POSNER, Chief Judge.
The government in 1995 indicted 39 members of the Gangster Disciples, a street gang operating in southwestern Chicago and the southern Chicago suburbs, on federal narcotics charges. Twelve of these individuals, convicted by a jury in the federal district court in Chicago after a three-month trial, appeal to us, challenging their convictions and heavy sentences — three of the defendants were sentenced to life in prison, and only four received sentences shorter than 20 years. The government has cross-appealed, complaining that several of the sentences were too short. Many issues are raised but few have sufficient merit to warrant discussion. The evidence of each defendant’s guilt was clearly sufficient and the alleged trial errors were for the most part — though with an important exception — either nonexistent or clearly harmless.
Taking the evidence as favorably to the government as the record permits, as we are required to do, we have a gang some 6,000 strong engaged mainly in the sale of crack and powder cocaine, led by an Illinois state prison inmate named Larry Hoover. By the early 1990s the gang had revenues of some $100 million a year. As befits an operation of such magnitude, the gang had an elaborate structure. Hoover was assisted by a board of directors, and below the board were governors and regents having territorial jurisdictions, along with assistant governors, treasurers, security chiefs, and other officials all with defined responsibilities. The defendants in this case are drawn mainly from the leadership ranks (and include governors, assistant governors, and regents), although some of them merely assisted the leaders.
Some of the government’s strongest evidence was obtained by electronic surveillance of Hoover. Microphones were concealed in the visitors’ badges of Hoo[914]*914ver’s visitors — many of whom were gang officials — and the conversations captured on those microphones were relayed from the prison, which is in southern Illinois, to Chicago, and there recorded, and listened to, by federal agents. Two of the discussion-worthy issues raised by the defendants concern this electronic surveillance. A third issue relating to electronic surveillance (though not of Hoover) requires only the briefest of mentions. It is whether a warrant that authorizes “roving surveillance,” such as the interception authorized here, pursuant to 18 U.S.C. § 2518(11), of calls to and from any cellular phones that one of the Gangster Disciples (Darryl Johnson) might use, violates the Fourth Amendment’s requirement of particularity of description of the place to be searched. Cellular phones have no fixed locus and here were not even identified by a telephone number. But the cases hold that such roving surveillance is constitutional, United States v. Gaytan, 74 F.3d 545, 553 (5th Cir.1996); United States v. Bianco, 998 F.2d 1112, 1120-25 (2d Cir.1993); United States v. Petti, 973 F.2d 1441, 1443-45 (9th Cir.1992); see also Michael Goldsmith, “Eavesdropping Reform: The Legality of Roving Surveillance,” 1987 U. Ill. L. Rev. 401, 415-25, and we have noth-, ing to add to their analysis of the issue.
The first issue we do want to discuss is whether the chief judge of the federal district court in the Northern District of Illinois (which is mainly Chicago) had jurisdiction to authorize the surveillance. Title III, the federal statute regulating electronic surveillance, authorizes an interception order by a judge “within the territorial jurisdiction of the court in which the judge is sitting.” 18 U.S.C. § 2518(3). Hoover’s prison is in the Southern District of Illinois and the defendants argue that therefore the judge lacked the power to issue the order. If this is right, the evidence obtained by the surveillance was inadmissible. 18 U.S.C. §§ 2515, 2518(10)(a); see also United States v. Ojeda Rios, 495 U.S. 257, 260 n. 1, 110 S.Ct. 1845, 109 L.Ed.2d 224 (1990). The government points out that so far as bears on this case “interception” is defined as “the aural or other acquisition” of the contents of a communication, 18 U.S.C. § 2510(4), and that an “acquisition” took place in the Northern District, since the agents first listened to the conversations in Chicago. This is literally true and has persuaded the other courts in which the issue has arisen to uphold the government’s position, United States v. Denman, 100 F.3d 399, 402-04 (5th Cir.1996); United States v. Rodriguez, 968 F.2d 130, 135-36 (2d Cir.1992); see also United States v. Tavarez, 40 F.3d 1136, 1138 (10th Cir.1994); cf. United States v. Ramirez, 112 F.3d 849, 852 (7th Cir.1997), but it creates, as the government’s lawyer acknowledged with refreshing candor at argument, a potential for abuse that resembles the familiar problem of “judge shopping” for conventional search and arrest warrants. Candace McCoy, “The Good-Faith Warrant Cases — What Price Judge-Shopping?,” 21 Crim. L. Bull. 53, 62 (1985); see also United States v. Leon, 468 U.S. 897, 918, 104 S.Ct. 3405, 82 L.Ed.2d 677 (1984). This is true even though the cases gloss “acquisition” to mean “first acquisition.” United States v. Denman, supra, 100 F.3d at 403; United States v. Rodriguez, supra, 968 F.2d at 136. The government still could ask any federal judge in the United States to issue an interception order, and simply arrange for the intercepted communications to be relayed to the judge’s district and listened to there by federal agents. The judge might be in Hawaii, the intercepted communication in Florida, and the investigation to which the interception pertained in Maine.
Although the potential for abuse is undeniable, it does not authorize us to rewrite the statute, especially because the defendants do not argue that the potential has ever become actual and because their position, while curing one problem, would create another — namely that interception orders would often have to be obtained from judges at locations wholly adventitious in relation to the investigation to which the interception pertained. Admittedly this is [915]*915a feature of discovery practice as well, and so is not quite the anomaly that the government depicts it as. See, e.g., Fed. R.Civ.P. 45(a)(2). Still, it is sheer accident that Hoover was imprisoned in southern Illinois rather than in the Northern District of Illinois, or for that matter in Colorado or Indiana; the location of his prison bears no relation to the location of his and his confederates’ crimes and of the government’s investigation of those crimes. And this means that the interest in privacy that the statute seeks to protect is likely to be better protected under the government’s interpretation, because the judge who is familiar with the investigation is in a better position to appraise the materiality of the communications that the government wants to intercept.
Furthermore, although the parties have assumed that the reference to “the territorial jurisdiction of the court” is to the district in which the judge sits, this is not certain, since for many purposes the jurisdiction of a district court extends beyond the boundaries of the district. For example, the personal jurisdiction of a federal district court often extends beyond the district and even state boundaries, and indeed to the nation as a whole under statutes that provide for nationwide service of process. Some districts are coterminous with entire states that are much larger than other districts; compare the District of Montana with the Southern District of New York. The position for which the defendants contend would not cure the abuse that concerns them. This is a problem for Congress to solve if the problem is serious enough to warrant solution.
The next and most troublesome issue concerns the requirement of prompt judicial sealing of recordings of intercepted communications. Because tape recordings of conversations are powerful evidence yet susceptible to tampering that may be extremely difficult to discover, they must “be done in such way as will protect the recording from editing or other alterations.” 18 U.S.C. § 2518(8)(a). To this end, the recording must be judicially sealed “immediately upon the expiration of the period of the order, or extensions thereof,” id.; see United States v. Ojeda Rios, supra, 495 U.S. at 263, 110 S.Ct. 1845; United States v. Plescia, 48 F.3d 1452, 1463 (7th Cir. 1995); United States v. Wong, 40 F.3d 1347, 1375 (2d Cir.1994), and if it is not sealed immediately it can’t be used in evidence unless the government offers a “satisfactory explanation” for the delay in sealing. 18 U.S.C. § 2518(8)(a); United States v. Ojeda Rios, supra, 495 U.S. at 262-65, 110 S.Ct. 1845; United States v. Plescia, supra, 48 F.3d at 1463; United States v. Maxwell, 25 F.3d 1389, 1393 (8th Cir.1994); United States v. Pedroni, 958 F.2d 262, 265 (9th Cir.1992).
The recordings of Hoover’s intercepted conversations were not sealed until 32 days after the expiration of the surveillance warrant. That was much too long to qualify as an immediate sealing, United States v. Williams, 124 F.3d 411, 429-30 (3d Cir.1997); United States v. Wilkinson, 53 F.3d 757, 759-60 (6th Cir.1995); United States v. Wong, supra, 40 F.3d at 1375; United States v. Pitera, 5 F.3d 624, 627 (2d Cir.1993), and so we must consider whether the government’s explanation was adequate. The warrant was to expire on January 2,1994, but on December 19 Hoover had discovered the concealed microphone, interrupting the surveillance. The government wanted to continue recording but with a smaller microphone that Hoover would be less likely to discover. It needed, or rather thought it needed, access to the existing recordings in order to compare them with recordings made with the new microphone to make sure the new recordings were acoustically no worse than the old. The old ones had been so poor, United States v. Parks, 100 F.3d 1300, 1302 (7th Cir.1996), that if the new were worse (as they eventually turned out to be), there was no use installing them in the visitors’ badges. Since the government wanted to have access to the old tapes for purposes of comparison, it didn’t want them sealed.
[916]*916If this were all there was to the government’s explanation for failing to have the recordings judicially sealed immediately, it wouldn’t be enough. A satisfactory explanation is one that is “objectively reasonable,” United States v. Ojeda Rios, supra, 495 U.S. at 266-67, 110 S.Ct. 1845; United States v. Quintero, 38 F.3d 1317, 1328-30 (3d Cir.1994); United States v. Carson, 969 F.2d 1480, 1497-98 (3d Cir. 1992); cf. Tuke v. United States, 76 F.3d 155, 156-57 (7th Cir.1996), as well as sincere. United States v. Quintero, supra, 38 F.3d at 1326-27; United States v. Vastola, 989 F.2d 1318, 1325 (3d Cir.1993). Hoover’s conversations had been recorded in triplicate, so the government could have sealed one set of recordings and used one of the others to compare with recordings made with the new microphone. It is surprising, to say the least, that the Assistant U.S. Attorney in charge of the investigation, a long-time senior member of the U.S. Attorney’s office, Ronald Safer, was unaware that there were duplicate record ings of the intercepted conversations. (It would be reckless not to record in duplicate or, as here, triplicate, since it is so easy for recording equipment to fail.) Such an oversight could not be thought reasonable. One of the sets of tapes could have been sealed, leaving two others for purposes of comparison. And it is better, from the standpoint of minimizing the risk of tampering, to seal the tapes and then unseal them as needed than to leave them unsealed for an indefinite time (though the statutory permission to leave the original tapes unsealed during any extension of the original surveillance warrant, 18 U.S.C. § 2518(8)(a); United States v. Carson, supra, 969 F.2d at 1487, makes the requirement of sealing a rather porous prophylactic against tampering). Nor is it obvious why comparison was a necessary or useful method for determining audibility. Recordings made with the new microphone either were, or were not, audible. Listening to the old tapes would cast little if any light on the new.
But there is more. First and least, Safer believed that he didn’t have to have the tapes sealed as soon as the interception warrant expired, because he anticipated seeking an extension of the warrant within what he thought a reasonable time (30 days) after its expiration. He was confident that the new recording system would be up and running by then and he thought that during this period he would need the original recordings for purposes of comparison. When toward the end of this period he realized it wouldn’t be ready in time, he had them sealed at last. But thirty days is merely the maximum period for which electronic surveillance can be authorized, 18 U.S.C. § 2518(5); it has no relevance to the period within which surveillance recordings must be sealed. There is no basis in the statute or the case law (nor was there when Safer had to make the decision whether to have the recordings sealed, which is the relevant time, United States v. Ojeda Rios, supra, 495 U.S. at 266, 110 S.Ct. 1845), for a rule that the government can leave surveillance recordings unsealed for up to 30 days while it ponders whether to seek an extension. The government must have a reason for such a delay. It is true that months later, in May, the government obtained a new authorization to record Hoover’s conversations. But “an order authorizing surveillance of the same subject, at the same location, regarding the same matter as an earlier authorized surveillance, constitutes an ‘extension’ of the earlier authorization for purposes of section 2518(8)(a) if, but only if, the new authorization was obtained as soon as administratively practical or any delay is satisfactorily explained, i.e., is shown to have occurred without fault or bad faith on the part of the government.” United States v. Carson, supra, 969 F.2d at 1488.
So large a mistake of law as thinking that one has an automatic 30 days to seal surveillance tapes, and so large a mistake of fact as not realizing that multiple tapes were cut, are difficult to describe as being “without fault.” Safer’s affidavit, the only evidence the government tendered with regard to the reasonableness of the delay, states that he believed that “30 days was [917]*917well within that reasonable period of time given the nature of this extension, i.e., the same place of intercept, same criminal conduct, same interceptees,” but the affidavit gives no reason for picking 30 days; nor is the fact that good grounds existed for the extension a rational basis for delay in seeking it — the opposite might well be argued. The affidavit adds that Safer “wanted to have the original tapes available for comparison to tapes produced by the new device,” but does not explain why this was necessary when there were three sets of original tapes.
The government has an alternative ground for affirmance on this point— that the recordings didn’t have to be sealed because an order extending the original interception order had not yet expired. 18 U.S.C. § 2518(8)(a). This ground was not presented to the district court, but the government asks us to overlook the forfeiture because the facts underlying the argument are not contested. But we cannot do that when the case is before us after a trial. We can affirm a judgment on any ground that was not waived or forfeited in the district court, unless one of the conditions for relieving a party from a waiver or forfeiture is present. Jenkins v. Nelson, 157 F.3d 485, 497 (7th Cir.1998); Door Systems, Inc. v. Pro-Line Door Systems, Inc., 83 F.3d 169, 173-74 (7th Cir. 1996); Singletary v. Continental Illinois National Bank & Trust Co., 9 F.3d 1236, 1240 (7th Cir.1993); cf. Rowe v. Schreiber, 139 F.3d 1381, 1382 and n. 2 (11th Cir. 1998). The qualification that we have italicized is not always explicit, see, e.g., Massachusetts Mutual Life Ins. Co. v. Ludwig, 426 U.S. 479, 481, 96 S.Ct. 2158, 48 L.Ed.2d 784 (1976) (per curiam), though we have found only two cases in which it was explicitly rejected (but without discussion). African American Voting Rights Legal Defense Fund, Inc. v. Villa, 54 F.3d 1345, 1356 (8th Cir.1995); International Ore & Fertilizer Corp. v. SGS Control Services, Inc., 38 F.3d 1279, 1286 (2d Cir. 1994). Plenty of other cases, it is true, have broad language that might be thought to imply the rejection of the qualification. E.g., Hernandez v. Starbuck, 69 F.3d 1089, 1093 (10th Cir.1995). But that language is probably inadvertent, or influenced by the exceptions built into waiver doctrine, such as the plain-error doctrine, of which more shortly. The qualification that the language of these cases occludes (that the ground not have been waived or forfeited in the district court) is easily overlooked because a party is not required to advance all its possible grounds in a motion for judgment on the pleadings or summary judgment, with the result that the failure to advance a ground, and the resulting failure of the district court to address it, do not work a forfeiture. Door Systems, Inc. v. Pro-Line Door Systems, Inc., supra, 83 F.3d at 173-74; cf. Curran v. Kwon, 153 F.3d 481, 487 and n. 11 (7th Cir.1998). But a ground not raised at trial is forfeited and therefore cannot be used on appeal. “[A] defendant can move to dismiss or for summary judgment on fewer than all possible grounds without waiving the others, ... but if the case goes to trial he cannot hold some of his grounds in reserve for use should he lose on the grounds he does present.” Smith v. Richert, 35 F.3d 300, 305 (7th Cir.1994).
Perhaps, though, it would be a plain error to reject the government’s ground; and while it is unusual for the government to be arguing plain error in a criminal case, there is nothing to prevent its doing so. United States v. Brown, 164 F.3d 518, 522 (10th Cir.1998); United States v. Zeigler, 19 F.3d 486, 494 (10th Cir.1994); United States v. Sprei, 145 F.3d 528, 533-34 (2d Cir.1998); United States v. Barajas-Nunez, 91 F.3d 826, 833-34 (6th Cir.1996). (See also Fed. R. Crim. P. 52(b), which draws no distinction between the government and the defendant.) It is true that the government has failed to argue plain error to us; it has argued error, but has failed to argue that the error was plain, not realizing, apparently, that it had forfeited the error in the district court and thus had to invoke the plain-error rule to prevail. But when an [918]*918error is plain, the interests of justice require the court, if it can, to notice the error without prompting rather than to perpetrate an unjust decision. Unhappily for the government, the error is not plain; it is not an error at all. For remember that the new authorization to conduct surveillance of Hoover’s conversations was not sought until May, and the previous order had expired the previous January. Recordings cannot be left unsealed indefinitely just because months or years later the government is able to convince a judge to allow the surveillance to resume. Allowing such a hiatus would defeat the purpose of the requirement of sealing. See United States v. Ojeda Rios, supra, 495 U.S. at 263-64, 110 S.Ct. 1845; United States v. Carson, supra, 969 F.2d at 1488, 1497-98.
If the requirement is violated without reasonable excuse, evidence obtained in violation of it must be excluded, period; there is no mitigation beyond what the excuse provision itself allows. 18 U.S.C. § 2518(8)(a); United States v. Ojeda Rios, supra, 495 U.S. at 260, 110 S.Ct. 1845. The harmless-error rule is applicable, but the government does not and could not argue harmless error here, since the Hoover conversations were the linchpin of its case.
The government has one last string to its bow. Although not in Safer’s affidavit, the government argues in its brief to us that the real reason for the delay was that it expected the new bugging apparatus to be completed sooner. Remember that Hoover discovered the original bug on December 19, at which point the government had two weeks to obtain either an extension or a judicial seal. If on January 2 the government reasonably expected the new bug to be completed and in working condition within a few days, this would be a reasonable basis for delaying the seeking of an extension for a few days. At some point it became clear that “a few days” were going to stretch on indefinitely; and then the government, having no immediate use for an extension (which depended on the new apparatus), did seek to have the recordings placed under judicial seal. If the technicians kept assuring the prosecutors that the bug was a day away from completion, naturally the prosecutors would think they could wait another day. As we say, this was argued in the government’s brief but does not appear in Safer’s affidavit — which doesn’t mean it’s untrue, especially since it was one of the reasons the district judge gave for allowing the recordings of Hoover’s conversations to be admitted into evidence; and the defendants do not argue that the government waived the point in the district court. There is no suggestion that the government postponed the sealing of the tapes in order to tamper with them, and in the absence of any such suggestion we have no reason to doubt that the delay did result from a mistake about when the new bugging device would be available. The defendants reply that Safer and the technicians should have communicated with each other more effectively, which is true; but the failure of communication does not strike us, any more than it struck the district judge, as so wanton a blunder as not to constitute a (barely) satisfactory explanation within the meaning of the statute.
A few more issues require discussion. Defendant Yates complains about the absence of his lawyer from the instructions conference. Such a denial, if it is deemed as Yates asks us to deem it an abandonment by the lawyer of his client, leaving the client without representation, rather than merely a failure to come up to a minimum standard of legal professionalism, would require reversal irrespective of prejudice. Roe v. Flores-Ortega, — U.S. -,-, 120 S.Ct. 1029, 1038-39, 145 L.Ed.2d 985 (2000); United States v. Cronic, 466 U.S. 648, 658-60, 104 S.Ct. 2039, 80 L.Ed.2d 657 (1984); United States v. Morrison, 946 F.2d 484, 503 (7th Cir. 1991) (dictum); cf. Neder v. United States, 527 U.S. 1,-, 119 S.Ct. 1827, 1833, 144 L.Ed.2d 35 (1999); United States v. Santos, 201 F.3d 953, 959-60 (7th Cir.2000). [919]*919But in the particular setting of this case, with multiple defendants and multiple counsel, we do not think the lawyer’s missing one instructions conference constituted abandonment. Yates was one of twelve defendants. All the other defendants’ lawyers were present at the conference and with one exception he is unable either to specify a defense peculiar to him that might have warranted a special instruction had his lawyer been there to urge it on the judge or to indicate any respect in which the other lawyers failed to protect the interests common to all the defendants including himself. With that exception, he had virtual representation by the other lawyers of a kind that is commonplace in multidefendant criminal cases, as illustrated by the rule that in such a case an objection by one defendant’s lawyer preserves the objection for the other defendants. E.g., United States v. Gatling, 96 F.3d 1511, 1521 (D.C.Cir.1996). The exception has to do with Yates’s defense that he had withdrawn from the gang in time to avoid liability for the acts that were attributed to him as a member of the conspiracy. That was his defense alone and the lawyers for the other defendants did not press it at the conference. But there was a later instructions conference, which, though abbreviated, gave Yates (by now representing himself by his own choice) the opportunity, which he took advantage of, to press for such an instruction. A criminal defendant who decides to represent himself will not be heard to complain that he was denied the effective assistance of counsel. Faretta v. California, 422 U.S. 806, 834-35 n. 46, 95 S.Ct. 2525, 45 L.Ed.2d 562 (1975); United States v. Chapman, 954 F.2d 1352, 1363 (7th Cir.1992).
Several of the defendants press on us Richardson v. United States, 526 U.S. 813, 119 S.Ct. 1707, 143 L.Ed.2d 985 (1999), which was decided after the trial in this case and holds that a conviction for participation in a continuing criminal enterprise requires that the jury agree unanimously on the specific acts that are the predicate for such a conviction. Our cases prior to Richardson imposed no such requirement and so the judge didn’t give such an instruction. The jury, however, found all three defendants guilty of many more than three predicate offenses relating to the drug conspiracy. The jury thus unanimously agreed that each of the defendants had committed three of the predicate offenses with which he was charged, showing that the omission of the instruction was a harmless error. Lanier v. United States, 205 F.3d 958, 964-65 (7th Cir.2000); Murr v. United States, 200 F.3d 895, 904-06 (6th Cir.2000); United States v. Long, 190 F.3d 471, 476 n. 3 (6th Cir. 1999); United States v. Escobar-de Jesus, 187 F.3d 148, 161-62 (1st Cir.1999); compare United States v. Brown, 202 F.3d 691, 699-703 (4th Cir.2000).
The only other issues that merit discussion concern sentencing. First is whether Yates was properly sentenced to life imprisonment for being “one of several ... principal administrators, organizers, or leaders” of a continuing criminal enterprise, namely the Gangster Disciples. 21 U.S.C. § 848(b)(1). Yates was a “governor,” one of about ten, which put him two levels below the top of the Disciples hierarchy. The top level was occupied by Hoover and the second level by the board of directors, for unlike the conventional corporation the Disciples board reported to its CEO rather than vice versa. We do not know how large the board was (in fact there were two boards, one for Disciples in prison and the other for those at large, though we can ignore this detail), and it rather strains the ordinary meaning of the word “several” to describe Yates as one of “several” administrators of the enterprise. The government asks us to count from the bottom up rather than from the top down, pointing out that since the Disciples had about 6,000 members during the period at issue, Yates belonged to a relatively quite tiny layer of top-level supervisors and the evidence is that he had six regents and 411 rank and file Disciples under his command.
[920]*920The statute’s drafters probably did not envisage such a large criminal enterprise, for the minimum annual receipts of a continuing criminal enterprise that are necessary to make a principal administrator, organizer, or leader subject to mandatory life imprisonment is $10 million, 21 U.S.C. § 848(b)(2)(B), which is less than a tenth of the annual gross receipts that the judge could and did attribute to the Gangster Disciples. We think the literal meaning is not strained overmuch by construing “several” in relative rather than absolute terms, the better to carry out the purpose behind the provision, although we cannot find a case dealing with the issue. Of course there are limits to the relativity of “several.” If we assume there were 30 GDs at Yates’s level or higher, that would be one-half of one percent of the total number of conspirators, and it would be distinctly odd to think that a reference to “several” Americans could be to 1,375,000 people; but we think “several” will stretch to 30, bearing in mind the statute’s objective.
Next is the vexing question, made urgent by the Supreme Court’s recent decision in Jones v. United States, 526 U.S. 227, 119 S.Ct. 1215, 143 L.Ed.2d 311 (1999), as well as by dicta in Edwards v. United States, 523 U.S. 511, 515, 118 S.Ct. 1475, 140 L.Ed.2d 703 (1998), and cases such as United States v. Dale, 178 F.3d 429, 432-34 (6th Cir.1999), whether type and quantity of drugs are elements of the federal drug offense that is created by 21 U.S.C. § 841 and so must be proved at trial beyond a reasonable doubt, or are mere sentencing factors to be determined by the judge, applying a lower standard of proof, at the sentencing hearing. Jones construed a federal carjacking statute that appeared to make the infliction of grave bodily injury in the course of a carjacking a sentencing factor as making it an element of the crime. It did so in order to avoid the constitutional problem that would be presented if Congress tried to skirt the requirement of proof beyond a reasonable doubt and the right of trial by jury in criminal cases by redefining elements of a crime as sentencing factors. One can imagine in the limit replacing the separate statutes for assault and murder by a single statute in which the violator would be punished by probation if he committed an assault that caused no injury at all and by death if the assault consisted in the intentional killing of the victim.
The avoidance of the constitutional issue by statutory construction is not available in the case of section 841, because the division between the elements of the crime and factors relating to how severely to punish offenders is much clearer than in the statute interpreted in Jones. Subsection (a), captioned “Unlawful acts,” defines the offense of distributing, etc. a controlled (or counterfeit controlled) substance, while subsection (b), captioned “Penalties,” specifies how the “person who violates subsection (a) ... shall be sentenced” — namely more severely depending on the type and quantity of the drug. The defendants in this case were convicted of distributing a variety of drugs, including marijuana, the distribution of which calls for a much lighter sentence in section 841(b) than other drugs, notably crack cocaine, one of the major commodities sold by the Gangster Disciples. It is apparent that Congress intended the type and quantity of the drugs distributed by a defendant convicted under section 841(a) to be determined at sentencing, unlike the situation in Jones, and Congress’s determination of the appropriate allocation of decisional responsibilities carries a presumption of constitutionality.
We adhere to our decisions holding that the allocation is constitutional. United States v. Arango-Montoya, 61 F.3d 1331, 1338-39 (7th Cir.1995) (per curiam); United States v. Trujillo, 959 F.2d 1377, 1381-84 (7th Cir.1992); cf. United States v. Edwards, 105 F.3d 1179, 1180 (7th Cir.1997), aff'd. on other grounds, 523 U.S. 511, 118 S.Ct. 1475, 140 L.Ed.2d 703 (1998); see also United States v. Thomas, No.98-1051, 204 F.3d 381, 384 (2d Cir.2000) (per curiam); United States v. Swiney, 203 F.3d 397, 404 n. 5 (6th Cir.2000); United States [921]*921v. Hester, 199 F.3d 1287, 1291-93 (11th Cir.2000); United States v. Jones, 194 F.3d 1178, 1183-86 (10th Cir.1999); United States v. Williams, 194 F.3d 100, 104-07 (D.C.Cir.1999). We emphasize a reason that is practical rather than traditional, although it may explain the tradition. A glance at section 841(b) reveals numerous and minute gradations. For example, a heavier punishment is prescribed for distributing 50 or more grams of a mixture or substance containing crack than for distributing 5 or more grams, 21 U.S.C. §§ 841(b)(1)(A)(iii), (B)(iii), and a heavier punishment for distributing 5 or more grams than for distributing fewer than 5 grams, §§ 812 Schedule II, 841(b)(1)(C), even though the differences among these quantities is slight (there are about 28 grams to an ounce). If a jury were required to determine whether the defendant had distributed 3, 6, 49, or 52 grams of mixture or substance containing crack, its attention would be deflected from the question at once more fundamental to culpability and more manageable by a lay factfinder whether the defendant had distributed a forbidden substance. Similar problems would attend a requirement that the jury discriminate among particular controlled substances, such as powder and crack cocaine.
The defendants’ argument amounts to saying that the federal sentencing guidelines must be administered by juries, with the exception of the criminal history provisions, which the defendants concede, as they must, Jones v. United States, supra, 526 U.S. at 246-50, 119 S.Ct. at 1226-27, identify proper sentencing considerations. But the guidelines are too complicated to be applied by lay persons; even lawyers and judges cannot apply them without training and experience. The Constitution does not require the impossible. The practical effect of the defendants’ argument would be the elimination of most gradations in criminal punishment. We are reluctant to embark upon a path that leads to such a dubious destination. We grant that our position is less compelling when the issue is the type rather than the amount of the drug, but note that in this case, even if type were a jury issue, the failure to instruct the jury that it had to decide whether the defendants were selling cocaine or marijuana would be harmless, as the evidence is overwhelming that it was the former. United States v. Barnes, 158 F.3d 662, 668 (2d Cir.1998). We add that the due process clause protects defendants from being sentenced on the basis of unreliable evidence, albeit it does not give them all the protections that the Constitution has been interpreted to give criminal defendants at the guilt phase of their trials.
Last we consider the government’s cross-appeal. The district judge properly increased the offense levels of four of the defendants — “regents,” each of whom supervised more than a hundred Gangster Disciples — three steps under a provision of the guidelines commanding such a punishment bonus for managers or supervisors of a criminal activity involving five or more participants. U.S.S.G. § 3Bl.l(b). But then he reduced their offense levels two steps under § 3B1.2(b), which provides for such a reduction for a minor participant, defined as one “less culpable than most other participants.” Id., Application Note 3. The judge — who made this reduction though not requested to do so by the defendants — considered regents minor participants in relation to some of the other defendants, who were governors, and to some of the other members of the conspiracy, such as Hoover and the members of his boards of directors. The government argues that a section 3B1.1 sentencing bonus and a 3B1.2 sentencing reduction are not possible in the same case, pointing to an introductory comment to chapter 3 of the guidelines that describes these as alternatives: “When an offense is committed by more than one participant, § 3B1.1 or § 3B1.2 (or neither) may apply.”
The argument was not made to the district judge, but the government argues [922]*922that his error was plain. The government could, but does not, argue that it should not have to show a plain error — that because the judge made the reduction without any forewarning, the government was surprised and should not be held to have forfeited its objection to the judge’s action. E.g., United States v. Muzika, 986 F.2d 1050, 1055 (7th Cir.1993); United States v. Alba, 933 F.2d 1117, 1120 (2d Cir.1991). We think there was error here and that it was plain, though the government goes too far in arguing that there can never be a situation in which a defendant could receive both a punishment bonus for being a manager or supervisor and a punishment discount for being a minor participant. Section 3B1.2 does not say that a manager or supervisor cannot be a minor participant; all that is required is that he be less culpable than most of the other participants. In a case such as United States v. Tsai, 954 F.2d 155, 166-67 (3d Cir.1992), involving a top-heavy conspiracy in which the managers outnumbered the rank and file, it is possible for one of the managers to be less culpable than most of the participants though more culpable than the (few) foot soldiers, and then both adjustments would be possible. But the present case involves a conspiracy with 6,000 participants, and since the defendants in question were, as regents, members of a small supervisory layer consisting of no more than 2 percent of the membership, they clearly were not less culpable than “most” of the participants, and so they were not entitled to the section 3B1.2 reduction. So clear is this that the judge’s sentencing error must be deemed plain, provided that the error is prejudicial. United States v. Olano, 507 U.S. 725, 732-35, 113 S.Ct. 1770, 123 L.Ed.2d 508 (1993). But we think it was, as it is evident that the judge would have sentenced these defendants more heavily if he had not given them a minor-participant discount. In the obverse situation, where a criminal defendant establishes that the judge by virtue of committing a clear error gave him a heavier sentence, the error is deemed plain and the defendant is ordered resentenced. E.g., United States v. Spears, 159 F.3d 1081, 1088 (7th Cir.1998); United States v. Szabo, 147 F.3d 559, 561-62 (7th-Cir.1998); United States v. Whiting, 28 F.3d 1296, 1310-12 (1st Cir.1994).
The four regents must be resentenced; in addition the government concedes that the conspiracy convictions of three of the defendants must be vacated in accordance with Rutledge v. United States, 517 U.S. 292, 307, 116 S.Ct. 1241, 134 L.Ed.2d 419 (1996). With these modifications, the judgments are
APFIRMED.