United States v. Vladimir Rodriguez

[PUBLISH] IN THE UNITED STATES COURT OF APPEALS FOR THE ELEVENTH CIRCUIT ________________________ No. 04-12676 FILED _______________________ U.S. COURT OF APPEALS ELEVENTH CIRCUIT D.C. Docket No. 03-00217-CR-T-23-EAJ April 19, 2005 THOMAS K. KAHN CLERK UNITED STATES OFAMERICA, Plaintiff-Appellee, versus VLADIMIR RODRIGUEZ, Defendant-Appellant. _______________________ Appeal from the United States District Court for the Middle District of Florida _______________________ ON REQUEST FOR POLL FOR REHEARING EN BANC Before EDMONDSON, Chief Judge, TJOFLAT, ANDERSON, BIRCH, DUBINA, BLACK, CARNES, BARKETT, HULL, MARCUS, WILSON, and PRYOR, Circuit Judges. O R D E R: The Court having been polled at the request of one of the members of the Court and a majority of the Circuit Judges who are in regular active service not having voted in favor of it (Rule 35, Federal Rules of Appellate Procedure; Eleventh Circuit Rule 35-5), rehearing en banc is DENIED. /s/ J. L. EDMONDSON CHIEF JUDGE 2 CARNES, Circuit Judge, concurring in the denial of rehearing en banc: With its denial of rehearing en banc in this case, this Court has left intact our circuit law on Booker plain error as it is laid out by our panel decision in this case, United States v. Rodriguez, 398 F.3d 1291 (11th Cir. 2005), and by our decisions in United States v. Duncan, 400 F.3d 1297 (11th Cir. 2005), United States v. Shelton, 400 F.3d 1325 (11th Cir. 2005), and United States v. Curtis, 400 F.3d 1334 (11th Cir. 2005) (per curiam). Our Rodriguez decision, which was followed in Duncan and Curtis, establishes that the use of extra-verdict enhancements under the pre-Booker mandatory guidelines scheme is Sixth Amendment error that is plain. Rodriguez, 398 F.3d at 1298–99; Duncan, 400 F.3d at 1304; Curtis, 400 F.3d at 1335. Shelton adds to our circuit law the rule that while pre-Booker sentencing free of any extra-verdict enhancement is not a violation of the Sixth Amendment, it is statutory error under the remedial part of the Booker decision. Shelton, 400 F.3d at 1330–31. The upshot of our four decisions is that the first two prongs of the four-prong plain error test are met in all pre-Booker sentencing cases.1 To that 1 In United States v. Olano, 507 U.S. 725, 113 S. Ct. 1770 (1993), the Supreme Court held that in order to satisfy the plain error test, the defendant must demonstrate (1) error, (2) that is plain, and (3) that affects substantial rights. Id. at 732–37, 113 S. Ct. at 1777–79. “If all three conditions are met, an appellate court may then exercise its discretion to notice a forfeited error, but only if (4) the error ‘seriously affects the fairness, integrity, or public reputation of judicial proceedings.’” Johnson v. United States, 520 U.S. 461, 467, 117 S. Ct. 1544, 1549 (1997) 3 common holding Shelton effectively adds that where the third prong of the plain error test is met in these cases, the fourth one will be also. See id. at 1333–34. Because the effect of Booker error is the same regardless of the type, our decisions make no functional distinction between constitutional and statutory error. For purposes of the plain error rule, unpreserved error is unpreserved error. Under our decisions, where the Booker issue is raised for the first time on appeal the third prong of the plain error test will be the decisive one. As we explained in our panel opinion, the Supreme Court has instructed us that the third prong requires that an error have “affect[ed] substantial rights,” which almost always means that the error “must have affected the outcome of the district court proceedings.” Rodriguez, 398 F.3d at 1299 (quoting United States v. Olano, 507 U.S. 725, 734, 113 S. Ct. 1770, 1778 (1993)). The standard for showing prejudice is the familiar reasonable probability of a different result formulation, which means a probability “‘sufficient to undermine confidence in the outcome’ of the proceeding.” United States v. Dominguez Benitez, 542 U.S. 74, ___, 124 S. Ct. 2333, 2340 (2004) (quoting Strickland v. Washington, 466 U.S. 668, 694, 104 S. Ct. 2052, 2068 (1984)). Of critical importance, “[i]t is the defendant rather than the [g]overnment who bears the burden of persuasion with respect to prejudice.” (internal marks omitted) (quoting Olano, 507 U.S. at 732, 113 S. Ct. at 1776). 4 Olano, 507 U.S. at 734, 113 S. Ct. at 1778. Our four decisions do not adopt a per se rule about whether the third prong of the plain error test will be met in pre-Booker sentencing cases. Instead, the result depends, as it should, on the facts of the case. For that reason, it is entirely consistent for Rodriguez, Duncan, and Curtis to have concluded that the defendants in those three cases did not carry their burden of establishing the third prong of the plain error test, while Shelton concluded that the defendant in that case did. I. Judge Tjoflat would have this Court adopt a per se rule that the third prong of the plain error test is met in every case of pre-Booker constitutional error, and he would do it in a way that would also preclude application of the harmless error doctrine even in the most extreme case. His thesis is that a Booker constitutional error is a structural error or defect, and for that reason there is no need for the defendant to show third-prong prejudice for plain error purposes. Judge Tjoflat brands all pre-Booker sentences in which there was constitutional error “illegal,” and he offers no plausible reason why the fourth prong of the test would not be met if the third prong were. The bottom line of his approach is automatic reversal of every pre-Booker sentence in which there was an extra-verdict enhancement. 5 That approach does offer the attraction of reducing this Court’s workload, because nothing is easier to apply than an automatic rule that dictates the same result regardless of the facts. Ease of application aside, the proposed rule is not legally or logically appropriate in the pre-Booker area. No other judge has ever even suggested this theory, except in the course of rejecting it. Contrary to Judge Tjoflat’s belief, it simply is not true that “the only real difference” between his approach and that of the Third, Fourth, and Sixth Circuits is that his “offers a more satisfactory rationale for its result.” See Tjoflat, J., dissenting, at 70-71. While the situation in the Third Circuit is unclear, under the Fourth and Sixth Circuits’ approach, sentences involving Booker constitutional error may be upheld under the harmless error doctrine. See, e.g., United States v. Tate, No. 02-4382, 2005 WL 513491, *6 (6th Cir. Mar. 3, 2005) (unpub.); United States v. Bethea, No. 04-5022, 2005 WL 807016, *1 (4th Cir. Apr. 8, 2005) (unpub.). As Judge Tjoflat concedes, his structural error approach precludes application of that harmless error doctrine. See Tjoflat, J., dissenting, at 48 (“[S]tructural errors ‘defy analysis by ‘harmless-error’ standards’ and are per se reversible if an objection is made at trial.” (quoting Arizona v. Fulminante, 499 U.S. 279, 309, 111 S. Ct. 1246, 1265 (1991)). No matter how clear it may be from the record that the defendant would not have received a lesser sentence, a pre- 6 Booker sentence that involved an extra-verdict enhancement could not be affirmed under harmless error analysis if we followed his suggestion. None of the eleven other circuits to address Booker plain error issues have taken such an extreme approach.2 The First Circuit has expressly rejected Judge Tjoflat’s structural theory of Booker plain error. This is what that Court said about it: Nor is this structural error. In certain structural error cases, those which “undermin[e] the fairness of a criminal proceeding as a whole,” errors can be corrected regardless of an individualized showing of prejudice to the defendant. Dominguez Benitez, 124 S. Ct. at 2339; Olano, 507 U.S. at 735, 113 S. Ct. 1770; see Arizona v. Fulminante, 499 U.S. 279, 309–310, 111 S. Ct. 1246, 113 L.Ed.2d 302 (1991) (providing examples of structural error). Because sentencing under a mandatory system is not an error that “undermines the fairness of a criminal proceeding as a whole,” as we discuss above, a Booker type error is not a structural error; the defendant must convince us of prejudice. Indeed, had the majority in Booker thought there was structural error, it would have said so. United States v. Antonakopoulos, 399 F.3d 68, 80 n.11 (1st Cir. 2005); see also United States v. Gonzalez-Huerta, ___ F.3d ___, 2005 WL 807008, *4 (10th Cir. Apr. 8, 2005) (en banc) (holding that Booker statutory error is not structural); Knox v. United States, 400 F.3d 519, 523 (7th Cir. 2005) (explaining why Apprendi error is not structural). 2 The eleven circuits in this count include the Ninth, although its Booker plain error decision has been vacated for rehearing en banc. See United States v. Ameline, 400 F.3d 646, 654–55 (9th Cir.), reh’g en banc granted, ___ F.3d ___, 2005 WL 612710 (Mar. 11, 2005). 7 In response to the First Circuit’s reasoning that if Booker error were structural the Supreme Court would have said so, Judge Tjoflat says that the Court did. According to him, the last sentence of Justice Breyer’s Booker majority opinion about remedy tells us that we are dealing with structural error. To see that message in the last sentence of that opinion requires not just a set of reading glasses but also a vivid imagination. Rather than take the sentence out of context, it is best to set out the entire last paragraph of the opinion: As these dispositions indicate, we must apply today’s holdings—both the Sixth Amendment holding and our remedial interpretation of the Sentencing Act—to all cases on direct review. That fact does not mean that we believe that every sentence gives rise to a Sixth Amendment violation. Nor do we believe that every appeal will lead to a new sentencing hearing. That is because we expect reviewing courts to apply ordinary prudential doctrines, determining, for example, whether the issue was raised below and whether it fails the “plain-error” test. It is also because, in cases not involving a Sixth Amendment violation, whether resentencing is warranted or whether it will instead be sufficient to review a sentence for reasonableness may depend upon application of the harmless-error doctrine. United States v. Booker, 543 U.S. ___, ___, 125 S. Ct. 738, 769 (2005) (internal citations omitted). Focusing on the last sentence, Judge Tjoflat reasons that: 1) because the Supreme Court explicitly said that the harmless error doctrine applies to statutory Booker error, the Court must have meant that the harmless error doctrine does not 8 apply to constitutional Booker error; 2) because the harmless error doctrine does not apply to constitutional Booker error, this type of error must be structural in nature; and 3) because constitutional Booker error is structural in nature, all constitutional Booker error amounts to the deprivation of substantial rights for plain error purposes. Never has so much been inferred from so little. There are all kinds of problems with this theory. To begin with, it is based upon a negative pregnant, and as Judge Becker once observed for the Third Circuit, “drawing instruction from Supreme Court passages through the use of the negative pregnant is risky and unsatisfactory.” Brooks v. Kyler, 204 F.3d 102, 108 (3d Cir. 2000). Here doing so is particularly unsatisfactory because it has the effect of contradicting a more direct statement in the same paragraph of the same opinion. In the two sentences immediately preceding the one in question, the Supreme Court said that every Booker appeal will not lead to a new sentencing hearing “because we expect reviewing courts to apply ordinary prudential doctrines, determining, for example, whether the issue was raised below and whether it fails the ‘plain-error’ test.” Booker, 543 U.S. at ___, 125 S. Ct. at 769. In that sentence the Supreme Court expressed its belief that some sentences involving Booker error would be affirmed because the defendant could not satisfy the plain error test. The Court’s stated belief that the rigors of the plain error test 9 would weed out some Booker error cases is not surprising given that it has previously instructed us that the power to notice plain error should be exercised “sparingly,” Jones v. United States, 527 U.S. 373, 389, 119 S. Ct. 2090, 2102 (1999), and only “in those circumstances in which a miscarriage of justice would otherwise result,” Olano, 507 U.S. at 736, 113 S. Ct. at 1779 (internal quotation omitted). Judge Tjoflat’s position directly contradicts the Supreme Court’s belief that some Booker error cases, even those involving Sixth Amendment violations, will be affirmed because the defendants cannot satisfy the rigorous requirements of the plain error test. Judge Tjoflat’s structural theory contradicts that stated belief of the Supreme Court, because under his theory the third prong of the plain error test will always be satisfied where there is Booker constitutional error. And the third prong is the decisive one. Every court to address the matter has agreed that Booker error satisfies the first two prongs of the plain error test. Likewise, all of the courts to address the issue have found the fourth prong met in Booker constitutional error cases where the third prong is satisfied. See United States v. Oliver, 397 F.3d 369, 380–81 (6th Cir. 2005); United States v. Hughes, 396 F.3d 374, 380–81 (4th Cir.), amended on reh’g by ___ F.3d ___, 2005 WL 628224 (Mar. 16, 2005); United States v. Ameline, 400 F.3d 646, 654–55 (9th Cir.), reh’g en banc granted, ___ 10 F.3d ___, 2005 WL 612710 (Mar. 11, 2005); Shelton, 400 F.3d at 1333–34; United States v. Coles, ___ F.3d ___, 2005 WL 783069, *3 (D.C. Cir. Apr. 8, 2005) (per curiam); see also United States v. Davis, 397 F.3d 173, 183 (3d Cir. 2005).3 Judge Tjoflat does not provide any persuasive basis for believing that the fourth prong of the plain error test will not be met as a matter of course if, as he contends, Booker constitutional error is structural. The Supreme Court said in Dominguez Benitez that structural errors “undermin[e] the fairness of a criminal proceeding as a whole.” 542 U.S. at ___, 124 S. Ct. at 2339. Quoting another Supreme Court opinion, Judge Tjoflat insists that “[b]ecause structural error affects the framework of the trial itself, its ‘consequences . . . are necessarily unquantifiable and indeterminate.’” See Tjoflat, J., dissenting, at 48 (quoting Sullivan v. Louisiana, 508 U.S. 275, 282, 113 S. Ct. 2078, 2083 (1993)). In explaining the impact of structural errors, the Supreme Court stated: “Without 3 A panel of the Sixth Circuit skipped the third prong and held that the fourth prong of the plain error test was not met in one case. United States v. Bruce, 396 F.3d 697, 719–20 (6th Cir. 2005). However, the Sixth Circuit has since decided that the Bruce decision is not to be followed because it conflicts with the earlier decision of another Sixth Circuit panel in the Oliver case. See United States v. Milan, 398 F.3d 445, 452 n.3 (6th Cir. 2005) (“To the extent Bruce conflicts with Oliver, we note that we must follow Oliver because it was decided first.”). The Tenth Circuit’s en banc majority opinion, like that of the Sixth Circuit’s Bruce panel, also skipped the third prong and went straight to the fourth in its recent plain error decision. See Gonzalez- Huerta, 2005 WL 807008, at *6–9. 11 these basic protections, a criminal trial cannot reliably serve its function as a vehicle for determination of guilt or innocence, and no criminal punishment may be regarded as fundamentally fair.” Rose v. Clark, 478 U.S. 570, 577–78, 106 S. Ct. 3101, 3106 (1986) (internal citation omitted). Because structural error, where it exists, renders a criminal punishment fundamentally unfair, it would be difficult to justify a conclusion that an error that is structural does not “seriously affect[] the fairness, integrity or public reputation of judicial proceedings,” Olano, 507 U.S. at 732, 113 S. Ct. at 1776 (quotation omitted). See United States v. Recio, 371 F.3d 1093, 1103 n.7 (9th Cir. 2004) (“We note that structural error is particularly likely to satisfy Olano’s fourth prong.”). So far as can be discovered, no court has ever actually held that an error is structural but fails to meet the fourth prong of the plain error test. However much it may detract from the appeal of the Booker structural error theory, the fact is that adopting that theory will almost certainly result in reversal without regard to the facts of an individual case. That is what the structural error approach generally does and is one reason it is so rarely applied. The result of Judge Tjoflat’s position that Booker constitutional error is structural error, then, is that every Booker constitutional error will satisfy the plain 12 error test and require reversal and resentencing. Thus, in his view, the Supreme Court’s belief about the effect of applying “ordinary prudential doctrines, determining, for example, whether the issue was raised below and whether it fails the ‘plain-error’ test,” Booker, 543 U.S. at ___, 125 S. Ct. at 769, is wrong. Applying the plain error test to unpreserved Booker error will not, as the Court thought, prevent every appeal from leading to a new sentencing hearing. See id. Seeking to free his theory from the clutches of the Supreme Court’s statement that not every Booker error will meet the requirements of the plain error test, Judge Tjoflat posits that the Court must have meant only Booker statutory error. Beyond satisfying the needs of his theory, there is no reason to interpret the Court’s statement that way. The Court did not say that it expected reviewing courts to apply ordinary prudential doctrines, such as the plain error test, only in cases of statutory error. Nor is there any reason in law or logic to distinguish between the two types of Booker error for purposes of the plain error rule. In order to make Judge Tjoflat’s post-Booker plain error theory work, we would have to write into our law a rule that statutory errors could not be treated as structural regardless of their structural effect. No decision of the Supreme Court or of this 13 Court has ever held that, and there is no reason for this Court to do so now.4 The heart of Judge Tjoflat’s structural error theory is that Booker constitutional errors affect the sentencing framework in ways that are “necessarily unquantifiable and indeterminate.” Tjoflat, J., dissenting, at 48 (quoting Sullivan, 508 U.S. at 282, 113 S. Ct. at 2083). Yet, as he commendably concedes, Booker constitutional and statutory errors cannot be distinguished based on the effect they 4 The language that Judge Tjoflat quotes from a footnote in our decision in United States v. Sanchez, 269 F.3d 1250, 1272 n.41 (11th Cir. 2001) (en banc), is pure dicta. See Tjoflat, J., dissenting, at 54. The error in that case was constitutional, not statutory. Judge Tjoflat expresses some surprise at the refusal of the panel opinion and its author to be impressed by that dictum from the Sanchez opinion. Tjoflat, J., dissenting, at 76 n.18. He shouldn’t be surprised, because it is well-established that dicta does not bind anyone. E.g., McDonald’s Corp. v. Robertson, 147 F.3d 1301, 1315 (11th Cir. 1998) (Carnes, J., concurring) (“For these reasons, among others, dicta in our opinions is not binding on anyone for any purpose.”), quoted with approval, Shinn ex rel. Shinn v. Comm’r of Soc. Sec., 391 F.3d 1276, 1285 (11th Cir. 2004) (Tjoflat, J.). The only thing that is truly surprising is Judge Tjoflat’s new-found respect for dicta, and his reliance on the dictum in the Sanchez opinion, Tjoflat, J., dissenting, at 54., which comes after years of proclaiming himself not to be bound by dicta in this Court’s, or even the Supreme Court’s, opinions. See, e.g., United States v. Smith, ___ F.3d __, 2005 WL 628686, *19 n.7 (11th Cir. Mar. 18, 2005) (Tjoflat, J.) (“[T]he prior panel rule does not extend to dicta.” (internal citation and quotation omitted)); United States v. Santa, 236 F.3d 662, 672 n.14 (11th Cir. 2000) (Tjoflat, J.) (“It is well settled, however, that no opinion can be considered as binding authority unless the case calls for its expression. As the court’s statement in Nixon was unnecessary to its decision, it is a dictum and does not control our decision in this case.” (internal citation and quotation omitted)); Klay v. United Healthgroup, Inc. 376 F.3d 1092, 1101 n.12 (11th Cir. 2004) (Tjoflat, J.) (stating that “[t]his holding, moreover, was pure dicta, as we had previously concluded that we lacked jurisdiction in the matter,” and concluding that “we are not bound to adhere to it”); Mobley v. Head, 306 F.3d 1096, 1102 (11th Cir. 2002) (Tjoflat, J., dissenting) (stating that because a decision on an issue by a prior panel was mere dicta, that issue remains open for later review); Armstrong v. Martin Marietta Corp., 138 F.3d 1374, 1396 (11th Cir. 1998) (Tjoflat, J.) (stating that he was “mindful of dicta” in a Supreme Court decision “that appeared to be in tension with the conclusions I reach today,” but concluding that he is not bound by Supreme Court dicta). 14 may have had on the sentence proceeding. Id. at 53-54. Everything that can be said about the effect Booker constitutional error may have on a sentencing proceeding applies with equal force to Booker statutory error. Every difference between what Judge Tjoflat calls the “old model” of sentencing and the “new model” that would be in place on remand exists regardless of whether there is an extra-verdict enhancement in a case. The presence or absence of prejudice, and the difficulty of determining it, is the same regardless of whether the Booker error is constitutional or statutory. Judge Tjoflat concedes that treating one type of unpreserved Booker error as structural and the other not, even though they have the same effect on the structure of the proceeding, is “intuitively . . . odd,” id. at 53. That is an understatement. In arguing for the distinction, Judge Tjoflat relies on a dissenting opinion from Booker to dismiss statutory errors as “merely a byproduct of Booker’s ‘unnecessarily broad remedy.’” Tjoflat, J., dissenting, at 75 (citing Booker, 543 U.S. at ___, 125 S. Ct. at 788 (Stevens, J., dissenting)). That would be fine, except dissenting opinions express dissenting views. In applying the Booker decision, we cannot adopt the dissenting view on the remedy but must instead accept the majority opinion as the law and follow it. Judge Tjoflat’s theory that Booker constitutional error is structural error is 15 also contrary to this Court’s own en banc decision in United States v. Sanchez, 269 F.3d 1250 (11th Cir. 2001), where we discussed at length and rejected the contention that any of the various types of Apprendi error were structural error. Id. at 1272–75. Our decision that Apprendi error is not structural answers the question of whether Booker error is structural, because Booker is an application of Apprendi once removed. See Booker, 543 U.S. at ___, 125 S. Ct. at 755–56 (Stevens, J., maj. op.) (“our holding in Blakely applies to the Sentencing Guidelines”); Blakely v. Washington, 542 U.S. __, 124 S.Ct. 2531, 2536 (2004) (“This case requires us to apply the rule we expressed in Apprendi . . . .”). Even if our Sanchez decision did not exist, we still should not hold that Booker error is structural error. The Supreme Court itself has been careful to note that, “[w]e have found structural errors only in a very limited class of cases.” Johnson v. United States, 520 U.S. 461, 468, 117 S. Ct. 1544, 1549 (1997). In describing how limited that class of cases is, the Court in Arizona v. Fulminante, listed the following decisions finding structural error: Gideon v. Wainwright, 372 U.S. 335, 83 S. Ct. 792 (1963) (the total deprivation of the right to counsel at trial); Tumey v. Ohio, 273 U.S. 510, 47 S. Ct. 437 (1927) (a biased judge); Vasquez v. Hillery, 474 U.S. 254, 106 S. Ct. 617 (1986) (unlawful exclusion of members of the defendant’s race from a grand jury); McKaskle v. Wiggins, 465 16 U.S. 168, 104 S. Ct. 944 (1984) (denial of the right to self-representation at trial); and Waller v. Georgia, 467 U.S. 39, 104 S. Ct. 2210 (1984) (denial of the right to a public trial). Arizona v. Fulminante, 499 U.S. 279, 309–10, 111 S. Ct. 1246, 1265 (1991) (Rehnquist, C.J., maj. op.). The only case since Fulminante where the Supreme Court has found a structural error is Sullivan v. Louisiana, 508 U.S. 275, 278–82, 113 S. Ct. 2078, 2081–83 (1993), which involved an improper jury instruction on the “beyond a reasonable doubt” standard. After listing those structural error cases, the Supreme Court in Fulminante explained the common thread: “‘Without these basic protections, a criminal trial cannot reliably serve its function as a vehicle for determination of guilt or innocence, and no criminal punishment may be regarded as fundamentally fair.’” Id. at 310, 111 S. Ct. at 1265 (quoting Rose v. Clark, 478 U.S. 570, 577–78, 106 S. Ct. 3101, 3106 (1986)). By contrast, Booker error does not prevent a trial from serving its function as a vehicle for the determination of guilt or innocence, and it does not render criminal punishment fundamentally unfair. The pre-Booker regime was the law of the land that applied in determining hundreds of thousands of federal sentences in this country over a period of almost two decades. While Booker changes some of the rules, it does not compel the conclusion that the sentencing process used in every federal court in this country 17 for the last twenty years was so defective that it resulted in thousands of fundamentally unfair sentences. The Supreme Court reiterated the rarity of structural error in Neder v. United States, stating that: “[I]f the defendant had counsel and was tried by an impartial adjudicator, there is a strong presumption that any other [constitutional] errors that may have occurred are subject to harmless-error analysis.” 527 U.S. 1, 8, 119 S. Ct. 1827, 1833 (1999) (quotation omitted, brackets in original). The same point about the rarity of structural error was made in Chief Justice Rehnquist’s majority opinion in Fulminante. There he listed all of the types of error the Court had held to be subject to harmless error analysis and therefore not structural in nature: Clemons v. Mississippi, 494 U.S. 738, 752–54, 110 S. Ct. 1441, 1450–51 (1990) (unconstitutionally overbroad jury instructions at the sentencing stage of a capital case); Satterwhite v. Texas, 486 U.S. 249, 108 S. Ct. 1792 (1988) (admission of evidence at the sentencing stage of a capital case in violation of the Sixth Amendment Counsel Clause); Carella v. California, 491 U.S. 263, 266, 109 S. Ct. 2419, 2421 (1989) (jury instruction containing an erroneous conclusive presumption); Pope v. Illinois, 481 U.S. 497, 501–04, 107 S. Ct. 1918, 1921–23 (1987) (jury instruction misstating an element of the offense); Rose v. Clark, 478 U.S. 570, 106 S. Ct. 3101 (1986) (jury instruction containing an erroneous rebuttable presumption); Crane v. Kentucky, 476 U.S. 683, 691, 106 S. Ct. 2142, 2147 (1986) (erroneous exclusion of defendant’s testimony regarding the circumstances of his confession); Delaware v. Van Arsdall, 475 U.S. 673, 106 S. Ct. 1431 (1986) (restriction on a defendant’s right to cross-examine a witness for bias in violation of the Sixth Amendment Confrontation Clause); Rushen v. Spain, 18 464 U.S. 114, 117–18 & n.2, 104 S. Ct. 453, 454–55 & n.2 (1983) (denial of a defendant’s right to be present at trial); United States v. Hasting, 461 U.S. 499, 103 S. Ct. 1974 (1983) (improper comment on defendant’s silence at trial, in violation of the Fifth Amendment Self-Incrimination Clause); Hopper v. Evans, 456 U.S. 605, 102 S. Ct. 2049 (1982) (statute improperly forbidding trial court’s giving a jury instruction on a lesser included offense in a capital case in violation of the Due Process Clause); Kentucky v. Whorton, 441 U.S. 786, 99 S. Ct. 2088 (1979) (failure to instruct the jury on the presumption of innocence); Moore v. Illinois, 434 U.S. 220, 232, 98 S. Ct. 458, 466 (1977) (admission of identification evidence in violation of the Sixth Amendment Counsel Clause); Brown v. United States, 411 U.S. 223, 231–32, 93 S. Ct. 1565, 1570–71 (1973) (admission of the out-of-court statement of a nontestifying codefendant in violation of the Sixth Amendment Counsel Clause); Milton v. Wainwright, 407 U.S. 371, 92 S. Ct. 2174 (1972) (confession obtained in violation of Massiah v. United States, 377 U.S. 201, 84 S. Ct. 1199 (1964)); Chambers v. Maroney, 399 U.S. 42, 52–53, 90 S. Ct. 1975, 1981–82 (1970) (admission of evidence obtained in violation of the Fourth Amendment); Coleman v. Alabama, 399 U.S. 1, 10–11, 90 S. Ct. 1999, 2003–04 (1970) (denial of counsel at a preliminary hearing in violation of the Sixth Amendment Confrontation Clause). Fulminante, 499 U.S. at 306–07, 111 S. Ct. at 1263 (Rehnquist, C.J., maj. op.) (some citations omitted). The Fulminante Court itself held that the admission of a coerced confession was not a structural error. Id. at 310, 111 S. Ct. at 1265. Since the Fulminante decision, the Court has added to the list of nonstructural errors in the following decisions: United States v. Dominguez Benitez, 542 U.S. 74, ___, 124 S. Ct. 2333, 2339 n.6 (2004) (omission of a Fed. R. Crim. P. 11 warning is not a structural error); Neder, 527 U.S. at 4, 14, 119 S. Ct. at 1831, 1836 (judicial usurpation of the jury’s function of determining whether an element of the crime 19 existed is not structural error); Brecht v. Abrahamson, 507 U.S. 619, 628–30, 113 S. Ct. 1710, 1716–17 (1993) (the government’s use of the defendant’s post- Miranda-warnings silence to impeach the defendant is not structural error). Against this flood of decisions holding that various and sundry constitutional and statutory errors are not structural, Judge Tjoflat’s proposition that Booker error is structural is unpersuasive. It would not be enough to make Booker error structural even if, as he argues, defendants had no incentive before Booker to put forward evidence relating to the 18 U.S.C. § 3553(a) factors. Tjoflat, J., dissenting, at 62. We know that would not be enough because the Supreme Court has held a number of errors not to be structural, even where the error may have prevented or discouraged the defendant from putting forward additional evidence or arguments. For example, in Carella v. California, 491 U.S. 263, 266, 109 S. Ct. 2419, 2421 (1989), the error was application of a conclusive presumption about the element of intent, and in Neder, 527 U.S. at 14–15, 119 S. Ct. at 1836–37, the error was taking an element of the crime away from the jury. In both of those cases, the error may have influenced the evidence the defendants put forward or the arguments they made. Nonetheless, the Supreme Court held implicitly in Carella and explicitly in Neder that the error was not structural. The same is true when constitutional error is committed by precluding 20 lesser included offenses in capital cases. A defendant prevented by state law from having the jury consider a lesser offense very well might refrain from putting forward evidence that he was guilty of it instead of the greater offense. Be that as it may, the Supreme Court in Hopper v. Evans determined that this type of constitutional error is not structural and affirmed the conviction because the unconstitutional preclusion provision had no prejudicial effect in that case. 456 U.S. 605, 613–14, 102 S. Ct. 2049, 2054 (1982) (rejecting the argument that the preclusion provision “so infected respondent’s trial that he must be retried” and reversing the grant of relief because that provision did not prejudice the respondent). For the reasons just discussed, it would not be enough to make Booker error structural if defendants in pre-Booker sentencing proceedings had no incentive to introduce mitigating evidence or arguments relating to § 3553(a) factors. However, Judge Tjoflat’s structural theory does depend on the assumption that before Booker there was no such incentive, while the reality is that there was. This is yet another problem with his theory. The assumed lack of incentive is his theory’s explanation for why the pre-Booker sentencing record in any given case would not contain evidence indicating the presence of any § 3553(a) factors that might make a difference in the post-Booker advisory sentencing regime. 21 Defendants in pre-Booker sentencing proceedings lacked any incentive to offer § 3553(a) evidence or argument, Judge Tjoflat says, because it would have been irrelevant unless it fit within a permissible ground for downward departure, which not all § 3553(a) factors do. Tjoflat, J., dissenting, at 62. Chief Judge Boggs of the Sixth Circuit has pointed out the flaw in this argument: This argument ignores a fundamental feature of the Guidelines: they present a sentencing court with a range, from which it must select a sentence. In this case the range was nearly five years—57 months. Counsel already had every reason and every opportunity to present any mitigating circumstance that might possibly have saved Barnett from an additional five years in prison. Any arguments that might be raised post-Booker about culpability, future dangerousness, offsetting good works, family obligations, or any other mitigating circumstance were also fair game pre-Booker, and these arguments for mitigation have been regularly invoked by defense counsels in pre-Booker sentencing proceedings. United States v. Riascos-Suarez, 73 F.3d 616, 627–28 (6th Cir. 1996) (finding reversible error when the defendant was not offered the opportunity to give mitigating evidence at sentencing). The Guidelines never placed any limits on the ability of the district court to consider these factors, so there is no reason to remand so the district court may consider additional circumstances. United States v. Barnett, 398 F.3d 516, 537–38 (6th Cir. 2005) (Boggs, C.J., concurring in part and dissenting in part); see also Gonzalez-Huerta, 2005 WL 807008, at *6. That’s a good point. For example, in this case the guidelines range was 97 to 121 months, a spread of two full years. Rodriguez, 398 F.3d at 1296. In our Shelton case, the 22 guidelines range was effectively 190 to 222 months, a spread of more than two- and-a-half years. 400 F.3d at 1328. The guideline ranges in Rodriguez and Shelton exemplify the fact that most pre-Booker sentencing defendants, including the one in this case, had plenty of incentive to put forth mitigating evidence and arguments relating to the § 3553(a) factors. Those factors were not, as Judge Tjoflat suggests, “irrelevant in all but the most unusual cases” in pre-Booker sentencing. Tjoflat, J., dissenting, at 62. To the contrary, they were relevant in virtually every case. (The Duncan case exemplifies the exception, because there the guidelines dictated a life sentence. 400 F.3d at 1300 n.2.). Judge Tjoflat says that “we must assume, if anything, that a defendant such as Rodriguez would have done something different under the new model.” Tjoflat, J., dissenting, at 65-66. There is no justification for that assumption, especially with a defendant such as Rodriguez who had two years worth of incentive built into the guidelines range to come forward with any § 3553(a) evidence or arguments at his pre-Booker sentencing proceeding. Even if we were to disregard the guidelines range incentive, the crucial assumption behind Judge Tjoflat’s structural theory ignores the fact that, if the guidelines had been only advisory, the probation officer in compiling the PSR and the government in advocating a longer sentence could have put forth evidence and arguments that the 23 § 3553(a) factors favored a sentence above the guideline range. The question is not what the defendant would have done. The question is what the district court would have done after hearing the evidence and arguments of both sides. There is no basis for any assumption that the differences between an advisory and a mandatory guidelines system will favor defendants in general or, more importantly, a particular defendant. The fact is that in most cases, as in Rodriguez, 398 F.3d at 1301, we just don’t know. But there are cases like Shelton, 400 F.3d at 1332–33, where the record shows a reasonable probability of a different result in favor of the defendant if the guidelines had not been mandatory. That is why we sent that case back for resentencing. Where, however, we have no indication either way, the party with the burden loses. Rodriguez, 398 F.3d at 1301. When it comes to the third prong of the plain error test, that party is the defendant. Id. II. Judge Tjoflat is correct about the Fourth and Sixth Circuits’ approach to this issue being wrong. See Tjoflat, J., dissenting, at 78 n.19 (citing United States v. Hughes, ___ F.3d ___, 2005 WL 628224 (4th Cir. Mar. 16, 2005), and United States v. Oliver, 397 F.3d 369 (6th Cir. 2005)). As he explains, it is simply inaccurate to say, as those two circuits do, a defendant has carried his burden of 24 showing that a Booker error has affected his substantial rights when we can only guess what the result would be without the error. Id. Instead of saying that there is a burden and then applying it in a way that is not burdensome, it is more forthright to argue directly, as Judge Tjoflat does, that the burden ought not exist. See id. (“In short, these courts require the defendant to show that his substantial rights were affected only to find that they were in every case involving constitutional error, whereas I would simply say that no such showing is required.”). For the reasons already discussed, this Court did not adopt Judge Tjoflat’s approach, but at least his reasoning has the virtue of analytical transparency. Since our Rodriguez opinion, the Fourth Circuit panel that issued the original Hughes decision has taken another look at it. That second look resulted in the issuance of a new opinion that reaches the same conclusion. See Hughes, 2005 WL 628224. We have already discussed at some length the flaws in the reasoning on display in the first Hughes opinion, Rodriguez, 398 F.3d at 1301–04, but the new opinion illuminates those flaws in ways that warrant additional comment. See also United States v. Coles, ___ F.3d ___, 2005 WL 783069, *4 (D.C. Cir. Apr. 8, 2005) (per curiam) (discussing the problems with the Hughes approach and concluding that it “flies in the face of the Supreme Court’s remedial order in 25 Booker”). There is no longer any reason to doubt that the Fourth Circuit’s erroneous approach has at its core the premise that the mandatory nature of the guidelines is not an essential part of a Booker constitutional violation. This is the paragraph of its new Hughes opinion that lays bare the substructure of its reasoning: Stated differently, the act of mistakenly treating the guidelines as mandatory is not part of the Sixth Amendment error before us, despite the fact that the former mandatory nature of the guidelines set the stage for the constitutional violation in Booker. That the erroneous treatment of the guidelines as mandatory is not part of the constitutional error can be seen most clearly in a post-Booker context. Suppose a district court, post-Booker, erroneously treats the guidelines as mandatory when imposing a sentence that rests in part on extra-verdict enhancements. Such a sentence would certainly be erroneous, but there would be no Sixth Amendment error because, regardless of what the district court thought, the guidelines post-Booker are in fact advisory and the sentence imposed did not exceed the maximum authorized by the jury verdict (which is, of course, the maximum set forth in the statute of conviction). Hughes, 2005 WL 628224, at *11; see also id. at *10 (There is “a separate class of error, namely, the error of treating the guidelines as mandatory at sentencing. Such an error is distinct from the Sixth Amendment claim that gave rise to the decision in Booker, and it is non-constitutional in nature.”). Wrong, wrong, wrong. Treating the guidelines as mandatory is an essential part of the constitutional error, not a non-constitutional error distinct from the one involved 26 in the Booker decision. The Supreme Court was unanimous about that. All nine Justices joined one or both of the majority opinions, each of which stated that the use of extra-verdict enhancements in an advisory regime is permissible and that if the guidelines are not applied mandatorily there is no constitutional error. Booker, 543 U.S. at ___, 125 S. Ct. at 750 (Stevens, J., maj. op.) (“If the Guidelines as currently written could be read as merely advisory provisions that recommended, rather than required, the selection of particular sentences in response to differing sets of facts, their use would not implicate the Sixth Amendment.”); id. at 764 (Breyer, J., maj. op.) (“[W]ithout this provision—namely the provision that makes the relevant sentencing rules mandatory and imposes binding requirements on all sentencing judges—the statute falls outside the scope of Apprendi’s requirement.” (internal quotation and markings omitted)). The Hughes opinion manages to reject the one principle that all nine Justices agreed on in Booker. Admittedly, the Justices did not add to their statements “. . . and we really mean it,” but what else short of that could the Court have said to convey that it was serious? The hypothetical laid out in the part of the Hughes opinion quoted earlier highlights that Court’s misunderstanding of Booker. If a district court in the post- Booker world applies an extra-verdict enhancement and treats the guidelines as mandatory, of course the resulting sentence will be unconstitutional. The violation 27 and its impact on the defendant will be identical to what occurred in the Booker case itself. That the constitutional violation and its impact occurs not in obedience to a statutory mandate but as a result of ignorance, negligence, or defiance does not make it any less a constitutional violation. Constitutional error can be committed free of statutory command. The effect, if any, on the defendant is the same if the same act is committed in the same circumstances regardless of the source of the error. If the guidelines are applied in a mandatory way when there are extra-verdict enhancements, the result is a pattern-perfect Booker constitutional violation. The Fourth Circuit’s disregard of what the Supreme Court said in Booker about the mandatory application of the guidelines being the source of the constitutional violation led that circuit to give the wrong answer to the question it posed in the latest Hughes opinion. The prejudice inquiry, Hughes says, “is whether the district court could have imposed the sentence it did without exceeding the relevant Sixth Amendment limitation. If the answer to this inquiry is ‘yes,’ . . . then the defendant has failed to demonstrate an effect on substantial rights . . . .” Hughes, 2005 WL 628224, at *8. The answer, of course, is “yes.” The district court could have applied the same extra-verdict enhancement and imposed the same sentence it did without violating the Sixth Amendment by 28 consulting the guidelines but treating them as advisory. That non-mandatory application of the guidelines would have violated the Sentencing Reform Act as it then existed, specifically 18 U.S.C. § 3553(b)(1), but it would not have violated the Sixth Amendment. That is why the Supreme Court’s Booker decision instructs district courts to find and apply any extra-verdict enhancements in future cases but to treat the resulting guidelines range as advisory. In addition to resting on a premise that is contrary to the stated belief in the Booker decision of every single Justice, the Fourth Circuit never adequately explains why the decision about whether a particular defendant has been prejudiced should not turn on the one factor that will be changed on remand in order to avoid the constitutional violation. It never tells us how its rule about the third prong of the plain error test serves the main purpose of the prong’s prejudice requirement, which is to avoid wasteful reversals. See Dominguez Benitez, 542 U.S. at ___, 124 S. Ct. at 2340 (application of the prejudice standard “should enforce the policies that underpin Rule 52(b) generally, to encourage timely objections and reduce wasteful reversals by demanding strenuous exertion to get relief for unpreserved error”). By divorcing the question of whether to reverse from the question of what would change on remand, the Fourth Circuit’s Hughes rule does not avoid wasteful reversals but is instead an engine for generating them. 29 III. While cheering on Judge Tjoflat’s efforts to construct a structural error theory, Judge Barkett’s dissenting opinion also suggests that the Court has applied the wrong prejudice standard in these post-Booker plain error cases, or perhaps that we applied the correct standard in the wrong way. To the contrary, we have applied the correct standard correctly. This is what we said in the panel opinion: The third prong of the plain error test, however, is another matter. It requires that an error have “affect[ed] substantial rights,” which almost always requires that the error “‘must have affected the outcome of the district court proceedings.’” Cotton, 535 U.S. at 632, 122 S. Ct. at 1786 (quoting Olano, 507 U.S. at 734, 113 S. Ct. at 1778). The standard for showing that is the familiar reasonable probability of a different result formulation, which means a probability “‘sufficient to undermine confidence in the outcome.’” United States v. Dominguez Benitez, ___ U.S. ___, 124 S. Ct. 2333, 2340 (2004) (quoting Strickland v. Washington, 466 U.S. 668, 694, 104 S. Ct. 2052, 2068 (1984)). In regard to this third prong, “[i]t is the defendant rather than the [g]overnment who bears the burden of persuasion with respect to prejudice.” Olano, 507 U.S. at 734, 113 S. Ct. at 1778. Rodriguez, 398 F.3d at 1299. We repeated and applied the reasonable probability of a different result standard—not a preponderance standard—at least five more times throughout our opinion. See id. at 1301 (“Therefore, in applying the third prong, we ask whether there is a reasonable probability of a different result if the guidelines had been applied in an advisory instead of binding fashion by the 30 sentencing judge in this case.” (emphasis added)); id. (“[W]here the effect of an error on the result in the district court is uncertain or indeterminate—where we would have to speculate—the appellant has not met his burden of showing a reasonable probability that the result would have been different but for the error; he has not met his burden of showing prejudice; he has not met his burden of showing that his substantial rights have been affected.” (emphasis added)); id. at 1302 (“The important function of the third prong of the plain error test is to prevent a remand for additional proceedings where the defendant, who failed to make a timely objection, cannot show that there is a reasonable probability that a do-over would more likely than not produce a different result.” (emphasis added)); id. (“Therefore in order to satisfy the third prong of the plain error test, pre-Booker defendants must establish a reasonable probability that if the district court had considered the guidelines range it arrived at using extra-verdict enhancements as merely advisory, instead of mandatory, and had taken into account any otherwise unconsidered § 3553 factors, the court would have imposed a lesser sentence than it did.” (emphasis added)); id. at 1304 (“As we have explained, where the record does not provide ‘any indication’ that there would have been a different sentence if the court had recognized and exercised its § 3553(a) discretion and treated the guidelines range as merely advisory, the party with the burden of showing a 31 reasonable probability of a different result loses.” (emphasis added)). The difference in views between this Court and Judge Barkett is not about the standard to apply. We all agree that the reasonable probability standard applies. The difference is how that standard applies when we cannot tell whether the error caused prejudice. The Court’s position, which is set out in Rodriguez, Curtis, and Duncan, is that the defendant loses in this situation. That is so because the burden is on the defendant, and the Supreme Court in Jones v. United States, 527 U.S. 373, 119 S. Ct. 2090 (1999), told us that when we cannot tell whether the error caused prejudice the defendant loses. Judge Barkett does not dispute that the defendant has the burden of showing prejudice; she just says that it does not really amount to any burden at all. Her view is that once there is error, the defendant carries his burden of establishing prejudice by showing that we don’t know if the error prejudiced him. Barkett, J., dissenting, at 82-83. In other words, nothing equals something; the burden is no burden at all. The adjective “oxymoronic” does not do justice to this “no-burden burden” concept. Aside from robbing a perfectly good word (“burden”) of its plain meaning, this key premise of Judge Barkett’s position is also contrary to a number of our prior decisions applying the reasonable probability of a different result standard. 32 See Straight v. Wainwright, 772 F.2d 674, 680 (11th Cir. 1985) (defendant fails to carry his burden of showing a reasonable probability of a different result where “[t]here is absolutely no evidence” on the issue); Henry v. Wainwright, 743 F.2d 761, 763 (11th Cir. 1984) (per curiam) (where there is nothing to show that an omitted instruction would have applied to the jury’s deliberations, there is “nothing upon which a finding of prejudice could be based,” and the existence of “only a possibility of prejudice” is not enough to satisfy the defendant’s burden of showing a reasonable probability of a different result); Adams v. Wainwright, 764 F.2d 1356, 1369 (11th Cir. 1985) (where there is no proof on the issue, the defendant has failed to carry his burden of showing a reasonable probability of a different result). More importantly, Judge Barkett’s position cannot be reconciled with the specific holding and instruction of the Supreme Court’s decision in Jones. As we explained in the panel opinion, Jones is directly applicable to plain error third- prong situations in which it is unclear what effect, if any, the error had. Rodriguez, 398 F.3d at 1301. The Jones decision came in a capital case where the defendant had failed to object in the district court to the omission of a jury instruction about the consequences of a deadlock on the sentence issue. In rejecting the defendant’s plain error argument, the Supreme Court explained that 33 the situation was one in which it could not be determined what effect, if any, omission of that instruction had on the outcome of the capital sentence proceeding. It instructed us that where the effect of an error may as likely have cut one way as the other, where the effect it had on the outcome is indeterminate, the “defendant cannot meet his burden of showing that the error actually affected his substantial rights.” 527 U.S. at 394–95, 119 S. Ct. at 2105. We are not to speculate in this kind of situation. The defendant loses because the defendant has the burden. Id.; see also Rodriguez, 398 F.3d at 1300 (“The lesson of the Supreme Court’s Jones decision is that the burden truly is on the defendant to show that the error actually did make a difference . . . . Where errors could have cut either way and uncertainty exists, the burden is the decisive factor in the third prong of the plain error test, and the burden is on the defendant.”).5 5 The Court in Jones was faced with the question of whether the defendant, who failed to properly preserve his objection, was entitled to a new sentence proceeding because of the district court’s failure to give a jury instruction. 527 U.S. at 387–88, 119 S. Ct. at 2101–02. The Court held that the error the defendant claimed was not plain error for two alternative reasons. It was not plain error because the jury instructions were not error, id. at 389–90, 119 S. Ct. at 2102, and also because the defendant had failed to carry his burden of satisfying the third prong of the plain error test by showing that the failure to give the additional instructions had prejudiced him, id. at 394–95, 119 S. Ct. at 2105. The Supreme Court’s decision in Jones that the defendant had failed to make the required showing of prejudice for plain error purposes provided as much support for the result in that case, which was affirmance of the sentence, as did its decision that there was no error. Either holding would have been adequate. Each is an alternative holding, and each alternative holding is binding. See Massachusetts v. United States, 333 U.S. 611, 623, 68 S. Ct. 747, 754 (1948); Richmond Screw Anchor Co. v. United States, 275 U.S. 331, 340, 48 S. Ct. 194, 196 (1928) (“It does not make a reason given for a conclusion in a case obiter dictum, because it is only one of 34 Judge Barkett apparently does not dispute that the rule stated in Jones would compel a finding that Rodriguez has failed to satisfy the third-prong of the plain error test. Instead, her position is that we need not and should not follow the Supreme Court’s decision in Jones. She gives several reasons, none of which are persuasive. Judge Barkett argues that what the Supreme Court said in Jones is “in direct conflict” with what it said later in United States v. Dominguez Benitez, 542 U.S. 74, 124 S. Ct. 2333 (2004). Barkett, J., dissenting, at 86. There is no conflict, direct or indirect. Even if there were, we would be required to follow the more specific commandment from the Jones case. All that Dominguez Benitez held is that the third-prong prejudice standard requires the defendant to show a reasonable probability of a different result, which means a probability sufficient to undermine confidence in the outcome of the proceeding. 542 U.S. at ___, 124 S. Ct. at 2338–40. The decision did not hold or say anything about what the result of applying that standard should be where there is no indication whether the error actually did have an adverse effect on the outcome of the proceeding. The Court two reasons for the same conclusion.”); United States v. Title Ins. & Trust Co., 265 U.S. 472, 486, 44 S. Ct. 621, 623 (1924); Union Pac. Ry. Co. v. Mason City & Fort Dodge Ry. Co., 199 U.S. 160, 166, 26 S. Ct. 19, 20 (1905); Johnson v. DeSoto County Bd. Comm’rs, 72 F.3d 1556, 1562 (11th Cir. 1996) (“we are bound by alternative holdings”); McLellan v. Mississippi Power & Light Co., 545 F.2d 919, 925 n.21 (5th Cir. 1977) (en banc) (“It has long been settled that all alternative rationales for a given result have precedential value.”). 35 remanded the Dominguez Benitez case to the court of appeals for application of the standard. Id. at ___, 124 S. Ct. at 2342. By contrast, in Jones this very issue was presented, addressed, and decided. The Supreme Court could not have been more specific in telling us that where the error just as likely could have worked in the defendant’s favor as against him, where the effect is indeterminate, where we simply cannot tell, the defendant has failed to carry his burden on the third prong of the plain error test. Dominguez Benitez states the general standard, and Jones speaks to a specific situation, the one before us in most of the pre-Booker sentence cases. The two decisions are not in conflict. Reduced to its essence, Judge Barkett’s theory for disregarding the direct application of Jones is that it was implicitly overruled by the later decision in Dominguez Benitez, a premise based on her belief that Jones rests on reasons rejected in Dominguez Benitez. The problem with her approach is that the Supreme Court has repeatedly told us not to take it. The Court has instructed us: “If a precedent of this Court has direct application in a case, yet appears to rest on reasons rejected in some other line of decisions, the Court of Appeals should follow the case which directly controls, leaving to this Court the prerogative of overruling its own decision.” Rodriguez de Quijas v. Shearson/Am. Express, Inc., 36 490 U.S. 477, 484, 109 S. Ct. 1917, 1921–22 (1989); see also Tenet v. Doe, 544 U.S. ___, ___, 125 S. Ct. 1230, 1237 (2005) (quoting Shearson/Am. Express, Inc., 490 U.S. at 484, 109 S. Ct. at 1921–22); Am. Trucking Ass’ns, Inc. v. Smith, 496 U.S. 167, 180, 110 S. Ct. 2323, 2332 (1990) (plurality op.) (same); Agostini v. Felton, 521 U.S. 203, 237, 117 S. Ct. 1997, 2017 (1997) (“We do not acknowledge, and we do not hold, that other courts should conclude our more recent cases have, by implication, overruled an earlier precedent.”); Hohn v. United States, 524 U.S. 236, 252–53, 118 S. Ct. 1969, 1978 (1998) (“Our decisions remain binding precedent until we see fit to reconsider them, regardless of whether subsequent cases have raised doubts about their continuing vitality.”). We have been careful to heed the Supreme Court’s admonition about following its decisions until that Court explicitly overrules them. See Fla. League of Prof’l Lobbyists, Inc. v. Meggs, 87 F.3d 457, 462 (11th Cir. 1996) (“We take this admonition to heart, and we decline to take any step which might appear to overrule [prior Supreme Court decisions].”); Eng’g Contractors Ass’n of S. Fla. Inc. v. Metro. Dade County, 122 F.3d 895, 908 (11th Cir. 1997) (“Of course, we take that admonition seriously.”); Scala v. City of Winter Park, 116 F.3d 1396, 1399 n.2 (11th Cir. 1997) (heeding the admonition and following an earlier Supreme Court decision even though later decisions of the Court had “cast some 37 doubt” on it); Brisentine v. Stone & Webster Eng’g Corp., 117 F.3d 519, 525 (11th Cir. 1997) (“It may be that the Supreme Court has cut [an earlier decision] back so far that it will not survive. Perhaps, but we are not convinced we are authorized to sing the dirge of [that earlier decision]. We will leave that to the Supreme Court, which has admonished courts of appeals [to do so].”). Likewise, the panel has heeded the Supreme Court’s admonition in this case. Failure to do so would not be reasoning, it would be rebellion. Judge Barkett also argues that our approach violates the Federal Sentencing Act’s direction that the district court should determine in the first instance the sentence that should be imposed. Barkett, J., dissenting, at 89. Not so. The district court has determined in the first instance the sentence in all pre-Booker sentence cases. The question now is not what the defendant’s sentence should be, but whether the defendant has carried his burden of establishing the prejudice required by the third prong of the plain error test. Judge Barkett’s argument that Williams v. United States, 503 U.S. 193, 205, 112 S. Ct. 1112, 1121 (1992), requires that we disregard the lack of a prejudice showing is also unpersuasive. The Williams decision involved preserved error. It did not purport to hold that the contemporaneous objection rule was inapplicable to sentencing errors. It did not say that the plain error rule was different for 38 sentencing cases. It did, however, note that the harmless error rule applies to sentence errors, id. at 203, 112 S. Ct. at 1120–21, which is inconsistent with the notion that sentencing errors require remand regardless of ordinary prudential doctrines.6 IV. Neither Judge Tjoflat nor Judge Barkett proposes that this Court adopt the Second, Seventh, and D.C. Circuits’ novel approach to third-prong prejudice determinations. See United States v. Crosby, 397 F.3d 103 (2d Cir. 2005); United States v. Williams, 399 F.3d 450 (2d Cir. 2005); United States v. Paladino, 401 F.3d 471 (7th Cir. 2005); United States v. Coles, ___ F.3d ___, 2005 WL 783069 (D.C. Cir. Apr. 8, 2005) (per curiam). Although unwilling to embrace that approach, Judge Tjoflat takes some comfort in the Seventh Circuit’s criticism of our application of conventional plain error analysis. See Tjoflat, J., dissenting, at 6 Judge Barkett argues in a footnote that this Court’s Booker plain error position is somehow inconsistent with a footnote in United States v. Fuente-Kolbenschlag, 878 F.2d 1377, 1379 n.7 (11th Cir. 1989) (per curiam). Barkett, J., dissenting, at 91 n.9. It is not. The defendant in that case had raised the alleged error by timely, specific objection during sentencing. The opinion in the case could not have decided, and did not purport to say anything about, what happens when there is a failure to object. At most Fuente-Kolbenschlag stands for the unremarkable proposition that in a case of preserved error the fact that overlapping ranges might have resulted in the same sentence being imposed anyway did not insulate the error from review on appeal. See Fuente-Kolbenschlag, 878 F.2d at 1379 n.7. That is another way of saying that if we don’t know what would have happened but for the error, the party with the burden on the prejudice issue loses. (In Fuente-Kolbenschlag that party was the government, but it won anyway because there was no error. Id. at 1378–79.) 39 63 (“Or, as the Seventh Circuit put it, I ‘cannot fathom why the Eleventh Circuit wants to condemn some unknown fraction of criminal defendants to serve an illegal sentence.’” (quoting Paladino, 401 F.3d at 484)). Only three of the eleven circuits that have taken a position on the proper way to decide whether plain error exists in pre-Booker situations have chosen the Crosby/Paladino model. What the Second, Seventh, and D.C. Circuits do is remand every pre- Booker sentence (where the defendant wants a remand) in which it is not clear that the unpreserved Booker error was harmless. Williams, 399 F.3d at 460 (“We remanded only after we determined that, on the record in Crosby, we could not say with sufficient confidence that the same sentence would have been imposed.”); Paladino, 401 F.3d at 484 (describing the court’s intention to remand in those “Booker cases in which it is difficult for us to determine whether the error was prejudicial”); Coles, 2005 WL 783069, at *1 (same). In the Second Circuit, if the district court determines on remand that it would have sentenced the defendant to less time under Booker, it is to pronounce that plain error exists, set aside the prior judgment, and proceed to conduct the real resentence proceeding. See Crosby, 397 F.3d at 117–18, 120; Williams, 399 F.3d at 459–61. But if the defendant fails to get the appellate relief he is seeking in the district court, he apparently is free to appeal to the real appellate court urging it to 40 reverse the district court’s third-prong plain error determination. And if the district court decides that it would have sentenced the defendant differently and does so, the government apparently can appeal that determination as well as any additional errors embedded in the new sentence. Crosby, 397 F.3d at 119. The Seventh Circuit’s procedure is similar, except that it formally retains jurisdiction of the case during the remand and will be the court that actually vacates the pre-Booker sentence following the district court’s determination that it would have reached a different result under the Booker regime. See Paladino, 401 F.3d at 484. Then the case will go back down for the actual resentencing. Id. While retaining for itself the responsibility to enter the order vacating the initial sentence after remand, it is only a pro forma reservation in the sense that the actual decision making will be vested entirely in the district court. See id. (“Under our procedure, since we retain jurisdiction throughout the limited remand, we shall vacate the sentence upon being notified by the judge that he would not have imposed it had he known that the guidelines were merely advisory.”). The D.C. Circuit’s procedure is identical to the Seventh’s. See Coles, 2005 WL 783069, at *7. The Crosby/Paladino model essentially delegates to the district court the appellate function of determining whether there is prejudice necessary for 41 correction of unpreserved error. There is no basis in any of the relevant Supreme Court decisions for that delegation, and every indication is that we should make the decision at the appellate level from the record as it exists. The language of those decisions, to the extent they touch on the matter, indicates what one would assume, which is that determining whether plain error exists is an appellate function to be decided from the record on appeal, not something to be decided by the district court after remand. See Dominguez Benitez, 542 U.S. at ___,124 S. Ct. at 2340 (“[Under plain error review, a] defendant must . . . satisfy the judgment of the reviewing court, informed by the entire record, that the probability of a different result is sufficient to undermine confidence in the outcome of the proceeding.” (internal quotation omitted)); United States v. Vonn, 535 U.S. 55, 59, 122 S. Ct. 1043, 1046 (2002) (“[A] silent defendant has the burden to satisfy the plain-error rule and . . . a reviewing court may consult the whole record when considering the effect of any error on substantial rights.”); see also Gonzalez- Huerta, 2005 WL 807008, at *3 n.4 (concluding that the Crosby/Paladino procedure is “inconsistent with plain error doctrine”). The emotional punch behind the Seventh Circuit’s criticism of our approach derives from its charge that we are “condemn[ing] some unknown fraction of criminal defendants to serve an illegal sentence.” Paladino, 401 F.3d at 484. That 42 charge defines “illegal sentence” as one in which a decision that would otherwise result in reversal is not applied in a case because of the contemporaneous objection rule. By equating enforcement of the contemporaneous objection rule, and the recognition that the plain error exception is “circumscribed,” Jones, 527 U.S. at 389, 119 S. Ct. at 2102, and should be exercised “sparingly,” id., with countenancing illegality, the charge proves too much. In every criminal case in which it matters that the contemporaneous objection rule was applied, and there have been thousands of cases where it has mattered, some defendant under this view will have been condemned to serve an “illegal sentence” or, worse still, will have been condemned to suffer an “illegal conviction.” If “illegality” is defined this way and the goal is to prevent “illegal” sentences and convictions, the only way to achieve that goal is to abolish the contemporaneous objection rule. Not just in Booker error cases but in all criminal cases. We also will have to abolish and repeal the procedural bar doctrine which a half century of decisional law and federal statutory development has put into place. After all, by applying that doctrine, as we do in countless cases, we are condemning some unknown fraction of criminals to serve “illegal” sentences or to suffer from “illegal” convictions. And what of the thousands of prisoners condemned to suffer sentences that violate the Sixth Amendment under Booker, 43 because their cases finished the direct review process before that decision was announced instead of afterwards. No circuit, including the Seventh, has yet to suggest that Booker is retroactively applicable to collateral proceedings, and in light of Schriro v. Summerlin, 542 U.S. ___, 124 S. Ct. 2519 (2004), it is highly unlikely that any will. The only way to rescue all of those prisoners from their “illegal sentences” is to throw out the Teague decision and its progeny. See Teague v. Lane, 489 U.S. 288, 109 S. Ct. 1060 (1989). The parts of the AEDPA that are found in 28 U.S.C. § 2255 ¶¶ 6 & 8 probably will have to go as well. Only when all of these decisional and statutory doctrines and the important values they serve are thrown on the trash heap will we be able to reduce the numerator of that unknown fraction of defendants suffering from “illegal” convictions or serving “illegal” sentences to zero. Our legal system does not simply require that the government comply with the Constitution. It also makes parties, defendants as well as the government, comply with procedural rules, such as the contemporaneous objection rule, on pain of forfeiting legal rights they could otherwise enforce. Requiring rights to be asserted in a timely and appropriate fashion furthers interests that are vital to the proper functioning of our judicial system. See Wainwright v. Sykes, 433 U.S. 72, 90, 97 S. Ct. 2497, 2508 (1977); United States v. Pielago, 135 F.3d 703, 709 44 (11th Cir. 1998); Esslinger v. Davis, 44 F.3d 1515, 1525 & n.36 (11th Cir. 1995); United States v. Sorondo, 845 F.2d 945, 948–49 (11th Cir. 1988). The narrowness of the plain error exception to the contemporaneous objection rule reflects the importance of those interests. Broadening that exception, or constructing ways to circumvent its restrictions on an issue-by-issue basis, lessens the effect of the rule and undermines the interests it serves. If the matter is to be addressed in “illegality” terms, then put it this way: Failure of a defendant to comply with clear procedural rules during the judicial process is itself a type of illegality that may block consideration of his claim that he has suffered an illegality. 45 TJOFLAT, Circuit Judge, dissenting from the denial of rehearing en banc: A jury in the United States District Court for the Middle District of Florida convicted Vladimir Rodriguez for distributing, or possessing with the intent to distribute, a “detectable amount” of MDMA, also known as ecstasy, in violation of 21 U.S.C. § 841(a)(1), and for conspiring to do the same, in violation of 21 U.S.C. § 846. At sentencing, the district court initially set Rodriguez’s base offense level at 30 based on its own determination that the offense involved 30,000 ecstasy tablets. It then added two levels because Rodriguez testified falsely under oath during his trial that he had no involvement in the offenses for which he was convicted, U.S.S.G. § 3C1.1, and awarded a two-level reduction based on Rodriguez’s minor role in the offense, U.S.S.G. § 3B1.2(b). Because Rodriguez had no prior convictions, his guideline sentencing range was 97 to 121 months in prison. The court imposed a sentence of 109 months in prison. After Rodriguez was sentenced, the Supreme Court held that the Federal Sentencing Guidelines violate the Sixth Amendment right to a trial by jury to the extent that they permit a judge to increase a defendant’s sentence based on facts that are neither found by the jury nor admitted by the defendant. United States v. Booker, __ U.S. __, 125 S. Ct. 738, 756, 160 L. Ed. 2d 621 (2005). To remedy this constitutional defect, the Court declared “the Guidelines effectively advisory.” 46 Id. at 757. As a result, a sentencing court must still “consider Guidelines ranges,” but it may “tailor the sentence in light of other statutory concerns as well.” Id. (citing 18 U.S.C. § 3553(a)). We review sentences imposed under this new model for “reasonableness” only. Id. at 765-67; see also McReynolds v. United States, 397 F.3d 479, 481 (7th Cir. 2005) (“District judges must [apply the Sentencing Guidelines] as guidelines, with appellate review to determine whether that task has been carried out reasonably.”). Because Rodriguez did not raise a constitutional objection at sentencing, the Booker claim he raises on appeal must pass the “‘plain-error’ test.” Booker, 125 S. Ct. at 769. In general, this requires the defendant to show that “(1) an error occurred, (2) the error was plain, (3) the error affected substantial rights, and (4) the error seriously affects the fairness, integrity or public reputation of judicial proceedings.” United States v. Duncan, 400 F.3d 1297, 1301 (11th Cir. 2005) (citing United States v. Olano, 507 U.S. 725, 732-36, 113 S. Ct. 1770, 1777-79, 123 L. Ed. 2d 508 (1993)). Under Booker, it is clear that error occurred, and that error is now plain. The panel holds, however, that Rodriguez fails the plain-error test because he cannot show that his substantial rights were affected. The panel does not suggest that Rodriguez’s substantial rights were not affected; rather, it rejects his claim because “[w]e just don’t know” what sentence he would have 47 received if the district court had treated the Guidelines as advisory only. United States v. Rodriguez, 398 F.3d 1291, 1301 (11th Cir. 2005). The panel’s error is its failure to recognize that Booker error of a constitutional dimension is one of a “very limited class” of “structural errors.” Johnson v. United States, 520 U.S. 461, 468-69, 117 S. Ct. 1544, 1549-50, 137 L. Ed. 2d 718 (1997).1 “Each of these constitutional deprivations is a . . . structural defect affecting the framework within which the trial proceeds, rather than simply an error in the trial process itself.” Arizona v. Fulminante, 499 U.S. 279, 310, 111 S. Ct. 1246, 1265, 113 L. Ed. 2d 302 (1991). Because structural error affects the framework of the trial itself, its “consequences . . . are necessarily unquantifiable and indeterminate.” Sullivan v. Louisiana, 508 U.S. 275, 281-82, 113 S. Ct. 2078, 2083, 124 L. Ed. 2d 182 (1993).2 As such, structural errors “defy analysis by ‘harmless-error’ standards” and are per se reversible if an objection is made at trial. Fulminante, 499 U.S. at 309, 111 S. Ct. at 1265. 1 This “very limited class” includes (1) a total deprivation of the right to counsel, (2) the lack of an impartial trial judge, (3) the unlawful exclusion of grand jurors of the defendant’s race, (4) the denial of the right to self-representation at trial, (5) the denial of the right to a public trial, and (6) an erroneous reasonable-doubt instruction. Johnson, 520 U.S. at 468-69, 117 S. Ct. at 1549-50. 2 In contrast, the Court has described ordinary “trial error” as “error which occurred during the presentation of the case to the jury, and which may therefore be quantitatively assessed in the context of other evidence presented in order to determine whether its admission was harmless beyond a reasonable doubt.” Fulminante, 499 U.S. at 307-08, 111 S. Ct. at 1264. 48 A harmless error is simply “[a]ny error . . . that does not affect substantial rights.” Fed. R. Crim. P. 52(a). As such, the Supreme Court has said that harmless-error analysis and the substantial-rights prong of the plain-error test essentially “require[] the same kind of inquiry.” Olano, 507 U.S. at 734, 113 S. Ct. at 1778 (stating that there is only “one important difference: It is the defendant rather than the Government who bears the burden of persuasion with respect to prejudice” under the plain-error test). Therefore, if structural errors “defy analysis” by harmless-error standards, they must also defy analysis under the third prong of the plain-error test. It would make no sense to say that the same prejudice inquiry that is impractical as harmless-error analysis is somehow practical under the plain-error test. Structural errors are not per se reversible in the plain-error context because a defendant who fails to object at trial must still satisfy the second and fourth prongs of the plain-error test. But it is clear that it makes no more sense to require a case-specific showing of prejudice under the plain-error test than it does under the harmless-error standard. A Booker error that involves an actual Sixth Amendment violation is a structural error. Booker dramatically alters the very “framework within which [sentencing] proceeds,” Fulminante, 499 U.S. at 310, 111 S. Ct. at 1265, and the effect of Booker error is “necessarily unquantifiable and indeterminate,” Sullivan, 49 508 U.S. at 281-82, 113 S. Ct. at 2083. Moreover, the logical implication of the concluding paragraph of the Booker remedial opinion is that “cases . . . involving a Sixth Amendment violation,” Booker, 125 S. Ct. at 769, are not subject to the harmless-error doctrine and, therefore, must involve structural error. See infra Part I. For these reasons, I disagree with the panel opinion in this case. The panel’s error is a serious one and warrants rehearing en banc. The Supreme Court instructed that “we must apply [Booker’s] holdings—both the Sixth Amendment holding and [its] remedial interpretation of the Sentencing Act—to all cases on direct review.” Booker, 125 S. Ct. at 769. If, however, we require defendants like Rodriguez to prove an effect on their substantial rights, applying Booker will be a meaningless formality in all but the rarest of cases subject to plain-error analysis.3 This is a sizable class of defendants. Until the Supreme Court’s decision last June in Blakely v. Washington, __ U.S. __, 124 S. Ct. 2531, 159 L. Ed. 2d 403 (2004), the constitutionality of the Federal Sentencing Guidelines was beyond question in this circuit and every other. See id. at 2547 n.1 (O’Connor, J., dissenting) (collecting cases); United States v. Sanchez, 269 F.3d 1250, 1262 (11th Cir. 2001) (en banc) (“Apprendi does not apply to judge-made determinations pursuant to the Sentencing Guidelines.”). Defendants sentenced 3 See infra note 13. 50 prior to Blakely had no reason to raise a constitutional objection to their sentence because such an objection would have appeared futile. As such, the number of cases impacted by the panel’s decision is significant. Because the panel’s application of the plain-error test is fundamentally flawed,4 and because this question is of substantial importance, I dissent from the denial of rehearing en banc. Part I of this opinion explains why Booker itself requires that we treat Booker errors of a constitutional dimension as structural errors. Part II briefly describes how dramatically Booker has altered federal sentencing and why this supports the conclusion that Booker errors involving an actual violation of the Sixth Amendment are structural. Part III explains how the plain-error test applies to structural errors. Part IV responds to Judge Carnes’s opinion concurring in the denial of rehearing en banc. Part V concludes. I. At the outset, it is important to understand that there are two different types of Booker error: “there is a constitutional error (based in the Sixth Amendment) when a judge enhances a sentence in a mandatory sentencing system based on facts not admitted by the defendant or proved to a jury beyond a reasonable 4 The First and Fifth Circuits have committed the same error. See United States v. Antonakopoulos, 399 F.3d 68, 77-80 (1st Cir. 2005); United States v. Mares, __ F.3d __, 2005 WL 503715, at *8-9 (5th Cir. Mar. 4, 2005). 51 doubt,” and there is “a statutory error (based in the severability principles) when a federal judge applied the guidelines as mandatory rather than advisory.” Douglas A. Berman, Sorting through the Circuit circus, Sentencing Law and Policy, at http://sentencing.typepad.com/sentencing_law_and_policy/2005/02/sorting_throu gh.html (Feb.14, 2005). “Notably, only some pre-Booker sentencings involved constitutional error, since not every pre-Booker guideline sentence depended upon judicial fact-finding. But every pre-Booker sentencing involved statutory error, since every pre-Booker guideline sentence was imposed based on the assumption that the guidelines were mandatory . . . .” Id. The instant case involves both constitutional and statutory error. See Rodriguez, 398 F.3d at 1298 (“Under a mandatory guidelines system, Rodriguez’s sentence was enhanced as a result of findings made by the judge that went beyond the facts admitted by the defendant or found by the jury.”). Therefore, the precise question presented is whether, in a case involving both constitutional and statutory error under Booker, the defendant must establish that his substantial rights were affected—that is, “a reasonable probability of a different result if the guidelines had been applied” as required by Booker (Rodriguez, 398 F.3d at 1301)—in order to pass the plain-error test. The distinction between Booker constitutional errors and Booker statutory errors is significant because the Booker Court indicated that the two must be 52 treated differently. In Booker, the Court stated that “in cases not involving a Sixth Amendment violation, whether resentencing is warranted or whether it will instead be sufficient to review a sentence for reasonableness may depend upon application of the harmless-error doctrine.” 125 S. Ct. at 769.5 If all Booker errors were ordinary trial errors, this statement would be superfluous because it is well-settled that the harmless-error doctrine applies to all ordinary trial errors—even constitutional ones. Chapman v. California, 386 U.S. 18, 22, 87 S. Ct. 824, 827, 17 L. Ed. 2d 705 (1967). Thus, there must be something more to this statement. The logical implication is that “cases . . . involving a Sixth Amendment violation” are not subject to harmless-error review. In other words, Booker constitutional error is structural error, for only structural errors defy harmless-error analysis. Fulminante, 499 U.S. at 306-10, 111 S. Ct. at 1263-65. Intuitively, it may seem odd to say that Booker constitutional errors are structural while Booker statutory errors are not. After all, the “framework within which [sentencing] proceeds,” id. at 310, 111 S. Ct. at 1265, has changed because the Guidelines are no longer mandatory, and this development applies fully even if the trial court made no extra-verdict (or extra-plea) findings of fact. Moreover, in 5 In the same paragraph, the Court also emphasized that “not . . . every sentence gives rise to a Sixth Amendment violation.” Booker, 125 S. Ct. at 769. 53 general, the effects of statutory errors are likely to be just as “unquantifiable and indeterminate,” Sullivan, 508 U.S. at 281-82, 113 S. Ct. at 2083, as the effects of constitutional errors. Even so, the distinction is valid. To begin with, it has clear support in our own precedent. In Sanchez, we specifically stated that “[t]here is no separate category of structural error apart from constitutional error. The only question is whether any constitutional errors . . . rise to the level of structural error.” 269 F.3d at 1272 n.41.6 This statement from Sanchez finds implicit 6 This question whether non-constitutional errors can ever be structural errors has engendered much disagreement within several other circuits. Compare United States v. Curbelo, 343 F.3d 273, 280 n.6 (4th Cir. 2003) (“Despite occasionally suggesting in dicta that structural errors must implicate constitutional rights, the Supreme Court has clearly held that structural errors need not be of constitutional dimension.” (citation omitted)), with id. at 289 (Wilkins, C.J., dissenting) (“The Supreme Court and this court have repeatedly made clear that structural errors necessarily must affect a defendant’s constitutional rights.”). Compare Green v. United States, 262 F.3d 715, 717-79 (8th Cir. 2001) (holding that the denial of a federal prisoner’s statutory right to counsel in collateral proceedings is structural and therefore not subject to harmless-error analysis), with id. at 719 (Bye, J., dissenting) (“Structural errors appear to be confined to the constitutional sphere because Congress has mandated the application of harmless error review by statute. Presumably, only grave constitutional errors could surmount the statutory default rule that harmless error analysis applies.” (citation omitted)). Compare United States v. Annigoni, 96 F.3d 1132, 1144 (9th Cir. 1996) (“Although we conclude that the erroneous denial of the right of peremptory challenge is not amenable to harmless-error analysis, we need not decide whether such error rises to the level of structural error. The error in this case—the erroneous denial of a right of peremptory challenge—is simply not amenable to harmless-error analysis.”), with id. at 1150 (Kozinski, J., dissenting) (“The right to a certain number of peremptory strikes—or to any at all—is not guaranteed by the Constitution . . . . Since the Court has allowed for the possibility of harmless error even when important constitutional rights are violated, I find it hard to believe the Court would now conclude that it’s always reversible error to deny a defendant a mere statutory right.” (citations omitted)). In Sanchez, we appear to have answered this question in the negative. But see McGriff v. Dep’t of Corr., 338 F.3d 1231, 1235 (11th Cir. 2003) (holding that failure to appoint counsel as required by the rules governing proceedings under 28 U.S.C. § 2254 is “structural error”); Shepherd v. United States, 253 F.3d 585, 588 (11th Cir. 2001) (holding that failure to appoint counsel as required by the rules governing proceedings under 28 U.S.C. § 2255 “is not subject to harmless error analysis”). In any event, we would not need to address this 54 support in a number of Supreme Court cases.7 It also seems (at least) reasonable to say that if a defendant’s Sixth Amendment right to a trial by jury was not actually violated, we should not presume prejudice. In any event, it is enough to say that the distinction is one that Booker itself implies. II. Booker’s implication that the constitutional error it corrects is structural is fully consistent with the Supreme Court’s structural error cases. As noted above, Fulminante defines structural error as a “constitutional deprivation[] . . . affecting the framework within which the trial proceeds, rather than simply an error in the broad issue to decide this case. It would be enough to say that Booker constitutional error is structural, but Booker statutory error is not. 7 See, e.g., Neder v. United States, 527 U.S. 1, 7, 119 S. Ct. 1827, 1833, 144 L. Ed. 2d 35 (1999) (“Although [the harmless error rule] by its terms applies to all errors . . . , we have recognized a limited class of fundamental constitutional errors that ‘defy analysis by ‘harmless error’ standards.’” (emphasis added) (quoting Fulminante, 499 U.S. at 309, 111 S. Ct. at 1265)); id. at 14, 119 S. Ct. at 1836 (“Under our cases, a constitutional error is either structural or it is not.” (emphasis added)); Sullivan, 508 U.S. at 282, 113 S. Ct. at 2083 (Rehnquist, C.J., concurring) (“In [Fulminante], we divided the class of constitutional violations that may occur during the course of a criminal proceeding, be it at trial or sentencing, into two categories: [trial errors and structural errors].” (emphasis added)); Brecht v. Abrahamson, 507 U.S. 619, 629-30, 113 S. Ct. 1710, 1717, 123 L. Ed. 2d 353 (1993) (discussing trial error and structural error as part of the “spectrum of constitutional errors”); Fulminante, 499 U.S. at 310, 111 S. Ct. at 1265 (“Since our decision in Chapman, other cases have added to the category of constitutional errors which are not subject to harmless error . . . . Each of these constitutional deprivations is a similar structural defect . . . . ” (emphasis added)); Chapman, 386 U.S. at 23, 87 S. Ct. at 827-28 (“Although our prior cases have indicated that there are some constitutional rights so basic to a fair trial that their infraction can never be treated as harmless error, this statement . . . itself belies any belief that all trial errors which violate the Constitution automatically call for reversal.” (emphasis added and footnote omitted)). 55 trial process itself.” 499 U.S. at 310, 111 S. Ct. at 1265. It is difficult to overstate the extent to which Booker affects the federal sentencing framework. The starting point under both the old model (pre-Booker) and new model (post-Booker) of federal sentencing is 18 U.S.C. § 3553(a), which states that [t]he court shall impose a sentence sufficient, but not greater than necessary, . . . (A) to reflect the seriousness of the offense, to promote respect for law, and to provide just punishment for the offense; (B) to afford adequate deterrence to criminal conduct; (C) to protect the public from further crimes of the defendant; and (D) to provide the defendant with needed educational or vocational training, medical care, or other correctional treatment in the most effective manner[.] 18 U.S.C. § 3553(a)(2). In other words, the underlying goals of the statute and the Guidelines are “retribution, general deterrence, incapacitation, and rehabilitation.” United States v. Mogel, 956 F.2d 1555, 1558 n.2 (11th Cir. 1992). In selecting an appropriate sentence, the sentencing court must also consider “the nature and circumstances of the offense and the history and characteristics of the defendant”; “the kinds of sentences available”; all relevant guidelines and policy statements issued by the Sentencing Commission; “the need to avoid unwarranted sentence disparities among defendants with similar records who have been found guilty of similar conduct; and the need to provide restitution to any victims of the offense.” 18 U.S.C. § 3553(a)(1), (3)-(7). Under the old model, the district court was bound to sentence the defendant 56 within the applicable guideline range unless it determined that an aggravating or mitigating circumstance of a kind, or to a degree, not adequately taken into consideration by the Sentencing Commission in formulating the guidelines . . . should result in a sentence different from that described. In determining whether a circumstance was adequately taken into consideration, the court [could] consider only the sentencing guidelines, policy statements, and official commentary of the Sentencing Commission. 18 U.S.C. § 3553(b)(1). Stated differently, the only way the defendant was going to get a downward departure under the old model was by showing that his case was not within the “heartland of typical cases.” Koon v. United States, 518 U.S. 81, 94, 116 S. Ct. 2035, 2044, 135 L. Ed. 2d 392 (1996). This was a difficult task. To start with, “the Sentencing Commission has already considered, and the Sentencing Guidelines have already factored in, many if not all circumstances that are arguably relevant to criminal sentencing . . . . The Guidelines are, as Congress intended them to be, comprehensive . . . .” Kate Stith & Jose A. Cabranes, Fear of Judging: Sentencing Guidelines in the Federal Courts 102 (1998). Thus, it was quite unlikely that the defendant would be able to identify a circumstance that the Commission had altogether failed to consider.8 And once the sentencing court 8 Professor Stith and Judge Cabranes further observe that “with respect to commonly occurring circumstances that are not explicitly addressed by the Guidelines,” courts generally “assume[] that the Commission has already taken the matter into account.” Stith & Cabranes, supra, at 102. That is, if the circumstance is not rare, the Commission must be aware of it and therefore its omission must mean that the Commission does not think it significant to sentencing policy. Thus, “[t]he real question is not whether the Commission actually took some factor into 57 determined that the Commission had taken a circumstance into account, the defendant could only argue that the circumstance was present in his case to some atypical degree. This is because the Commission’s substantive judgment as to the significance of the circumstance in a heartland case was unassailable. The comprehensiveness of the Guidelines and the invulnerability of the policy judgments on which they rest essentially rendered any evidence a defendant might present regarding the need for “adequate deterrence” or the need for “just punishment” irrelevant under the old model. The reason is simply that although reasonable minds may differ as to the level of punishment needed to adequately deter or justly punish a particular offense, once the Commission announced its views on the subject, it was difficult to argue that any particular case was atypical or outside the heartland with respect to these issues.9 Rather, a party seeking a account, but whether the factor is rare enough to overcome a presumption that the Commission has taken it into account.” Id. at 102-03 (citing Koon, 518 U.S. at 110-11, 116 S. Ct. at 2052). 9 See, e.g., United States v. Davis, 98 F.3d 141, 145 (4th Cir. 1996) (“Davis argues that a sentence of less than the 30 years . . . would provide just punishment . . . . Davis’ claim . . . that less than 30 years would be sufficient is misplaced. The commentary to the sentencing guidelines states: ‘dissatisfaction with the available sentencing range or a preference for a different sentence than that authorized by the guidelines is not an appropriate basis for a sentence outside the applicable guideline range.’” (citation omitted)); United States v. Frazier, 979 F.2d 1227, 1231 (7th Cir. 1992) (“[District courts may not] depart based on their perception of a lack of a need for general deterrence. ‘It would be difficult to imagine a finding that the Sentencing Commission failed to adequately consider the general deterrent effect of the criminal law. . . . District courts must justify their departures by reference to factors particular to the defendant that the Guidelines inadequately considered.’ [United States v. Thomas, 906 F.2d 323, 327 (7th Cir.1990)] (emphasis added).”). 58 departure generally accepted his base offense level as a starting point and then attempted to show that the case was atypical because it was committed in an unusual manner or for some unanticipated and unaccounted-for reason warranted a downward departure. Because the Guidelines are merely advisory under the new model, the defendant is no longer limited to arguing that his case is somehow atypical. Such arguments, of course, remain viable, but under the new model the defendant can also simply concede that his case is typical and challenge the wisdom of the Commission’s judgment regarding the appropriate punishment in heartland cases.10 That is, the defendant may simply argue that the applicable guideline 10 See, e.g., United States v. Jaber, __ F. Supp.2d __, 2005 WL 605787, at *3-9, 13-15 (D. Mass. March 16, 2005); United States v. Carvajal, 2005 WL 476125, at *5-6 (S.D.N.Y. Feb. 22, 2005); United States v. Biheiri, 356 F. Supp. 2d 589, 594 n.6 (E.D. Va. 2005) (“No individual [§ 3553(a)] factor is singled out as having greater weight; instead, the richness of factual diversity in cases calls on sentencing judges to consider all of the factors and to accord each factor the weight it deserves under the circumstances. Thus, the Guidelines sentencing range is not entitled to ‘heavy weight,’ but it is a useful starting point in fashioning a just and appropriate sentence.”); United States v. Nellum, 2005 WL 300073, at *4 (N.D. Ind. Feb. 3, 2005) (“The defendant is also an Army veteran, who was honorably discharged. Under the guidelines, . . . military service is not ordinarily relevant in arriving at an appropriate sentence. Yet, this Court finds it very relevant that a defendant honorably served his country when considering his history and characteristics. See § 3553(a)(1).” (citation omitted)); United States v. Galvez-Barrios, 355 F. Supp. 2d 958 (E.D. Wis. 2005); United States v. Huerta-Rodriguez, 355 F. Supp. 2d 1019, 1025-28, 1029-30 (D. Neb. 2005); United States v. Myers, 353 F. Supp. 2d 1026, 1027-32 (S.D. Iowa 2005); United States v. Jones, 352 F. Supp. 2d 22, 25-26 (D. Me. 2005); United States v. Ranum, 352 F. Supp. 2d 984, 987 (E.D. Wis. 2005) (“The guidelines are not binding, and courts need not justify a sentence outside of them by citing factors that take the case outside the ‘heartland.’ Rather, courts are free to disagree, in individual cases and in the exercise of discretion, with the actual range proposed by the guidelines, so long as the ultimate sentence is reasonable and carefully supported by reasons tied to the § 3553(a) factors.”). 59 sentence is “greater than necessary” to achieve the purposes of the Sentencing Reform Act. 18 U.S.C. § 3553(a). “[T]he Guidelines . . . are the product of policy decisions by the Sentencing Commission . . . . If those policy decisions are no longer mandatory, the sentencing judge is free to disagree with them.” Booker, 125 S. Ct. at 790 n.3 (Scalia, J., dissenting).11 11 Courts may be particularly likely to disagree with guideline ranges deriving from base offense levels tied to drug quantities or dollar amounts, which have long been the subject of criticism. Post-Booker, several district courts have already indicated as much. For example, one district court has announced, [T]he Guidelines are based in part on statistical analyses of pre-Guidelines sentencing practices. Based on those statistics, the Commission established the offense levels for each crime, linked to a recommended imprisonment range. Accordingly, in many cases, the Guidelines represent a reasonable estimation of a fair sentencing range. However, for policy reasons and because statutory mandatory minima dictated many terms of the drug-trafficking guidelines, the Commission departed from past practices in setting offense levels for such crimes as fraud and drug trafficking. Consequently, and based also on its own experience and familiarity with state court sentencing, the court finds Guideline ranges of imprisonment for those crimes are less reliable appraisals of fair sentences. .... The Guidelines’ quantity-driven, “market-oriented” approach also contributes to disparity and unreliability in drug sentencing. The quantity system was developed to punish bigger distributors more harshly, but practices of charging conspiracies over a long period of time has the result of aggregating many small distributions so as to make a long-term small quantity distributor look like a large-quantity distributor. For example, a distributor responsible for selling one gram at a time a hundred times gets the same sentence as the dealer caught selling 100 grams only once. Huerta-Rodriguez, 355 F. Supp. 2d at 1025-26 & n.6 (citations omitted). Similarly, another district judge has observed, [I]t is worth pointing out that under the guidelines, the weight of the narcotics is the driving force behind the sentence. The government is well aware that for every controlled buy that is made, the quantity of drugs is increased, and so is the sentence. There is a randomness to this in the following sense: [the defendant’s] guideline range would have been significantly decreased if he was arrested after 60 Subject to review for reasonableness, district judges are now free to apply their “own perceptions of just punishment, deterrence, and protection of the public even when these differ from the perceptions of the Commission members who drew up the Guidelines.” Id. at 790. Although “judges must still consider the sentencing range contained in the Guidelines, . . . that range is now nothing more than a suggestion that may or may not be persuasive . . . when weighed against the numerous other considerations listed in [§ 3553(a)].” Id. at 787 (Stevens, J., dissenting). Indeed, as one district judge has already observed, the remedial majority in Booker[] direct[s] courts to consider all of the § 3353(a) factors, many of which the guidelines either reject or the first [undercover] buy . . . and not after the fourth buy. Under the guidelines, this fortuity increased [his] guidelines range from 87-108 months . . . to 168-210 months. Indeed, on the other hand, if the officers wanted to, they probably could have made additional controlled buys from [the defendant] and the total weight of the drugs attributable to him would have been even higher and so too his sentence under the guidelines. It is . . . difficult to ignore the random nature of how the system plays out in reality. Nellum, 2005 WL 30073, at *5; see also Jaber, 2005 WL 605787, at *12 (“Drug quantity may well be a kind of accident, depending on the fortuities of law enforcement or even the market, as much as it reflects the defendant’s culpability.”). The district judge in Nellum also noted that the guideline sentencing disparity between crack cocaine and powder cocaine was likely to be a controversial and recurring issue in sentencing hearings conducted under the new model, although he concluded that he could impose an appropriate sentence without addressing it directly in that case. Nellum, 2005 WL 30073, at *4; cf. United States v. Wilson, 350 F. Supp. 2d 910, 918 (D. Utah 2005) (arguing that the Guidelines generally reflect public opinion, with one notable exception being “the Guidelines’ differentially harsh treatment of distribution of crack cocaine (as compared to powder cocaine)”). Recently, in United States v. Smith, __ F. Supp. 2d __, 2005 WL 549057 (E.D. Wis. March 3, 2005), another district court imposed a lower sentence than that called for by the Guidelines to avoid an “unwarranted disparity between defendants convicted of possessing powder cocaine and defendants convicted of possessing crack cocaine.” Id. at *9. 61 ignore. For example, under § 3553(a)(1) a sentencing court must consider the “history and characteristics of the defendant.” But under the guidelines, courts are generally forbidden to consider the defendant’s age, his education and vocational skills, his mental and emotional condition, his physical condition including drug or alcohol dependence, his employment record, his family ties and responsibilities, his socio-economic status, his civic and military contributions, and his lack of guidance as a youth. The guidelines’ prohibition of considering these factors cannot be squared with the § 3553(a)(1) requirement that the court evaluate the “history and characteristics” of the defendant. United States v. Ranum, 353 F. Supp. 2d 984, 986 (E.D. Wis. 2005) (citations omitted). Thus, mitigating circumstances and substantive policy arguments that were formerly irrelevant in all but the most unusual cases12 are now potentially relevant in every case. That constitutional Booker errors are “constitutional deprivations . . . affecting the framework within which [sentencing] proceeds,” Fulminante, 499 12 See, e.g., United States v. Lacy, 99 F. Supp. 2d 108, 112 (D. Mass. 2000) (footnotes omitted): [W]hile the Guidelines’ emphasis on quantity and criminal history drives these high sentences, sadly, other factors, which I believe bear directly on culpability, hardly count at all: Profound drug addiction, sometimes dating from extremely young ages, the fact that the offender was subject to serious child abuse, or abandoned by one parent or the other, little or no education. Nor may I consider the fact that the disarray so clear in the lives of many of these defendants appears to be repeating itself in the next generation: Many have had children at a young age, and repeat the volatile relationships with their girlfriends that their parents may have had. And I surely cannot evaluate the extent to which lengthy incarceration will exacerbate the problem, separating the defendant from whatever family relationships he may have, or the impact on communities when these young men return. 62 U.S. at 310, 111 S. Ct. at 1265, is thus inescapable. It is equally clear that their “consequences . . . are necessarily unquantifiable and indeterminate.” Sullivan, 508 U.S. at 282, 113 S. Ct. at 2083. Indeed, the panel recognizes this when it “ask[ed] whether there is a reasonable probability of a different result if the guidelines had been applied in an advisory instead of binding fashion”: The obvious answer is that we don’t know. If the district court judge in this case had the liberty of increasing or decreasing Rodriguez’s sentence above or below the guidelines range, he might have given Rodriguez a longer sentence, or he might have given a shorter sentence, or he might have given the same sentence. The record provides no reason to believe any result is more likely than the other. We just don’t know. Rodriguez, 398 F.3d at 1301. I do not see how we can dismiss Rodriguez’s Sixth Amendment claim on the ground that “[t]he record provides no reason to believe” that he would have received a lesser sentence under the new model. Or, as the Seventh Circuit put it, I “cannot fathom why the Eleventh Circuit wants to condemn some unknown fraction of criminal defendants to serve an illegal sentence.” United States v. Paladino, 401 F.3d 471, 484 (7th Cir. 2005). Before adopting such a rule, we should at least consider what in the record might have “provide[d] . . . reason to believe” that a lesser sentence would have been imposed had Rodriguez been sentenced under the new model. At the sentencing hearing, the defendant might have argued that his guideline sentence 63 was “greater than necessary” to achieve the sentencing purposes identified in § 3553(a). More specifically, the defendant might have presented the district judge with law-enforcement data tending to show that the need for general deterrence is low with respect to his particular offense and in his particular community, and that his guideline range is therefore unnecessarily high. Alternatively, the defendant might have argued that his personal history or circumstances somehow render him less culpable or less likely to commit future crimes than the Guidelines suggest. Of course, none of this type of evidence or argument is likely to be found in the record of a pre-Booker hearing.13 Any attempt to assess prejudice resulting from Booker error is, therefore, pure guesswork. In cases involving statutory error only, it is necessary guesswork, for the Supreme Court has clearly indicated that the harmless-error and plain-error doctrines apply to such cases. Again, this is reasonable given that these 13 Alternatively, if “the district court during sentencing expressed several times its view that the sentence required by the Guidelines was too severe,” stated that it was “unfortunate[]” that the Guidelines overstated the seriousness of the defendant’s criminal history, imposed the lowest possible sentence under the Guidelines, and remarked that even that sentence was “more than appropriate,” then the defendant will satisfy the Rodriguez standard. United States v. Shelton, 400 F.3d 1325, 1331-33 (11th Cir. 2005). Shelton involved only statutory error, so I agree with the panel that a prejudice inquiry was necessary. I also agree that the defendant established prejudice. Indeed, I do not understand why the court felt the need to emphasize that “[a]ll of these comments taken together convince[d] [it] that there [was] a reasonable probability the district court would have imposed a lesser sentence . . . if it had not felt bound by the Guidelines.” Id. at 1332-33. I should think that any one of these circumstances standing alone would be sufficient. 64 defendants have suffered no actual constitutional deprivation. But when, as in this case, the Booker error is constitutional, the fact that its effect is wholly “unquantifiable and indeterminate,” Sullivan, 508 U.S. at 282, 113 S. Ct. at 2083, reinforces its structural nature. In this sense, Booker error is similar to other types of error that fit within the very limited class of structural errors. See supra note 1 (listing structural errors). It is, for example, similar to Gideon error in that just as we cannot know what evidence a capable attorney might have uncovered, what favorable testimony he might have elicited, or what persuasive arguments he might have made, we also cannot know what evidence or argument a defendant might have presented had he been sentenced under the new sentencing model. It is also similar to an erroneous reasonable-doubt instruction, for just as such a misinstruction “vitiates all the jury’s findings” so that an appellate court can do no more than “engage in pure speculation—its view of what a reasonable jury would have done,” Sullivan, 508 U.S. at 281, 113 S. Ct. at 2082—all we can do is speculate as to what a reasonable sentencing judge might have done under the new model. Indeed, we are in one sense worse off because a court assessing an erroneous reasonable-doubt instruction can at least assume that the remaining record was not affected by the instruction error, whereas I think we must assume, if anything, that a defendant such as Rodriguez would have done something 65 different under the new model. I am thus firmly convinced that Booker constitutional error is structural error. In the next Part, I explain the significance of this conclusion in the plain- error context. III. It is clear that structural errors are not subject to harmless-error analysis and therefore always require reversal if a timely objection is made. Fulminante, 499 U.S. at 309-10, 111 S. Ct. at 1264-65. Rodriguez did not raise a constitutional objection to his sentence, however, so the question here is how structural errors are to be treated under the plain-error test.14 The Supreme Court has never squarely addressed this issue, but on several occasions it has suggested that structural errors are not subject to the substantial-rights prong of the plain-error test. See United States v. Dominguez-Benitez, 542 U.S. 74, __, 124 S. Ct. 2333, 2339-40, 159 L. Ed. 2d 157 (2004); United States v. Cotton, 535 U.S. 625, 632-33, 122 S. Ct. 1781, 1785-86, 152 L. Ed. 2d 860 (2002); Johnson, 520 U.S. at 468-69, 117 S. Ct. at 1549-50; Olano, 507 U.S. at 736, 113 S. Ct. at 1778. Similar 14 Rodriguez did, however, object to the district court’s calculation of the number of ecstasy tablets involved in the offense. The Eighth Circuit “has held that when a defendant objects to a District Court’s determination of drug quantity at sentencing, the defendant preserves a Booker-based challenge to his sentence and is entitled to a new sentencing proceeding.” United States v. Sdoulam, 398 F.3d 981, 995 (8th Cir. 2005). 66 statements can be found in a number of our opinions, see, e.g., McCoy v. United States, 266 F.3d 1245, 1252 n.9 (11th Cir. 2001) (“Apprendi error is a constitutional error, subject to plain- or harmless-error review, and does not create a structural error.”); United States v. Smith, 240 F.3d 927, 930 (11th Cir. 2001) (“[F]ailure to submit the issue of drug quantity to the jury did not affect Defendants’ substantial rights. Apprendi did not create a structural error that would require per se reversal.”); United States v. Frost, 139 F.3d 856, 859 (11th Cir. 1998) (discussing Johnson, supra), as well as those of other circuits, see, e.g., United States v. Bruno, 383 F.3d 65, 79 (2d Cir. 2004); United States v. Adams, 252 F.3d 276, 285 & n.6 (3d Cir. 2001); United States v. Wilson, 240 F.3d 39, 44 (D.C. Cir. 2001). Moreover, those circuits that have squarely addressed this issue have held that structural errors are not subject to substantial-rights analysis under the plain-error test—or, stated differently, are presumed to have affected the defendant’s substantial rights. See United States v. Recio, 371 F.3d 1093, 1103 (9th Cir. 2004); United States v. David, 83 F.3d 638, 646-47 (4th Cir. 1996). That structural errors should not be subject to substantial-rights analysis seems obvious to me. Indeed, Federal Rule of Criminal Procedure 52, read in light of the Supreme Court’s structural error cases, compels this conclusion. Rule 52(a) defines harmless error as “[a]ny error . . . that does not affect substantial rights.” 67 And Rule 52(b) says that we may correct “[a] plain error that affects substantial rights.” Thus, when the Supreme Court said that structural errors “defy analysis by ‘harmless-error’ standards,” Fulminante, 499 U.S. at 309, 111 S. Ct. at 1265, it meant that structural errors “defy” substantial-rights analysis, for substantial-rights analysis and harmless-error analysis are one in the same. Olano, 507 U.S. at 734, 113 S. Ct. at 1778 (stating that the only difference between the two is that the defendant bears the burden of persuasion under the former and the Government under the latter). I cannot imagine how an error would “defy” substantial-rights analysis under Rule 52(a) but not under Rule 52(b). Therefore, structural errors cannot be subject to the third prong of the plain-error test. This does not mean that structural errors are per se reversible under the plain-error standard, see Johnson, 520 U.S. at 466, 117 S. Ct. at 1548; Recio, 371 F.3d at 1100 n.4; David, 83 F.3d at 647-48, but it does mean that we cannot affirm simply because the defendant is unable show that his substantial rights were affected. Because structural errors are not subject to substantial-rights analysis, the panel’s reliance on Jones v. United States, 527 U.S. 373, 119 S. Ct. 2090, 144 L. Ed. 2d 370 (1999), is misplaced. In Jones, the defendant argued that the jury was erroneously “led to believe that if it could not reach a unanimous sentence recommendation he would receive a judge-imposed sentence less severe than life 68 imprisonment.” Id. at 384, 119 S. Ct. at 2100. The Court first held that no error was committed. Id. at 389-90, 119 S. Ct. at 2102. It then explained that “[e]ven assuming, arguendo, that an error occurred (and that it was plain),” and further assuming that the jurors were confused over the consequences of deadlock, petitioner cannot show the confusion necessarily worked to his detriment. It is just as likely that the jurors, loath to recommend a lesser sentence, would have compromised on a sentence of life imprisonment as on a death sentence. Where the effect of an alleged error is so uncertain, a defendant cannot meet his burden of showing that the error actually affected his substantial rights. Id. at 394-95, 119 S. Ct. at 2105. The dissent in Jones responded only that such uncertainty should satisfy the substantial-rights prong of the plain-error test: “I would demur . . . to that position. It should suffice that the potential to confuse existed, i.e., that the instructions could have tilted the jury toward death.” Id. at 416, 119 S. Ct. at 2115-16 (Ginsburg, J., dissenting). In other words, the dissent argued only that the defendant had made a sufficient showing that the error affected his substantial rights, not that he was not required to do so. Neither the majority nor the dissent ever mentioned the concept of “structural error.” Nor, for that matter, did any of the parties in Jones, either in the briefs or at oral argument. It is clear that Jones did not involve structural error and, therefore, does not resolve this case. Instead, Jones announces a general rule of thumb that the defendant loses if we can only speculate as to whether he was prejudiced by an 69 ordinary trial error to which did not object. Jones would not apply if, for example, the judge forgot to give a reasonable-doubt instruction and the defendant did not object; that is, we could not say, “Well, the defendant did not object, and we just don’t know what, if any, prejudice this omission caused him, so he fails the plain- error test.” Because Booker constitutional error is also structural error, Jones does not apply here either. IV. As Judge Carnes says, no federal court has explicitly held, as I would, that Booker constitutional error is structural error. Ante, at 6. While technically accurate, this statement is at least a little misleading. After all, the Third, Fourth, and Sixth Circuits appear to think that every Booker constitutional error affects substantial rights, which is really all that calling such error “structural” means in the plain-error context. See, e.g., United States v. Spivey, 2005 WL 647345, at *5 (3d Cir. Mar. 22, 2005); United States v. Hughes, __ F.3d __, 2005 WL 628224, at *5-12 (4th Cir. Mar. 16, 2005); United States v. Oliver, 397 F.3d 369, 379-80 (6th Cir. 2005). The Ninth Circuit also took this position initially, though it has since granted rehearing en banc. United States v. Ameline, 400 F.3d 646, 654 (9th Cir. 2005), reh’g en banc granted, __ F.3d __, 2005 WL 612710 (9th Cir. Mar. 11, 2005). Indeed, the only real difference between these opinions and mine is that 70 mine offers a more satisfactory rationale for its result. See infra note 19. In addition, the Eighth is still yet to weigh in on this issue with a significant opinion, and the Tenth Circuit has addressed only statutory Booker error. See United States v. Gonzalez-Huerta, __ F.3d __, 2005 WL 807008, at *3 (10th Cir. Apr. 8, 2005) (en banc) (“This case presents us with a non-constitutional Booker error.”). Thus, I do not believe that my opinion is nearly so radical as Judge Carnes suggests. In Booker, the Court instructed that “in cases not involving a Sixth Amendment violation, whether resentencing is warranted or whether it will instead be sufficient to review a sentence for reasonableness may depend upon application of the harmless-error doctrine.” Booker, 125 S. Ct. at 769 (emphasis added). Judge Carnes fairly summarizes what I infer from this instruction: 1) the harmless- error doctrine does not apply to “cases . . . involving a Sixth Amendment violation”; 2) ergo, Booker errors “involving a Sixth Amendment violation”—i.e., constitutional Booker errors—must be structural errors, for only structural errors defy harmless-error analysis; and, 3) ergo, all constitutional Booker errors affect substantial rights for plain-error purposes. See ante, at 8-9. Judge Carnes says, “Never has so much been inferred from so little.” Id. Propositions 2 and 3 in this chain are, however, relatively uncontroversial, and I do not understand Judge 71 Carnes to disagree with them—that is, it is clear that only structural errors defy harmless-error analysis and that structural errors always affect substantial rights. As such, Judge Carnes really only takes issue with proposition 1. I believe that I make a sound case in Part I that the Court’s statement regarding cases not involving Sixth Amendment error would be superfluous if it does not mean what I conclude it means, but my main point here is that the extent of my inference should not be overstated: propositions 2 and 3 follow logically if proposition 1 is accepted. Judge Carnes then argues that “[t]here are all kinds of problems with [my] theory.” Ante, at 9. He says that it contradicts the Booker Court’s expectation that we “apply ordinary prudential doctrines, determining, for example, whether the issue was raised below and whether it fails the ‘plain-error’ test.” Booker, 125 S. Ct. at 769. This simply is not so. In cases involving only statutory error, Judge Carnes and I would, I think, apply the plain-error test in much the same way. And even in cases involving constitutional error, I would require the defendant to satisfy the fourth prong of the test.15 Thus, my approach is in no sense in conflict 15 Judge Carnes says that my approach to the plain-error test will result in all sentences involving constitutional error being vacated. This is so, he says, because “Shelton effectively adds that where the third prong of the plain error test is met in [Booker] cases, the fourth one will be also.” Ante, at 3-4 (citing Shelton, 400 F.3d at 1333-34). That is not how I read Shelton. If that case does, in fact, hold that the third and fourth prongs of the test always go hand-in-hand in Booker cases, it certainly does so subtly. The passage Judge Carnes cites states: 72 A plain error affecting substantial rights does not, without more, satisfy the plain-error test, for otherwise the fourth prong and the discretion afforded by the fourth prong would be illusory. We conclude that the fourth prong is established here and that an exercise of our discretion is warranted in this particular case. The district court in this case indicated an express desire to impose a sentence lesser than the low end of the Guidelines range of 130 months’ imprisonment, and the Supreme Court in Booker plainly indicated that the district court now has the discretion to do so, provided the resulting sentence is reasonable in light of the § 3553(a) factors. Under these circumstances, defendant Shelton has carried his burden to establish the fourth prong and has shown that the plain error that affected his substantial rights also seriously affected the fairness, integrity or public reputation of the judicial proceedings in his particular case. Accordingly, we vacate Shelton’s sentence and remand for resentencing consistent with Booker. Shelton, 400 F.3d at 1333-34. I do expect that in most cases the defendant will also satisfy the fourth prong of the plain- error test if he is able to satisfy the third prong. But if, for example, the Government could show that the district court expressed a strong desire to impose a more severe sentence than the Guidelines permitted because of circumstances not easily countered by mitigating evidence, declining to correct the plain error would not lead to a “miscarriage of justice.” Of course, under the panel’s approach, such a defendant would never get past the third prong of the plain-error test anyway. As such, part of the reason Judge Carnes thinks that my approach would mean that all sentences involving constitutional error would satisfy the plain-error test is a direct result of an aspect of the panel opinion with which I disagree and would not follow. I also note that, addressing statutory error only, the Tenth Circuit recently rejected the proposition for which Judge Carnes reads Shelton to stand: As a preliminary matter, we note that in the wake of Booker several courts of appeals have collapsed the third and fourth prong analyses. That is to say, if these courts find the third prong satisfied, they conclude that the fourth prong is met as a matter of course. We cannot subscribe to this approach. The Court in Olano clearly held that “a plain error affecting substantial rights does not, without more, satisfy the . . . standard, for otherwise the discretion afforded by Rule 52(b) would be illusory.” Gonzalez-Huerta, __ F.3d at __, 2005 WL 807008, at *6 (citation omitted) (quoting Olano, 507 U.S. at 737, 113 S. Ct. at 1779); see also Gonzalez-Huerta, __ F.3d at __, 2005 WL 807008, at *12 n.2 (Ebel, J., concurring) (“I agree with both the majority opinion and with Judge Hartz that we cannot conflate the third and fourth prongs of the plain-error analysis. Rather, we must address these two inquiries separately.”). Judge Carnes further argues that it would be “difficult to justify a conclusion that an error that is structural does not ‘seriously affect[] the fairness, integrity or public reputation of judicial proceedings,’” and that, “[s]o far as can be discovered, no court has ever actually” reached such a conclusion. Ante, at 12 (quoting Olano, 507 U.S. at 732, 113 S. Ct. at 1776). The Supreme 73 with the Supreme Court’s expectation regarding the applicability of the plain-error test. For the same reasons, it is entirely consistent with the Court’s statement that not “every appeal will lead to a new sentencing hearing.” Id.16 Judge Carnes next attacks my distinction between constitutional errors and statutory errors, arguing that it lacks support “in law or logic.” Ante, at 13. Apparently, he thinks that there is no difference between a case in which the defendant’s constitutional rights were violated and a case in which they were not. I think that there is. The Supreme Court seems to think so, too. See Booker, 125 S. Ct. at 769 (“[W]e must apply today’s holdings—both the Sixth Amendment Court, however, has twice concluded that a defendant failed to satisfy the fourth prong of the plain-error test even assuming that the error claimed was both plain and structural. See Cotton, 535 U.S. at 632-34, 122 S. Ct. at 1786-87; Johnson, 520 U.S. at 468-70, 117 S. Ct. at 1549-50. Although the Court did not actually say that the errors it addressed in these cases were structural, it necessarily held that, in general, structural errors do not per se affect the fairness, integrity, or public reputation of judicial proceedings. Otherwise, the Court would have been required to reach the defendants’ structural-error arguments. For the same reason, Cotton and Johnson also necessarily hold that the plain-error test applies to even structural errors, and, so far as I have discovered, no court of appeals has held otherwise. Judge Carnes’s opinion is inconsistent with this holding, as he all but says that structural errors are per se reversible even in the plain-error context. 16 Moreover, in the same paragraph, the Court also stated, “[W]e must apply today’s holdings—both the Sixth Amendment holding and our remedial interpretation of the Sentencing Act—to all cases on direct review. That fact does not mean that we believe that every sentence gives rise to a Sixth Amendment violation.” Booker, 125 S. Ct. at 769. The panel’s approach renders the observation that not “every sentence gives rise to a Sixth Amendment violation” superfluous because Rodriguez and progeny treat all Booker errors exactly the same. See ante, at 4 (“Because the effect of Booker error is the same regardless of the type, our decisions make no functional distinction between constitutional and statutory error. For purposes of the plain error rule, unpreserved error is unpreserved error.”). 74 holding and our remedial interpretation of the Sentencing Act—to all cases on direct review. That fact does not mean that we believe that every sentence gives rise to a Sixth Amendment violation. . . . [And] in cases not involving a Sixth Amendment violation, whether resentencing is warranted or whether it will instead be sufficient to review a sentence for reasonableness may depend upon application of the harmless-error doctrine.” (emphasis added and citations omitted)). Stated differently, statutory Booker errors are merely a byproduct of Booker’s “unnecessarily broad remedy,” id. at 788 (Stevens, J., dissenting), whereas constitutional Booker errors are violations of the Sixth Amendment requirement that, “[o]ther than the fact of a prior conviction, any fact that increases the penalty for a crime beyond the prescribed statutory maximum must be submitted to a jury, and proved beyond a reasonable doubt.” Apprendi, 530 U.S. at 490, 120 S. Ct. at 2362-63.17 Moreover, as Judge Carnes concedes, see ante, at 14 n.4, this court has 17 The Tenth Circuit has also noted the distinction between constitutional and non- constitutional Booker errors. See Gonzalez-Huerta, __ F.3d at __, 2005 WL 807008, at *4. It “[held] that non-constitutional Booker error is not structural error,” id., but did not reach the issue of constitutional Booker error. In concluding, as I do, that non-constitutional (or statutory) Booker error is not structural, it observed that the Supreme Court has implied that “generally speaking structural errors must, at a minimum, be constitutional errors.” Id.; see supra note 7 & accompanying text (collecting cases and observing the same). Later in its opinion, the Tenth Circuit also noted the normative difference between constitutional and statutory Booker errors: The error of which [the defendant] complains is not the substantive error first recognized in Blakely and which Booker sought to eliminate—namely, that the Sixth Amendment is violated when a judge, rather than a jury, finds facts that mandatorily increase a defendant’s sentence. Rather, the error in [this] case—that the District Court applied the Guidelines mandatorily—is only error insofar as it 75 already stated once that only constitutional errors can be structural errors. See Sanchez, 269 F.3d at 1272 n.41.18 Judge Carnes then relies on Sanchez, arguing that its holding “that Apprendi error is not structural answers the question of whether Booker error is structural, because Booker is an application of Apprendi once removed.” Ante, at 16. This statement is obviously wrong. Because the Guidelines were still binding post- Apprendi, a court reviewing Apprendi error could simply determine whether the evidence supporting any judicial fact-finding was so clear that failure to submit the issue to a jury was harmless; if it was, then the error was harmless precisely because the Guidelines were still binding. But post-Booker, even if the evidence supporting all facts found by the judge is overwhelming, we can still only guess as to what sentence the judge might have imposed under an advisory-guideline runs afoul of the Court’s remedy for the unconstitutional implications of the Guidelines. This disconnect between the constitutional violation and the remedy makes Booker unique. . . . The fortuity of the Court’s choice to excise 18 U.S.C. § 3553(b)(1), instead of a remedy more directly related to the underlying constitutional problem, is key to our determination that the District Court’s erroneous—although not constitutionally erroneous—mandatory application of the Guidelines is not particularly egregious or a miscarriage of justice. Gonzalez-Huerta, __ F.3d at __, 2005 WL 807008, at *8. Although I do not agree with all of the Tenth Circuit’s analysis, I do agree that there is an important distinction between constitutional and statutory Booker errors. 18 Judge Carnes may be correct that Sanchez’s statement is “pure dicta.” Ante, at 14 n.4. As such, I do not suggest that he is “bound” by it. Id. My only surprise is that what this court said—in an en banc opinion that Judge Carnes joined—less than four years ago is today thought so obviously wrongheaded. 76 regime. We, therefore, cannot know whether the error affected the defendant’s substantial rights. This is why Booker constitutional error is structural even though Apprendi error was not. Unlike Booker, Apprendi did not fundamentally alter the sentencing framework, and the effects of Apprendi error were just as determinable as the effects of any other sort of trial error. Indeed, a panel of this court, in an opinion Judge Carnes joined, has already observed that “Blakely error in the Sentencing Guidelines context . . . was judicial versus jury fact-finding of sentencing enhancements and is entirely different from the error we now know to exist under Booker as to the Sentencing Guidelines.” United States v. Shelton, 400 F.3d 1325, 1333 n.12 (11th Cir. 2005) (emphasis added). Finally, Judge Carnes argues that most mitigating evidence or argument a defendant might present under the post-Booker sentencing model should already be in the record because it was arguably relevant (pre-Booker) to determining the defendant’s sentence within his guideline-mandated range. See ante, at 21-24. This may be the case with respect to some types of mitigating evidence, but it ignores reality to suggest that it is true across the board. For example, although it was theoretically possible under the old model for a defendant to seek a sentence at the low end of a mandatory guideline range by presenting statistical evidence that the Sentencing Commission had overestimated the need for general deterrence 77 with respect to his offense in his particular community, I personally never heard of such an argument being made. Now, however, such Sentencing Commission policy judgments are open to attack in any given case. In sum, my opinion is not nearly so novel as Judge Carnes suggests, it does not depend on any monumental inferential leap, and it is entirely consistent with the Supreme Court’s opinion in Booker. The distinction I draw between constitutional and statutory Booker errors has ample support in law and logic; in fact, it is not even unprecedented in this circuit. Finally, Judge Carnes’s opinion is grounded in large part on the faulty assumptions that Apprendi error and Booker error are essentially the same and that Booker really won’t do that much to change the type of the evidence and argument presented at federal sentencing hearings. V. I tend to agree with the panel in this case that we cannot automatically equate prejudice with the imposition of extra-verdict sentencing enhancements, since these same enhancements must be considered under the new model.19 I 19 This is essentially the position taken by the Fourth and Sixth Circuits. E.g., United States v. Hughes, __F.3d__, 2005 WL 628224, at *5-12 (4th Cir. March 16, 2004); United States v. Oliver, 397 F.3d 369, 379-80 (6th Cir. 2005); see also United States v. Ameline, 400 F.3d 646, 654 (9th Cir. 2005), vacated and reh’g en banc granted, __ F.3d __, 2005 WL 612710 (9th Cir. March 11, 2005). In practice, this approach reaches the same result as mine in cases involving constitutional error. My disagreement goes only to its reasoning. It is inaccurate to say that a defendant has actually established an effect on his substantial rights if we can only guess as to what sentence he might get under the new sentencing model. In short, these courts require the 78 completely agree with the panel that we cannot know what effect Booker error had on Rodriguez’s sentence. I take a different lesson from this uncertainty, though. Because its effects “are necessarily unquantifiable and indeterminate,” Sullivan, 508 U.S. at 281-82, 113 S. Ct. at 2083, and because Booker dramatically “affect[s] the framework within which [sentencing] proceeds,” Fulminante, 499 U.S. at 310, 111 S. Ct. at 1265, Booker error is structural in nature. Moreover, treating Booker constitutional error as structural error would give effect to the distinction Booker draws between constitutional and statutory error. See 125 S. Ct. at 769. defendant to show that his substantial rights were affected only to find that they were in every case involving constitutional error, whereas I would simply say that no such showing is required. To the extent that some courts are automatically vacating sentences that involve only statutory error, e.g., United States v. McCraven, __ F.3d __, 2005 WL 608263, at *6 (6th Cir. March 17, 2005), this cannot be correct because it renders meaningless the Supreme Court’s instruction that we apply the plain-error test to cases pending on direct review. The Second, Seventh, and D.C. Circuits have adopted the novel approach of remanding all Booker cases to allow the district court to determine whether the defendant’s substantial rights were affected so as to require a new sentencing hearing. See United States v. Coles, __ F.3d __, 2005 WL 783069, at *1 (D.C. Cir. Apr. 8, 2005); United States v. Crosby, 397 F.3d 103, 117- 118 (2d Cir. 2005); Paladino, 401 F.3d at 483-85. To the extent that this approach fails to distinguish between constitutional and statutory error, it is inconsistent with the Booker remedial opinion. Moreover, given the significant differences between the old and new sentencing models, “the only way to know whether a different sentence would have been imposed under advisory guidelines” is to actually hold a full hearing. Paladino, 401 F.3d at 488 (Kanne, J., dissenting from the denial of rehearing en banc). If this is what the district court is to do, then this approach is no more efficient than simply remanding for a new hearing, as I would do in all cases involving constitutional error. If, however, the district court is only going to take a “quick look” on remand, id. at 486 (Ripple, J., dissenting from the denial of rehearing en banc), then it overlooks the fundamental change Booker has wrought in the federal sentencing framework and fails to fully remedy structural constitutional errors. Finally, by remanding even cases involving only statutory error, this approach fails to relieve district courts of any of the burden the limiting final paragraph of the Booker remedial opinion seeks to avoid. 79 The panel errs by accepting Booker’s invitation to “apply ordinary prudential doctrines” without following its implicit instruction as to how they should be applied. Id. As a result, it erroneously requires Rodriguez to prove an effect on his “substantial rights.” This an important issue because the panel’s opinion imposes a virtually impossible burden on a large class of defendants whose cases are still in the pipeline on direct appeal. Our rule will “condemn some unknown fraction of criminal defendants to serve an illegal sentence.” Paladino, 401 F.3d at 484. The panel does not deny that this is the case. It concedes that it has no way of knowing whether Rodriguez is serving an illegal sentence. Rodriguez, 398 F.3d at 1301. At a minimum, we know that his sentence was imposed in an unconstitutional manner. Nevertheless, the panel refuses to grant a new sentencing hearing. For the foregoing reasons, I dissent from the denial of rehearing en banc. 80 BARKETT, Circuit Judge, dissenting from denial of rehearing en banc: As an initial matter, I agree with Judge Tjoflat that the issue presented here is eminently worthy of en banc review. Furthermore, I find his reasoning extremely persuasive. However, even assuming that constitutional Booker error is not “structural,” I believe that the panel erroneously applies Jones v. United States, 527 U.S. 373 (1999), instead of United States v. Dominguez Benitez, __ U.S. __, 124 S. Ct. 2333 (2004). In so doing, the panel erroneously requires the defendant to prove that his “substantial rights” were affected by a preponderance of the evidence, instead of requiring him to prove only a “reasonable probability” that, but for the error, the outcome of the district court proceedings would have been different. Rodriguez1 is a Booker “pipeline” case wherein the defendant was unconstitutionally sentenced before Booker was decided but did not preserve the error for appellate review. Conceding that the defendant had met the first two prongs of the “plain error” test, Rodriguez addressed the question posed by the third prong of that test, whether the defendant could prove that the constitutional violation affected his “substantial rights.” Rodriguez, 398 F.3d at 1299-1300. Rodriguez held that if the evidence is in equipoise, the defendant has not met his 1 United States v. Rodriguez, 398 F.3d 1291 (11th Cir. 2005). 81 burden of proof. Id. I believe this conclusion conflicts with the holding of Dominguez Benitez, and indeed, applies the preponderance of the evidence test, a standard of proof Dominguez Benitez specifically rejected. In Dominguez Benitez, the Supreme Court specifically clarified: (1) that a defendant proves that an error affects his substantial rights by establishing a “reasonable probability” that, but for the error, the outcome of the district court proceedings would have been different;2 (2) that a reasonable probability of a different result is one that is “sufficient to undermine confidence in the outcome of the proceeding;”3 and (3) that this “reasonable probability” standard differs from the “less defendant-friendly ‘more likely than not’” preponderance of the evidence standard.4 I agree that the defendant bears the burden of proving this “reasonable probability.” However, I believe the defendant meets this burden by showing that: (1) the guidelines were mandatory and that nothing in the record indicates that the 2 Dominguez Benitez, 124 S. Ct. at 2340. 3 Id. (internal quotation marks omitted) (citing Strickland v. Washington, 466 U.S. 668, 694 (1984), and United States v. Bagley, 473 U.S. 667, 682 (1985)). 4 Id. at 2342 (Scalia, J., concurring in the judgment) (emphasis added); see also id. at 2340 n.9 (“The reasonable-probability standard is not the same as, and should not be confused with, a requirement that a defendant prove by a preponderance of the evidence that but for error things would have been different.”). 82 district court applied the guidelines in a non-mandatory fashion;5 and (2) nothing in the record indicates that the district court would apply the same or a greater sentence on remand under the new, advisory guidelines. Although the panel repeatedly cites to the “reasonable probability” standard of Dominguez Benitez as the touchstone of its inquiry, talismanic repetition of the proper standard does not necessarily translate into its faithful application. Despite the panel’s repeated assertions to the contrary, id., 398 F.3d at 1299-1301, the test it applies appears identical to a “preponderance of the evidence” standard, a standard Dominguez Benitez explicitly rejects as excessively stringent. Rodriguez states that: if it is equally plausible that the error worked in favor of the defense, the defendant loses; if the effect of the error is so uncertain that we do not know which, if either, side it helped the defendant loses. Where errors could have cut either way and uncertainty exists, the burden is the decisive factor in the third prong of the plain error test, and the burden is on the defendant. Id. at 1300. Requiring a quantum of proof beyond equipoise is the preponderance of the evidence standard. Under the preponderance of the evidence test, the defendant 5 For example, in Booker, the district court in defendant Fanfan’s case did not apply applicable provisions of the guidelines even though they were mandatory at the time. United States v. Booker, __ U.S. __, 125 S. Ct. 738, 747 (2005). 83 loses when the evidence is in equipoise because he did not present that slight quantum of evidence necessary to tip the balance from equipoise to his favor. See, e.g., Nat’l Lime Ass’n v. Envtl. Prot. Agency, 627 F.2d 416, 453 n.139 (D.C. Cir. 1980) (“[T]he standard of ordinary civil litigation, a preponderance of the evidence, demands only 51% certainty.”); Black’s Law Dictionary 1201 (7th ed. 1999). However, Dominguez Benitez says that the “reasonable probability” standard is more lenient than the preponderance of the evidence test. Dominguez Benitez, 124 S. Ct. at 2342 (Scalia, J., concurring in the judgment); id. at 2340 n.9. Thus, equipoise must of necessity be sufficient to satisfy the lesser burden of a reasonable probability.6 6 Judge Carnes argues that my reading of the “reasonable probability” standard differs from prior Circuit case law. He states that we have held in several cases that a defendant failed to meet the “reasonable probability” standard when there was no evidence on the record that an alleged error prejudiced him. Ante, at 32-33 (Carnes, J., concurring in denial of reh’g en banc). However, I do not believe that any of the cases he cites are of particular significance to the inquiry here. In each case there was record evidence contradicting the defendant’s claims of prejudice. In Henry v. Wainwright, the judge failed to instruct the jury about the consequences of a 6-6 split over whether to sentence the defendant to death. 743 F.2d 761, 763 (11th Cir. 1984) (per curiam). However, the record contained evidence suggesting that the jury was never actually deadlocked, belying the defendant’s claim that the lack of an instruction prejudiced him. Id. Adams v. Wainwright presented a similar jury instruction claim in which the court felt “[b]ound by” Henry because there was no evidence that a deadlock ever existed in the first place, 764 F.2d 1356, 1369 (11th Cir. 1985), and also presented an ineffective assistance of counsel claim wherein the defendant failed to rebut a “strong presumption” that a jury is presumed to have followed the trial court’s express, and correct, instructions. Id. And in Straight v. Wainwright, the defendant’s theory of prejudice stemming from allegedly ineffective assistance of counsel was so attenuated as to border on the incredulous. 772 F.2d 674, 680 (11th Cir. 1985). There, the defendant claimed that his attorney rendered ineffective assistance by failing to “rehabilitate” jurors who had stated “either that they could not return a death penalty in any case or that they 84 To support this de facto preponderance of the evidence standard, Rodriguez reaches back to the 1999 decision Jones v. United States, 527 U.S. 373, in which the Supreme Court noted that when it was “just as likely” that an error prejudiced the defendant as not, the defendant cannot meet his burden of showing that the error affected his substantial rights. Rodriguez, 398 F.3d at 1299-1301 (citing Jones, 527 U.S. at 394-95). However, Jones was decided well before Dominguez Benitez specifically enunciated the standard by which the defendant must prove prejudice in the plain-error context. Obviously, we cannot hold that the Supreme Court has implicitly overruled Jones in Dominguez Benitez. Rodriguez de Quijas v. Shearson/American Express Inc., 490 U.S. 477, 484 (1989).7 However, we are confronted with language in could not evaluate the evidence fairly, knowing that the defendant was charged with a capital crime.” Id. He contended that if his attorney had successfully coaxed out more ambiguous answers about jurors’ sentiments with respect to the death penalty through further questioning, the prosecution would not have been able to strike them for cause, and would thus have had to expend “expensive” peremptory challenges to eliminate them. Id. The result, apparently, would have been a jury somewhat less inclined to sentence the defendant to death. Id. In Rodriguez, in contrast, the defendant’s theory of prejudice requires no such flights of fancy. We are certain that the district court erred, and the chance that this error prejudiced the defendant is quite real — just as likely as not, as the panel itself admits. Further, there is no evidence on the record in Rodriguez contradicting a showing of prejudice. 7 Indeed, that doctrine applies to situations, unlike this one, where one Supreme Court case is unmistakably on point, but where the rationale underlying that decision has been undermined indirectly by subsequent cases. See, e.g., Tenet v. Doe, __ U.S. __, 125 S. Ct. 1230, 1238 (2005); Hohn v. United States, 524 U.S. 236, 252-53 (1998); Agostini v. Felton, 521 U.S. 203, 237 (1997); Am. Trucking Ass’ns Inc. v. Smith, 496 U.S. 167, 180 (1990) (plurality opinion). 85 Jones that is in direct conflict with the “reasonable probability” standard of Dominguez Benitez, stating that a defendant cannot prove prejudice when it is “just as likely” as not that an error was prejudicial. The question, therefore, is not whether Dominguez Benitez implicitly overruled Jones by undermining its underlying rationale, but rather how to resolve these seemingly irreconcilable cases.8 The Supreme Court offers little guidance, if any, about the role of a Court of Appeals when faced with conflicting precedent, although Justice Scalia has recognized that such situations do arise. Kaiser Aluminum & Chem. Corp. v. Bonjorno, 494 U.S. 827, 841 (1990) (Scalia, J., concurring). However, the Seventh Circuit, when faced with the conflict Justice Scalia acknowledged in Bonjorno, offered some helpful advice. It stated that when we are faced with conflicting Supreme Court precedent, we should attempt to reconcile the two cases in a manner which comports with their underlying policies. Mozee v. Am. Commercial Marine Svc. Co., 963 F.2d 929, 935 (7th Cir. 1992) (citing Rodriguez 8 It would appear that we are encountering the difficulty of applying the various standards for analyzing “prejudice” that Justice Scalia predicted in Dominguez Benitez. 124 S. Ct. at 2342 (Scalia, J., concurring) (“By my count, this Court has adopted no fewer than four assertedly different standards of probability relating to the assessment of whether the outcome of trial would have been different if error had not occurred….Such ineffable gradations of probability seem to me quite beyond the ability of the judicial mind (or any mind) to grasp….That is especially so when they are applied to the hypothesizing of events that never in fact occurred.”). 86 de Quijas, 490 U.S. at 484). In particular, it reasoned that we should begin by asking “which line of cases embodies the general rule of construction and which line of cases has more limited applicability.” Id. An estimation of which holding the Supreme Court would likely apply if presented with the case may also suggest the preferred precedent. Easter House v. Felder, 910 F.2d 1387, 1409 (7th Cir. 1990) (en banc) (Easterbrook, J., concurring) (“Inconsistent lines of precedent are common in American law.…Such hair-splitting leaves judges of the inferior federal courts in a difficult position, because any effort to reconcile and apply the cases will be met with a convincing demonstration…that there is a fly in the ointment.…I believe that despite the force of [the dissent’s] arguments, [the majority] offers the best estimate of the course a majority of the [Supreme] Court will take, and I therefore join [its] opinion.”). It is hard to see why we would not decide that Dominguez Benitez, and not Jones, is the controlling case here. Whereas Jones simply noted that, even if a constitutional violation did exist, the defendant’s showing would not satisfy the “substantial rights” prong under the facts in that case, 527 U.S. at 394-95, Dominguez Benitez undertook to clearly articulate a burden of proof for plain- error cases generally. 124 S. Ct. at 2336. Moreover, Jones could not, as Judge Carnes suggests, have applied the Dominguez Benitez standard to a situation 87 “where there is no indication whether the error actually did have an adverse effect on the outcome of the proceeding.” Ante at 35-36 (Carnes, J., concurring in denial of reh’g en banc). The simple fact that Jones was decided over five years before Dominguez Benitez established the “reasonable probability” standard belies any argument that Jones somehow constitutes an application of that standard. Thus, Dominguez Benitez, not Jones, embodies “the general rule of construction” relating to plain-error review. Moreover, it is more likely that a majority of the Supreme Court would determine that Dominguez Benitez applies to this case instead of Jones. Eight members of the Court signed the majority opinion in Dominguez Benitez, including the author of Jones and three of the four Justices who signed his opinion. In contrast, only five members of the Court made up the Jones majority. Further, in Jones the Supreme Court actually held that there was no error of law in the first place, and that any hypothetical error had been mitigated by the district court. Jones, 527 U.S. at 393-95. The Jones Court noted that the substantial rights prong would not be satisfied only assuming arguendo that such a constitutional error could still be found. Id. at 394-95. This represents a marked departure from the case at bar, where we are certain that the district court was operating under an erroneous premise of law. Essentially, in cases like Rodriguez, where the defendant can prove that 88 Booker error denied him a constitutionally-mandated process and that the outcome of that process cannot be known until the process actually takes place, a defendant sufficiently undermines our confidence in the outcome of his sentencing to meet the prejudice prong. See, e.g., United States v. Paladino, 401 F.3d 471, 486-87 (7th Cir. 2005) (Ripple, J., dissenting from denial of reh’g en banc); see also ante, at 58-60 (Tjoflat, J., dissenting from denial of reh’g en banc) (noting that the mandatory guidelines effectively foreclosed many avenues a defendant may now pursue in requesting a more lenient sentence, such as arguing that the recommended guidelines sentence is not a “just punishment,” or challenging the wisdom of the Sentencing Commission’s policy decisions themselves); see also United States v. Serrano-Beauvaix, 400 F.3d 50, 57-58 (1st Cir. 2005) (Lipez, J., concurring); United States v. Crosby, 397 F.3d 103, 115 (2d Cir. 2005). Moreover, I believe that the result the panel reaches impermissibly trenches upon the Federal Sentencing Act’s policy of allowing “the district court, not the court of appeals, to determine, in the first instance, the sentence that should be imposed in light of certain factors properly considered under the Guidelines.” Williams v. United States, 503 U.S. 193, 205 (1992) (citing a 1986 congressional amendment to the Act deleting provisions that authorized appellate courts to correct a sentence imposed as a result of an incorrect application of the 89 guidelines). The statute itself reflects this consideration: If the court of appeals determines that-- (1) the sentence was imposed in violation of law or imposed as a result of an incorrect application of the sentencing guidelines, the court shall remand the case for further sentencing proceedings with such instructions as the court considers appropriate[.] 18 U.S.C. § 3742(f)(1) (2005). The Williams Court underscored the importance of this language in the remand process, stating that “Section 3742(f)(1) does not call for a remand every time a sentencing court might misapply a provision of the Guidelines; rather, remand is required only if the sentence was ‘imposed as a result of an incorrect application’ of the Guidelines.” Williams, 503 U.S. at 202- 03 (quoting 18 U.S.C. § 3742(f)(1)) (emphasis in original). Even under this somewhat restrictive reading of the statute, in every case involving constitutional Booker error, the district court surely has imposed a sentence “as a result of” an incorrect application of the guidelines. Given the statutory language and the policies behind it, that error would require a remand in the absence of evidence in the record rendering it harmless. Moreover, the Third Circuit held, pre-Booker, that this policy consideration precluded speculation about the sentence a district court would impose upon remand when determining whether a guidelines sentencing error affected a defendant’s “substantial rights,” making such errors presumptively prejudicial. In 90 United States v. Knight, 266 F.3d 203 (3d Cir. 2001), the district court applied an incorrect sentencing range, but imposed a sentence that also fell within the correct guideline range. Id. at 205. The Knight Court refused to speculate about whether the district court would impose an equivalent sentence on remand for purposes of the plain-error “substantial rights” inquiry. Id. at 208. Instead, it remanded for resentencing. Id. at 210. Quoting Williams, it stated that this approach effectuated the Supreme Court’s determination of the federal sentencing statute’s intent to allow the district court to determine the proper guidelines sentence in the first instance. Id. at 208 (quoting Williams, 503 U.S. at 205 (1992)).9 Other pre-Booker precedent also presumed prejudice in cases when a sentencing error’s effect on the length of a defendant’s sentence was extraordinarily difficult to ascertain. See, e.g., United States v. Plaza-Garcia, 914 F.2d 345, 347-48 (1st Cir. 1990) (Breyer, C.J.) (vacating and remanding an illegal sentence also falling within the correct guidelines range under the plain-error doctrine because it “may well have been influenced by the [erroneous] sentencing recommendation”); United States v. Reyna, 358 F.3d 344, 351-52 (5th Cir.) (en 9 Indeed, this circuit has expressed similar concerns. In determining that a defendant could appeal a sentence enhancement where the sentence also fell within the guidelines range he had advocated before the district court, we said that we were “not willing to speculate as to whether appellant’s sentence would have been the same without the enhancement. In our view, that determination is for the district court in the first instance.” United States v. Fuente-Kolbenschlag, 878 F.2d 1377, 1379 n.7 (11th Cir. 1989). 91 banc), cert. denied, 124 S. Ct. 2390, 158 L. Ed. 2d 966 (2004) (presuming that denial of defendant’s right to allocution caused prejudice, because of the nature of the right and the difficulty of proving the violation affected a specific sentence); United States v. Adams, 252 F.3d 276, 287 (3d Cir. 2001) (same); United States v. Alba Pagan, 33 F.3d 125, 130 (1st Cir. 1994) (“[T]he impact of the omission [of the right to allocution] on a discretionary decision is usually enormously difficult to ascertain.”); United States v. Prouty, 303 F.3d 1249, 1252-53 (11th Cir. 2002) (presuming prejudice “where the defendant was not afforded the opportunity to allocute and the court did not impose the lowest sentence under the guidelines”). As Judge Tjoflat and Judge Lipez of the First Circuit have pointed out, the effects of constitutional Booker error are similarly hard to gauge. Ante, at 62-66 (Tjoflat, J., dissenting from denial of reh’g en banc); Serrano-Beauvaix, 400 F.3d at 56-61 (Lipez, J., concurring); see also Paladino, 401 F.3d at 486-87 (Ripple, J., dissenting from denial of reh’g en banc); United States v. Barnett, 398 F.3d 516, 526-29 (6th Cir. 2005); United States v. Crosby, 397 F.3d at 116-18. Accordingly, I respectfully dissent from the denial of rehearing en banc. 92