EASTERBROOK, Circuit Judge.

Judges and genies have this in common: by granting supplicants exactly what they wish for, they may produce misery and regret. Eakin v. Continental Illinois National Bank, 875 F.2d 114, 118 (7th Cir. 1989). Alan Masters, sentenced to 40 years in prison for racketeering activities that ran the gamut from protecting bookies to soliciting the murder of his wife, asked to be resentenced under the sentencing guidelines. He had received the maximum sentence for the crimes of which he had been convicted (consecutive 20-year terms) and believed that the guidelines required a lower sentence. We granted his wish. 924 F.2d 1362, 1369 (7th Cir.1991). Once again the district court selected a 40-year term— but with a big difference. Under the first sentence, Masters was eligible for parole after 10 years, and a well-behaved prisoner had to be released after serving two-thirds of his sentence, shortened by 120 days of good time credit per year. 18 U.S.C. §§ 4161, 4205(a), 4206(d) (1976 ed.). Imprisonment under the Sentencing Reform Act of 1984, by contrast, comes without possibility of parole, and a prisoner earns good time at a maximum rate of 54 days per year. 18 U.S.C. § 3624(b). Masters now rues his request and appeals again. So does his henchman James Keating, who on the first appeal adopted Masters’ arguments but now disclaims any desire for a guideline sentence.

On the first appeal Keating adopted all of Masters’ arguments, adding a few of his own. We understood him to seek re-sentencing under the guidelines and remanded so that the district judge could determine whether the conspiracy ended *283 before November 1, 1987, the date the new sentencing system took effect. Keating did not seek rehearing, and in the district court he clamored for the benefit of the guidelines — until he saw how the prosecutor calculated the' guideline sentence for Masters. Then he yanked his request and insisted that he had never wanted resen-tencing. Too late. By adopting all of Masters’ arguments on the first appeal, Keat-ing invited what has happened. If he believed that we misunderstood the effect of this adoption, he should have asked the first panel to rescind the portion of the judgment remanding his case. Instead he embraced the remand, and the recalculation, until the numbers turned sour.

Masters concedes that his criminal enterprise continued into the period covered by the Sentencing Reform Act. Keating submits that he dropped out before November 1, 1987. The jury convicted Keating on an indictment charging that the conspiracy continued through mid-1988. Never during trial, at the initial sentencing, or on the first appeal, did Keating say that he withdrew at any earlier date. In his initial filing after remand, Keating told the district court: “There is no evidence that your Petitioner withdrew from the conspiracy. Therefore, the conspiracy continued after November 1, 1987.” Keating changed his tune when he realized the effect of the guidelines, but he was right on the facts: there is no evidence that he withdrew, and the district judge so found. He took no step to end the criminal adventure or report its operations to the authorities, did not make a clean break of it and thus did not withdraw under the demanding standard of the criminal law. United States v. Patel, 879 F.2d 292 (7th Cir.1989). As Keating’s remaining arguments track Masters’, we need not mention him again.

Masters was convicted of racketeering and conspiracy to commit racketeering. Only the latter offense lasted into the guideline era. Masters contends that a court should start with § 2E1.1, which applies to racketeering conspiracies. That guideline provides a base offense level of 19 or the level appropriate to the substantive crime, whichever is greater. Because in Masters’ view that offense was solicitation to commit murder, the substantive guideline is § 2A2.1. (Rather, was § 2A2.1, as that guideline existed before amendment 311.) Guideline 2A2.1 provided a base offense level of 20. Masters submits that because § 2E1.1 lists no adjustments, none can be made. Combining an offense level of 20 with his criminal history category of I produces a guideline range of 33-41 months. Because he received more than that for the racketeering offense, Masters concludes that the guideline sentence for the conspiracy is effectively zero, running concurrently with the substantive RICO sentence. See U.S.S.G. § 5G1.2(c).

The district judge saw things differently. He calculated the sentence in three ways, each of which produced 40 years’ imprisonment. Each of the three depends to some degree on his finding that Masters “is directly responsible for the murder of his wife.” The judge so found even though the jury did not answer a special interrogatory asking about the murder. Masters contends that this omission produced an implicit acquittal, an argument to which we return.

1. Section 2E1.1(a)(2) calls for use of “the offense level applicable to the underlying racketeering activity.” Not the base offense level, as Masters would have it, but the whole offense level. Guideline 2X1.1(a) drives home this point, providing that in cases of attempt, solicitation, and conspiracy that fall outside other guidelines, the court should use .the “base offense level from the guideline for the substantive offense, plus any adjustments from such guideline for any intended offense conduct that can be established with reasonable certainty.” Then § 2Xl.l(b) provides some increases on top of the “base!’ level computed in this way. Starting from the base of 20 in the former version of § 2A2.1, the district judge added 2 levels for more than minimal planning (§ 2A2.1(b)(l)), 9 levels for the use of a firearm to inflict bodily injury on Dianne Masters (§ 2A2.1(b)(2), (3)), and 2 more levels for paying money to other persons in connection with the crime (§ 2A2.1(b)(4)). Thus the district judge ar *284 rived at 33 within § 2A2.1 itself. Then he turned to generally applicable enhancements: 2 levels because Mrs. Masters was physically restrained (§ 3A1.3), 2 levels because Masters was the organizer or leader of the conspiracy (§ 3Bl.l(c)), and 2 levels for obstruction of justice (§ 3C1.1). Level 39 yields a sentencing range of 262-327 months. Nothing in the computation thus far reflects the victim’s death. Because Dianne Masters died and Alan Masters was responsible, the judge concluded that an upward departure to 480 months is appropriate. See U.S.S.G. § 5K2.1, providing that departure is in order when a death results.

2. Section 2E1.1(a)(2) refers to “the offense level applicable to the underlying racketeering activity.” Instead of turning to the guideline for conspiracy to commit murder, as in the prior method, one may turn to the guideline for murder, § 2A1.1. The base offense level for murder is 43. Life imprisonment is the sentence prescribed for level 43 offenses. To come as close to life imprisonment as possible, the court gave consecutive maximum sentences on each count, for a total of 40 years. Guideline 5G1.2(d) requires this approach.

3. Am'endment 311; effective November 1, 1990, moved conspiracy to commit murder from § 2A1.1 (on which see Method 1) to a new § 2A1.5. This guideline says that “[i]f the offense resulted in the death of a victim, apply § 2A1.1 (First Degree Murder).” U.S.S.G. § 2A1.5(c)(l). That takes us straight to Method 2. Congress provided that courts must apply the guidelines “that are in effect on the date the defendant is sentenced”. 18 U.S.C. § 3553(a)(4). Masters was sentenced before amendment 311 but resentenced after its adoption. He contends- that the ex post facto clause of the Constitution prevents the court from' applying this amendment to him. But see United States v. Bader, 956 F.2d 708 (7th Cir.1992). Sidestepping this contention, the district court treated amendment 311 as clarifying rather than changing the law — as demonstrating, in other words, that Method 2 has been the right one all along.

Like the district judge, we conclude that Method 2 is proper. Methods 1 and 3 reinforce the result, demonstrating the (rough) internal consistency of the guidelines. Masters insists that the guidelines implement a charge-offense system and observes that-he was neither charged with nor convicted of murder. True enough, the Sentencing Commission rejected real-offense sentencing as impractical, see Stephen Breyer, The Federal Sentencing Guidelines and the Key Compromises Upon Which They Rest, 17 Hofstra L.Rev. 1, 8-12 (1988); Ilene H. Nagel, Structuring Sentencing Discretion: The New Federal Sentencing Guidelines, 80 J.Crim.L. & Criminology 883, 914-25 (1990). The guidelines start with the offense of conviction. U.S.S.G. § 1B1.2(a); United States v. Missick, 875 F.2d 1294, 1302 (7th Cir.1989). But real-offense principles influence not only the aggravating and mitigating factors but also the definition of relevant conduct under § 1B1.3. William W. Wilkins, Jr. & John R. Steer, Relevant Conduct: The Cornerstone of the Federal Sentencing Guidelines, 41 S.C.L.Rev. 495 (1990); Nagel, 80 J.Crim.L. & Criminology at 925-27. For example, the guidelines use the full quantity of drugs sold or loot stolen in a single scheme or course of conduct, not just the amount the prosecutor names in the indictment, to determine the seriousness of an offense. U.S.S.G. § 1B1.3(a)(2); United States v. White, 888 F.2d 490 (7th Cir.1989).

Cross-references within the guidelines introduce real-óffense principles into the charge-offense system. United States v. Davern, 970 F.2d 1490 (6th Cir.1992) (in banc). Section 2E1.1(a)(2), which requires the court to use “the offense level applicable to the underlying racketeering activity”, speaks of the underlying activity and not an underlying conviction. Often there will be no other conviction, and the existence of one in this case does not change the nature of the cross-reference. Section lB1.3(a)(iii) says that “cross references in Chapter Two ... shall be determined on the basis of the following: (1) all acts and omissions committed and aided or abetted by the defendant ... that occurred during *285 the commission of the offense of conviction”. The murder of Dianne Masters occurred during the racketeering conspiracy. So § 2El.l(a)(2), read from the perspective of § lB1.3(a), directs the court to the murder guideline rather than the solicitation guideline. At least it does this if the court’s finding that Masters is responsible for his wife’s murder is sustainable, and the process is otherwise consistent with the Constitution.

Masters tries to knock out the entire guideline sentence with the argument that consecutive terms for racketeering and racketeering conspiracy violate the double jeopardy clause. How could this be,, given the ancient principle that choate and inchoate crimes may be punished separately? United States v. Felix, — U.S.—, 112 S.Ct. 1377, 118 L.Ed.2d 25 (1992). An overlap in the evidence used to prove the two crimes does not matter; the elements of the offense, and not the evidence used to prove these elements, identify separate crimes. Id. — U.S. at —:-—, 112 S.Ct. at 1382-83. Then there is the further rule that cumulative punishment is always consistent with the double jeopardy clause, provided there is but a single trial. Missouri v. Hunter, 459 U.S. 359, 103 S.Ct. 673, 74 L.Ed.2d 535 (1983); Albernaz v. United States, 450 U.S. 333, 343-44, 101 S.Ct. 1137, 1144-45, 67 L.Ed.2d 275 (1981). According to Masters, because racketeering is an organized activity, it is impossible to be a racketeer and not be a conspirator. Provided there is only one trial, it is a matter of nomenclature whether the legislature says that a particular enterprise violates two laws, each punishable by 20 years’ imprisonment, or one law punishable by 40 years’ imprisonment. Masters does not contend that consecutive sentences for racketeering and racketeering conspiracy are impermissible under the “elements” approach used since Blockburger v. United States, 284 U.S. 299, 52 S.Ct. 180, 76 L.Ed. 306 (1932), to decide when criminal statutes establish separate offenses. He urges identity of evidence, not identity of elements. Felix, Albernaz, and Hunter, sink that contention.

To make headway, Masters needs to upset the district judge’s finding that he is responsible for the murder of his wife. Plenty of evidence supports the finding, but Masters insists that the judge had no business reexamining the jury’s acquittal. The judge submitted to the jury a set of verdict forms listing “racketeering activities” in which Masters may have engaged. The jurors checked off lines indicating findings that Masters planned and solicited the murder of his wife and participated in the overall racketeering enterprise. They left blank a line concerning the actual murder. Masters treats this omission as a finding that he is not responsible; the prosecutor replies that the jury, having found all it needed to, decided not to resolve the question.

Green v. United States, 355 U.S. 184, 78 S.Ct. 221, 2 L.Ed.2d 199 (1957), which uses an “implicit acquittal” approach similar to the one Masters espouses, could limit the prosecutor’s ability to try Masters for the crime of murder. But of course the prosecutor does not want to try him for that offense. Our dispute concerns the selection of a sentence for the crime of conspiracy. A sentence at the top of the statutory range does not punish Masters for a crime he didn’t commit; it uses all available information about his character and dangerousness in choosing the sentence for the crime of which he stands convicted. Judges have been considering defendants’ activities and character since long before there were guidelines, with consistent approval from the highest court. E.g., Roberts v. United States, 445 U.S. 552, 100 S.Ct. 1358, 63 L.Ed.2d 622 (1980); United States v. Grayson, 438 U.S. 41, 98 S.Ct. 2610, 57 L.Ed.2d 582 (1978); Williams v. Oklahoma, 358 U.S. 576, 79 S.Ct. 421, 3 L.Ed.2d 516 (1959); Williams v. New York, 337 U.S. 241 (1949); see also Chapman v. United States, — U.S. —, 111 S.Ct. 1919, 114 L.Ed.2d 524 (1991); United States v. Pinto, 875 F.2d 143 (7th Cir.1989). This is one reason-why we have held that judges may take other crimes into account when selecting a sentence under the guidelines, even if the defendant has been charged with and acquitted of those crimes. United States v. Fonner, 920 F.2d 1330, 1332-34 (7th Cir.1990); *286 see also, e.g., United States v. Morrison, 946 F.2d 484, 501 (7th Cir.1991); United States v. Welch, 945 F.2d 1378, 1385 (7th Cir.1991); United States v. Lawrence, 934 F.2d 868, 873-74 (7th Cir.1991), id. at 875 (concurring opinion). An acquittal means that the. charge was not proven beyond a reasonable doubt; it does not mean that the defendant didn’t do it. Dowling v. United States, 493 U.S. 342, 110 S.Ct. 668, 107 L.Ed.2d 708 (1990). We recognize that a divided panel of the ninth circuit has forbidden district judges to consider conduct of which a person has been acquitted, United States v. Brady, 928 F.2d 844 (9th Cir.1991), but we have previously announced our disagreement with that court’s conclusion. United States v. Smith, 953 F.2d 1060, 1066 (7th Cir.1992).

All that remains is Masters’ contention that in considering whether a defendant committed some other offense, courts should use a standard more exacting than the preponderance of the evidence. As a constitutional argument this goes nowhere. McMillan v. Pennsylvania, 477 U.S. 79, 106 S.Ct. 2411, 91 L.Ed.2d 67 (1986), holds that the preponderance standard satisfies the due process clause. We have applied this holding to sentencing under the federal guidelines. United States v. Ebbole, 917 F.2d 1495 (7th Cir.1990). Similarly, the Supreme Court has concluded that the preponderance standard is appropriate when deciding whether to use other-crimes evidence at trial. Huddleston v. United States, 485 U.S. 681, 690 (1988). The district judge used the preponderance standard when concluding that Masters is responsible for his wife's murder, and this was entirely proper. Conviction at trial supplies all of the justification the Constitution requires for depriving the defendant of liberty for any term up to the maximum prescribed by statute. How much of that time to grant back in sentencing, and what procedures to use when doing so, are legislative rather than constitutional choices given McMillan and its many predecessors.

Contemporaneously with our decision in Ebbole, the third circuit concluded that when the findings made at sentencing transform the offense of conviction into something much more grave, and with a much more severe penalty, the court should use an enhanced burden of persuasion. United States v. Kikumura, 918 F.2d 1084, 1098-1102 (3d Cir.1990). Kiku-mura was convicted of transporting explosives across state lines, with a guideline range of 27-33 months. The district judge concluded that Kikumura planned to use the bombs to kill scores of people and departed upward to the statutory maximum sentence, 30 years. This twelve-fold increase in the punishment, the court of appeals concluded, required greater confidence than the preponderance of the evidence.

Although Kikumura expressed this conclusion in constitutional terms, it is impossible to square such a holding with McMillan — or with the history of discretionary sentencing in the United States. Before November 1, 1987, some federal judges regularly sentenced defendants to the maximum terms provided by law, and tacked sentences consecutively, on hunch and whim. A few even acquired sobriquets such as “Maximum John” (Judge Sirica, of Watergate fame). Others rejected deterrence and desert as justifications of imprisonment and gave sentences reflecting disparate views about rehabilitation. States occasionally adopted indeterminate sentencing systems. For some years in California every felony carried a term of life, with release in the discretion of the Adult Authority. Zero-to-life sentences meant that everything lay in the hands of persons who were scarcely obliged to consider evidence, let alone to weigh it carefully. See Marvin Zalman, The Rise and Fall of the Indeterminate Sentence, 24 Wayne L.Rev. 45 (1977). The guidelines represent a reaction against vesting such discretion in the hands of persons who may have dramatically different ideas about punishment. Yet they hardly exemplify a consensus that in 1987 what had been the norm in 20th Century sentencing became constitutionally repulsive. See United States v. Silverman, 976 F.2d 1502 (6th Cir.1992) (in banc) (constitutional rules applicable to pre-guideline sentencing, including the preponderance standard, still govern); United States v. Galloway, 976 F.2d 414 (8th Cir.1992) (in banc) (same); United States v. Restrepo, 946 *287 F.2d 654 (9th Cir.1991) (in banc) (same); cf. United States v. Wise, 976 F.2d 393 (8th Cir.1992) (in banc) (confrontation clause does not apply at sentencing under the guidelines). While- sentencing practices have cycled, from fixed terms in the 19th Century to discretion in the middle 20th Century to guided discretion in the late 20th Century, the due process clause has remained the same. It is statutes, rules, practices, and ideas about criminology that have changed. Rather than treating the Constitution as paraffin that may be poured into a mold suiting current tastes, courts properly look to these statutes, rules, and practices. The living legal community, not an imagined decision imputed to those long departed, is responsible for contemporary sentencing practices.

No one can deny that the guidelines, coupled with recent increases in the maximum terms for many offenses, have increased the importance of sentencing vis-avis trials. We have indicated some sympathy with the conclusion that when sentencing becomes the dog and the trial the tail, judges should borrow some of the devices used at trial to protect the' defendant’s interests and improve accuracy (two objectives that are not always compatible). See United States v. Trujillo, 959 F.2d 1377, 1382 (7th Cir.1992); United States v. Schuster, 948 F.2d 313, 315 (7th Cir.1991). Cf. United States v. Jewel, 947 F.2d 224, 238 (7th Cir.1991) (concurring opinion). Neither the Sentencing Reform Act nor the guidelines specifies a burden of persuasion, so courts must devise their own. We are free to modify this standard as wisdom gained from experience suggests. But the preponderance standard remains the norm — in sentencing, as in almost all litigation, because it treats errors in either direction as equally costly. Grogan v. Garner, 498 U.S. 279, -, 111 S.Ct. 654, 659, 112 L.Ed.2d 755 (1991); Herman & MacLean v. Huddleston, 459 U.S. 375, 390, 103 S.Ct. 683, 691-92, 74 L.Ed.2d 548 (1983). An excessive sentence injures the defendant, but an insufficient sentence injures society, especially persons who may be victims of crime tomorrow if courts do too little to deter and incapacitate criminals today.

Nothing about Masters’ sentencing suggests any need for an enhanced burden of persuasion. Taking an offender’s other crimes into account is, as we have remarked, an old practice indeed. The district judge obviously took the murder into account when imposing consecutive 20-year terms in what he supposed was a preguideline sentence. On remand, the district judge reimposed these terms under the guidelines. Here we find no dramatic alteration of the roles of trial and sentence. True, the guideline sentence means no parole and lesser good-time credits, but these statutory changes do not alter the balance between trial and sentencing. Working through the guidelines after the fashion of Method 1 also shows the difference between our case and Kikumura: instead of departing upward by an order of magnitude, the district judge here used the finding of murder to support a four-level departure, from level 39 to level 43. So we may defer until another day the decision whether development in a common-law fashion ever would lead to the use of a clear-and-convincing standard in sentencing. Finding murder by a preponderance of the evidence violated none of Masters’ rights.

Affirmed.