Larry Peterson robbed a bank. Larry Willis put him up to it, furnished a gun, and helped with the getaway. The two divided the loot with one of the bank’s tellers, who helped them pull the job. All eventually were caught and pleaded guilty. The ex-teller did not appeal. Willis appealed but his lawyer moves to withdraw, contending that further argument would be frivolous. See Anders v. California, 386 U.S. 738, 87 S.Ct. 1396, 18 L.Ed.2d 493 (1967). Peterson’s lawyer has not thrown in the towel, because his client (unlike Willis) entered a conditional plea, reserving an issue for appellate review. See Fed.R.Crim.P. 11(a)(2). Later Peterson moved to withdraw the plea and go to trial. The district judge denied this motion, as he had denied Peterson’s earlier motion to suppress his confession. Logically the first question is whether the plea is valid, so we start there.

Peterson’s principal contention is that the plea is involuntary because counsel falsely promised him that his federal sentence would run concurrently with any punishment under state law. The district judge found, however, that Peterson had not entered his plea under any such misapprehension. During the proceedings that led to acceptance of his plea, Peterson was asked whether anyone had promised[*827] him anything not mentioned in the written plea agreement. He gave a negative answer. Later his lawyer said that he planned to argue for a concurrent sentence but that the prosecutor “has not agreed .to any of that with me.” The judge asked Peterson whether he understood that the sentence need not be concurrent; Peterson said that he did. If these representations to the judge are true, his current submission must be false. Judges need not let litigants contradict themselves so readily; a motion that can succeed only if the defendant committed perjury at the plea proceedings may be rejected out of hand unless the defendant has a compelling explanation for the contradiction. See, e.g., United States v. Stewart, 198 F.3d 984 (7th Cir.1999); United States v. Messino, 55 F.3d 1241, 1248 (7th Cir.1995); United States v. Ellison, 835 F.2d 687, 693 (7th Cir.1987).

Peterson does not present a compelling, or indeed any, explanation. He has tried instead to avoid self-contradiction. The motion to withdraw the plea has been supported by counsel’s say-so rather than evidence. Peterson’s current lawyer filed an affidavit saying that Peterson assured him that his former lawyer had promised a concurrent sentence, but Peterson has been unwilling to put his own name on such an affidavit and expose himself to a prosecution for making inconsistent declarations under oath. See 18 U.S.C. § 1623. The lawyer’s affidavit, being hearsay, is no evidence at all. At oral argument counsel insisted that Galbraith v. United States, 313 F.3d 1001 (7th Cir.2002), permits a defendant to secure an evidentiary hearing on the basis of hearsay routed through his lawyer. Galbraith establishes nothing of the kind. We remarked in passing that neither the defendant nor his lawyer in that case had submitted an affidavit; we did not say (let alone hold) that hearsay in an affidavit from counsel could be submitted in lieu of an affidavit by someone competent to testify about the disputed issue. Lawyers often have personal knowledge of material issues, such as what was said in the discussions leading to the plea. We cannot imagine any reason to relax the normal rules that limit judicial consideration to those statements in affidavits that would be admissible if made on the stand. Otherwise it would be all too easy to route bogus contentions through counsel and avoid prosecution for telling lies to the court.

Peterson’s other arguments in support of withdrawing the plea likewise blame counsel for one or another supposed shortcoming and have equally little record support, as the district judge observed. Moreover, because Peterson has not established that he would not have pleaded guilty had counsel acted differently (for example, spending more time discussing opportunities to impeach the prosecutor’s witnesses at trial), Peterson is not entitled to relief. See Hill v. Lockhart, 474 U.S. 52, 58-60, 106 S.Ct. 366, 88 L.Ed.2d 203 (1985). Further discussion is unnecessary.

Because the district judge did not abuse his discretion in denying Peterson’s motion to withdraw his plea, we must decide the issue reserved in the plea agreement: whether the confession was secured in violation of the principles announced in Miranda v. Arizona, 384 U.S. 436, 86 S.Ct. 1602, 16 L.Ed.2d 694 (1966). After Peterson’s arrest, federal agents laid out the evidence against him. The narration took about 50 minutes. Only then did the officers administer Miranda warnings, obtain waivers (written and oral), and secure Peterson’s confession (written and oral). Peterson contends that the long summary of the evidence was a form of interrogation because it was designed to provoke a re[*828] sponse, see Rhode Island v. Innis, 446 U.S. 291, 100 S.Ct. 1682, 64 L.Ed.2d 297 (1980), and that his statements must be suppressed because the interrogation preceded the warnings and waiver. The problem with this argument is that none of his own statements preceded the warnings. There is no abstract “right to be warned.” See United States v. Patane, 542 U.S. 630, 124 S.Ct. 2620, 159 L.Ed.2d 667 (2004). Miranda creates a rule of evidence: statements made during custodial interrogation by a person who has not been informed of, and waived, his privilege against compulsory self-incrimination (and the right to counsel designed to facilitate its invocation) are inadmissible in criminal proceedings.

Peterson did not make any pre-warning statement. He relies principally on Missouri v. Seibert, 542 U.S. 600, 124 S.Ct. 2601, 159 L.Ed.2d 643 (2004), which holds that post-warning statements are inadmissible if they duplicate pre-warning statements intentionally elicited in an effort to evade Miranda. (This is the understanding of Seibert’s holding shared by Justices Kennedy and Breyer, whose votes were essential to that fractured decision. See United States v. Stewart, 388 F.3d 1079, 1086-90 (7th Cir.2004).) The thinking behind this is that a suspect who has been induced to make a statement may see little point in clamming up after warnings have been-given; he may think that the cat is out of the bag. Moreover, eliciting a statement before the suspect has been informed of his rights implies that the warnings and rights are charades, which reduces the chance that the suspect will invoke his constitutional privilege. The tactic used in Peterson’s case — a long speech during which the officers told the suspect to be silent and repeated that admonition every time he tried to open his mouth — does not undermine Miranda in either of these ways. By the time Peterson received the Miranda warnings he had not said one word; none of his rights had been diluted.

The officers’ approach is risky; if Peterson had spoken out of turn, then Seibert would pose a problem for the prosecutor. (Whether the problem would be insuperable depends on the officers’ mental states; the Justices could not agree in Seibert which had what consequences and how the officers’ thinking would be demonstrated.) Here the risks did not come to pass, because Peterson heeded the instruction not to talk prematurely. Seibert does not require the exclusion of his statements.

As for Willis: His only potential appellate issues concern the sentence. The pre-sentence report proposed an enhancement for leadership. See U.S.S.G. § 3Bl.l(c). Defense counsel opposed that in part because, he asserted, Blakely v. Washington, 542 U.S. 296, 124 S.Ct. 2531, 159 L.Ed.2d 403 (2004), precluded the judge from making the essential factual findings. The judge decided not to increase the sentencing range under § 3Bl;l(e) and inquired whether Willis had any other objections based on Blakely or its predecessors back to Apprendi v. New Jersey, 530 U.S. 466, 120 S.Ct. 2348, 147 L.Ed.2d 435 (2000). Counsel said no. The judge then asked whether it was fair to state that Willis was “waiving any rights ... under Blakely and Apprendi ” other than his objection to a leadership enhancement. Counsel answered: “We are, because we had agreed to those [other] enhancements in the plea agreement.” This waiver — plus the fact that the sentence is within a properly constructed Guideline range — are why counsel now deems an appeal frivolous.

Rights under the sixth amendment may be waived. Blakely says so, observing that “a defendant ... may consent to judicial factfinding as to sentence enhancements, which may well be in his[*829] interest if relevant evidence would prejudice him at trial.” 124 S.Ct. at 2541. That’s what Willis did, agreeing to certain enhancements as part of the plea bargain. See also Shepard v. United States, — U.S. -, n. 5, 125 S.Ct. 1254, 1263 n. 5, 161 L.Ed.2d 205 (2005). The parties did not anticipate the outcome of United States v. Booker, — U.S. -, 125 SiCt. 738, 160 L.Ed.2d 621 (2005), which solved the sixth amendment problem by making the Guidelines non-mandatory. But waivers are not undone by subsequent legal developments, including those in Booker. See United States v. Bownes, 405 F.3d 634 (7th Cir.2005). Willis waived all of his rights under this line of decisions; the waiver did not depend on prevision about exactly what those rights would turn out to be. Any attempt to resurrect a sixth amendment argument now, or to seek the Booker remedy, would be vain.

In Willis the motion to withdraw is granted, and the appeal is dismissed as frivolous. In Peterson the judgment is affirmed.