concurring:
I write separately to express my concern that, under our case law, harmless-error review based on post-conviction factual submissions could swallow up the rule of Apprendi v. New Jersey, 530 U.S. 466, 120 S.Ct. 2348, 147 L.Ed.2d 435 (2000). We need to re-think our doctrine on this important point.
I.
First, some context:
After Guerrero-Jasso filed a sentencing memorandum objecting to the Pre-Sen-tence Report’s conclusion that he was subject to a mandatory twenty-year statutory maximum sentence, the government asked for more time to check Guerrero-Jasso’s assertion that he had not, in fact, admitted each date of removal alleged in the con-junctively phrased information. Guerrero-Jasso agreed to continue sentencing to give the government the time it sought. After receiving the transcript of the change of plea proceeding (which confirmed that Guerrero-Jasso did not admit to each date of removal) the government “request[ed] additional time to compare [the] transcripts with case law.” The district court granted the government’s request. The day before the continued sentencing hearing, the government filed a “Supplemental Exhibit” in support of its sentencing memorandum. The exhibit comprises three one-page documents, each entitled “Warrant of Removal/Deportation” and each purporting to show that Guerrero-Jasso was removed to Mexico on the date indicated.
At sentencing, Guerrero-Jasso renewed his position that he was not subject to the twenty-year maximum penalty because he had not admitted the requisite removal date. The government argued, and the court responded, as follows:
THE GOVERNMENT: [T]here’s enough in the record for the Court to find that the date was proven. The Supreme Court says that any Apprendi error is reviewed for harmlessness.
THE COURT: That’s when the Circuit gets their hands on it — I mean after Pm, through. I mean, if you tell me that if I commit an error, “It’s okay. Don’t worry about it because the Ninth Circuit is not going to pay any attention to it.” To me that doesn’t impress me in terms of whether I should commit an error or not. So I don’t think I should be committing any errors even if I know that if I commit an error, the Ninth Circuit isn’t going to care.
THE GOVERNMENT: The cases that say Apprendi error is reviewed for harmlessness don’t draw a distinction between whether the district court judge is aware of this error or not. What they ask the district court to consider is whether there is overwhelming and un-controverted evidence of, in this case, the fact of the prior deport.
And the Government submitted exhibits to this Court that prove the prior deport, the warrants of deport. The Government notes that the PSR — the Court is allowed to look at the failure to *1197object to [the date of] deport in the PSR....
And maybe most tellingly, the defendant is never telling this Court he wasn’t deported on those dates. There’s no chance that the result in this Court would be different. And that’s what Apprendi error asks this Court and asks the Ninth Circuit to look at: Would the result be different if the error this Court is worried about committing didn’t occur? In this case the result wouldn’t be different.1
Combined with the timing of the government’s evidentiary submission, this exchange makes it obvious that the sole point of introducing the warrants of removal was to “prove” the essential removal date, thereby justifying a twenty-year statutory ceiling.
Even though the district court’s sentence violated Apprendi, the government asks us to uphold the sentence on the basis of the “warrant of removal/deportation” purporting to show that Guerrero-Jasso was removed to Mexico on January 19, 2011. As our opinion in this case indicates, there is support in this court’s case law for permitting such postconviction governmental submissions as proof of Appren-di harmlessness — although, as the result reached in this case demonstrates, such after-the-fact submissions can establish Apprendi harmlessness only in limited circumstances.
One would think that a constitutional protection designed to assure that juries rather than judges decide facts essential to determining the potential maximum sentence could never be satisfied by post-conviction evidentiary submissions, reviewed by a judge and directed at demonstrating that had the submissions been introduced at a jury trial, a jury would have found the facts in the government’s favor beyond a reasonable doubt. Such an approach is entirely different from the usual harmless-error analysis, which reviews the record of an actual trial to determine what the actual jury in that case would have decided on the record before it. See, e.g., Chapman v. California, 386 U.S. 18, 25-26, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967). It is, after all, out of respect for “the jury-trial guarantee,” that the Supreme Court “instructs ... reviewing court[s] to consider ... not what effect [a] constitutional error might generally be expected to have upon a reasonable jury, but rather what effect it had upon the guilty verdict in the case at hand.” Sullivan v. Louisiana, 508 U.S. 275, 113 S.Ct. 2078, 2081, 124 L.Ed.2d 182 (1993).
This case, of course, involved a guilty plea, so there was no jury trial. But that circumstance makes the postconviction submission of evidence in an effort to demonstrate harmless error even more questionable: The defendant waived a jury trial, so there is no pre-conviction factual record at all. To create one after the fact is not only to undermine Apprendi but to broaden the defendant’s waiver, by allowing courts to make factual determinations concerning a trial that never occurred.
Here, for example, the government neglected to elicit necessary admissions at the plea hearing. How can we, or the district court, know that it would have done better had a trial occurred — that is, that it would properly have submitted the necessary removal documents? And why is Guerrero-Jasso’s waiver of jury trial to be taken as waiving a jury trial as to that *1198question, which did not arise until after the waiver?
This state of affairs leads me to wonder: How did we get here? And can, and should, we reconsider?
II.
A review of our case law shows that we have not always treated Apprendi’s protections so carelessly:
• United States v. Tighe, 266 F.3d 1187, 1195 (9th Cir.2001), held that if the fact of a juvenile adjudication is used to increase the maximum statutory penalty, that fact must be charged in an indictment and found by a jury beyond a reasonable doubt. “Because Tighe properly preserved his ... Ap-prendi claim for appeal, his sentence [could not] stand unless the district court’s constitutional error was harmless beyond a reasonable doubt.” Id. We then held, without further inquiry, that the error in that case was not harmless, simply because the defendant’s sentence exceeded “the applicable statutory maximum.” Id. Like Guerrero-Jasso, Tighe’s conviction was the result of a guilty plea. Id. at 1190.
• United States v. Velasco-Heredia, 319 F.3d 1080, 1085-86 (9th Cir.2003), considered whether an Apprendi error resulting from judicial fact-finding as to drug quantity was harmless. The drug quantity found at sentencing increased what was otherwise a thirty-seven-to-forty-six month guidelines range and a statutory maximum of five years, to a mandatory five-year sentence and a maximum forty-year sentence. Id. at 1083-84, 1086. We reasoned that therefore “not only was the error not harmless, it was demonstrably harmful.” Id. at 1086. We found it “too clever by half to permit the government in the guilt phase of a case to prove beyond a reasonable doubt that only one kilogram of marijuana was involved in the offense, and then at sentencing to prove 101 kilograms by a preponderance of the evidence and claim that such a finding ... requires the maximum sentence of five years.” Id. Rather, in our view, the increased penalty provision could not apply “unless and until 100 kilograms or more of marijuana are properly on the sentencing table,” which, “[a]fter Apprendi, ... cannot happen until the jury, or the court in a bench trial, finds beyond a reasonable doubt that this is the quantity involved in the violation.” Id. Recognizing that “the Fifth Amendment to our Constitution does not permit Velasco-Heradia to be tried twice for the same offense[,]” we vacated his sentence and remanded for resentencing in accordance with an unspecified drug quantity. Id. at 1086-87.
• United States v. Banuelos, 322 F.3d 700, 705 (9th Cir.2003), held that the district court erred by making a finding, by a standard lower than beyond a reasonable doubt, as to the amount of drugs for which a defendant involved in a drug conspiracy was personally responsible, where the finding increased the defendant’s maximum sentence exposure. The defendant in that case had entered a guilty plea to a drug conspiracy charge, but he never waived his right to have a jury determine the drug quantity attributable to him, and he refused to waive the beyond-a-reasonable-doubt standard as to drug quantity. Id. at 703. We held that, “because Baneuelos did not allo-cute to drug quantity at the change of plea hearing or admit to drug quantity in a written plea agreement,” he was properly convicted of only the “general offense ... charged in the indictment ... the only offense for which there *1199was a factual basis for conviction.” Id. at 706-07. Because he challenged his sentence but not his conviction, we explained that the proper course on remand was for the district court to resentence him “subject to the maximum sentence supported by the facts found by the [fact-finder] beyond a reasonable doubt,” — the facts necessarily included in his plea to the general drug-conspiracy offense. Id. at 706 (internal quotation marks omitted; alteration in original). And, although Banuelos did not raise the issue on appeal, we noted that “[t]he court’s finding of drug quantity attributable to Banuelos by any standard, without first advising Banuelos that he had a right to jury determination of that fact beyond a reasonable doubt, also violated Apprendi” Id. at 705 n. 3.
Finally, we rejected the suggestion that the court should scour the record at sentencing to determine if the error was harmless beyond a reasonable doubt. Id. at 706 & n. 4. As “the very finding of an Apprendi violation means that it was improper for the district court to determine drug quantity attributable to Banuelos at sentencing without first informing Banuelos of his right to a jury determination of drug quantity to him beyond a reasonable doubt[,]” we deemed ourselves “prohibited] ... from considering admissions made at sentencing in evaluating an Apprendi violation for harmless error.” Id. at 706 n. 4.
• United States v. Thomas, 355 F.3d 1191 (9th Cir.2004), followed Banuelos and Velasco-Heredia. In that case, the defendant pled guilty to a drug offense, but refused to accept responsibility for any specific drug quantity during the plea colloquy. Id. at 1192-93. Nevertheless, the judge sentenced him in accordance with a statutory maximum correlating to a specified drug quantity. Id. at 1194. Because the error increased the defendant’s guidelines offense level and required a mandatory minimum sentence, we deemed the error “clearly harmful.” Id. at 1201. We rejected the government’s argument that, on remand, the judge could impose a sentence in accordance with the elevated statutory maximum penalty if the judge found the drug quantity beyond a reasonable doubt: As the defendant “did not ... admit to possessing any specific quantity, nor ... knowingly waive his right under Apprendi ... to have a jury determine quantity beyond a reasonable doubt,” we held, “the district judge cannot determine any particular drug quantity that would affect the maximum statutory sentence to which [the defendant] is exposed.” Id. at 1202.
• United States v. Patterson, 381 F.3d 859 (9th Cir.2004), considered the effect of a guilty plea to manufacturing an unspecified amount of marijuana on the defendant’s sentence. The defendant entered, and the district court accepted, a guilty plea that did not admit to a specified amount of marijuana, with the understanding that the judge would determine the amount of marijuana at the time of sentencing. Id. at 861-62. After the defendant entered his plea, but before sentencing, Apprendi issued, holding such judicial determinations unconstitutional, and the district court, on the government’s motion, vacated the defendant’s plea. Id. at 862. The defendant was then tried and convicted by a jury, and the jury made a finding as to the specific drug quantity. Id. This court reversed, reasoning that jeopardy attached the moment the defendant entered and the court accepted the guilty plea; that the court was not free to vacate the plea on the government’s motion; and that the trial therefore *1200violated double jeopardy principles. Id. at 864-65. Although it would have been a simple matter to review the trial record to determine whether there was evidence to support a finding that the Apprendi violation was harmless beyond a reasonable doubt, we instead remanded with instructions to resentence the defendant in accordance with the maximum penalty allowable under the plea that he entered. Id. at 866.
• Finally, United States v. Lococo, 514 F.3d 860, 865 (9th Cir.2007), held that a district court violated Apprendi by sentencing a defendant according to the statutory maximum penalty for knowing involvement in a conspiracy to distribute crack cocaine, because the defendant admitted only to knowing involvement in the conspiracy’s distribution of powder cocaine when he entered his guilty plea. In remanding for resentencing, we cited our prior decision in Banuelos, and instructed the district court that it could only base the defendant’s sentence on the amount of powder cocaine involved in the conspiracy, “because it is only powder cocaine that Lococo admits he knew about.” Id. at 866.
Our interpretations of the Apprendi harmless-error review standard in these cases do not correspond with the “overwhelming and uncontroverted” standard described in Neder v. United States, 527 U.S. 1, 9, 119 S.Ct. 1827, 144 L.Ed.2d 35 (1999), outside the Apprendi context, and adopted in United States v. Zepeda-Martinez, 470 F.3d 909, 913 (9th Cir.2006), and United States v. Hunt, 656 F.3d 906, 913 (9th Cir.2011), as applicable in Apprendi cases. The difference, in my view, rests largely on a critical distinction between post-conviction, harmless-error review of a preserved Apprendi claim and plain-error review of an unpreserved claim. See United States v. Minore, 292 F.3d 1109, 1122 n. 12 (9th Cir.2002).
United States v. Minore considered two potential methods, first outlined by United States v. Nordby, 225 F.3d 1053, 1060 (9th Cir.2000), overruled in part by United States v. Buckland, 289 F.3d 558, 567-68 (9th Cir.2002) (en banc), for reviewing an unpreserved Apprendi violation for plain error. 292 F.3d at 1121-22. We termed the two Nordby methods the “less stringent” approach and the “more stringent” approach. Id. The more stringent approach followed Neder, asking “whether it was clear beyond a reasonable doubt that a rational jury would have found the defendant guilty absent the error.” Id. at 1122 (internal quotation marks omitted). We held the more stringent approach properly applicable to plain-error cases, where it is the defendant’s burden to prove the error “affected his substantial rights.”2 Id. at 1123; accord United States v. Covian-Sandoval, 462 F.3d 1090, 1098 (9th Cir.2006); Buckland, 289 F.3d at 569-70 (9th Cir.2002) (concluding that defendant’s substantial rights were not affected by Ap-*1201prendí error in light of the unchallenged amount of narcotics attributed to him and therefore refusing to reverse under plain error review).
Minore described the less stringent approach as simply asking whether the defendant received a sentence greater than authorized, absent the error. See 292 F.3d at 1121-22. And although we held the less stringent standard not applicable to plain-error cases, we recognized that we had previously applied it in harmless-error cases, and were careful “not [to] suggest that the less stringent approach is no longer available on harmless error review.” Id. at 1122 n. 12. We then continued to review properly preserved Apprendi claims for harmless error by analyzing only the error’s effect on the sentence received, not combing the record for overwhelming, uncontroverted evidence of the sentencing-enhancing fact. See, e.g., Thomas, 355 F.3d at 1201.
Zepeda-Martinez, however, apparently deeming this long line of cases incompatible with Washington v. Recuenco, 548 U.S. 212, 126 S.Ct. 2546, 165 L.Ed.2d 466 (2006), held that Neder’s brand of harmles-serror review applies to “properly preserved” Apprendi violations, too. 470 F.3d at 913. Accepting that “our prior case law may suggest otherwise,” Zepeda-Martinez cites Recuenco as holding that “Apprendi errors are reviewed under the harmless error standard as applied in Neder [.]” Id.
In my view, Recuenco did not so hold, and Zepeda-Martinez erred in assuming that it did. Recuenco affirmed a rule that already existed in our circuit, i.e., that Apprendi violations are reviewable for harmless error. But it did not mandate any particular method for conducting that review, and so did not require us to overrule our circuit precedents regarding the nature of that review where there is preserved Apprendi error.
In Recuenco, the jury returned a guilty verdict for assault and a special verdict finding that the defendant was armed with a deadly weapon. The jury was not asked to find, and did not find, that the defendant was armed with a firearm. 548 U.S. at 215, 126 S.Ct. 2546. Nonetheless, at sentencing, the defendant received a three-year enhancement for being armed with a firearm. Id.
The Court held Neder’s harmless-error ruling applicable under these circumstances, reasoning that, “[b]ecause Neder’s jury did not find him guilty of each of the elements of the offenses with which he was charged, its verdict is no more fairly described as a complete finding of guilt of the crimes for which the defendant was sentenced than is the verdict here.” Id. at 221, 126 S.Ct. 2546. As the case was in this respect “indistinguishable from Neder,” id. at 220, 126 S.Ct. 2546, the Court held the Apprendi error in Recuenco reviewable for harmless error. The Court did not, however, explain how harmless-error review of Apprendi errors should be conducted. Neder’s mode of harmless-error analysis is not discussed in Recuenco, and Recuenco specifically declined to rule on the question of whether the error in the case before it was harmless.3 Id. at 217-*120218, 126 S.Ct. 2546. Recuenco held only that “[f]ailure to submit a sentencing factor to the jury, like failure to submit an element to the jury, is not structural error.” 548 U.S. at 222, 126 S.Ct. 2546.
Our rejection of our prior case law following Recuenco might not be so troubling if our mode of harmless-error review had been limited to reviewing the trial record with regard to what the jury would almost surely have found on the factual question essential to determining the maximum sentence. But we held in Nordby that review of an Apprendi violation should “encompass[ ] the ‘whole record,’ ” including evidence from sentencing proceedings. 225 F.3d at 1061 n. 6 (quoting Delaware v. Van Arsdall, 475 U.S. 673, 681, 106 S.Ct. 1431, 89 L.Ed.2d 674 (1986)). We derived the notion of “whole record” review from the Supreme Court’s opinion in Delaware v. Van Arsdall, 475 U.S. 673, 681, 106 S.Ct. 1431, 89 L.Ed.2d 674 (1986), which held that a violation of the constitutional right to confront an adversarial witness does not require reversal if the error was harmless beyond a reasonable doubt.
Van Arsdall, however, made no mention of sentencing proceedings. Instead, it concluded that “the Supreme Court of Delaware was wrong when it declined to consider whether [the adverse] ruling was harmless in the context of the trial as a whole.” 475 U.S. at 674, 106 S.Ct. 1431 (emphasis added). And none of the cases Van Arsdall cites as support for “whole record” review mention sentencing proceedings after a trial or a guilty plea either. See United States v. Hasting, 461 U.S. 499, 509, 103 S.Ct. 1974, 76 L.Ed.2d 96 (1983) (“[T]he Court has consistently made clear that it is the duty of a reviewing court to consider the trial record as a whole and to ignore errors that are harmless.” (emphasis added)); Moore v. Illinois, 434 U.S. 220, 232, 98 S.Ct. 458, 54 L.Ed.2d 424 (1977); Harrington v. California, 395 U.S. 250, 253-54, 89 S.Ct. 1726, 23 L.Ed.2d 284 (1969).
Nonetheless, our decision to consider sentencing proceedings in Nordby made a certain amount of sense, as we did so for the purpose of giving the defendant an opportunity to counter evidence that the government introduced at trial — evidence which the defendant might have failed to dispute because he was unaware of its relevance to the jury’s decision until it was declared relevant on appeal. We therefore looked to the sentencing proceedings for a limited purpose — “to assist us in determining what evidence Nordby would have introduced at trial [on the omitted fact] had that issue been relevant.” Nordby, 225 F.3d at 1061 n. 6 (emphasis added). But we refused to consider any post-conviction admissions or stipulations made by the defendant, because we deemed “new admissions by Nordby at sentencing, made after the jury had already rendered its verdict ... irrelevant to [the] inquiry.” Id.
Unfortunately, we later threw Nordby’s limitations to the wind. All that is left of Nordby’s careful excision of the relevant post-conviction material from the irrelevant is the rule that we do not consider the defendant’s post-conviction admissions or stipulations in Apprendi harmless-error review.4 See Butler v. Curry, 528 F.3d 624, 648 n. 16 (9th Cir.2008) (noting “our *1203long-standing rule that admissions at sentencing are not relevant to an Apprendi harmless error analysis”); Lococo, 514 F.3d at 864 (refusing to consider defendant’s statements at sentencing, even though they could be interpreted as an admission, in assessing the harmlessness of an Apprendi error in accepting a guilty plea); United States v. Salazar-Lopez, 506 F.3d 748, 755 (9th Cir.2007) (“[W]e do not consider new admissions made at sentencing in our harmless error inquiry”); United States v. Jordan, 291 F.3d 1091, 1097 (9th Cir.2002) (“A stipulation at sentencing does not address the jury’s finding and cannot be considered under Apprendi.”).
Thus, several opinions of this court since Nordby have considered evidence introduced at sentencing by the government as part of harmless-error, rather than — as in Nordby — plain-error review, see, e.g., Hunt, 656 F.3d at 913-16; Zepeda-Mar-tinez, 470 F.3d at 913, and other circuits have done likewise, see, e.g., United States v. Harakaly, 734 F.3d 88, 96-97 & n. 9 (1st Cir.2013) (declining to decide whether post-conviction concessions “would independently suffice to establish harmlessness,” but relying on them for corroboration of the defendant’s “earlier concessions at his Rule 11 hearing”); United States v. Williams, 493 F.3d 763, 767-68 (7th Cir. 2007). Moreover — again, unlike in Nord-by — we have applied this process of post-conviction evidentiary submission and fact-finding analysis to cases in which there was never a trial at all, even though Ned-er, Recuenco, and Nordby all involved jury verdicts. See Hunt, 656 F.3d at 913-16 (guilty plea); Zepeda-Martinez, 470 F.3d at 913 (same). In the process, we expanded the scope of postconviction evidence considered as to Apprendi harmlessness from evidence the defendant would have admitted at trial to evidence the parties would have admitted at trial, Zepeda-Martinez, 470 F.3d at 913 n. 3. The result was to pave the way for a post-conviction bench trial on a fact never conceded before conviction, even where no pre-conviction trial was ever held.
That is precisely what the government proposed here — a bench trial as a substitute for the trial by jury required by Ap-prendi, in which evidence never submitted to any jury is presented for the first and only time to a judge. To sanction such a procedure is to allow the protections accorded by Apprendi entirely to atrophy.
Were we to countenance governmental introduction of evidence for the sole, explicit purpose of defeating harmless-error review, we would be approving a court doing “just what [the Supreme Court] ha[s] said it cannot: relying on its own finding about a non-elemental fact to increase a defendant’s maximum sentence.” Descamps v. United States, — U.S. -, 133 S.Ct. 2276, 2288-89, 186 L.Ed.2d 438 (2013); see also Shepard v. United States, 544 U.S. 13, 25, 125 S.Ct. 1254, 161 L.Ed.2d 205 (2005) (plurality opinion); id. at 28, 125 S.Ct. 1254 (Thomas, J., concurring in part and concurring in judgment). Descamps applied this prohibition broadly, resoundingly rejecting our circuit’s prior approach, which had authorized “the court to try to discern what a trial showed, or a plea proceeding revealed, about the defendant’s underlying conduct,” and emphasizing that “[t]he Sixth Amendment contemplates that a jury — not a sentencing court — will find such facts, unanimously and beyond a reasonable doubt.” 133 S.Ct. at 2288.
Zepeda-Martinez purportedly used post-conviction evidence as a mere “guide” to determining what would have ensued in a jury trial and then predicting whether a reasonable jury necessarily would have found the facts essential to the crime charged beyond a reasonable doubt. But this mode of analysis is judicial fact-finding of precisely the sort Descamps proscribes: *1204It requires going behind the fact of conviction to establish what evidence the parties would have introduced at a trial, and then relies on evaluation of that evidence to impose a sentence greater than the maximum sentence that Congress intended. Moreover, as in United States v. Aguila-Montes de Oca, 655 F.3d 915, 918 (9th Cir.2011) (en banc), abrogated by Des-camps, 133 S.Ct. at 2282-83, in cases like Zepeda-Martinez and Hunt, and this case, no trial occurred, so guesses about what would have occurred at trial are entirely hypothetical.
III.
The government’s proposal that we affirm the instant sentence on harmless-error review would require us to hold that because the government produced an execution-of-warrant form at sentencing, Guerrero-Jasso’s constitutional right, absent waiver, to have the essential fact of his removal date put to a jury and proved beyond a reasonable doubt can go by the wayside. Quite aside from the reasons we give in the panel opinion for rejecting this proposition, it is one that, in my view, should not be entertained at all. Neither Neder nor Recuenco support such a result. And, by affording penalty provisions that increase a maximum statutory sentence the protections of the Sixth Amendment and reasonable-doubt standard, the Supreme Court has forbid it. See Apprendi 530 U.S. at 477, 120 S.Ct. 2348. Instead, harmless-error review in Apprendi cases must respect the principle that a court may not itself make a finding as to a disputed fact — as opposed to an assessment of harmless error on a preconviction trial record — in situations where fact-finding would increase the statutory maximum. If the defendant did not admit an essential fact during his plea colloquy and evidence concerning the fact was not put to a jury, it violates Apprendi for a court to allow the government, postconviction, to introduce new evidence, find that evidence of the fact would have been introduced in a hypothetical jury trial, and then determine that the essential fact would have been found by the hypothetical jury had the newly produced evidence been before it.
With our expansive approach to harmless error in Apprendi cases, especially in guilty plea cases, Apprendi’s protections could become protections in name only. Indeed, although I have no reason to think that the Apprendi error in this case was engineered by the government, our current Apprendi harmless-error methodology could encourage prosecutors to do exactly that — consciously allow Apprendi error, and then introduce the omitted evidence for the first time at sentencing, thereby bypassing the jury trial protection underlying Apprendi. We should stop this process in its tracks by reconsidering en banc our Apprendi harmless-error cases, particularly Zepeda-Martinez.
. The government also pointed to Guerrero-Jasso’s failure to object to the court’s representation that he was subject to the twenty-year maximum penalty as an admission of post-conviction, removal. For the reasons expressed in the majority opinion, Guerrero-Jasso's preliminary
. Similarly, United States v. Cotton held that failure to charge drug quantity in an indictment was error, but that where the evidence of drug quantity was "overwhelming and essentially uncontroverted,” the error did not "seriously affect[] the fairness, integrity or public reputation of judicial proceedings.” 535 U.S. 625, 632-33, 122 S.Ct. 1781, 152 L.Ed.2d 860 (2002) (internal quotation marks omitted). Perceiving that "[t]he real threat ... to the fairness, integrity, and public reputation of judicial proceedings would be if respondents, despite the overwhelming and un-controverted evidence that they were involved in a vast drug conspiracy, were to receive a sentence prescribed for those committing less substantial drug offenses because of an error that was never objected to at trial!,]” the Court determined reversal for plain error improper. Id. at 634, 122 S.Ct. 1781 (interned quotation marks omitted).
. The defendant-respondent in Recuenco had argued that at the time of his conviction, Washington state law did not provide a procedure whereby a jury could make the finding required by Apprendi. Id. at 217, 126 S.Ct 2546. He went on to maintain that the Washington Supreme Court's structural error holding regarding the Apprendi violation in his case thus rested on an adequate and independent state ground — the lack of a procedural mechanism allowing a jury to make the necessary factual finding. Id. at 216-17, 126 S.Ct. 2546. Recuenco rejected the argument that state law barred it from reaching the merits, but explained that the state's mechanisms for presenting sentence enhancements to a jury were relevant to harmlessness. If the state had no such mechanism, Recuenco *1202reasoned, "that ... suggests that respondent will be able to demonstrate that the ... violation in this particular case was not harmless.” Id. at 218, 126 S.Ct. 2546 (emphasis in original) (citing Chapman, 386 U.S. at 24, 87 S.Ct. 824). Recuenco thus suggests that a sentence above the statutory maximum could not be deemed harmless if the penalty provision that increased the maximum sentence was not, and could not have been, submitted to a jury.
. Even this rule has on occasion been breached. See Zepeda-Martinez, 470 F.3d at 913.