[649]*649SUTTON, J., delivered the opinion of the court, in which BATCHELDER, C.J., BOGGS, GILMAN, GIBBONS, COOK, McKEAGUE, GRIFFIN and KETHLEDGE, JJ., joined, and MOORE, J., joined in the result. MOORE, J. (pp. 660-69), delivered a separate opinion concurring in the judgment. MERRITT, J. (pp. 669-71), delivered a separate dissenting opinion, in which DONALD J., joined. COLE, J. (pp. 671-75), delivered a separate dissenting opinion. CLAY, J. (pp. 675-85), delivered a separate dissenting opinion, in which DONALD, J., joined. ROGERS, J. (pp. 685-90), delivered a separate dissenting opinion, in which WHITE and STRANCH, JJ., joined, MERRITT, J., joined in part, and COLE, J., joined except for the last paragraph. WHITE, J. (pp. 690-98), delivered a separate dissenting opinion.
OPINION
SUTTON, Circuit Judge.
The Anti-Drug Abuse Act of 1986 established mandatory minimum sentences for possessing crack cocaine with the intent to distribute it: 5 years for possessing 5 grams, 10 years for possessing 50 grams. 100 Stat. 3207. The Fair Sentencing Act of 2010 increased the amount of crack cocaine needed to activate the same mandatory minimums. It now takes 28 grams to trigger the 5-year penalty and 280 grams to trigger the 10-year penalty. 124 Stat. 2372. Through these changes, the Fair Sentencing Act significantly reduced, but did not eliminate, a sentencing disparity between offenses involving crack and powder cocaine. What used to be a 100:1 ratio between the amount of powder and crack needed to trigger the mandatory minimums has become an 18:1 ratio.
At issue in this case is whether the changes created by the Act apply to defendants sentenced five years before the new law took effect. Consistent with a 142-year-old congressional presumption against applying reductions in criminal penalties to those already sentenced, 1 U.S.C. § 109, consistent with the views of all nine Justices and all of the litigants in Dorsey v. United States, — U.S. -, 132 S.Ct. 2321, 2332, 183 L.Ed.2d 250 (2012), consistent with the decisions of every other court of appeals in the country, and consistent with dozens of our own decisions, we hold that the Act does not retroactively undo final sentences.
I.
In 2005, a federal court convicted cousins Cornelius Blewett and Jarreous Blewitt (the Blewetts) of possessing crack cocaine with the intent to distribute it. Neither side contests that Cornelius had 19.65 grams and Jarreous 26.92 grams. Under the Anti-Drug Abuse Act, those amounts normally would have produced a 5-year minimum sentence. But because the Blewetts each had a prior felony drug conviction, the Act doubled the minimum. See 21 U.S.C. §§ 841(b), 851. The court sentenced each cousin to 10 years, the lowest sentence possible under the law.
After Congress enacted the Fair Sentencing Act, the Blewetts moved for a sentence reduction. See 18 U.S.C. § 3582(c)(2). The Blewetts’ 19.65 and 26.92 grams of crack cocaine fall short of the 28 grams required for the Fair Sentencing Act’s new mandatory minimums to kick in. As a result, the Blewetts claimed that, even though they were sentenced before the Fair Sentencing Act took effect, they should benefit from Congress’s reduction of the statutory penalties. The district court denied their request.
The Blewetts appealed. While the appeal was pending, Jarreous’s prison sen-[650]*650tenee (though not his term of supervised release) ended, prompting the government to ask us to dismiss his appeal as moot. Because we no doubt have jurisdiction over one of the defendants in this consolidated appeal, we need not resolve the motion and may proceed to answer the merits question presented by this appeal. See Steel Co. v. Citizens for a Better Env’t, 523 U.S. 83, 98, 118 S.Ct. 1003, 140 L.Ed.2d 210 (1998); Norton v. Mathews, 427 U.S. 524, 531, 96 S.Ct. 2771, 49 L.Ed.2d 672 (1976).
II.
In our view: (1) the Fair Sentencing Act’s new mandatory mínimums do not apply to defendants sentenced before it took effect; (2) § 3582(c)(2) does not provide a vehicle for circumventing that interpretation; and (3) the Constitution does not provide a basis for blocking it.
A.
The Fair Sentencing Act. Under the general saving statute, enacted in 1871, a law that reduces statutory penalties for an offense presumptively does not alter penalties incurred before the new law took effect. 1 U.S.C. § 109. A statute may overcome this default rule but only expressly or by clear implication. Dorsey, 132 S.Ct. at 2332. The Fair Sentencing Act nowhere provides that it covers offenders sentenced before it became effective. Nor do the Act’s clear implications show a desire to apply the new law to offenders already sentenced. To the contrary, no feature of the Act is backward looking and many are forward looking. Congress, to be sure, disapproved of the old mandatory mínimums; that is why it changed them. But if rejection of the old penalty were all it took to make a new penalty retroactive to all individuals previously sentenced, every new penalty would be fully retroactive, shearing the general saving statute of meaning.
Dorsey reinforces this conclusion. Its core holding is that the Fair Sentencing Act applies to all individuals sentenced after its effective date, which of course defeats the Blewetts’ claim. The Blewetts were sentenced in 2005, five years before the Act became law.
Dorsey’s reasoning points in the same direction. The Court identified six factors that supported this cut-off date, one that made the Act applicable to individuals who had distributed crack cocaine before its effective date but who were sentenced after it. While all six factors favored the Dorsey outcome, the same is not remotely true here.
Factor one. Dorsey began by explaining that Congress may overcome the § 109 default rule by a clear statement or clear implication. Id. at 2331. This factor helped Dorsey: He could point to language in the Fair Sentencing Act suggesting that Congress wanted the new ratios to apply to future sentences. See id. at 2332-33 (discussing § 8 of the Act). But it hurts the Blewetts: They cannot find, indeed they do not even try to find, any similar textual foothold to support their claim that Congress wanted the new ratios to alter final sentences.
Factor two. Under the sentencing guidelines system created by the Sentencing Reform Act of 1986, Dorsey pointed out, a defendant’s sentence normally depends on the rules “in effect on the date the defendant is sentenced.” Id. at 2332. This rule helped Dorsey: When he was first sentenced, the new mínimums were in effect. But this rule hurts the Blewetts: When they were sentenced, the old míni-mums were in effect.
Factor three. Dorsey explained that the “language in the Fair Sentencing Act im[651]*651plies that Congress intended to follow the Sentencing Reform Act background principle here.” Id. In other words, Dorsey concluded that both the Fair Sentencing Act and the Sentencing Reform Act counsel in favor of applying the relevant sentencing law on the date of initial sentencing. This factor favored Dorsey (who was sentenced after the Act’s effective date) but hurts the Blewetts (who were sentenced five years before the Act’s effective date).
Factor four. If the new mandatory minimums did not apply to everyone sentenced after the Act’s passage, Dorsey cautioned, two offenders who committed the same crime — one say a week before the Act became law, one a week after — could receive “disparate sentences” despite “roughly contemporaneous sentencing, ie., the same time, the same place, and even the same judge.” Id. at 2333. While this date-of-the-crime consideration favored Dorsey, it means nothing here. Under our holding and the holdings of all courts in the country, everyone sentenced at the same time is treated the same way: The new minimums apply to everybody sentenced after the Act, and the old minimums apply to everybody sentenced before it. In this instance, disparities will arise only if our court picks a different cutoff date and begins applying the new law to old sentences.
Factor five. Congress required the Sentencing Commission to apply the new guidelines (reflecting the new ratios) to all future sentences. Id. at 2332-33. Dorsey thus reasoned that Congress would not have simultaneously insisted on using new guidelines for all future sentences but on using new statutory minimums for some future sentences. Id. at 2334. This factor helped Dorsey but it cuts the other way here. Congress, all agree, did not require the Commission to apply the new guidelines to past sentences.
Factor six. Dorsey found no “strong countervailing consideration” to using the date of sentencing for determining when the Act applied. Id. at 2335. While that factor helped Dorsey, it hurts the Blew-etts. This case involves a “countervailing consideration” of the weightiest sort: the government’s interest in the finality of its sentences. Not only is that principle codified in the § 109 saving statute, but “the principle of finality ... is essential to the operation of our criminal justice system.” Teague v. Lane, 489 U.S. 288, 309, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989) (plurality opinion); see Johnson v. United States, 544 U.S. 295, 309, 125 S.Ct. 1571, 161 L.Ed.2d 542 (2005).
Dorsey carefully marshaled all six factors before holding that, despite the § 109 presumption, the Fair Sentencing Act applied to defendants who offended before but were sentenced after the Act’s passage. And even with all six factors pointing one way, Dorsey remained a close case: The Court divided 5-4, and the five agreed that the issue was “difficult.” 132 S.Ct. at 2330. Where Dorsey had six pluses going for him, the Blewetts have (at best) two zeros and four minuses.
Dorsey’s outcome and six-factor test leave little doubt over what to do here. The Act simply does not apply to individuals sentenced before its effective date, let alone five years before its effective date. But one need not take our word for how to read Dorsey or how to apply the six-factor test here. Listen to how the Court itself reads the opinion. Congress, it acknowledged, intended to follow the “ordinary practice [of] applying] new penalties to defendants not yet sentenced, while withholding that change from defendants already sentenced.” Id. at 2335 (emphasis added). That is today’s case. Removing [652]*652any doubt about the point, the Court acknowledged the reality that its decision would create a “disparity ... between those pre-Act offenders already sentenced and those not yet sentenced as of [the Act’s date of effectiveness].” Id. That too covers today’s case. No disparity would exist if the Act applied to all previously sentenced defendants.
All of the participants in Dorsey, moreover, assumed that defendants sentenced before the Fair Sentencing Act’s passage could not benefit from its provisions. The Court made this assumption in the majority and dissenting opinions. See id.; id. at 2343 (Scalia, J., dissenting). So did the Justices at oral argument. See Transcript of Oral Argument at 34, 46. So did Dorsey. See Brief for Petitioner at 51. And so did the United States. See Brief for the United States at 23-24, 48-51. The Court-appointed amicus curiae summed it all up when he said that “everybody agrees” Congress did not want to reopen final sentences. Transcript of Oral Argument at 40. All of this may not amount to a binding holding from the Supreme Court. But according to a reliable source, circumstances like these “have much weight, as they show that this point neither occurred to the bar or the bench.” Bank of United States v. Deveaux, 9 U.S. 61, 5 Cranch 61, 88, 3 L.Ed. 38 (1809) (Marshall, C.J.).
Precedents from our circuit have long accepted this interpretation. We have held in published opinions that the Fair Sentencing Act does not apply to individuals sentenced before its effective date. See United States v. Bell, 731 F.3d 552 (6th Cir.2013); United States v. Hammond, 712 F.3d 333 (6th Cir.2013). And we have adhered to the same rule in dozens of unpublished cases — enough to add a page and a half to this opinion had we wished to inflict the burden on our readers. See, e.g., United States v. Williams, 522 Fed. Appx. 278 (6th Cir.2013); United States v. Gooch, 518 Fed.Appx. 473 (6th Cir.2013); United States v. Tillman, 511 Fed.Appx. 519 (6th Cir.2013).
Every other federal court of appeals— except the Federal Circuit, which does not hear criminal cases — has followed this path. See United States v. Santos-Rivera, 726 F.3d 17, 28 n. 2 (1st Cir.2013); United States v. Diaz, 627 F.3d 930, 931 (2d Cir. 2010) (per curiam); United States v. Turlington, 696 F.3d 425, 428 (3d Cir.2012); United States v. Mouzone, 687 F.3d 207, 222 (4th Cir.2012); United States v. Kelly, 716 F.3d 180, 181 (5th Cir.2013); United States v. Foster, 706 F.3d 887, 888 (7th Cir.2013); United States v. Reeves, 717 F.3d 647, 651 (8th Cir.2013); United States v. Augustine, 712 F.3d 1290, 1294-95 (9th Cir.2013); United States v. Lucero, 713 F.3d 1024, 1027 (10th Cir.2013); United States v. Hippolyte, 712 F.3d 535, 542 (11th Cir.2013); United States v. Swangin, 726 F.3d 205, 208 (D.C.Cir.2013). Notwithstanding hundreds if not thousands of opportunities, no other appellate court in the country has held to the contrary.
Last but not least, the Sentencing Commission, which is responsible for overseeing the guidelines and for implementing parts of the Fair Sentencing Act, has concluded that the changes to the statutory mandatory mínimums do not apply retroactively. Nothing the Commission has done, it acknowledges, “make[s] any of the statutory changes in the Fair Sentencing Act of 2010 retroactive.” 76 Fed.Reg. 41,-332, 41,333 (July 13, 2011).
Swimming against this tide, the Blew-etts target an alleged anomaly that arises from our conclusion. Suppose, the Blew-etts point out, that the district court had sentenced Cornelius to the minimum of 120 months but Jarreous to 125 months because Jarreous possessed more cocaine. In this scenario, the Blewetts correctly [653]*653observe, Jarreous could get a five-month sentence reduction (down to the mandatory minimum), but Cornelius, already at the minimum, would get nothing.
Yet this result says nothing about the forward-looking or backward-looking nature of the Fair Sentencing Act. It instead says something about, and is a potential consequence of, all statutory minimum and maximum sentences, whether in the Fair Sentencing Act or elsewhere. Suppose once more that the district court starts out intending to give Cornelius 120 months and Jarreous 125. But in this example, the district court becomes acquainted with the defendants’ efforts to rehabilitate themselves, and decides that each deserves a reduction. In this instance, Jarreous could profit from this mitigating circumstance, but Cornelius, already at the minimum, could not. It works the same way with statutory máximums. Suppose that the district court started out wishing to give both cousins the maximum sentence permitted by the statute yet one of them deserved an even higher sentence due to the aggravating circumstances of his crime. Here, too, the court would have to treat the cousins the same, even though one of them committed a worse crime. None of this is anomalous; it is a natural consequence of having cliff-creating mandatory minimum or maximum sentences in the first place. As the Supreme Court has acknowledged, “[wjhereas guidelines seek a smooth continuum, mandatory mínimums result in ‘cliffs.’ ” Neal v. United States, 516 U.S. 284, 292, 116 S.Ct. 763, 133 L.Ed.2d 709 (1996) (quoting United States Sentencing Commission, Special Report to Congress: Mandatory Minimum Penalties in the Federal Criminal Justice System iii (1991)).
What is more, the whole point of any retroactivity debate is whether to apply more lenient penalties to everyone or to apply them only to those sentenced after the new law takes effect. And the whole point of § 109 is to create a presumption in favor of a disparity between people who offended before a decrease in statutory penalties and people who offended after. Because § 109 mandates greater cliffs/mismatches/disparities between the two groups of individuals in the normal course, it hardly respects the saving statute to allow more modest disparities to drive the retroactivity analysis. The lead dissent authored by Judge Rogers mentions the saving statute once in a quotation from Dorsey, and it is fair to ask whether the dissent has flipped the § 109 presumption against retroactivity into a presumption in favor of it.
In a variation on this theme, the lead dissent finds the following two propositions irreconcilable: (1) that Congress wanted the new statutory minimums to be prospective (2) but it wanted the new guidelines to be retroactive. To eliminate the inconsistency between these two statements, the lead dissent concludes that the first statement must be false. But it offers no explanation for this question-begging choice. One could eliminate the inconsistency just as readily by saying that the second statement is false or by considering why both statements could be true.
The Fair Sentencing Act, for what it is worth, offers little guidance on whether the second statement is true or false. The Act, as Dorsey explains, affirmatively instructs the Sentencing Commission to use the new ratios for all future sentences. See 124 Stat. 2374, § 8; Dorsey, 132 S.Ct. at 2332, 2335. The Act does not tell the Commission to use the new ratios for all past sentences — that is, to make the new ratios retroactive. At first, indeed, the Commission adopted the new ratios only prospectively. See 75 Fed.Reg. 66,188 (Oct. 27, 2010). Months later, the Com[654]*654mission solicited public comments about whether to make the new ratios retroactive. See 76 Fed.Reg. 24,960 (May 3, 2011). Later still, almost a year after the Fair Sentencing Act became law, the Commission finally voted for retroactivity. See 76 Fed.Reg. 41,382 (July 13, 2011).
If the Commission’s year-later choice was legitimate, as the lead dissent assumes, that is not because Congress directed it to do so. The only congressional directive concerned new ratios for future sentences. All that could justify the Commission’s authority to make the new guidelines retroactive is its authority to fill the gap in the Fair Sentencing Act and to construe its own guidelines and policy statements. Let us assume for the sake of argument that the Commission indeed had such authority and thus permissively made the guidelines retroactive to individuals previously sentenced. But if that is true, no principle of administrative deference would allow us to accept the Commission’s judgment on that score and ignore its contrary judgment on a related score — whether the Fair Sentencing Act applied to individuals already sentenced. Its opinion on this last point is unequivocal: The Commission explained that its decision “only allows the guideline changes to be considered for retroactive application; it does not make any of the statutory changes in the Fair Sentencing Act of 2010 retroactive.” 76 Fed.Reg. at 41,333. If the lead dissent wishes to respect the Commission’s administrative authority to make the guidelines retroactive based on its interpretation of the Act, § 109 and § 3582, it should respect the Commission’s view that this authority ends at the door of congressional mandatory mínimums. To do otherwise is a bait and switch.
Nor was the Commission’s choice an unusual one or for that matter an “incoherent anomaly.” Dissenting Op. at 689 (Rogers, J.). The point of mandatory mín-imums and máximums, for better or worse, is to give Congress more control over sentencing. It makes considerable sense that the Commission would think it has discretion to operate outside the mínimums, including by making changes retroactive, but would respect Congress’s control over mín-imums and máximums by declining to make changes that cross this line. It merely shows that the Commission sees this the same way we do — and for that matter the way every appellate court in the country has seen it.
In focusing on the Sentencing Commission’s decision to make the new guidelines partially retroactive as the reason why the Fair Sentencing Act is retroactive, the lead dissent’s theory creates another problem. It would mean that the new mínimums were not retroactive when the Act took effect but became retroactive a year later when the Commission decided to make the corresponding guidelines retroactive. As the dissent puts it, “The Sentencing Commission ... made the Fair Sentencing Act-driven guidelines retroactive. Doing so provided the statutory key to making the statutory minimum changes [retroactive].” Dissenting Op. at 689 (Rogers, J.).
This theory of springing retroactivity has no antecedents. Under this reading, the Fair Sentencing Act gave the Sentencing Commission the power to decide which mínimums, new or old, would apply to offenders already sentenced and in the process allowed a retroactive alteration of the Act to spring forth like a jack-in-the-box at a date of the Sentencing Commission’s choosing. Besides its novelty, the theory raises serious constitutional doubts. The constitutionality of the Sentencing Commission depends on the premise that Congress did not “vest in the [Commission] the legislative responsibility for establishing minimum and maximum penal[655]*655ties.” Mistretta v. United States, 488 U.S. 361, 396, 109 S.Ct. 647, 102 L.Ed.2d 714 (1989).
The lead dissent adds that all six of Dorsey’s factors support treating the Fair Sentencing Act as fully retroactive. See Dissenting Op. at 685-86 (Rogers, J.). That is an aggressive reading of the decision, one disclaimed by Dorsey, 132 S.Ct. at 2335 (“[The Fair Sentencing Act creates a] new disparity ... between those pre-Act offenders already sentenced and those not yet sentenced as of [the Act’s effective date].”), and one yet to find support from a single other appellate court in the country.
Judge White goes one step beyond Judge Rogers, indeed a step beyond what the Blewetts or anyone else has argued in this case. She maintains that Congress affirmatively wanted the Sentencing Commission to decide whether to make the new mandatory mínimums retroactive — and that the Commission exercised this authority by making the mínimums retroactive. The short answer is: when and where? Nothing in the text of the Fair Sentencing Act or the Commission’s handiwork offers any support for this theory. Nor does the dissent reconcile this theory with the Supreme Court’s assurance that Congress did not want “to impose an unforeseeable, potentially complex application date” when it passed the Act. Id. at 2336. Nor does it explain why we should force upon the Commission a power that the Commission disclaims. See supra at 654. Nor does it explain why, as recently as September 25, 2013, the chair of the Sentencing Commission recommended to the Senate Judiciary Committee “that the provisions of the Fair Sentencing Act of 2010, which Congress passed to reduce the disparity in treatment of crack and powder cocaine, be made retroactive.” Reevaluating the Effectiveness of Federal Mandatory Minimum Sentences: Hearing Before the Sen. Comm. on the Judiciary, 113th Cong. 2 (2013) (statement of Patti B. Saris, Chair, U.S. Sentencing Comm’n). There would be no need for Congress today to make the mínimums retroactive if the Commission already did so at some still-unidentified date in the past.
The support for this theory apparently comes from a 1994 law establishing a “safety valve” from mandatory mínimums and providing that the new rules “shall apply to all sentences imposed on or after the 10th day beginning after the date of enactment.” 108 Stat. 1796, 1986. The dissent claims that the absence of a similar clause in the Fair Sentencing Act shows that the Act left its effective date up to the Sentencing Commission. This theory overlooks a simpler explanation for the inclusion of this clause in the 1994 statute. By default under the saving statute, the new safety valve would not have applied to people who offended before the statute’s effective date, even those sentenced after the effective date. See 1 U.S.C. § 109. In order to overcome this default rule, Congress had to provide that the safety valve “shall apply to all sentences imposed” after the effective date. The effective date provision in the 1994 law thus confirms the importance, not the irrelevance, of the saving statute, and that Congress knows how to overcome it.
In the last analysis, all indicators of meaning — the default rule that reductions in criminal penalties are not retroactive, the absence of any language in the Fair Sentencing Act that overcomes this presumption, the outcome, reasoning .and assumptions of Dorsey, a cascade of authority from our court and every other court of appeals, and the Sentencing Commission’s own interpretation — show that the Act does not apply to defendants sentenced before its effective date. That would be the end of this dispute but for another [656]*656law&emdash;18 U.S.C. § 3582(c)(2)&emdash;which the Blewetts claim allows them to sidestep this conclusion.
B.
Section 3582(c)(2). Enacted as part of the Sentencing Reform Act of 1984, § 3582(c)(2) permits inmates to seek a sentence reduction in a specific setting: “in the case of a defendant who has been sentenced to a term of imprisonment based on a sentencing range that has subsequently been lowered by the Sentencing Commission pursuant to 28 U.S.C. § 994(o).” This provision does not cover the Blewetts’ request, as Judge Gilman ably explained at the panel stage of this dispute. See United States v. Blewett, 719 F.3d 482, 496-97 (6th Cir.2013) (Gilman, J., dissenting). It applies only to sentencing ranges decreased “by the Sentencing Commission” in accordance with “ § 994(o).” In the Fair Sentencing Act, Congress, not the Sentencing Commission, lowered these mandatory minimum criminal penalties. Congress is not the Sentencing Commission. Compare U.S. Const, art. I, § 1 (establishing Congress), with 28 U.S.C. § 991(a) (establishing the Sentencing Commission). When Congress lowers a penalty, it acts under Article I of the Constitution, not 28 U.S.C. § 994(o). It follows that a mandatory minimum subsequently lowered by Congress is not, as § 3582(c)(2) requires, a “sentencing range ... subsequently ... lowered by the Sentencing Commission pursuant to 28 U.S.C. § 994(o).”
Not only does the straightforward language of § 3582(c)(2) limit its coverage to decreases in sentencing ranges made by the Sentencing Commission, but two features of its adoption cement this interpretation. The statute came into being through the Sentencing Reform Act of 1984, which created the Sentencing Commission and authorized the sentencing guidelines system. It is unsurprising that Congress would limit this “narrow” sentence-reduction option, Dillon v. United States, 560 U.S. 817, 130 S.Ct. 2683, 2691, 177 L.Ed.2d 271 (2010), to retroactive modifications of the sentencing guidelines by the Sentencing Commission, not to non-retroactive modifications of criminal laws by Congress.
That leads to the second point. Under the Sentencing Reform Act of 1984, the Sentencing Commission has no authority to contradict congressional mandatory minimum sentences. 28 U.S.C. § 994(b)(1). In upholding the Sentencing Reform Act against a non-delegation challenge, the Supreme Court relied on the premise that the Commission’s decisions had to “be consistent with pertinent provisions of Title 18 of the United States Code.” Mistretta, 488 U.S. at 375, 109 S.Ct. 647. If, as we conclude, the Fair Sentencing Act of 2010 does not apply to sentences imposed before its effective date, nothing in § 3582(c)(2) in particular or in the Reform Act of 1984 in general permits the Sentencing Commission to override that conclusion.
Section 3582(c)(2) thus permits criminal defendants to seek a sentence reduction when the Sentencing Commission lowers a guidelines range. But it does not permit criminal defendants to transform a non-retroactive reduction in penalties enacted by Congress into a retroactive one.
In response, the Blewetts offer a creative, but in the last analysis unconvincing, line of argument. It goes like this: (1) at their sentencing hearings in 2005, the Sentencing Guidelines recommended a sentencing range of 120-137 months for Cornelius and 120-125 months for Jarreous; (2) after passage of the Fair Sentencing Act in 2010, the Sentencing Commission adopted Amendment 750, which reduced [657]*657sentencing ranges for crack cocaine offenders and today would give Cornelius 63-78 months and Jarreous 70-87 months; (3) this amendment allows affected defendants to seek a sentence reduction under § 3582(c)(2); and (4) the court therefore must apply the new mandatory minimum statutory sentences for crack cocaine offenses, not the old ones, in deciding the § 3582(c)(2) motion.
The problem with this theory is that it treats § 3582(c)(2) as an alchemist’s stone, capable of transforming a prospective law into a retroactive one. All a defendant needs to do is somehow get into a sentence reduction proceeding. Once there, he can take advantage of any changes to mandatory minimum statutes — retroactive or not, related to the Fair Sentencing Act or not — made since his initial sentencing. At least four roadblocks stand in the way of this interpretation.
First, the argument overlooks “the limited scope and purpose of § 3582(c)(2).” Dillon, 130 S.Ct. at 2692. Sentence modification proceedings are not “plenary re-sentencing proceedings.” Id.; see Blewett, 719 F.3d at 497 (Gilman, J., dissenting). Section 3582(c)(2) allows consideration only of retroactive amendments to the sentencing guidelines, with other sentencing elements — mandatory mínimums included— held constant.
Second, the argument proves too much. Suppose the Fair Sentencing Act contained a clause saying, “This Act does not apply to offenders sentenced before its passage.” Even then, the Blewetts’ theory would mean that the Sentencing Commission’s retroactive reduction of the crack guidelines would allow them to enter a sentence reduction proceeding. And, even then, their theory would mean that, once there, they could benefit from the Act’s new mínimums. How odd to give a law retroactive effect even though it said it did not have retroactive effect. Nor does this problem go away because the current law lacks an anti-retroactivity clause. Instead of appearing in the Fair Sentencing Act, a clause with similar effect appears in a neighboring law that covers all laws that lower criminal sentences: § 109, the general saving statute.
Third, the Sentencing Commission, no fan of the crack-powder disparities, disavows the Blewetts’ interpretation. One of its policy statements — which § 3582(c)(2) requires courts to follow — states that “a reduction in the defendant’s term of imprisonment is not authorized” to the extent blocked by “the operation of another ... statutory provision (e.g., a statutory mandatory minimum term of imprisonment).” U.S.S.G. § 1B1.10 cmt. n.l. Here, “another ... statutory provision” — the Anti-Drug Abuse Act — blocks reduction of the Blew-etts’ sentence beyond 120 months.
The Blewetts reply that the phrase “statutory provision” in § 1B1.10 refers to the Fair Sentencing Act, not to the Anti-Drug Abuse Act. But why? As shown, the Fair Sentencing Act does not govern offenders sentenced before its enactment, making it implausible that a guideline of the Sentencing Commission would alter that statutory directive of Congress. The guideline instead refers to the applicable “statutory provision” — the Anti-Drug Abuse Act. We thus look to that law, not the Fair Sentencing Act, to figure out whether “another ... statutory provision” prevents the Blewetts’ sentences from getting any lower. No other circuit in the country questions this interpretation.
Ending any doubt about this conclusion, the Sentencing Commission itself has announced that Amendment 750 “only allows the guideline changes to be considered for retroactive application; it does not make any of the statutory changes in the Fair Sentencing Act of 2010 retroactive.” 76 [658]*658Fed.Reg. at 41,333. The very entity responsible for writing § 1B1.10, for lowering the guidelines sentencing ranges and for creating the pivot point of the Blew-etts’ argument thus disagrees that the retroactive lowering of the one (some guidelines ranges) has the effect of retroactively lowering the other (the mandatory minimums).
Fourth, the Blewetts’ interpretation would make a fool’s errand of Dorsey. Under their theory, even if Dorsey had lost the debate about the meaning of the Fair Sentencing Act, he still could have brought a § 3582(c)(2) motion to get the benefit of the new mandatory minimums. If that were true, why decide Dorsey? The Justices have better things to do than to resolve difficult 5^ issues that don’t matter. There remain more ways to say less, but we will leave it at that with respect to the statutory arguments.
C.
The Constitution. Even if the Fair Sentencing Act applies only to individuals sentenced after its effective date and even if § 3582(c)(2) does not convert the Act into a retroactive change to these mandatory minimums, the Blewetts claim that the Constitution’s equal-protection and cruel- and-unusual-punishment principles give them relief. Long before the passage of the Fair Sentencing Act, our court and others repeatedly rejected similar constitutional challenges to the crack and powder cocaine sentencing disparities. See United States v. Blair, 214 F.3d 690, 702 (6th Cir.2000) (“The law is well settled in this circuit that the [crack-powder] ratio withstands constitutional scrutiny.”); see also, e.g., United States v. Garcia-Carrasquillo, 483 F.3d 124, 134 (1st Cir.2007); United States v. Samas, 561 F.3d 108, 110 (2d Cir.2009); United States v. James, 78 F.3d 851, 853 n. 2 (3d Cir.1996); United States v. Burgos, 94 F.3d 849, 876-77 (4th Cir. 1996); United States v. Fisher, 22 F.3d 574, 579 (5th Cir.1994); United States v. Chandler, 996 F.2d 917, 918-19 (7th Cir. 1993) (per curiam); United States v. Watts, 553 F.3d 603, 604 (8th Cir.2009); United States v. Dumas, 64 F.3d 1427, 1429-30 (9th Cir.1995); United States v. Cornelius, 696 F.3d 1307, 1325 (10th Cir. 2012); United States v. Solomon, 848 F.2d 156, 157 (11th Cir.1988) (per curiam); United States v. Holton, 116 F.3d 1536, 1548 (D.C.Cir.1997). Congress’s mitigation of the crack and powder disparities does not weaken these precedents; it strengthens them. Besides, the Blewetts cite no cases that call these conclusions into question. “This is not due to a lack of diligent research; it is due to the lack of any such cases.” Blewett, 719 F.3d at 495 (Gilman, J., dissenting).
Observing that users of crack cocaine are predominantly black but users of powder are predominantly white, the Blewetts insist (along with the dissents authored by Judges Merritt, Cole, Clay and White, and with support from the concurrence in the judgment authored by Judge Moore) that a congressional failure to make the Fair Sentencing Act fully retroactive would violate the equal protection component of the Fifth Amendment’s Due Process Clause. But this guarantee forbids only intentional discrimination. A law that lacks a “racially discriminatory purpose” does not become unconstitutional “solely because it has a racially disproportionate impact.” Washington v. Davis, 426 U.S. 229, 239, 96 S.Ct. 2040, 48 L.Ed.2d 597 (1976). No evidence shows that Congress acted with the purpose to discriminate when it refused to apply the Fair Sentencing Act to defendants — all defendants, white or black — sentenced before the Act took effect.
[659]*659The Blewetts and Judge Clay add that Congress must have foreseen that its failure to make the Fair Sentencing Act fully retroactive would have a racially disproportionate impact. But as Judge Gilman rightly pointed out, see 719 F.3d at 495, that makes no difference. A disproportionate effect does not violate the Equal Protection Clause, even if it was foreseen. The Supreme Court has instructed us that the “intent” in “intentional discrimination” means more than “intent as awareness of consequences.” Pers. Adm’r of Mass. v. Feeney, 442 U.S. 256, 279, 99 S.Ct. 2282, 60 L.Ed.2d 870 (1979). A violation arises only when a public official takes an action “ ‘because of,’ not merely ‘in spite of,’ its adverse effects upon an identifiable group.” Id. No evidence exists that Congress refused to make the Fair Sentencing Act retroactive because this refusal would disproportionately harm black defendants.
This theory of discrimination also makes little sense. Is it really possible that the same Congress that was deeply concerned about racial justice when looking at future sentences suddenly became racist when contemplating past sentences? That is a heavy lift. A more basic explanation exists for what Congress did, and for what it failed to do, in the Fair Sentencing Act. The government has a powerful interest in avoiding the disruption of final sentences. See Johnson, 544 U.S. at 309, 125 S.Ct. 1571; Teague, 489 U.S. at 309, 109 S.Ct. 1060 (plurality opinion). Congress did nothing extraordinary or for that matter discriminatory when it respected this interest. It was merely sticking to what Dorsey called the “ordinary practice” in federal sentencing, “withholding [a] change from defendants already sentenced.” 132 S.Ct. at 2335.
The Blewetts, joined by several of the dissents, insist that Congress acted “irrationally” by failing to make the Fair Sentencing Act fully retroactive. But the government has a strong interest in the finality of sentences, and the Fair Sentencing Act advances that interest. Ruling otherwise means forgetting Armour v. City of Indianapolis, which endorsed “the rationality of the distinction ... the law often makes between actions previously taken and those yet to come,” — U.S. -, 132 S.Ct. 2073, 2082, 182 L.Ed.2d 998 (2012), and forgetting Dillon v. United States, which was “aware of no constitutional requirement of retroactivity that entitles defendants sentenced to a term of imprisonment to the benefit of subsequent ... amendments,” 130 S.Ct. at 2692.
Two of the dissents respond that, because Congress did not insist on the finality of sentences based on the old guidelines, it acted irrationally by insisting on the finality of sentences based on the old statutory mínimums. See Dissenting Op. at 673-75 (Cole, J.); Dissenting Op. at 696-98 (White, J.). That does not follow. Congress created the mínimums, and the Commission created the guidelines. It is eminently rational for Congress to decide the retroactivity of its creations and to allow the Commission to decide the retro-activity of its creations. This argument also overlooks the reality that § 3582(c)(2) grants the district courts discretion to reduce sentences; it does not impose a mandate on them. Dillon, 130 S.Ct. at 2692. If the dissents’ theory is correct, countless district court judges have violated the Equal Protection Clause by exercising this discretion to leave in place some old sentences but not others, and we apparently have aided and abetted those constitutional violations in countless decisions of our own that found no abuse of discretion. Doubtful.
Turning to another part of the Constitution, the Blewetts and Judge Merritt (with support from Judge Moore’s [660]*660opinion) contend that the Eighth Amendment prohibits the continued imprisonment of a defendant sentenced under the old mandatory minimum laws. Yet they cannot contend that their 10-year sentences were cruel and unusual when imposed. After all, the Supreme Court has upheld a mandatory minimum of life without parole for possession of 672 grams of cocaine without intent to distribute. Harmelin v. Michigan, 501 U.S. 957, 111 S.Ct. 2680, 115 L.Ed.2d 836 (1991). The Blewetts’ crimes are less serious in one respect (they possessed less cocaine), but more serious in two others (they possessed with intent to distribute and they had prior felony drug convictions). Harmelin precludes the conclusion that the Blewetts’ much shorter mandatory minimum sentences were cruel and unusual.
The Blewetts persist that their sentences became cruel and unusual when Congress passed the Fair Sentencing Act. But the Eighth Amendment is not a ratchet that makes a harsher system of penalties unconstitutional the moment a more lenient one is (prospectively) adopted, a theory that would have the perverse effect of discouraging lawmakers from ever lowering criminal sentences. Withholding the benefits of a change from previously sentenced defendants at any rate is not “unusual”; it is the general practice in federal sentencing, as Dorsey and § 109 confirm.
The Blewetts turn last of all to the principle that courts should interpret statutes to avoid serious constitutional doubts. See Crowell v. Benson, 285 U.S. 22, 62, 52 S.Ct. 285, 76 L.Ed. 598 (1932). Yet a weak constitutional argument and a weak statutory argument do not add up to a strong constitutional avoidance argument. Just the opposite, as Judge Gilman explained. See 719 F.3d at 495-96. Avoidance permits the adoption of an interpretation only if that reading of the statute is “fairly possible.” Crowell, 285 U.S. at 62, 52 S.Ct. 285. It is not “fairly possible” to read “Sentencing Commission” in § 3582(c)(2) to mean “Congress,” or to interpret the Fair Sentencing Act to contain a clear implication of full retroactivity sufficient to displace the general saving statute’s default rule. Avoidance comes into play when there is a “serious doubt,” not just any doubt, about a statute’s constitutionality. Id. The Supreme Court’s decisions in cases like Davis and Harmelin leave no serious doubt about the constitutionality of § 3582(c)(2) and the Fair Sentencing Act as written.
ifr Hí
At the end of the day, this is a case about who, not what — about who has authority to lower the Blewetts’ sentences, not what should be done with that authority. In holding that the courts lack authority to give the Blewetts a sentence reduction, we do not mean to discount the policy arguments for granting that reduction. Although the various opinions in this case draw different conclusions about the law, they all agree that Congress should think seriously about making the new mínimums retroactive. Indeed, the Fair Sentencing Act, prospective though it is, dignifies much of what the Blewetts are saying as a matter of legislative reform and may well be a powerful ground for seeking relief from Congress. Yet the language of the relevant statutes (the Fair Sentencing Act, § 109 and § 3582(c)(2)) and the language of the relevant decisions (Dorsey, Davis and Harmelin) leave us no room to grant that relief here. Any request for a sentence reduction must be addressed to a higher tribunal (the Supreme Court) or to a different forum altogether (the Congress and the President).