Ct. 1653, 32 L.Ed.2d 212 (1972). Nonetheless, the trial judge refused to order Smith to testify, observing that under Wis.Stat. 972.08(1) a judge could order a witness to testify only in exchange for transactional immunity, which the prosecutor was not offering. (The statute has since been amended to make use immunity sufficient.) Liegakos raised this argument on direct appeal in 1987, and the court of appeals held that the ruling at trial had been correct. On this issue, then, the amended § 2254(d)(1) applies, and to obtain relief Liegakos must show that the decision “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States”.

Liegakos argues that, once the prosecutor offered Smith a form of immunity constitutionally sufficient to lift the privilege, the sixth amendment entitled him to compel Smith to testify. A state may not deny him compulsory process by granting witnesses privileges (such as the right to demand transactional immunity) so difficult to overcome that defendants cannot obtain testimony. This would be clear enough if, for example, Wisconsin had enacted a statute reading: “No witness is required to testify unless the prosecutor promises him clemency for all transgressions he has ever committed.” Such a statute would be equivalent to one reading: “No defendant may call any witness unless the prosecutor approves.” Given the Supremacy Clause of the Constitution, neither statute could stand. Why, Liegakos asks, should § 972.08(1) be treated differently? Doubtless the state is right to say that Wis.Stat. § 972.08(1), as it read in 1987, could not be used to compel Smith’s testimony. That’s what the state’s court of appeals held. But Liegakos did not need a state statute when he had the compulsory process clause of the sixth amendment. The state makes a greater-includes-the-lesser argument. Smith had a bona fide constitutional privilege; the fifth amendment, not state law, entitled Smith to reject any demand for his testimony; because the prosecutor did not have to offer immunity of any kind, the statutory requirement that immunity be of the transactional variety did not deprive Liega-kos of any testimony to which he was otherwise entitled. This has some power, but like most arguments of this stripe is not entirely convincing. The prosecutor was willing to extend use immunity, so the obstacle in the end was the state statute, not the fifth amendment. Although the compulsory process clause does not override all privileges under state law, it is hard to put Wis.Stat. § 972.08(1) on a plane with the attorney-client or priest-penitent privileges.

None of this did Liegakos any good in the district court, which held that he had not exhausted this claim in state court. 928 F.Supp. at 809-10. If there is indeed an exhaustion problem then the district court should have dismissed the entire petition, rather than adjudicating the claims it deemed exhausted. Rose v. Lundy, 455 U.S. 509, 102 S.Ct. 1198, 71 L.Ed.2d 379 (1982). Although the Supreme Court once held that the state may forfeit, through inattention, the benefit of Rose’s complete-exhaustion rule, see Strickland, 466 U.S. at 684, 104 S.Ct. at 2062-63, the new § 2254(b)(3), added by the Antiterrorism and Effective Death Penalty Act, may require a new approach. This subsection provides that “[a] State shall not be deemed to have waived the exhaustion requirement or be estopped from reliance upon the requirement unless the State, through counsel, expressly waives the requirement.” A district court now may deny a claim despite lack of exhaustion, see § 2254(b)(2), but this is not what the district judge did. A remand is unnecessary, however, for two reasons. First, this claim has been exhausted. It was presented, if with incomplete argument, on direct appeal, and no state remedies remain open to Liegakos. Second, the claim cannot succeed on the merits under either Teague or the amended § 2254(d)(1), so we can write finis to this litigation ourselves.

Although Liegakos has a substantial logical argument, it is one unsupported by precedent in the Supreme Court, whose law governs under § 2254(d)(1). See Lindh, 96 F.3d at 868-71, 874-77. The law was not “clearly established” in 1987 (or today) that state transactional-immunity statutes violate the compulsory process clause of the sixth amendment. A defendant whose position depends on anything other than a straightforward application of established rules cannot obtain a writ of habeas corpus. Liegakos wants us to adopt a novel position, which we have been firmly instructed by Congress not to do on collateral attack. So whether or not Liegakos has a sound legal position about the operation of the compulsory process clause, he has not established a right to collateral relief from the judgment under which he was committed to prison.

Affirmed.