Before SMITH, ELROD, and OLDHAM, Circuit Judges. ANDREW S. OLDHAM, Circuit Judge:* A jury convicted Juan Rojelio Cabello of aiding and abetting his codefendant’s crime: the possession of methamphetamine with intent to distribute it. Cabello asks us to vacate his conviction for three reasons. But this case squarely implicates none of them. This is a case about plain error— and Cabello can’t establish it, so we affirm. I. A. Police arrested Cabello and his codefendant Cristoval Manuel Garcia in June 2020 for trying to sell drugs to an undercover police officer. After the arrest, law-enforcement officers interviewed Cabello. They recorded the interview. We base our summary of the facts on that recording. Cf. Scott v. Harris, 550 U.S. 372, 378–81 (2007). In response to the officers’ questions, Cabello told essentially the following story. Garcia, whom Cabello had first met just a few days earlier, asked for a ride in Cabello’s truck. Cabello agreed. While in the truck, Garcia used Cabello’s phone to contact people. It turns out that Garcia was arranging to sell crystal meth. After some errands not relevant here, the pair went to “Jorge’s” house. While Cabello waited, Garcia went into the house and picked up some meth. The recording is ambiguous about whether Cabello knew what Garcia was doing before Garcia entered the house. In any event, Cabello admitted that by the time the pair left Jorge’s house, he knew Garcia had “at least a half [a gram]” of meth, or “ a little bit more, maybe.” At some point, Garcia told Cabello (in Cabello’s words), “you’re gonna get a hundred dollars.” Cabello, who needed gas money at the time, then drove Garcia to the so-called “deal,” where undercover officers arrested both men. Garcia had about six grams of meth on his person at the time of arrest, and there was some in the truck’s console. Cabello did not have any meth on his person. B. The Government charged Cabello and Garcia in a single-count indictment for violating 21 U.S.C. § 841(a)(1) and (b)(1)(B). Section 841(a)(1) makes it “unlawful for any person knowingly or intentionally . . . to manufacture, distribute, or dispense, or possess with intent to manufacture, distribute, or dispense, a controlled substance.” The elements of this offense are: “(1) knowledge, (2) possession, and (3) intent to distribute the controlled substance.” United States v. Patino-Prado, 533 F.3d 304, 309 (5th Cir. 2008) (per curiam). Section 841(b) describes penalties for various forms of the basic § 841(a) offense. See 21 U.S.C. § 841(b)(1)(B)(viii) (concerning, relevant here, offenses involving “5 grams or more of methamphetamine”). Cabello’s case went to trial. At the close of the Government’s case, Cabello moved for a judgment of acquittal under Federal Rule of Criminal Procedure 29(a). See ibid. (“After the government closes its evidence or after the close of all the evidence, the court on the defendant’s motion must enter a judgment of acquittal of any offense for which the evidence is insufficient to sustain a conviction.”). But he didn’t reassert that motion at the close of all the evidence. See Blue Br. at 17 (conceding this). The jury deliberated for approximately seven hours before convicting Cabello. It began deliberating on September 29, 2020, at 11:25 a.m. and returned its guilty verdict at 6:29 p.m. The record reflects that the jury sent six notes to the judge during its deliberations. Only the fourth and sixth are relevant here. The fourth note read: “We are not going to be able to come up with a unanimous decision.” The judge proposed responding with, “you have your instructions. Please continue to deliberate.” Cabello’s attorney said: “No [objection], your Honor. I’m good with that.” The judge explained his view that “it’s too early to Allen charge them or to propose that” and went on to give the instruction he’d originally proposed. See Allen v. United States, 164 U.S. 492, 501 (1896). Cabello’s attorney once again registered his agreement with the judge’s response to the fourth note. The jury sent its sixth and final note at 6:00 p.m. It said: “We cannot come to a unanimous decision.”[1] After this note, the judge suggested an Allen charge and asked the attorneys what they thought. The prosecutor floated the idea of waiting until the following day to give the charge. The judge said, “I thought we would just do it tonight” and asked Cabello’s counsel what he thought. Cabello’s counsel responded: “Do it tonight. They’re here.” The judge then double-checked, asking Cabello’s counsel if he had “any objection.” Counsel said, “no, sir.” The judge gave the jury a modified Allen charge that reminded them of the importance of the case, reiterated the reasonable-doubt standard, and asked all the jurors to think over their views carefully. Thirteen minutes after the Allen charge, at 6:29 p.m., the jury returned its guilty verdict. C. Cabello raises three issues on appeal. First, he challenges the sufficiency of the indictment. Second, he challenges the sufficiency of the evidence underlying his conviction. And third, he argues the Allen charge coerced the jury into reaching a guilty verdict. The parties correctly agree that plain-error review applies to all three issues. Cabello entirely failed to raise the first and third issues—the indictment’s sufficiency and the Allen charge—in district court. See United States v. Muhammad, 14 F.4th 352, 363 (5th Cir. 2021) (“Muhammad makes this argument for the first time on appeal, so our review is for plain error.”). As for the sufficiency-of-the-evidence challenge, Cabello did make a Rule 29 motion at the close of the Government’s case-in-chief, but he concedes he did not reassert that motion at the close of all the evidence. So plain-error review applies there, too. United States v. Oti, 872 F.3d 678, 686 (5th Cir. 2017) (“Because Oti failed to renew her motion for judgment of acquittal after the jury’s verdict, we review her sufficiency challenge for plain error.”); cf. United States v. Dubin, 27 F.4th 1021, 1033–35 (5th Cir. 2022) (en banc) (Oldham, J., concurring). Thus, to prevail, Cabello must satisfy the strictures of Federal Rule of Criminal Procedure 52(b). That requires a showing “(1) that the district court committed an error (2) that is plain and (3) affects his substantial rights and (4) that failure to correct the error would ‘seriously affect the fairness, integrity or public reputation of judicial proceedings.’” United States v. Sanchez-Hernandez, 931 F.3d 408, 410 (5th Cir. 2019) (quoting Johnson v. United States, 520 U.S. 461, 466–67 (1997)). “The fourth requirement . . . is discretionary,” and “only particularly egregious errors will meet [its] rigorous standard.” Rosales-Mireles v. United States, 138 S. Ct. 1897, 1912 (2018) (Thomas, J., dissenting) (quotation omitted). In the sections that follow, we hold that Cabello cannot show plain error as to (II) the sufficiency of his indictment, (III) the sufficiency of the evidence, or (IV) the Allen charge. II. The indictment first. We (A) explain the legal standards that govern the indictment in this case. Then we consider and reject Cabello’s arguments that the indictment is insufficient because (B) it contains a typo and (C) it failed to allege mens rea. A. “An indictment is sufficient if it [1] contains the elements of the charged offense, [2] fairly informs the defendant of the charges against him, and [3] ensures that there is no risk of future prosecutions for the same offense.” United States v. Harms, 442 F.3d 367, 372 (5th Cir. 2006) (quotation omitted); see also Russell v. United States, 369 U.S. 749, 763–64 (1962) (explaining those requirements). As a corollary, [4] the indictment must allege an actual crime. See United States v. Meacham, 626 F.2d 503, 507 (5th Cir. 1980) (“It is axiomatic that the elements alleged must amount to an offense.”). Our court, however, has long rejected an overly technical approach to evaluating the sufficiency of an indictment. See, e.g., United States v. Rainey, 757 F.3d 234, 247–48 (5th Cir. 2014) (collecting cases spanning decades to this effect). Thus, an indictment that “closely tracks the language” of the statute “under which it is brought” will generally pass muster—”[n]o prescribed set of words are required.” United States v. Franco, 632 F.3d 880, 884–85 (5th Cir. 2011) (per curiam); see also United States v. Fitzgerald, 89 F.3d 218, 222 (5th Cir. 1996) (“The test of the validity of the indictment is not whether the indictment could have been framed in a more satisfactory manner, but whether it conforms to minimal constitutional standards.”). Now consider the text of the indictment in this case. Its charging language reads as follows: COUNT ONE [21 U.S.C. § 846] On or about June 9, 2020 in the Western District of Texas, the Defendants, CRISTOVAL MANUEL GARCIA JUAN ROJELIO CABELLO, Aided and abetted by each other, and others known and unknown to the Grand Jury to possess with intent to distribute a controlled substance, which offense involved five (5) grams or more of actual methamphetamine, contrary to Title 21, United States Code, Sections 841(a)(1) and 841(b)(1)(B). (Brackets in original.) The indictment’s caption clarified the alleged violation by including the following: “Vio: 21 U.S.C. § 841(a)(1) Aiding and Abetting Possession With Intent to Distribute a Controlled Substance.” B. Cabello correctly points out a grammatical defect in the indictment. The indictment said the defendants “to possess with intent to distribute a controlled substance.” (Emphasis added.) That is obviously wrong; it should have said the defendants “possessed with intent to distribute a controlled substance.” Cabello argues the typo caused plain error. Because of the typo, he says, the indictment merely alleges that Garcia and Cabello aided and abetted each other. But aiding and abetting is not itself a crime; it’s a theory of liability. See 18 U.S.C. § 2(a) (“Whoever commits an offense against the United States or aids, abets, counsels, commands, induces or procures its commission, is punishable as a principal.” (emphasis added)); United States v. Pearson, 667 F.2d 12, 13 (5th Cir. Unit B 1982) (per curiam) (“[A]iding and abetting is not a separate crime with elements of its own.”). Thus, says Cabello, the court should’ve dismissed the indictment sua sponte because “[i]t is axiomatic that the elements alleged [in the indictment] must amount to an offense.” Meacham, 626 F.2d at 507. Common sense and circuit precedent say otherwise. The indictment’s text and its caption each clearly referred to 21 U.S.C. § 841(a)(1) as the relevant offense. Those citations “fairly inform[ed Cabello] of the charges against him.” See Harms, 442 F.3d at 372 (quotation omitted). And it’s obvious that the same citations “alleged . . . an offense” against him. See Meacham, 626 F.2d at 507; see also Fitzgerald, 89 F.3d at 222–23 (holding that information provided in the caption of an indictment “can cure a defect in the body of the indictment”). Cabello, perhaps anticipating those points, relies heavily on our decision in Meacham. That case implicated 21 U.S.C. §§ 846 and 963. See 626 F.2d at 507–09. Those statutes provided, with near-identical wording, that “[a]ny person who attempts or conspires to commit any offense defined in this title” would be subject to penalties. 21 U.S.C. §§ 846, 963 (1980). Everyone agreed those statutes criminalized “attempts . . . to commit” certain offenses. See Meacham, 626 F.2d at 508. And everyone agreed those statutes criminalized “conspir[acies] to commit” certain offenses. See id. at 507. The Government, however, took things further. It argued the above- quoted language also criminalized conspiracies to attempt to commit substantive offenses. See ibid. Our court rejected that reading of the statute: “Acceptance of the government’s position would lead to the conclusion that §§ 846 and 963 describe four separate crimes apiece: conspiracy, attempt, conspiracy to attempt[,] and attempt to conspire. We do not believe Congress intended to create four discrete crimes with the three words ‘attempts or conspires.’” Id. at 508. And because the Government’s reading made no sense, the court held the indictment was insufficient because it failed to allege an offense. Id. at 508–09. Meacham thus stands for two propositions: First, you can’t fold a statute in on itself like a piece of origami. Second, if you try it, you’ve failed to allege an offense, and the indictment is therefore defective. Neither of those propositions is controversial, and neither of them has anything to do with this case. The indictment in this case had a typo. That’s it. No one reading the indictment—with the typo or without it—could reasonably wonder what the grand jury alleged. The grand jury alleged that Cabello aided and abetted possession with intent to distribute a controlled substance in violation of § 841. Neither the typo nor Meacham does anything to change that. Cabello therefore cannot show any error, much less can he establish the other requirements of Rule 52(b). C. Finally, Cabello points out that the indictment alleges possession “with intent to distribute” but does not expressly allege knowledge. See 21 U.S.C. § 841(a) (prohibiting only “knowingly or intentionally” possessing drugs). He says that omission required the district court to dismiss the indictment sua sponte. Cf. Harms, 442 F.3d at 372 (indictment must “contain[] the elements of the charged offense”). That argument is squarely foreclosed by United States v. Arteaga- Limones, 529 F.2d 1183 (5th Cir. 1976). The indictment in that case failed “to include ‘knowingly or intentionally’ in its language,” even though the charged statute had that mens rea requirement. Id. at 1199. But two factors overcame the omission: “[T]he language used was accompanied by specification of the statutory section numbers,” and “[t]he jury was charged that they must find knowledge or intent in order to convict.” Id. at 1200. That meant the indictment was not defective and “prevented any injustice to” the defendant. See ibid. And the court’s review in that case wasn’t even for plain error. See ibid. This case presents the same two features. The indictment accompanied its scienter-less allegation with “specification of the statutory section numbers.” Id. at 1199. And immediately after reading the indictment, the judge instructed the jury about the statute’s mens rea requirement. See ibid. The district court committed no error, much less plain error, by not dismissing the indictment sua sponte. III. For the first time on appeal, Cabello argues that § 841 requires proof that he knew Garcia intended to get meth before the pair went to Jorge’s house to get it. Because the Government failed to offer such advance-knowledge evidence, Cabello contends, insufficient evidence supports his conviction for aiding and abetting Garcia’s possession. Preserved challenges to the sufficiency of the evidence get de novo review, with a heavy thumb on the scale in favor of the verdict. See, e.g., United States v. Brown, 727 F.3d 329, 335 (5th Cir. 2013) (asking “whether any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt” (emphasis added) (quotation omitted)). But when a defendant doesn’t properly preserve his sufficiency challenge at trial, that rule combines with plain-error review to produce a super-deferential result. See United States v. Delgado, 672 F.3d 320, 330–31 (5th Cir. 2012) (en banc). Thus, to prevail, Cabello must show that “the record is devoid of evidence pointing to guilt or [that] the evidence is so tenuous that a conviction is shocking.” Id. at 331 (quotation omitted). We will reverse “only if there is a manifest miscarriage of justice.” Ibid. (quotation omitted). We (A) hold that Cabello cannot show any error because his interpretation of the statute is wrong. Then we (B) hold that, in any event, he cannot show plain error—much less a manifest miscarriage of justice. A. The plain text, statutory context, and precedent all confirm that “possession” is an ongoing act; therefore, an aider-and-abettor can have knowledge of the principal’s possession while the possession is ongoing. There’s no justification for Cabello’s argument that the aider-and-abettor must have such knowledge before the possession begins. First, consider the common usage of “possess.” See 21 U.S.C. § 841(a)(1) (making it a crime to “knowingly or intentionally . . . possess with intent to . . . distribute . . . a controlled substance”). “Possession” is a “chameleon-hued word” in legal contexts, and “it can have one of four senses.” Possession, BRYAN A. GARNER, GARNER’S DICTIONARY OF LEGAL USAGE 688 (3d ed. 2011). Yet none of those four senses suggests that possession is, like acquisition, a one-instant phenomenon. Instead, all four suggest possession is a continuing activity. See, e.g., ibid. (listing the first sense as “the fact of having or holding property in one’s power”). Non-legal dictionaries likewise distinguish the ongoing act of possession from the instant act of obtaining or acquiring. Compare Possess, WEBSTER’S NEW INTERNATIONAL DICTIONARY OF THE ENGLISH LANGUAGE 1926 (2d ed. 1941) (one listed sense: “To have and hold as property; to have a just right to; to be master of; to own; as, to possess lands, money, a horse, a watch”), with Obtain, id. at 1682 (one listed sense: “To get hold of by effort; to gain possession of; to procure; to acquire, in any way; as, to obtain one’s ends, wealth, another’s confidence”). Second, statutory context agrees. Congress used the phrase “acquire[] the controlled substance” and the similar “procure . . . the substance” in § 841 itself. See 21 U.S.C. § 841(g)(2)(B), (B)(ii). Thus, if Congress meant to criminalize the knowing or intentional “acquisition” or “procurement” of drugs in § 841(a)(1), it had the vocabulary to do so. It did not. It instead criminalized the “possess[ion]” of drugs. Id. § 841(a)(1). That word connotes a continuing activity—not an activity that ends the moment it begins. Related statutory provisions confirm that § 841 criminalizes the ongoing act of possession. The criminal venue statute, for example, provides that “any offense against the United States begun in one district and completed in another, or committed in more than one district, may be inquired of and prosecuted in any district in which such offense was begun, continued, or completed.” 18 U.S.C. § 3237(a). In order to apply that provision, courts must determine where a crime occurred. And that often requires figuring out the duration of the crime. For these purposes, possession-with-intent under 21 U.S.C. § 841(a)(1) is a “continuing offense,” which means the crime occurs in any district where the defendant, well, possessed the drugs. See United States v. Davis, 666 F.2d 195, 199 and n.5 (5th Cir. Unit B 1982) (“A continuing offense is a continuous, unlawful act or series of acts set on foot by a single impulse and operated by an unintermittent force, however long a time it may occupy. Where such an act or series of acts runs through several jurisdictions, the offense is committed and cognizable in each.”). Because possession is an ongoing action, there’s no justification for Cabello’s advance-knowledge rule. A defendant like Cabello can have knowledge of the possession, even without obtaining that knowledge until partway through the principal’s possession. Cabello’s real argument is that we should rewrite § 841 to criminalize “knowingly or intentionally obtaining with intent to distribute a controlled substance.” We obviously cannot do that. See Epic Sys. Corp. v. Lewis, 138 S. Ct. 1612, 1624 (2018) (“[I]t’s the job of Congress by legislation, not this Court by supposition, both to write the laws and to repeal them.”). Third, Cabello cannot avoid the statutory text and context based on precedent. To the contrary, our decision in United States v. Fischel, 686 F.2d 1082 (5th Cir. 1982), squarely forecloses his advance-knowledge reading of § 841. Fischel showed up to a deal in a motel room, together with another individual. Id. at 1084. They were working together to sell some drugs, but Fischel didn’t handle the drugs himself. He simply sat in a chair and watched as the other man laid them out on the bed in preparation for the deal. See ibid. A jury convicted Fischel of aiding and abetting possession-with-intent under 21 U.S.C. § 841(a)(1). We affirmed: Fischel’s shared control over the disposition of the cocaine when in the hotel room with Ludwig suffices to prove assistance to Ludwig in possession of the cocaine. Fischel need not have pulled the cocaine from his own pocket and maintained total control over it until the consummating of the sale to be found an aider and abettor of Ludwig’s possession. As we have noted, Fischel need only have helped Ludwig’s possession. The jury was warranted in concluding that Fischel aided and abetted both elements of the underlying crime. Id. at 1088–89. The Fischel court didn’t care one bit whether Fischel had foreknowledge of the principal’s possession. It was enough that he “helped [the principal's] possession.” Id. at 1089. Cabello has no convincing explanation for why his case is any different: If anything, his decision to cart around the drugs in his truck amounted to more direct assistance than that in Fischel. Nor is Rosemond v. United States, 572 U.S. 65 (2014), to the contrary. There the Court addressed the requirements for aiding and abetting a crime under 18 U.S.C. § 924(c). Section 924(c) “prohibits ‘us[ing] or carr[ying]‘ a firearm ‘during and in relation to any crime of violence or drug trafficking crime.’” Id. at 67 (quoting § 924(c)). The Court held that, to aid and abet a § 924(c) violation, a would-be accomplice must have “full knowledge of the circumstances constituting the charged offense”—that is, knowledge both that there would be a drug deal and that there would be a gun present. See id. at 77–78. Most relevant here, the Court also held the defendant must have “advance knowledge of a firearm’s presence.” Id. at 81 (emphasis added). Put differently, that means “knowledge that enables him to make the relevant legal (and indeed, moral) choice.” Id. at 78; see also ibid. (“[W]hen an accomplice knows nothing of a gun until it appears at the scene, he may already have completed his acts of assistance; or even if not, he may at that late point have no realistic opportunity to quit the crime. And when that is so, the defendant has not shown the requisite intent to assist a crime involving a gun.” (emphasis added)). The key difference between Rosemond and this case lies in the underlying offense. To aid and abet bringing a gun to a drug deal, the aider- and-abettor obviously must know about the gun beforehand. That’s because the moment the gun shows up, the § 924(c) offense is complete; if the would- be aider-and-abettor knew nothing about the gun before that moment, he could not have the requisite mens rea to aid and abet the offense. Possession, by contrast, is an ongoing offense. It starts the moment the principal takes possession of the drugs, but then it continues as long as he possesses them. And the moment the principal says, “hey, I’ve got some meth,” the aider- and-abettor has knowledge of the possession and the opportunity to make the relevant legal and moral choice: “Should I help the principal continue possessing the meth?” Cabello’s contrary reading of Rosemond leads to absurd results. Imagine Case 1, where A says to B: “Please drive me to Jorge’s house so I can get some meth. I’ll give you $100 for the ride.” Then imagine Case 2, where X says to Y: “I just got some meth from Jorge’s house. I’ll give you $100 if you drive me around to find a buyer.” Cabello would read Rosemond to impose aider-and-abettor liability only on B, and not on Y, because only B had foreknowledge of the possession. That’s absurd. The legal and moral choices facing B and Y are identical. B. Even if Cabello were correct that Rosemond should be extended to § 841, the district court’s failure to do so sua sponte did not constitute plain error. As the Supreme Court has explained: [A] new rule of law, set forth by an appellate court, cannot automatically lead that court to consider all contrary determinations by trial courts plainly erroneous. Many such new rules . . . concern matters of degree, not kind. And a lower court ruling about such matters (say, the nature of a closing argument), even if now wrong (in light of the new appellate holding), is not necessarily plainly wrong. [Rule 52(b)'s] requirement that an error be “plain” means that lower court decisions that are questionable but not plainly wrong (at time of trial or at time of appeal) fall outside the Rule’s scope. Henderson v. United States, 568 U.S. 266, 278 (2013) (emphasis added). Or, in our court’s words, though a “defendant need not show that the specific factual and legal scenario has been addressed” in a prior case, he “must at least show error in the straightforward applications of” existing cases. United States v. Vargas-Soto, 700 F.3d 180, 182 (5th Cir. 2012) (quotation omitted). And of particular relevance here: “[A]n error is not plain if it requires the extension of precedent.” Ibid. (emphasis added); see also United States v. Trejo, 610 F.3d 308, 319 (5th Cir. 2010) (“An error is not plain under current law if a defendant’s theory requires the extension of precedent.” (quotation omitted)); United States v. Rodriguez-Flores, 25 F.4th 385, 390 (5th Cir. 2022) (holding an error was plain precisely because it was obvious under a “straightforward application of” the relevant Supreme Court case, even though this court had not addressed the precise issue since the Court’s decision (quotation omitted)). Cabello’s argument rests, not on “straightforward applications of” precedent, but rather on his quest for “the extension of” that precedent. Vargas-Soto, 700 F.3d at 182 (quotation omitted). Our precedent says that “aiding-and-abetting liability requires knowledge of all elements of the underlying crime.” United States v. McDowell, 498 F.3d 308, 315 (5th Cir. 2007). One of the elements of a 21 U.S.C. § 841(a)(1) offense, of course, is possession. See Patino-Prado, 533 F.3d at 309. But we’ve never—at the time of Cabello’s trial or after it—extended McDowell to require foreknowledge of possession. Performing that extension would be anything but a “straightforward application” of the case. See Rodriguez-Flores, 25 F.4th at 390 (quotation omitted). Likewise, the Supreme Court’s precedent requires “advance knowledge of a firearm’s presence” in the 18 U.S.C. § 924(c) context. See Rosemond, 572 U.S. at 77–81. And it’s anything but “straightforward” to extend that rule to the drastically dissimilar “factual and legal scenario” at play in this case under § 841(a)(1). See Vargas-Soto, 700 F.3d at 182. The trial court did not commit plain error by failing to go where neither we nor the Supreme Court has ventured before.[2] Cabello’s principal counterargument is that we did venture there in United States v. Jackson, 526 F.2d 1236 (5th Cir. 1976). Jackson introduced one codefendant to another codefendant, with the intent that the pair would do a drug deal together. Id. at 1237–38. But Jackson never “exercise[d] dominion or control over” the drugs. Id. at 1237. A jury convicted Jackson of aiding and abetting possession-with-intent under § 841(a)(1). Ibid. This court reversed, holding Jackson hadn’t assisted his codefendants’ possession of the drugs. Id. at 1238. “There was no participation by Jackson in the possession aspect of the transaction on which his conviction of aiding and abetting possession with intent to distribute can be sustained.” Ibid. Note that Jackson didn’t actually state Cabello’s proposed foreknowledge rule. Nor have we ever interpreted Jackson to require foreknowledge of the drug possession. The closest Cabello can get is an out-of-circuit case that, in dicta, characterized Jackson as resting on the fact “that Jackson did not aid or abet his colleague until after the cocaine had come into the co-defendant’s possession.” United States v. Bascaro, 742 F.2d 1335, 1364 (11th Cir. 1984), overruled on other grounds by United States v. Lewis, 492 F.3d 1219, 1221–22 (11th Cir. 2007) (en banc). That is far too thin a reed upon which to rest plain error. If Cabello had objected at trial, he could’ve urged the district court to extend Rosemond or Jackson to require foreknowledge for aider-and-abettor liability in § 841 cases. But the trial judge’s failure to innovate, imagine, and adopt Cabello’s far-reaching proposed extension of precedent sua sponte was not even close to plainly wrong. See Henderson, 568 U.S. at 278; Vargas-Soto, 700 F.3d at 182; FED. R. CRIM. P. 52(b). Much less was it a manifest injustice. See Delgado, 672 F.3d at 330. IV. Finally, Cabello argues the district court’s Allen charge coerced the jury into convicting him. We (A) explain the operative rules; then we (B) hold that Cabello waived his Allen claim. A. Precedent requires us to apply a two-pronged analysis to Allen-charge claims. First, we ask “whether any semantic deviation from approved Allen-charge language was so prejudicial that it requires reversal.” United States v. Andaverde-Tiñoco, 741 F.3d 509, 516 (5th Cir. 2013). Second, even if the charge’s wording was permissible, we ask “whether the circumstances surrounding the use of the charge were coercive” based on “the totality of the circumstances.” Id. at 517 (quotation omitted). The first prong—which we call the semantics prong—is squarely rooted in Supreme Court precedent. See, e.g., Jenkins v. United States, 380 U.S. 445, 445–46 (1965) (per curiam) (holding an Allen charge was impermissible in a case where the judge told the jury, “[y]ou have got to reach a decision in this case”). At most, Cabello raises only a half-hearted challenge to the semantics of his Allen charge,[3] so we’ll say nothing more about prong one. The second prong—which we call the “coercion” prong—is the focus of Cabello’s appeal. And our coercion doctrine is, to put it gently, a mess. Over the years, we’ve developed a loose totality-of-the-circumstances “test” (if you can call it that) for assessing the coerciveness of an Allen charge. See United States v. Eghobor, 812 F.3d 352, 359 (5th Cir. 2015) (surveying cases and mentioning “[f]actors that weigh against coercion include,” then listing the three factors we discuss below). And like so many such tests, ours feels a lot like “judging whether a particular line is longer than a particular rock is heavy.” Bendix Autolite Corp. v. Midwesco Enters., Inc., 486 U.S. 888, 897 (1988) (Scalia, J., concurring in judgment). The first two factors are mutually inconsistent. The first one asks whether “the charge was . . . given prematurely.” Eghobor, 812 F.3d at 359 (quotation omitted). The second asks whether “the jurors were . . . required to deliberate for an unreasonable length of time before the charge was given.” Ibid. (quotation omitted). Put together, these two factors always point in opposite directions: If the first factor tips in favor of a finding of coercion, the second factor necessarily goes against that same finding. Similarly, if the second factor suggests coercion, the first factor necessarily suggests the opposite. Or perhaps our cases embrace some Goldilocks principle for timing an Allen charge: not too soon, not too late, but just right. Either way, the first two factors are hopelessly indeterminate because they either cannot be satisfied simultaneously, or they can be—but only under Goldilocksian circumstances that are unknowable to a district judge ex ante and unreproducible ex post. The third factor in our sub-”test” for coercion doesn’t have anything to do with what the judge did. Instead, it asks whether “the time lapse between the charge and the jury’s decision was . . . unduly short.” Eghobor, 812 F.3d at 359. What’s the time cut-off? We cannot say. All we can tell a hapless district judge contemplating an Allen charge is that we’ll evaluate the timing between the charge and a guilty verdict to make sure it’s not “unduly short.” What’s worse, we cannot explain why the timing between an Allen charge and a guilty verdict should matter at all. The idea behind our timing factor seems to be that a short turnaround between the Allen charge and a guilty verdict suggests the jury felt pressured to convict. But why should we assume that a short turnaround increases the likelihood of a conviction? It’s no answer to cite cases where the jury convicted after brief deliberations: Because the Double Jeopardy Clause prevents the Government from appealing acquittals, this court never sees the cases where a jury acquits after brief deliberations. That sampling bias means we can’t infer anything from the pool of quick-conviction cases that reach our court—and certainly not that brevity tends toward conviction. See George L. Priest & Benjamin Klein, The Selection of Disputes for Litigation, 13 J. LEGAL STUD. 1 (1984) (the seminal work on this topic). And our factors aren’t firmly rooted in Supreme Court holdings, either. Start with the fact that not a single Supreme Court case has ever held a properly phrased Allen charge unconstitutionally coercive. Cf. Jenkins, 380 U.S. at 445–46 (holding a charge was coercive “in its context and under all the circumstances,” but in a situation where the judge flatly said “[y]ou have got to reach a decision in this case” after the jury said it had “insufficient evidence” to reach a verdict (quotation omitted)); Allen, 164 U.S. at 501–02 (discussing only the wording and “substance” of the eponymous Allen charge, without discussing the totality of the circumstances at all). So where did we get our factors? Well, the third factor—the quick-verdict consideration—seems to come from dicta in two Supreme Court decisions. See Lowenfield v. Phelps, 484 U.S. 231, 240 (1988) (explaining, “[w]e are mindful that the jury returned with its verdict soon after receiving the supplemental instruction, and that this suggests the possibility of coercion,” but going on to find no coercion); United States v. U.S. Gypsum Co., 438 U.S. 422, 462 (1978) (noting that “this swift resolution of the issues in the face of positive prior indications of hopeless deadlock, at the very least, gives rise to serious questions,” but only after discussing and condemning the trial judge’s highly unusual decision to meet one-on-one with the jury’s foreman off the record). The origin of the first two factors is even murkier; our best guess is that they came from out-of-circuit, not Supreme Court, precedents. Cf. Eghobor, 812 F.3d at 358–59 (listing the factors and citing to circuit precedents whose genealogy doesn’t obviously trace to the Supreme Court). But cf. United States v. Fossler, 597 F.2d 478, 484–85 (5th Cir. 1979) (ostensibly rooting its analysis in “the Supreme Court’s language in” Jenkins, but going on to hold an Allen charge coercive based on a free-floating review that seems based more in out-of-circuit precedents than anything the Supreme Court did). So our three factors are internally inconsistent and empirically questionable—and they’re largely homegrown rather than grounded in Supreme Court precedent. At least they bring a measure of determinacy to a “coercion” inquiry that might otherwise be totally unguided. Except they don’t even do that. Instead of limiting ourselves to our three coercion factors, our court has repeatedly looked beyond them to various other considerations. For example, in United States v. Betancourt, we approved an Allen charge—but not without noting the jury had had “a long day” and that it “did not return its verdict until 10:23 o’clock of a stormy night.” 427 F.2d 851, 854 (5th Cir. 1970). In United States v. Bottom, we reached the same result and mentioned that: “The time of the day was not late. The day was not Friday or the day before a holiday. The weather was not alleged to be inclement.” 638 F.2d 781, 788 (5th Cir. Unit B Mar. 1981). And in Andaverde-Tiñoco, among other points, we discussed the time of the day the charge was given, the day of the week, the weather, and even whether the jury had decided to skip lunch before convicting. See 741 F.3d at 517–18. To be sure, our court must address arguments raised on appeal—even meritless ones. But how our court handles these arguments is telling. Rather than holding that defendants need to fit their arguments into our three-factor test—flawed though it may be—our court has repeatedly intimated that our three factors are not exhaustive. It’s no wonder defendants keep making weather-, calendar-, and lunch-based arguments in our court. Assessing the state of precedent in this area in 2010, Justice Alito concluded: “About all that can be said is that coercive instructions are unconstitutional, coerciveness must be judged on the totality of the circumstances, and the facts of Lowenfield (polling a deadlocked jury and reading a slightly modified Allen charge) were not unconstitutionally coercive.” Wong v. Smith, 562 U.S. 1021, 1023 (2010) (Alito, J., dissenting from denial of certiorari). Or, to put a finer point on it, our instructions are as follows: “Don’t jump the gun on the Allen charge—the jurors might feel like you’re rushing them. But don’t be too cautious before giving the charge— the jurors might feel like you’re icing them. Beware of Allen charges that spur the jury to convict quickly—but you can ignore those that spur the jury to acquit quickly. One last thing: Make sure to double-check the calendar, the weather forecast, and the juror’s lunch options—you never know what might be relevant on appeal.” B. We need not consider Cabello’s coercion challenge to his Allen charge because he waived it. Longstanding precedent distinguishes between forfeiture of a right and waiver of it. Forfeiture is the “failure to make the timely assertion of a right.” United States v. Olano, 507 U.S. 725, 733 (1993). Waiver, in contrast, “is the intentional relinquishment or abandonment of a known right.” Ibid. (quotation omitted). That distinction matters because “[p]lain-error review is available only for forfeitures—not waivers.” Sanchez-Hernandez, 931 F.3d at 411 n.2; see also United States v. Aparicio, 963 F.3d 470, 473 (5th Cir. 2020) (“To prevail on plain-error review, Aparicio must show (1) an error that has not been affirmatively waived, (2) that is clear or obvious, and (3) that affected his substantial rights.” (emphasis added)). This distinction is as old as Olano itself, where the Court explained that “[d]eviation from a legal rule is ‘error’ [for plain-error purposes] unless the rule has been waived.” 507 U.S. at 732–33 (emphasis added); see also ibid. (going on to give waiver of the right to a trial as an example). That means a defendant who merely fails to object at trial gets plain-error review—but a defendant who waives a right at trial can’t play take-backs on appeal. Cabello waived his Allen-charge claim. The district judge suggested giving the charge in the evening of the jury’s first day of deliberations. The prosecutor hinted that it might be best to hold off until the next morning. It was Cabello’s own attorney who said: “Do it tonight. They’re here.” The judge, playing it safe, then gave him an out by asking, “any objection?” The defense’s answer was unambiguous: “No, sir.” That conversation was an “intentional relinquishment or abandonment” of any right not to have an Allen charge given at that time. See Olano, 507 U.S. at 733 (quotation omitted). Cabello can’t ask the court to give an Allen charge at a given time, get his wish, and then fault the court for doing exactly what he asked it to do. Cf. New York v. Hill, 528 U.S. 110, 113, 118 (2000) (holding a defendant waived even when his defense counsel merely said, “[t]hat will be fine, Your Honor,” and explaining waiver doesn’t require “affirmative conduct” over and above that (quotation omitted)). True, Cabello’s waiver happened by way of counsel. But as a general rule, because counsel is the defendant’s agent, the defendant “must accept the consequences of the lawyer’s decision[s].” Taylor v. Illinois, 484 U.S. 400, 418 (1988); see also ibid. (going on to explain that “[t]he adversary process could not function effectively if every tactical decision required client approval”). That’s true even when counsel’s “tactical decision” ends up impacting a defendant’s constitutional rights, as long as the decision was not so wrong as to amount to ineffective assistance of counsel. See ibid.; Peretz v. United States, 501 U.S. 923, 936 (1991) (“[T]he most basic rights of criminal defendants are . . . subject to waiver.”). Thus, in Taylor, the Court upheld a trial court’s exclusion of a witness—even though the exclusion was a sanction for the defense attorney‘s failure to disclose the witness. See 484 U.S. at 401–02. Neither the Sixth Amendment’s right to present witnesses, see id. at 409, nor the fact that the defendant himself hadn’t made the culpable decision, see id. at 417–18, changed that result. See also id. at 418 (explaining this same rule applies to “the lawyer’s decision to forgo cross-examination, to decide not to put certain witnesses on the stand, or to decide not to disclose the identity of certain witnesses in advance of trial”). Cabello, like the defendant in Taylor, must live with the decision his attorney made. Whether to request an Allen charge at any given moment is a tactical decision, akin to the decisions “to forgo cross-examination, to decide not to put certain witnesses on the stand, or to decide not to disclose the identity of certain witnesses in advance of trial.” See ibid.; see also Hill, 528 U.S. at 111, 115 (holding that a defendant was bound by his counsel’s agreement to a trial date and explaining that “[s]cheduling matters are plainly among those for which agreement by counsel generally controls”). Cabello does not argue that his attorney rendered ineffective assistance by requesting the charge. Nor does he suggest that the timing of an Allen charge is one of the “basic trial choices [that] are so important that an attorney must seek the client’s consent in order to waive the right.” Gonzalez v. United States, 553 U.S. 242, 250 (2008). Accordingly, Cabello must live with the choice that his attorney made on his behalf. It makes no difference that the Government didn’t raise the waiver argument in its brief. As this court has explained, “the government cannot waive the proper interpretation of Rule 52.” Sanchez-Hernandez, 931 F.3d at 411 (applying that rule to explain why it didn’t matter that the Government “concede[d] the first two prongs” of the plain-error analysis). “When an issue or claim is properly before the court, the court is not limited to the particular legal theories advanced by the parties, but rather retains the independent power to identify and apply the proper construction of governing law.” Kamen v. Kemper Fin. Servs., 500 U.S. 90, 99 (1991). Here, both parties ask us to apply Rule 52 and plain error. But we cannot do so where the defendant waived his claim in the district court. See Olano, 507 U.S. at 732–33. AFFIRMED.
