dissenting.
The Sixth Amendment right to effective assistance of counsel is an instrumental right designed to ensure a fair trial. By his own admission, Michael Williams received just such a trial, at the end of which he was convicted of first degree murder by a jury of his peers. We have no authority to disturb this outcome. The majority says that counsel’s deficient performance in the plea bargain process denied Mr. Williams a lesser second degree murder conviction, and that this justifies voiding the result of Mr. Williams’s fair trial. But the due process clauses of the Constitution’s Fifth and Fourteenth Amendments do not encompass a right to receive or accept plea offers. As the Supreme Court has repeatedly held, plea bargains are matters of executive discretion, not judicially enforceable entitlement; due process guarantees a fair trial, not a good bargain.
Without challenging any of this, the majority nonetheless recognizes a constitutional right to accept a plea offer grounded in the Sixth Amendment’s guarantee of effective assistance of counsel, creating indirectly what the Supreme Court’s precedents preclude it from recognizing directly as a matter of due process under the Fifth and Fourteenth Amendments. The practical upshot? So long as a defendant can claim his lawyer mishandled a plea offer, he can take his chances at a fair trial and, if dissatisfied with the result, still demand and receive the benefit of the forgone plea. The majority’s holding has already been rejected by a number of other courts and been the subject of a grant of certiorari (later dismissed when the parties mooted *1075the question). See Hoffman v. Arave, 455 F.3d 926, 942-43 (9th Cir.2006), cert. granted, — U.S. -, 128 S.Ct. 532, 532-33, 169 L.Ed.2d 371 (2007), vacated as moot, 552 U.S. 117, 128 S.Ct. 749, 169 L.Ed.2d 580 (2008). Respectfully, I dissent.
I
The OCCA held that counsel’s failure to encourage or allow Mr. Williams to accept the plea bargain violated his Sixth Amendment right to effective representation. But, by way of remedy, the OCCA afforded Mr. Williams only a reduced sentence— life with, rather than -without, the possibility of parole. Mr. Williams contends this remedy failed to redress fully the constitutional violation. The Sixth Amendment, he submits, requires us to return his case to the Oklahoma courts with directions either to release him or afford him the ten-year sentence the prosecution originally offered. We granted a certificate of appealability to address this question.
Of course, the adequacy of the OCCA’s chosen remedy under the Sixth Amendment cannot be evaluated without first determining the scope, if any, of the constitutional violation. Remedies, after all, exist to vindicate rights. If no Sixth Amendment violation took place, there need be, can be, no further remedy: “Absent [some prejudicial] impact on the criminal proceeding ... there is no basis for imposing a remedy.” United States v. Morrison, 449 U.S. 361, 365, 101 S.Ct. 665, 66 L.Ed.2d 564 (1981). That is, if no Sixth Amendment violation took place, we must reject Mr. Williams’s demand for greater relief than the OCCA provided because, as it happens, even that much wasn’t constitutionally compelled. We would, thus, deny Mr. Williams’s request for further relief, if for different grounds than those the OCCA offered.1 At the same time, if we do find a constitutional violation has occurred, ascertaining the nature and scope of that violation remains a critical and unavoidable task because any remedy must be “tailored to the injury suffered from the constitutional violation and should not unnecessarily infringe on competing interests.” Id. at 364,101 S.Ct. 665.
So, first things first, what exactly is the right allegedly violated here? Mr. Williams argues that he was entitled by the Sixth Amendment to effective representation by counsel during his plea negotiations with the prosecution; that his lawyer provided grossly deficient performance when he threatened to withdraw unless Mr. Williams rejected the proffered plea; and that this deficient performance resulted in prejudice to Mr. Williams because he was unable to accept a plea bargain offered him by the prosecution. I accept, at least for argument’s sake, all but the last link in this chain.
The Sixth Amendment guarantees that “[i]n all criminal prosecutions, the accused shall enjoy the right ... to have the Assistance of Counsel for his defence.” To be meaningful, the right to counsel necessarily embraces a right to effective counsel. McMann v. Richardson, 397 U.S. 759, 771 n. 14, 90 S.Ct. 1441, 25 L.Ed.2d 763 (1970). In Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), the Supreme Court established the familiar two-prong test courts use to evaluate whether a lawyer’s assistance was constitutionally ineffective. First, the defendant must show that his or her lawyer’s per*1076formance was “deficient.” To establish deficiency, the defendant must overcome “a strong presumption that counsel’s conduct falls within the wide range of reasonable professional assistance,” and show that it was objectively unreasonable. Id. at 687-89, 104 S.Ct. 2052. Second, the defendant must show that the lawyer’s deficiency caused prejudice to his or her defense. Id. at 687, 104 S.Ct. 2052. Prejudice is normally established by showing that “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Id. at 694, 104 S.Ct. 2052. “A reasonable probability is a probability sufficient to undermine confidence in the outcome.” Id.
Applying these principles here, I do not doubt that Mr. Williams’s right to counsel attached by the time of his plea negotiations, which occurred after judicial proceedings commenced. See Brewer v. Williams, 430 U.S. 387, 398-401, 97 S.Ct. 1232, 51 L.Ed.2d 424 (1977); Gideon v. Wainwright, 372 U.S. 335, 83 S.Ct. 792, 9 L.Ed.2d 799 (1963). Because Mr. Williams’s right to counsel embraces a right to effective counsel, I also accept that he had a right to effective representation in his plea negotiations with the government. See Brady v. United States, 397 U.S. 742, 748 n. 6, 90 S.Ct. 1463, 25 L.Ed.2d 747 (1970) (“[A]n intelligent assessment of the relative advantages of pleading guilty is frequently impossible without the assistance of an attorney .... ”); see also Bordenkircher v. Hayes, 434 U.S. 357, 363, 98 S.Ct. 663, 54 L.Ed.2d 604 (1978) (advice of competent counsel is a procedural safeguard enabling voluntary and intelligent plea bargaining).
Nor do I question the OCCA’s conclusion that Mr. Williams’s counsel performed deficiently in the plea negotiation process. The OCCA held that, by threatening to withdraw on the eve of trial if Mr. Williams accepted the plea offer, counsel usurped Mr. Williams’s right to decide how to plead to the charges against him. So it was that the lawyer, rather than the client, effectively decided whether or not to accept the government’s proffered plea. And this, the OCCA held, represented an affront not merely to professional rules of conduct, but to the rightful allocation of authority between lawyer and client — between the agent who advises and assists in the criminal process and the principal who must endure any sentence that process yields. See Florida v. Nixon, 543 U.S. 175, 187, 125 S.Ct. 551, 160 L.Ed.2d 565 (2004) (“A defendant ... has the ultimate authority to determine whether to plead guilty, waive a jury, testify in his or her own behalf, or take an appeal.”) (quotations omitted); see also Oklahoma Professional Conduct Rule 1.2(a) (“In a criminal case, the lawyer shall abide by the client’s decision, after consultation with the lawyer, as to a plea to be entered.”); ABA Model Rule of Professional Conduct 1.2(a) (same); ABA Standard for Criminal Justice 4-5.2(a) (“Certain decisions relating to the conduct of the case are ultimately for the accused .... including] (i) what pleas to enter; (ii) whether to accept a plea agreement; (iii) whether to waive a jury trial”).
At the same time, though improper, counsel’s actions were apparently undertaken in good faith. Counsel refused to aid Mr. Williams in accepting the plea because he was absolutely convinced by his client’s initial protestations of innocence. He saw no reason for an innocent man to plead guilty to the charges against him and spend ten years in prison. He also seemed to fear that securing the plea bargain might involve him and Mr. Williams in perjury, because Mr. Williams would *1077have to admit guilt to a crime he (counsel thought) didn’t commit.2
Ultimately, though, counsel’s good faith is beside the point. If the lawyer knew his client were innocent, or at least had reasonable grounds to believe it, his concerns about participating in a potential perjury might have some currency. See ABA Rules 1.2(d); 3.3(a)(3); 1.16(b)(2). But here, particularly in light of the substantial evidence of Mr. Williams’s guilt, counsel had no such grounds. I cannot disagree with the OCCA that
it would be foolish to deny the fact that criminal defendants often lie to their own attorneys concerning their criminal culpability____ Unless defense counsel personally witnessed the crime or has hard proof as to what precisely happened there, he cannot really take the position that his client is committing perjury by pleading guilty.
OCCA Op. at 23. This is not to say that I think Mr. Williams initially deceived his lawyer. But I do agree with the OCCA that a lawyer cannot reasonably discount that possibility without real evidence to the contrary. Mr. Williams’s lawyer had no such evidence. In that light, I see no reason to disturb the OCCA’s conclusion that counsel’s refusal to help secure the plea and threat to quit on the eve of trial amounted to deficient performance.
II
Prejudice is the critical question in this case. Mr. Williams argues that, once the State made a plea offer, he had a right, emanating from the Sixth Amendment’s right to effective representation, to be appropriately counseled about the consequences of accepting or rejecting the plea. Counsel’s failure to advise him properly, Mr. Williams submits, amounts to prejudice within the meaning of the Sixth Amendment. At least one court, People v. Curry, 178 Ill.2d 509, 227 Ill.Dec. 395, 687 N.E.2d 877 (1997), has adopted this line of reasoning. In Curry, the Supreme Court of Illinois held that the Sixth Amendment contains a right “to be reasonably informed as to the direct consequences of accepting or rejecting that offer,” id. at 888, and that the loss of this right qualifies as prejudice within the meaning of Strickland. In concluding that Mr. Williams had suffered a Sixth Amendment violation, the OCCA seemed to take much the same view.
This logic, however, collapses deficient performance and prejudice. Under Curry ’s rationale and the OCCA’s, a violation of the Sixth Amendment is complete the moment a lawyer fails to advise his client adequately about the consequences of a plea. That is, the lawyer’s bad advice (deficient performance) deprives the client of a right to good advice (prejudice). The Supreme Court has repeatedly warned against this fallacy, reminding us that some independent showing of prejudice must be made. “The requirement that a defendant show prejudice in effective representation cases arises from the very nature of the specific element of the right to counsel at issue there — effective (not mistake-free) representation. Counsel cannot be ‘ineffective’ unless his mistakes have harmed the defense.... [A] violation of the Sixth Amendment right to effective representation is not ‘complete’ until the defendant is prejudiced.” United States v. Gonzalez-Lopez, 548 U.S. 140, 147, 126 S.Ct. 2557, 165 L.Ed.2d 409 (2006); see *1078also Strickland, 466 U.S. at 691, 104 S.Ct. 2052 (“An error by counsel, even if professionally unreasonable, does not warrant setting aside the judgment of a criminal proceeding if the error had no effect on the judgment.”). Put differently, bad advice isn’t enough. Some harm to the defense must follow. Calling a lawyer’s incompetence both deficient performance and prejudice misconstrues the Supreme Court’s test.
Given that the OCCA’s prejudice analysis is not defensible, we might ask whether there is some other basis for finding prejudice in Mr. Williams’s case that warrants remedy. The usual ineffective assistance claim in the plea bargaining context involves a defendant who has pled guilty as a result of poor legal advice. In such cases, we readily find prejudice because the waiver of the right to trial must be a “voluntary ... knowing, intelligent act[ ] done with sufficient awareness of the relevant circumstances and likely consequences.” Brady, 397 U.S. at 748, 90 S.Ct. 1463. If counsel provided unsound advice, and that advice resulted in a guilty plea, the defendant’s waiver of the right to trial cannot be said to have been made freely, knowingly, and intelligently. Hill v. Lockhart, 474 U.S. 52, 56, 106 S.Ct. 366, 88 L.Ed.2d 203 (1985). Prejudice accrues because “there is a reasonable probability that, but for counsel’s errors, he would not have pleaded guilty and would have insisted on going to trial.” Id. at 59,106 S.Ct. 366. That is, the defendant is prejudiced by the lost right to test the government’s evidence at trial.
This case is different. Mr. Williams pled not guilty to the charge against him. And a not-guilty plea is a waiver of nothing; it is an invocation of the constitutional right to a trial, and it is effective whether or not it is made knowingly and voluntarily. So Mr. Williams cannot and does not challenge the validity of his not-guilty plea; he must identify some other source of prejudice. What might that be?
Mr. Williams argues that prejudice can be shown from the fact that he lost a good chance at a ten-year sentence, and instead received a life sentence. Though at first blush this sounds like a complaint that his sentence is the result of ineffective assistance, that isn’t quite right. It is undisputed that the ten-year sentence Mr. Williams desires is unavailable for a person convicted of first degree murder in Oklahoma; life with the possibility of parole is the statutory minimum for that offense. Thus, it would have been unlawful for the Oklahoma court to convict Mr. Williams of first degree murder, as it did, and then impose a sentence less than life with the possibility of parole. Because Mr. Williams cannot legally obtain a ten-year sentence for first degree murder, he cannot complain that his attorney was ineffective for failing to secure such a result. A defendant, of course, cannot claim prejudice from the loss of something the law forbids. See, e.g., Nix v. Whiteside, 475 U.S. 157, 106 S.Ct. 988, 89 L.Ed.2d 123 (1986) (no prejudice from failure to present perjured testimony).
Mr. Williams’s claim of prejudice must begin with his conviction for first degree murder. His argument is and has to be that his lawyer’s ineffectiveness rendered his first degree murder conviction invalid; that it should be vacated; and that he should be permitted to plead to a second degree murder charge instead — only that way would he become eligible for the ten-year sentence he seeks. But here, too, there is a twist. In the ordinary Strickland challenge to a conviction, the defendant argues that, because of counsel’s incompetent acts or omissions, he did not get a fair or reliable trial; therefore, the resulting conviction is unlawful. Mr. *1079Williams’s claim is just the opposite. He claims that because of his lawyer’s incompetence, he did get a fair and reliable trial leading to his first degree murder conviction. So the question Mr. Williams poses now becomes clear: Can a defendant be prejudiced, in the Sixth Amendment sense, by the loss of a plea bargain if he is later convicted only after an entirely fair trial?
I think not. The Sixth Amendment right to counsel, along with the right to counsel of one’s choice, form the “root meaning of the constitutional guarantee” of the Sixth Amendment. Gonzalez-Lopez, 548 U.S. at 147-48, 126 S.Ct. 2557. When a defendant is deprived of these rights, no further showing of prejudice is necessary — the deprivation of counsel altogether, or the deprivation of counsel of one’s choice, is the loss of a constitutional right. See id. at 148, 126 S.Ct. 2557. By contrast, the Supreme Court has told us, the Sixth Amendment “right to the effective assistance of counsel is recognized not for its own sake, but because of the effect it has on the ability of the accused to receive a fair trial.” United States v. Cronic, 466 U.S. 648, 658, 104 S.Ct. 2039, 80 L.Ed.2d 657 (1984). The right to effective representation originated in the Due Process Clause, which prohibits the government from depriving any person of liberty by fundamentally unfair or unreliable procedures. Gonzalez-Lopez, 548 U.S. at 147, 126 S.Ct. 2557; McMann, 397 U.S. at 771 n. 14, 90 S.Ct. 1441; Powell v. Alabama, 287 U.S. 45, 70-71, 53 S.Ct. 55, 77 L.Ed. 158 (1932). Because the right to effective assistance exists to serve the underlying purpose of ensuring a fair trial, a violation of the right requires some showing that counsel’s deficiency impacted the fair trial right. Gonzalez-Lopez, 548 U.S. at 147, 126 S.Ct. 2557. As the Court has put it, “[hjaving derived the right to effective representation from the purpose of ensuring a fair trial, we have, logically enough, also derived the limits of that right from that same purpose.” Id. “Absent some effect of challenged conduct on the reliability of the trial process, the Sixth Amendment guarantee is generally not implicated.” Cronic, 466 U.S. at 658, 104 S.Ct. 2039.
A fair trial process, in turn, is one in which the defendant is afforded the right “to require the prosecution’s case to survive the crucible of meaningful adversarial testing.” Id. at 656, 104 S.Ct. 2039. "When the outcome of the trial may have been altered by a defense lawyer’s failure to subject the government’s case to appropriate adversarial testing, then the resulting conviction was not the result of a fair trial. See id. at 656-57, 104 S.Ct. 2039; see also Strickland, 466 U.S. at 694-96, 104 S.Ct. 2052. But “[w]hen a true adversarial criminal trial has been conducted— even if defense counsel may have made demonstrable errors — the kind of testing envisioned by the Sixth Amendment has occurred,” and no constitutionally cognizable prejudice can be said to exist. Cronic, 466 U.S. at 656, 104 S.Ct. 2039. Accordingly, “the benchmark for judging any claim of ineffectiveness must be whether counsel’s conduct so undermined the proper functioning of the adversarial process that the trial cannot be relied on as having produced a just result.” Strickland, 466 U.S. at 686, 104 S.Ct. 2052. When, as here, “a defendant challenges a conviction, the question is whether there is a reasonable probability that, absent [counsel’s] errors, the factfinder would have had a reasonable doubt respecting guilt.” Id. at 695,104 S.Ct. 2052.
Mr. Williams cannot show prejudice under this standard. He does not say that his lawyer should have called this witness, won the exclusion of that piece of evidence, or pressed this legal defense. Put in Strickland’s terms, he does not suggest *1080that anything his lawyer could have done would have induced the jury to have “a reasonable doubt respecting [his] guilt” for first degree murder. Id. Rather, he says the jury would never have been empaneled in the first place.
No doubt that would have been better for Mr. Williams, but it has nothing to do with the question posed by Strickland. Prejudice is not established simply because, but for some accident or misfortune, the fair trial leading to a conviction would have been somehow averted. It must be that some doubt is raised about whether the conviction was obtained only because the prosecution’s case was not competently contested. A conviction for first degree murder at trial is no less the result of a meaningful adversary contest just because the defendant’s lawyer failed to allow his client to be convicted of some other crime by some other procedure.
Put differently, rather than have us review the fairness of the trial that did deprive him of liberty, as Strickland requires, Mr. Williams’s claim would have us examine the plea negotiation that did not. But that is not the relevant inquiry. In fact, Mr. Williams’s attempt to shift our focus from the fairness of the trial that took place to an abandoned plea deal is foreclosed by the Supreme Court’s decisions in Mabry v. Johnson, 467 U.S. 504, 104 S.Ct. 2543, 81 L.Ed.2d 437 (1984), abrogated on other grounds by Puckett v. United States, — U.S. -, 129 S.Ct. 1423, 173 L.Ed.2d 266 (2009), and Weatherford v. Bursey, 429 U.S. 545, 97 S.Ct. 837, 51 L.Ed.2d 30 (1977).
In Mabry, the government offered a plea on a Friday and the defendant pondered it over the weekend. When defense counsel called on Monday to accept the deal, the prosecutor said the offer was a mistake and withdrew it. In its place, the government offered a new bargain with a harsher sentence, which the defendant accepted. The defendant later sought the benefit of the original bargain in court, arguing that due process principles precluded the government from reneging on its deal. The Sixth Circuit agreed, holding that “fairness” prevented the government from withdrawing the original plea offer after it had been accepted. Mabry, 467 U.S. at 506, 104 S.Ct. 2543. But the Supreme Court reversed, stressing that it was not the original, abandoned plea bargain that deprived the defendant of his liberty; it was the guilty plea he actually entered. And the Court explained that only the plea that deprived the defendant of his liberty mattered:
A plea bargain standing alone is without constitutional significance; in itself it is a mere executory agreement which, until embodied in the judgment of a court, does not deprive an accused of liberty or any other constitutionally protected interest. It is the ensuing guilty plea that implicates the Constitution. Only after respondent pleaded guilty was he convicted, and it is that conviction which gave rise to the deprivation of respondent’s liberty at issue here.
Id. at 507-08, 104 S.Ct. 2543. The sole question implicating the Constitution, the Court held, was whether the eventual guilty plea that did deprive the defendant of his liberty was voluntary and intelligent. Id. at 508-09, 104 S.Ct. 2543. Even if the prosecutor was “negligent or otherwise culpable” in rejecting the original plea, the Constitution’s concern is only “with the manner in which persons are deprived of their liberty.” Id. at 511, 104 S.Ct. 2543.
The story of Weatherford is similar. There, the defendant complained that government misconduct (an undercover government agent, pretending to be a co-defendant, sat in on discussions with defense counsel) prevented him from having *1081even the opportunity to bargain for a plea. This court agreed that due process was offended, but the Supreme Court again reversed. While the prosecution likely engaged in misconduct, and while its misconduct may have denied the defendant the opportunity to engage in plea bargaining, the defendant “was not denied a fair trial.” Weatherford, 429 U.S. at 560, 97 S.Ct. 837. A lost opportunity to pursue a negotiated plea was of no moment.
The majority says our case is distinguishable because the prosecutor here neither declined to extend an offer nor revoked an offer already extended — that is, a different sort of problem kept Mr. Williams’s plea deal from being embodied in a court’s judgment. Maj. Op. at 1091. But the question we face isn’t whether the particular accident befalling Mr. Williams is the same particular accident that befell Mr. Mabry and Mr. Weatherford. That one petitioner has brown eyes and another blue usually makes no difference in the eyes of the law. Our job isn’t to ask whether cases are factually identical but whether, by their terms, the rules of law set forth by the Supreme Court do or do not apply to our case. And the rule of Mabry and Weatherford — that constitutional scrutiny is applied only to the procedure that deprives the defendant of his liberty, and that executory plea bargains do not — is no less applicable to Mr. Williams than it was to the defendants in those cases. The historical happenstance of why Mr. Williams’s plea remained executory is neither here nor there.
The only other conceivable difference one might imagine between our case and Mabry and Weatherford is that, in those cases, the defendant complained of prosecutorial misconduct under the due process clause, while the defendant in our case complains of defense counsel misconduct under the Sixth Amendment. But this distinction makes no more difference than the court’s. The Sixth Amendment right to effective representation is an instrumental right whose purpose is to serve the very same due process ends at issue in Mabry and Weatherford — to ensure that the defendant’s liberty is taken, if at all, only after a fundamentally fair process. If the prosecution does not trench on that objective by irrationally withdrawing a proffered plea when the defendant is later fairly convicted, defense counsel cannot do so by irrationally insisting that a client reject such a plea when the defendant is later convicted at a fair trial. If Mabry and Weatherford remain the law and the Constitution remains concerned only with “the manner in which persons are deprived of their liberty,” not with how they are not, then we must conclude that no prejudice to a constitutional interest took place here.
Ill
One might ask why Mabry and Weather-ford hold as they do. Why do we care about trials that deprive defendants of their liberty but not botched plea deals that don’t? The most obvious answer of course is that the latter don’t result in incarceration, and the Constitution cares only about processes that do. But there is even more to it than that. As both Mabry and Weatherford intimate, a lost plea deal implicates neither Strickland’s concern with ensuring a reliable result nor its concern with ensuring fair treatment. A fair trial’s outcome is as reliable an outcome as we can hope to achieve. And because the plea bargain is a matter of prosecutorial grace, not a matter of legal entitlement, a defendant who loses the chance for a deal cannot be said to have been treated unfairly.
A
In the terms of Strickland’s prejudice prong, the loss of a plea bargain does not *1082“undermine confidence” in the outcome of a subsequent fair trial. The American Constitution, our Bill of Rights, and our common law tradition place faith in the trial as the best means of protecting a defendant’s rights, testing the government’s case, and ensuring a reliable result. It is a foundational principle of our system of justice that a fair trial, imperfect though it and all human affairs surely are, offers the greatest assurance we have yet devised of a reliable resolution of a criminal charge. The guarantee of a fair trial is the bulwark of our national liberty.
This isn’t to say plea bargaining has no place. But unlike a trial, plea bargaining is not designed to subject the prosecution’s case to testing or to ensure that a defendant goes to jail only if no reasonable doubt remains about his or her guilt. In fact, plea bargains guarantee that some persons who would have been acquitted never receive a trial at all and instead serve time in prison. In part for this reason, the common law was hostile even to the uncompensated guilty plea, fearing that the innocent (and perhaps especially the innocent) would feel pressured to convict themselves either out of fear or hope of mercy.3 The Supreme Court, in more recent years, has acknowledged the value of plea bargains in efficiently disposing of criminal cases. Santobello v. New York, 404 U.S. 257, 261, 92 S.Ct. 495, 30 L.Ed.2d 427 (1971). According to the Court, plea bargaining
leads to prompt and largely final disposition of most criminal cases; it avoids much of the corrosive impact of enforced idleness during pre-trial confinement for those who are denied release pending trial; it protects the public from those accused persons who are prone to continue criminal conduct even while on pretrial release; and, by shortening the time between charge and disposition, it enhances whatever may be the rehabilitative prospects of the guilty when they are ultimately imprisoned.
Id.; see also George Fisher, Plea Bargaining’s Triumph: A History of Plea Bargaining in America (2003). But notably missing from the Court’s recitation of the plea bargain’s virtues is any suggestion that, compared to the trial as evolved through history and guaranteed by law, plea bargaining enhances the reliability of the outcome. When it comes to that — when it comes to the confidence we repose in the outcome of a criminal proceeding — the trial remains our gold standard.4
The fact that some lost opportunities in the criminal process do not undermine con*1083fidence in the ultimate result (and therefore do not constitute prejudice), so long as the defendant receives a fair trial, is nothing novel. Consider, for example, the lawyer who is ineffective at a trial that (fortuitously) results in a hung jury. If, on retrial, the lawyer performs perfectly and the defendant is convicted, we would not direct our gaze to the prior ineffectual proceeding and provide a remedy for counsel’s deficiency. The first trial did not deprive the defendant of liberty, and the trial that did was fair. Accordingly, our confidence in the outcome is unshaken by counsel’s prior deficient performance. That would end the matter in such a case, much as it did in Mabry and Weatherford, and much as it should here.
Consider, too, the grand jury setting. Like a pre-trial plea negotiation, proceedings before a grand jury usually determine whether there will be a trial at all. In United States v. Mechanik, 475 U.S. 66, 106 S.Ct. 938, 89 L.Ed.2d 50 (1986), the government violated Fed. R.Crim. 6(d) by permitting two witnesses to testify simultaneously before the grand jury. This was misconduct, and the Court assumed it might have warranted dismissal of the indictment had it been brought to the district court’s attention before trial. Id. at 69-70, 106 S.Ct. 938. But instead the trial proceeded to conviction, and the Supreme Court held that this fact rendered “any error in the grand jury proceeding ... harmless beyond a reasonable doubt.” Id. at 70, 106 S.Ct. 938. That was so because “the petit jury’s subsequent guilty verdict means not only that there was probable cause to believe that the defendants were guilty as charged, but also that they are in fact guilty as charged beyond a reasonable doubt.” Id. Thus, even when a criminal trial might have been averted through the assertion of an entitlement to dismiss the indictment outright, the fact that the defendant subsequently received a fair trial was enough to preclude reversal. It seems more than unlikely that the Constitution could be offended by a fair trial that occurs because of the loss of a plea bargain to which the defendant had no entitlement, but not by a fair trial that occurs only after counsel failed to pursue defendant’s entitlement to a dismissal of the indictment.
B
Just as a trial without plea bargaining is no less reliable, neither is it any less fair. A fair criminal process is one in which the defendant was not deprived of “any substantive or procedural right to which the law entitles him.” Terry Williams v. Taylor, 529 U.S. 362, 393 n. 17, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (quoting Lockhart v. Fretwell, 506 U.S. 364, 372, 113 S.Ct. 838, 122 L.Ed.2d 180 (1993)). Fairness means that the defendant is treated according to law, with the full panoply of rights and entitlements it affords. The difficulty for Mr. Williams is that, under the separation of powers embodied in our Constitution, plea bargaining is a matter of prosecutorial grace, not right or entitlement.
It has been long settled that, in the absence of deliberate discrimination on the basis of some protected equal protection status, such as race or religion, “the Executive Branch has exclusive authority and absolute discretion” to select its charge. Greenlaw v. United States, — U.S. -, 128 S.Ct. 2559, 2565, 171 L.Ed.2d 399 (2008) (quotation omitted); see also Bordenkircher, 434 U.S. at 364-65, 98 S.Ct. 663. This means that a court may neither compel a prosecutor to bring a particular charge, see, e.g., Inmates of Attica Correctional Facility v. Rockefeller, 477 F.2d 375 (2d Cir.1973), nor refuse to dismiss a charge when the prosecutor requests it, *1084Rinaldi v. United, States, 434 U.S. 22, 98 S.Ct. 81, 54 L.Ed.2d 207 (1977). Likewise, a defendant cannot compel the prosecutor to bargain at all if he does not want to, Weatherford, 429 U.S. at 561, 97 S.Ct. 837, or even prevent the prosecutor from reneging on an agreement after it is accepted. Mabry, 467 U.S. at 510, 104 S.Ct. 2543. Neither may a defendant force the trial judge to adopt an accepted bargain if the prosecutor does choose to abide the agreement. Santobello, 404 U.S. at 262, 92 S.Ct. 495. In short, plea bargains are products of executive prerogative, not legal entitlement.5
Absent a loss of a legal entitlement, the defendant can claim no prejudice. So, for example, in Fretwell the Supreme Court held that, even though a defendant’s lawyer failed to raise a point of law at sentencing that would have spared his client the death penalty, there could be no prejudice because the legal proposition was overruled by the time the case reached habeas review. 506 U.S. at 367-68, 113 S.Ct. 838. Despite the fact that the defendant could have secured a life sentence at the time of his hearing by raising a then-effective legal argument, and despite the greatest possible difference in outcome produced by counsel’s omission — a sentence of death rather than life — the Supreme Court ruled that giving the defendant the benefit of a legal argument later held erroneous would have improperly “grant[ed][him] a windfall to which the law does not entitle him.” Id. at 370,113 S.Ct. 838. There was, in the Court’s view, nothing unfair about the fact that the defendant was sentenced to death under the proper legal standard, even if his life could have been saved by raising an erroneous legal argument available at the time of his trial. The absence of legal error cannot undermine confidence in a trial. See id. at 369, 113 S.Ct. 838 (“[A]n analysis focusing solely on mere outcome determination, without attention to whether the result of the proceeding was fundamentally unfair or unreliable, is defective.”). As Justice Blackmun explained, when the defendant has been “deprived of neither a fair trial nor any of the specific constitutional rights designed to guarantee a fair trial, he has suffered no prejudice.” Nix, 475 U.S. at 186-87, 106 S.Ct. 988 (Blackmun, J., concurring in the judgment).
No different conclusion is fairly possible here. Mr. Williams cannot claim any substantive or procedural right to accept or enforce the prosecutor’s plea offer. To the contrary, because the plea was never embodied in the court’s judgment, it remained a matter of pure executive discretion and could be withdrawn at whim. Without any entitlement to the plea, Mr. Williams cannot complain that the loss of it was unfair. As the Seventh Circuit noted in the face of a claim similar to Mr. Williams’s, a defendant has no “entitlement to a discount of any kind, no right to be treated more leniently than other criminals unlucky enough to be without information for which the prosecutor is willing to pay.” United States v. Springs, 988 F.2d 746, 749 (7th Cir.1993).6
*1085Mr. Williams would surely respond that, while the law did not entitle him to accept or enforce the plea bargain, neither did the law forbid it; he at least had the opportunity to pursue the prosecutor’s offer. The problem with this line of argument is that constitutional prejudice arises only from the loss of a legal entitlement, not from the loss of a merely lawful opportunity. That this is so is illustrated not only by the Fretwell line of cases, but by Strickland itself. In Fretwell, the defendant surely lost the opportunity to take advantage of a favorable legal rule that would have spared him a death sentence; pursuing that opportunity was legally permissible at the time of trial, yet the Court held no prejudice attached from counsel’s failure to do so. In Strickland, the Court emphasized that Sixth Amendment prejudice analysis can take no account of “the possibility of arbitrariness, whimsy, caprice, ‘nullification,’ and the like. A defendant has no entitlement to the luck of a lawless decisionmaker.” 466 U.S. at 695, 104 S.Ct. 2052 (emphasis added). So, for example, the Court noted that nothing prevents defense counsel from appealing to a judge’s known tendency to be lenient in sentencing first-time offenders. But, while a lawyer’s failure to exploit that tendency at trial or sentencing might be relevant to the question of deficient performance, it is “irrelevant to the prejudice inquiry.” Id. This is because a sentencing judge’s “unusual propensities toward harshness or leniency,” must be excluded from consideration. Id. This sort of thing is “lawless” decisionmaking — not in the sense that it is illegal or even unseemly, but in the sense that it is not governed by legal standards, not part of the process that is “due” according to law. See id. A defendant has no “entitlement” to it.
Plea bargaining is an even more discretionary activity than sentencing. Suppose that, rather than a judge, it is the prosecutor who is known to be sympathetic to first-time offenders. A defendant’s lawyer knows that if he goes to see the prosecutor and makes a pitch, it is very likely the prosecutor will drop the charge entirely. If the lawyer fails to do this, he or she will have unreasonably failed to take advantage of a legally permissible opportunity for the client. But that is hardly “prejudice” within the meaning of the Sixth Amendment. For what would the rationale be for granting relief for counsel’s failure to take advantage of the lawless leniency of a prosecutor, in an out-of-court setting, but not that of the judge at the sentencing hearing that leads to the judgment being attacked in habeas corpus? In both scenarios, the lawyer’s incompetence makes a difference to the client. A lawful opportunity was lost. But in neither case could the defendant complain that he or she was deprived of a legal entitlement such that the judgment of conviction was unfair.
The majority mistakes the nature of this analysis in two ways. First, it suggests that I read the Sixth Amendment to protect only “trial rights in connection with guilt or innocence.” Maj. Op. at 1093. *1086But I have asked not only whether counsel’s conduct affected the reliability of Mr. Williams’s criminal proceeding (what the majority calls the “guilt or innocence” question). I have also asked whether, even if Mr. Williams’s criminal proceeding was reliable, counsel’s deficient performance nonetheless rendered it unfair by denying him any substantive or procedural right to which the law entitles him. The difference between these questions can be illuminated by an example. A defendant’s right to exclude evidence under the Fourth Amendment has nothing to do with “guilt or innocence,” which is one reason why the trial court’s failure to vindicate this right is not a basis for habeas corpus. Stone v. Powell, 428 U.S. 465, 486, 96 S.Ct. 3037, 49 L.Ed.2d 1067 (1976). But, because defendants enjoy a legal entitlement to suppress evidence unlawfully seized, a lawyer’s failure to raise a good exclusionary rule objection might still render the result of his criminal proceeding unfair and thus prejudicial in a Sixth Amendment sense. See Owens v. United States, 387 F.3d 607 (7th Cir.2004). Contrary to the majority’s suggestion, it is of no moment for purposes of my analysis whether the particular legal entitlement lost by a defendant exists to ensure a reliable verdict or to deter police misconduct or for any other purpose. The relevant question is whether Mr. Williams received the legal protections to which he was entitled at law. He did because he cannot identify any source of law that entitles him to the plea deal he lost.
Second, the majority claims that I have “superimposed] another requirement on the prejudice inquiry” by asking whether Mr. Williams’s lost plea implicates either the reliability or the fairness of his criminal proceeding. Maj. Op. at 1094. This appears to be an allusion to a species of error certain courts fell prey to in the wake of Fretwell: requiring an additional, quantitative inquiry into how large the difference in outcome produced in any particular case probably was, and denying relief if, in the court’s judgment, the difference wasn’t sufficiently large. So, for example, in Durrive v. United States, 4 F.3d 548, 550-51 (7th Cir.1993), the Seventh Circuit refused to afford a Strickland remedy when counsel’s deficient performance produced a marginal but real effect on the defendant’s sentence. The Supreme Court rejected this approach as error in Glover v. United States, 531 U.S. 198, 203, 121 S.Ct. 696, 148 L.Ed.2d 604 (2001). See also Terry Williams, 529 U.S. at 394-95, 120 S.Ct. 1495. It is not an error repeated here. The point of Fretwell, Nix, Glover, and Terry Williams is that certain acts or omissions by counsel may well affect the outcome, but categorically do not cause constitutionally cognizable prejudice because they have no effect on either the reliability or the fairness of the proceeding. We are thus told to ask a qualitative, not quantitative, question: Did “unreliability or unfairness result” by dint of counsel “depriving] the defendant of a substantive or procedural right to which the law entitles him?” Glover, 531 U.S. at 203, 121 S.Ct. 696 (quoting Terry Williams, 529 U.S. at 393, 120 S.Ct. 1495). And this question, the Supreme Court has insisted, is part of Strickland, not an adjunct to it.
It is also a question that, again, surely must be answered “no.” Mr. Williams was deprived of his liberty only after a proceeding in which, in the words of the Magna Carta, “the judge or jury acted according to law.” Strickland, 466 U.S. at 694, 104 S.Ct. 2052. The outcome of his trial was not just reliable but also fair, according him all protections to which he was entitled by law. The Sixth Amendment requires no more. As Justice White explained for the Supreme Court, “[i]t is a novel argument that constitutional rights are infringed by trying the defendant rath*1087er than accepting his plea of guilty.” Weatherford, 429 U.S. at 561, 97 S.Ct. 837.
C
The majority’s alternative prejudice analysis does not confront the central concern I have just laid out — namely, that prejudice cannot exist unless our confidence in the reliability or fairness of Mr. Williams’s trial is undermined, and the loss of a plea bargain categorically undermines neither. In one place, the majority suggests that prejudice arises because “had [Mr. Williams] been adequately counseled, there is a reasonable probability he would have accepted the plea offer.” Maj. Op. at 1094. But to the extent this might be read as suggesting that bad legal advice constitutes prejudice, it falls in to Curry’s trap, mixing Strickland’s deficient performance and prejudice prongs. See supra Part II. And to the extent this might be read as suggesting there is a legal entitlement to accept a plea offer, it fails to address the welter of legal authority from the Supreme Court holding otherwise. See supra Part III.B.
Elsewhere, the majority asserts flatly that “counsel can be ineffective where his mistakes have harmed the defense” and “[s]urely, the plea process is part of the defense.” Maj. Op. at 1092 (internal quotation marks omitted). But here again the majority does not explain the source of any legal entitlement to receive or accept a plea offer. It does refer us to Hill v. Lockhart and certain cases from other circuit courts. As I have already described, however, Hill explains merely that poor advice from counsel can render a guilty plea involuntary; it does not speak at all to the question of a defendant’s legal entitlement to a good plea offer. And the circuit cases the majority cites do no more to fill the analytical void. In fact, the search for reasons to support a prejudice finding in these cases is something like a scavenger hunt. One case holds prejudice exists but cites us to another case for the reasons supporting this result. We are promised that, if only we look over there, all will be clear. When we get there, though, we are only told again that we need to look somewhere else. And so we go on and on, bouncing from case to case, hoping that our persistence will eventually pay off with a prize. But opinions without reasons do not persuade; for what is the use of a book without pictures or conversations? Lewis Carroll, Alice’s Adventures in Wonderland 9 (1898). Though it’s possible I’ve taken a wrong turn somewhere in the hunt, at the end of it all the only prize seems to be an outdated 1982 case, United States ex rel. Caruso v. Zelinsky, 689 F.2d 435 (3d Cir.1982). Unlike the later cases that cite it and then cite each other, Caruso does offer reasons for its result. It explains in some detail how a lawyer’s failure to advise his client about a plea bargain can amount to deficient performance. (I agree. Who doesn’t?) The trouble is that the case does not even address the question of prejudice. Because it was decided before Strickland and Cronic, you might not fault Caruso for failing to contain a prejudice analysis. But neither would you expect it to be especially relevant to the prejudice analysis Strickland and Cronic now compel.7 By contrast, *1088Judge Easterbrook’s analysis in Springs is compelling (whether or not the Seventh Circuit has since forgotten about it), and the Utah Supreme Court’s rejection of prejudice in State v. Greuber, 165 P.3d 1185, 1190 (Utah 2007), follows a careful and thorough analysis. See also Bryan v. State, 134 S.W.3d 795, 802 (Mo.Ct.App.2004) (finding no prejudice); State v. Monroe, 757 So.2d 895, 898 (La.Ct.App.2000) (same); Commonwealth v. Mahar, 442 Mass. 11, 809 N.E.2d 989, 996 (2004) (Sosman, J., concurring). This case forces us to pick between competing lines of authority already once the subject of a grant of certiorari; the line of authority on which the court so heavily depends for its result is thin gruel compared to the substantial analysis offered by competing courts.
IV
Having found a violation of Mr. Williams’s constitutional rights, and having held the OCCA’s remedy inadequate to remedy that violation, the majority stops short of imposing a remedy of its own, instead telling the district court to devise one “that comes as close as possible to remedying the constitutional violation.” Maj. Op. at 1093. But there should be no confusion, the part of this case that matters is over, finished, done. According to the majority, the Constitution is offended, a right guaranteed by the Sixth Amendment is infringed, when an attorney’s deficient performance costs his client a plea bargain that appears in hindsight, after the trial’s results are known, to have been a good one. The question that has divided lower courts for a decade is irrevocably decided in this circuit and binding on future panels. No further proceedings can alter this result — at least not in this court. See Mazurek v. Armstrong, 520 U.S. 968, 975, 117 S.Ct. 1865, 138 L.Ed.2d 162 (1997); Eugene Gressman et al., Supreme Court Practice § 4.18 (9th ed.2007).
The majority today not only finally resolves the critical merits question whether a constitutional violation has occurred, it requires a pointless proceeding on remand. In fact, the majority’s instruction to the parties that they should submit briefing to the district court on the proper remedy will surely engender a strange feeling of deja vu. We granted a COA, after all, specifically to resolve what remedy is constitutionally compelled in these circumstances; the parties’ briefs and arguments before us have been largely devoted to this question; and it is of course a pure question of law. Neither, in fairness, is there much we can expect the district court to do to illuminate the matter further. Only two real options have been identified by the *1089parties in their briefs and arguments or by all the many appellate courts who have trod this path before: giving Mr. Williams specific performance of the deal he lost or ordering some variation of a new trial. The district court will have to select one of these two options, the loser will appeal, we will read the same briefs again, and then we will have to make our choice.
Any examination of the remedial question, moreover, serves only to underscore one thing: the absence of anything in need of remedying in the first place. As the majority implicitly recognizes, the traditional habeas remedy of releasing the petitioner is obviously inappropriate. Mr. Williams himself hasn’t asked for unconditional release, and with good reason: he admits to being guilty of second degree murder and wishes to be sentenced to ten years in prison for it. Of course, writs of habeas corpus are often conditionally drawn, specifying conditions by which the government may cure the constitutional problem with a prisoner’s conviction and confinement short of release (albeit on pains of release if the government fails to meet those conditions). But what condition might be appropriate here?
As the majority recognizes, one thing is certain: the OCCA’s chosen remedy can’t be it. Constrained by a state-law statutory minimum for those convicted of first degree murder, the best the OCCA said it could offer Mr. Williams was a discount to the lowest sentence authorized by Oklahoma law for that offense — life with parole. But Mr. Williams’s federal habeas claim is that his first degree murder conviction was secured in violation of the Sixth Amendment. If, as the court says, he is right about that — if his conviction is the result of a Strickland violation — then whatever else must be done, his conviction must be vacated. Whatever new sentence ultimately might be imposed, a wrongful conviction visits an intolerable harm on a criminal defendant. Whether state law allows the OCCA to provide this remedy, to vacate Mr. Williams’s conviction, is neither here nor there: It is the whole point of habeas that convictions secured in compliance with state law must be vacated if they violate the Constitution. And once Mr. Williams’s conviction for first degree murder goes by the boards, so goes any sentence he received for that (unlawfully secured) conviction. He must be re-charged by Oklahoma and found guilty in some lawful proceeding before he can be sentenced anew to any term of incarceration.
Rightly dissatisfied with the OCCA’s remedy rejected by the majority, Mr. Williams urges us to order specific performance of the lost plea bargain, which would involve vacating his first degree murder conviction and give him a ten-year sentence for second degree murder. Some of the courts to have preceded us down this road have offered such a remedy. E.g., Hoffman, 455 F.3d at 942-43; Blaylock, 20 F.3d at 1468-69. But they must be wrong, and wrong in a way that shows the absence of any constitutional right at stake. Even if Mr. Williams had managed to accept the plea offer, under Supreme Court precedent the prosecutor would have been free to withdraw it on a whim, and the state trial judge would have been free to reject it or to impose a different sentence than the one contemplated by the parties’ plea. So ordering specific performance of plea agreement would not restore Mr. Williams to his original position; it would confer on him something he never had to begin with: a legal entitlement, as opposed to a chance, to obtain a ten-year sentence on a charge for second degree murder never brought by the State. Ordering such a remedy would effectively permit Mr. Williams to use the rubric of ineffective assistance to upend the plea bargaining process, transforming it from *1090its historic role as an act of executive discretion to one of judicially enforceable right. We, rather than the State, would assume the job of selecting charges and offering plea terms.
More precisely restoring Mr. Williams to his original position would require something less: vacating the first degree murder conviction, ordering a new trial on that charge, while at the same time forcing the prosecutor to offer the original plea bargain. Some courts have pursued variations of this new trial remedy. E.g., Turner, 940 F.2d 1000. But this remedy still suffers the same problem that plagues the specific performance option: Mr. Williams doesn’t have, and never did have, a right to the plea offer. Unless we decide to assume control of the executive prerogatives of the State of Oklahoma and force the prosecution to keep the offer open, the government would be free to alter or withdraw the plea offer the moment it is extended— or even after it is accepted — for any reason, or for no reason. The trial court, too, would be free to reject any plea. All we could guarantee Mr. Williams at the end of the day is a new trial for first degree murder. And this leads us to another problem: a trial is exactly what Mr. Williams wishes to avoid. He has already received an error-free trial, and it did not end as he hoped. We do not know whether a second prosecution would end any differently than the first one did. Perhaps Mr. Williams would get life without parole again, an even worse result for him than giving him no federal habeas remedy at all; perhaps, if witnesses’ memories have faded or the prosecution makes missteps, he may be acquitted. But we do know this: if Mr. Williams fared better the second time around, it would not be because a more reliable or fair process was used than the one the Constitution provided the first time. If a fair trial is the right remedy, then in a real sense he’s already received it.
And this tells us all we need to know. The fact that a defendant who complains his conviction was secured as a result of a Strickland violation believes a new, fair trial would be inadequate to remedy his complaint strongly suggests that the Strickland right — aimed at securing just such a trial — was not infringed. The disarming reason why square pegs do not fit into round holes is that they are not round.8 Strickland seeks to assure defendants of reliable and fair criminal proceedings, conducted according to law; that is what Mr. Williams received. Mr. Williams would have us follow him through the looking glass, to a world where a fair trial is called “prejudice”; where the results of a fair trial are void because of a lost opportunity rather than an infringed legal entitlement; and where a lawyer’s incompetence transforms the executive plea bargain prerogative into a judicially enforceable entitlement. I do not believe the Sixth Amendment permits us to accompany him there. Respectful of my colleagues’ contrary conclusion, I must dissent.
. While the Antiterrorism and Effective Death Penalty Act of 1996 prohibits us, in certain circumstances, from issuing a habeas writ unless the state court's ruling was unreasonably wrong; it does nothing to disturb the usual rule that we may affirm a state court's denial of a writ for any reason supported by the record. See 28 U.S.C. § 2254(a) & (d).
. While a plea pursuant to North Carolina v. Alford, 400 U.S. 25, 91 S.Ct. 160, 27 L.Ed.2d 162 (1970), might have permitted Mr. Williams to maintain his factual innocence while pleading guilty, the prosecutor's offer in this case was, all sides agree, contingent on a guilty plea in which Mr. Williams had to admit responsibility for the crime.
. See Albert W. Alschuler, Plea Bargaining and its History, 13 Law & Soc’y Rev. 211, 223 (1979) ("[Tjhe poor, friendless, helpless man is most apt to become the one who helps swell the record of convictions.”) (internal quotation marks omitted). Mr. Williams’s counsel was, of course, convinced that Mr. Williams’s wish to plead guilty represented an example of just this phenomenon.
. On one famous account, plea bargains remain nothing more than coerced guilty pleas. See generally John H. Langbein, Torture and Plea Bargaining, 46 U. Chi. L.Rev. 3 (1978). Others respond that the plea bargain reflects the probability of conviction and various possible sentences, each discounted for their likelihood, in a sort of net present value analysis. See Frank H. Easterbrook, Criminal Procedure as a Market System, 12 J. Leg. Stud. 289, 316-17 (1983). On either account, however, the rational plea bargain combines two things the law keeps separate: the risk of conviction at trial and the probable sentence upon conviction. As the length of possible sentences increases, the number of guilty pleas goes up as well — even though the number of acquittals at trial presumably would stay the same. This is why whatever degree of reliability is produced by plea bargaining will always be less than that produced by a trial. And it is why the loss of a plea bargain can never undermine confidence in the result produced by a more reliable procedure.
. The point here is that even an accepted plea bargain confers no constitutionally protected interest on the defendant until it is reduced to a judgment of the court — and even then it is the court's judgment, not the agreement itself, that forms the basis of a defendant's entitlement. Mabry, 467 U.S. at 507-08, 104 S.Ct. 2543. This is not to suggest that a defendant has no remedies to prevent the government from gaining an unfair advantage from his reliance on a plea agreement before the court accepts it. It's just to say that the defendant has no entitlement to the agreement itself before it is reduced to a judgment.
. Perhaps Mr. Williams could show prejudice by demonstrating that his counsel denied him, not a federal right, but some entitlement aris*1085ing from state law. When it comes to plea agreements, however, Oklahoma law appears to track federal law in all pertinent respects. According to the OCCA, even when a plea bargain has been agreed upon by both prosecutor and defense counsel, the prosecutor is free to withdraw it and the defendant may not compel compliance with the deal. Ross v. State, 717 P.2d 117, 122 (Okla.Crim.App.1986). Because “plea bargaining is a discretionary matter with the prosecutor,’’ the OCCA has held that Oklahoma courts may not compel the district attorney to reinstate a plea offer once withdrawn. State ex rel. Stout v. Craytor, 753 P.2d 1365, 1368 (Okla.Crim.App.1988); see also Gray v. State, 650 P.2d 880, 883 (Okla.Crim.App.1982) (“The recognition of plea bargaining as an essential component of the administration of justice ... does not elevate it to a constitutional right.”).
. The maze of authority emanating from Caruso looks something like this: The Third Circuit relied on it in United States v. Day, 969 F.2d 39 (3d Cir. 1992), when that court held, without further analysis, that the right to effective assistance guarantees more than a fair trial. Id. at 44-45. In turn, Day contributed to the three Ninth Circuit cases on which the majority relies today. The first, United States v. Blaylock, 20 F.3d 1458, 1466 (9th Cir.1994), relied on Day for the proposition that a lost plea bargain can cause prejudice. *1088There was no further explanation. Cf. id. at 1465 (citing Caruso for finding deficient performance). The next case, Nunes v. Mueller, 350 F.3d 1045, 1052-53 (9th Cir.2003), principally relies on Blaylock and Day, and repeats a variation on the argument advanced in Caruso. Finally, Hoffman v. Arave, 455 F.3d at 941-42, simply cites Nunes and a few other cases that either can themselves be traced back to Day, e.g., Wanatee v. Ault, 259 F.3d 700 (8th Cir.2001), or do not confront the question whether a subsequent fair trial precludes a finding of prejudice, e.g., Cullen v. United States, 194 F.3d 401, 403 (2d Cir.1999). Like the majority today, Wanatee simply cites Hill v. Lockhart, and then holds out the promise that persuasive reasons for applying Hill in this context will be found somewhere else. 259 F.3d at 703-04. The case to which the reader is referred, of course, relies on Day and a familiar list of other precedents that similarly contain no real analysis. See Engelen v. United States, 68 F.3d 238, 241 (8th Cir.1995). Other lines of cases supporting the majority's result, such as the Sixth Circuit's, appear to amount to the repetition of what started as ipse dixit. See Turner v. Tennessee, 858 F.2d 1201, 1206 (6th Cir.1988) (finding prejudice), vacated on other grounds, 492 U.S. 902, 109 S.Ct. 3208, 106 L.Ed.2d 559 (1989), reaffirmed on remand, 940 F.2d 1000 (6th Cir.1991).
. The majority claims I agree with it that the "OCCA’s remedy is insufficient for the federal constitutional violation it found.” Maj. Op. at 1093. Of course, my point is a different one — that there was no "federal constitutional violation” to remedy in the first place and so the OCCA's remedy was more than "sufficient.”