dissenting.
I join all but the last paragraph of Judge Rogers’s persuasive dissent. I write separately to explain my view that applying the 100-to-l mandatory mínimums to deny eligibility for § 3582(c)(2) proceedings violates equal protection principles.
There is no question now that the 100-to-l ratio disproportionately affects African Americans. African-American crack users are more likely to be convicted of federal crack offenses than their White counterparts. Last year, for example, African Americans made up 30% of reported crack users but 83% of federal crack convicts.1 White people, in contrast, ac*672counted for 62% of users but only 7% of convicts. Federal crack convicts, substantially more likely to be African-American, receive longer sentences than powder cocaine convicts. In 2007, under the 100-to-l ratio and before the Sentencing Commission sought to mitigate modestly the disparity, the average crack sentence was 50% longer than the average powder cocaine sentence.2 And more generally, the 100-to-l ratio yields a sentencing guideline range three to six times longer for crack offenders than powder cocaine offenders with the same drug quantities.3 It is hardly a revelation that the ratio negatively affects African Americans more than White people. The Sentencing Commission reported so to Congress in 1995 (“The 100-to-l crack cocaine to powder cocaine quantity ratio is a primary cause of the growing disparity between sentences for Black and White federal defendants.”), and again in 1997 (“[S]entences appear to be harsher and more severe for racial minorities than others as a result of this law.”), and again in 2002 (“[T]o the extent that the 100-to-1 drug quantity ratio is shown to result in unduly severe penalties for most crack cocaine offenders, the impact of that severity falls primarily upon black offenders.”), and again in 2007 (“The current severity of crack cocaine penalties mostly impacts minorities.”).4
There is no question now that the 100-to-l ratio is unjustified. The Sentencing Commission issued four separate reports telling Congress that the ratio was too high and unjustified because, in part, the “current quantity-based penalties overstate the relative harmfulness of crack cocaine compared to powder cocaine” and the “current severity of crack cocaine penalties mostly impacts minorities.”5 Congress enacted the Fair Sentencing Act in response to these concerns. See Dorsey, 132 S.Ct. at 2329. As the Supreme Court described it, Congress “specifically found in the Fair Sentencing Act that [a 100-to-ll sentence was unfairly long.” Id. at 2333. Other federal actors too recognized that the 100-to-l ratio is arbitrary. Representative Daniel Lungren, for example, explained the ratio’s genesis: “We initially came out of committee with a 20-to-l ratio. By the time we finished on the floor, it was 100-to-l. We didn’t really have an evidentiary basis for it, but that’s what we did, thinking we were doing the right thing at the time.”6 In congressional hearings addressing the 100-to-l ratio, the Department of Justice urged Congress not to “ignore the mounting evidence that the current cocaine sentencing *673disparity is difficult to justify based on the facts and science.”7 And the U.S. Drug Enforcement Administration’s former Administrator testified that “[t]he facts and the passage of time have built a growing consensus that the sentencing disparity is fundamentally unfair, has a disparate racial impact and undermines the perception of fairness and the integrity of our criminal justice system.”8
There is no question now that using the 100-to-l mandatory mínimums to deny eligibility for § 3582(c)(2) sentence modifications continues to treat African-American offenders more harshly than White offenders. African Americans not only are overrepresented among those prosecuted and sentenced under the 100-to-l ratio, they also were more likely to be subject to the 100-to-l mínimums at sentencing. In fiscal year 2010, for example, African Americans composed 79% of those subjected to the crack mandatory mínimums.9 Only 4.4% of that group was White. And the Sentencing Commission estimates that if the 100-to-l mínimums do not prevent § 3582(c)(2) sentence modifications, 88% of inmates eligible for a reduction would be African-American.10
Why, then, are African Americans otherwise eligible for sentence modifications still imprisoned by a racially discriminatory, unjustified, repealed law? “Finality,” the majority and the government reply. They argue that Congress intended to keep these sentences final.
Judge Rogers’s analysis cogently explains that Congress did not, in fact, intend to keep these sentences final when it passed the Fair Sentencing Act. Instead, the Act’s 18-to-l ratio applies retroactively to reduce sentences driven by the 100-to-l guidelines and those driven by the 100-to-l mínimums. But the bigger problem, to my mind, is that this claim of finality cannot withstand even rational basis scrutiny under equal protection principles.
Take those everyone agrees are eligible for § 3582(c)(2) proceedings: crack convicts sentenced above the 100-to-l minimums. The majority and the government concede that an individual in this category could receive a sentence reduction. See Majority Op. at 652-53. In other words, a district court could hold a hearing and recalculate the inmate’s sentence, thereby disrupting its finality and incurring the time and costs of doing so. But then the majority says something odd: a district court still must apply the racially discriminatory, unjustified, repealed mínimums to deny a full reduction to the new guidelines, all in the name of finality — finality that already has been set aside. This approach is irrational; it does not rationally further finality. Applying the minimums in this way would fail basic rational basis scrutiny. See Nordlinger v. Hahn, 505 U.S. 1, 10, 112 S.Ct. 2326, 120 L.Ed.2d 1 (1992) (equal protection requires that “the classification rationally further a legitimate state interest”).
*674What about using the 100-to-l minimums to deny eligibility for § 3582(e)(2) altogether? This at least has the appearance of being rationally related to finality by preventing some crack convicts from accessing the statutory mechanism specifically designed to disrupt their sentences. But the appearance proves deceptive upon a closer look.
First, this case calls for a more searching form of review than we apply in the typical rational basis case. This unusual scenario, involving a clear racially disparate impact, Congress’s repudiation of the 100-to-l guidelines and minimums as discriminatory and unjustified, and retroactive application of the 18-to-l guidelines, gives us reason to be suspicious of a simultaneous effort to deny retroactive application of the 18-to-1 minimums. See, e.g., United States v. Windsor, — U.S.-, 133 S.Ct. 2675, 2692, 186 L.Ed.2d 808 (2013) (“Discriminations of an unusual character especially suggest careful consideration to determine whether they are obnoxious to [equal protection principles].”); see also Romer v. Evans, 517 U.S. 620, 632-35, 116 S.Ct. 1620, 134 L.Ed.2d 855 (1996); City of Cleburne, Tex. v. Cleburne Living Ctr., 473 U.S. 432, 447-50, 105 S.Ct. 3249, 87 L.Ed.2d 313 (1985); U.S. Dep’t of Agric. v. Moreno, 413 U.S. 528, 534-38, 93 S.Ct. 2821, 37 L.Ed.2d 782 (1973). As these cases demonstrate, our system of tiered scrutiny need not be unyielding in the face of recognized injustice.
Second, arguing that the Fair Sentencing Act rationally furthers finality makes little sense when the Act itself permits a slew of similarly situated crack offenders to disrupt finality. As of October, in fact, 7,400 crack offenders have received sentence reductions based on retroactive application of the 18-to-1 guidelines.11 Contrary to the majority’s suggestion, Op. at 653-54, Congress clearly gave the Sentencing Commission proper authority to apply retroactively the guidelines changes made in response to the Fair Sentencing Act. 28 U.S.C. § 994(u); Dillon v. United States, 560 U.S. 817, 130 S.Ct. 2683, 2688, 177 L.Ed.2d 271 (2010). The Act permits those initially sentenced above the 100-to-1 minimums — the more serious offenders — to intrude on finality. No persuasive justification exists for prohibiting those initially sentenced at the 100-to-l minimums — the less serious offenders — from doing the same.
Third, § 3582(c)(2) in part already addresses finality interests by restricting relief to offenders whose sentences were based on ranges lowered and made retroactive by the Sentencing Commission, among other limitations. See 18 U.S.C. § 3582(c)(2); U.S.S.G. § 1B1.10; 18 U.S.C. § 3553(a).
Finality is easily used as a post-hoc explanation that makes past unjustified discrimination permanent. But these three considerations, taken together, undercut the majority’s view of finality as inviolable. The Fair Sentencing Act cannot be said to further finality interests rationally when it permits modification of sentences above, but not at, the racially discriminatory, unjustified, repealed 100-to-l minimums.
That we have previously rejected equal protection challenges to the 100-to-l ratio in the § 3582(c)(2) context does not mean we must do so now. Simply put, our knowledge and Congress’s knowledge have changed. The ratio disproportionately affects African Americans. The ratio is unjustified. Congress repealed the law be*675cause the ratio is unjustified, with full awareness of its discriminatory effects. Using the ratio to deny sentence modifications continues to treat African-American offenders more harshly than White offenders, despite Congress’s aim to the contrary. Because of Congress’s acknowledgment of these problems and its explicit attempt to address them, “constitutional arguments that were unavailing in the past may not be foreclosed in the future.” United, States v. Then, 56 F.3d 464, 467 (2d Cir.1995) (Calabresi, J., concurring); see also United States v. Reddrick, 90 F.3d 1276, 1284 (7th Cir.1996) (Cudahy, J., concurring) (“The rationality of the 100:1 crack/powder ratio and its implications for equal protection will no doubt be the subject of continuing examination both within the judiciary and without.”); United States v. Smith, 73 F.3d 1414, 1422 (6th Cir.1996) (Jones, J., concurring) (“Continued use of the law to perpetuate a result at variance with rationality and common sense — even in a war on drugs — is indefensible.”).
I do not argue lightly that what once was rational has been made irrational by changed circumstances. But this is the rare case deserving such treatment. Our court mistakenly paid no heed to Judge Nathaniel Jones years ago in Smith, 73 F.3d at 1418. We would be wise to mark his words today: “As judges, we should no longer remain wedded to that which experience shows is neither rational nor fair.”
I respectfully dissent.
. See Substance Abuse & Mental Health Servs. Admin., Results from the 2012 National Survey on Drug Use and Health, tbl. 1.38A; *672U.S. Sentencing Comm’n, 2012 Sourcebook of Federal Sentencing Statistics, tbl. 34.
. Unfairness in Federal Cocaine Sentencing: Hearing Before Subcomm. on Crime, Terrorism, and Homeland Security of the H. Comm, on the Judiciary, 111th Cong. 50 (2009) (statement of Judge Ricardo H. Hinojosa, Acting Chair, U.S. Sentencing Comm’n).
. U.S. Sentencing Comm’n, Report to the Congress: Cocaine and Federal Sentencing Policy 11 (2002).
. See U.S. Sentencing Comm’n, Special Report to Congress: Cocaine and Federal Sentencing Policy ch. 7 (1995); U.S. Sentencing Comm’n, Special Report to Congress: Cocaine and Federal Sentencing Policy 8 (1997); U.S. Sentencing Comm’n, Report to the Congress: Cocaine and Federal Sentencing Policy 103 (2002); U.S. Sentencing Comm’n, Report to the Congress: Cocaine and Federal Sentencing Policy 8 (2007).
. U.S. Sentencing Comm’n, Report to the Congress: Cocaine and Federal Sentencing Policy 8 (2007); Dorsey v. United States, - U.S. -,-, 132 S.Ct. 2321, 2328, 183 L.Ed.2d 250 (2012).
. 156 Cong. Rec. H6196-01 (2010) (statement of Rep. Lungren).
. Restoring Fairness to Federal Sentencing: Addressing the Crack-Powder Disparity: Flearing Before the Subcomm. on Crime and Drugs of the S. Comm, on the Judiciary, 111th Cong. 99 (2009) (statement of U.S. Dep't of Justice through Lanny A. Breuer, Assistant Attorney General).
. Id. at 146-47 (statement of Asa Hutchinson).
. U.S. Sentencing Comm'n, Report to the Congress: Mandatory Minimum Penalties in the Federal Criminal Justice System 192 (2011).
. Reevaluating the Effectiveness of Federal Mandatory Minimum Sentences: Hearing Before S. Comm, on the Judiciary, 113th Cong. (2013) (statement of Judge Patti B. Saris, Chair, U.S. Sentencing Comm’n).
. U.S. Sentencing Comm'n, Preliminary Crack Retroactivity Data Report Fair Sentencing Act 6 tbl. 3 (Oct.2013).