Skip to main content


Reset A A Font size: Print

United States Court of Appeals, Eleventh Circuit.


No. 16-14556-J

Decided: July 18, 2016

Before: MARTIN, ROSENBAUM, and JILL PRYOR, Circuit Judges.

Application for Leave to File a Second or Successive Motion to Vacate, Set Aside, or Correct Sentence, 28 U.S.C. § 2255(h)


Charles Clayton seeks permission to file a 28 U.S.C. § 2255 motion based on Johnson v. United States, 135 S. Ct. 2551 (2015). Because Mr. Clayton previously filed a § 2255 motion, his new motion must be “certified as provided in section 2244 by a panel of the appropriate court of appeals to contain ․ a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court, that was previously unavailable.” 28 U.S.C. § 2255(h)(2). “The court of appeals may authorize the filing of a second or successive application only if it determines that the application makes a prima facie showing that the application satisfies the requirements of this subsection.” Id. § 2244(b)(3)(C). Mr. Clayton was sentenced in 2010 using United States Sentencing Guidelines § 4B1.1. This court has held that Johnson does not apply to sentences that were based on USSG § 4B1.1. See United States v. Matchett, 802 F.3d 1185, 1196 (11th Cir. 2015). Mr. Clayton has therefore not made “a prima facie showing” that his motion will meet § 2255(h)'s requirements for second or successive § 2255 motions. 28 U.S.C. § 2244(b)(3)(D).


Six years ago Charles Clayton was sentenced to 30 years in prison for possessing a kilogram of cocaine with intent to distribute. The criminal law he violated required a sentence of at least 10 years. But beyond the statute, Mr. Clayton faced the United States Sentencing Guidelines, which call for longer prison sentences for defendants who had been convicted of certain crimes earlier in their lives. Specifically, Mr. Clayton was sentenced based on a guideline that sets a longer sentence for defendants whose earlier crime “involves conduct that presents a serious potential risk of physical injury to another.” USSG § 4B1.2(a)(2). After Mr. Clayton was sentenced in 2010, the Supreme Court told us that these identical 13 words in the Armed Career Criminal Act (ACCA) are so vague as to violate the Due Process Clause of our Constitution. See Johnson v. United States, __ U.S. __, 135 S. Ct. 2551 (2015).

Since Johnson was decided, prisoners sentenced based on these words in the Sentencing Guidelines have come into federal courts seeking the same relief Johnson has given to prisoners sentenced based on the same words in the statute. Every other court of appeals has either held or assumed that Johnson makes the language in § 4B1.2(a)(2) of the Sentencing Guidelines unconstitutional.1 Our court alone has held otherwise. See United States v. Matchett, 802 F.3d 1185 (11th Cir. 2015). Over nine months ago Mr. Matchett asked this court to rehear his case, but we have yet to rule on his petition for rehearing. Because no ruling has issued on his petition and because the Supreme Court has now granted certiorari in a case that will evaluate this court's ruling in Matchett,2 I will write in Mr. Clayton's case to highlight the problems Matchett has caused people like him since it issued on September 21, 2015.

While I'm at it, Mr. Clayton's case also gives an opportunity to describe other ways our court has limited the reach of Johnson for people who may be serving unlawful sentences imposed in the federal courts of Alabama, Florida, and Georgia. Generally a person who finds himself serving an illegal sentence can seek relief by filing what is known as a § 2255 motion. Mr. Clayton filed a § 2255 motion in 2013. The statute governing § 2255 motions, the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), restricts a prisoner's ability to file more than once. Specifically, AEDPA allows a prisoner who already filed one § 2255 motion to file another (what the statute refers to as a “second or successive” motion) only if he first applies for and gets permission from the court of appeals. Mr. Clayton filed one of these applications. When courts of appeals get these applications AEDPA directs us to “certif[y]” whether the applicant made “a prima facie showing” that his § 2255 motion will “contain ․ a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court.” 28 U.S.C. §§ 2244(b)(3)(C), 2255(h). In the last couple of months, this court has received hundreds of these applications from prisoners who want relief based on the Supreme Court's ruling in Johnson.

In deciding these applications, we have been doing far more than what the statute directs. The judges of this court, myself included, have been combing through sealed records from the prisoner's original sentence hearing and going ahead to make a decision about whether the prisoner will win if we let him file his § 2255 motion in district court. We are making these decisions “without briefing or argument from a lawyer.” In re McCall, __ F.3d __, 2016 WL 3382006, at *2 (11th Cir. June 17, 2016) (Martin, J., concurring). Most troubling of all, none of these decisions can be appealed, even if it turns out we decided wrong. See 28 U.S.C. § 2244(b)(3)(E).


I first address this court's Matchett ruling, which has had the effect of locking prisoners into harsh sentences that I understand every other circuit to be reexamining. Matchett affects a lot of people. In fiscal year 2014 alone, 2,269 people in this country were sentenced using USSG § 4B1.1, known as the “career offender” guideline.3 No matter how short a sentence the guidelines might otherwise call for a defendant to get, if he is designated as a career offender then he is placed in the worst class of offenders and the judge is expected to impose the maximum sentence. See United States v. LaBonte, 520 U.S. 751, 754, 117 S. Ct. 1673, 1675 (1997) (“Pursuant to that Guideline, each defendant who qualifies for career offender status is automatically placed in criminal history ‘Category VI,’ the highest available under the Guidelines.”). To be clear, this drastic increase is required by statute. See 28 U.S.C. § 994(h) (“The Commission shall assure that the guidelines specify a sentence to a term of imprisonment at or near the maximum term authorized for categories of defendants ․ convicted of two or more prior felonies, each of which is ․ a crime of violence.”).

The guidelines define the term “crime of violence” from 28 U.S.C. § 994(h) in three ways, one of which uses the 13 words the Supreme Court ruled unconstitutional in Johnson. See USSG § 4B1.2. The term “crime of violence” is also used in several other places in the Sentencing Guidelines, so these 13 words can trigger higher sentences by way of those provisions as well. See, e.g., id. § 2K1.3 (the guidelines section for crimes involving explosives); id. § 2K2.1 (the section for firearm crimes, which can double a defendant's guidelines range based on one “crime of violence” or triple or quadruple it based on two 4 ); id. § 2S1.1 (the section for laundering crimes); id. § 7B1.1 (the section for probation and supervised release violations).

Since 2005, the Sentencing Guidelines are no longer mandatory, meaning that sentencing judges can now impose sentences above or below the sentence called for by the guidelines. Still, experience has shown us that sentencing judges rarely use their discretion to impose sentences up to the level called for by the career offender penalty (§ 4B1.1) when that penalty doesn't apply.5 Although I am sure he would prefer otherwise, Mr. Clayton's case is a good example of the enormous impact § 4B1.1 can have. Mr. Clayton received a § 4B1.1 sentence based on a 1994 Florida conviction. Back in 1994, the Florida court punished Mr. Clayton for this crime by giving him a sentence of probation and a $250 fine. When Mr. Clayton was sentenced in federal court sixteen years later, this same Florida conviction had a far more harsh effect. Without that 1994 crime, Mr. Clayton's sentencing range would have been 120 to 150 months in prison. But because the 1994 conviction required that he be sentenced as a career offender, his sentencing range became 360 months to life. The judge sentenced Mr. Clayton to 360 months.


In explaining why the Supreme Court's holding in Johnson doesn't apply to the guidelines, the Matchett panel looked to the Seventh Circuit's ruling in United States v. Tichenor, 683 F.3d 358 (7th Cir. 2012). Quoting Tichenor, the panel said: “Since the Guidelines are merely advisory, defendants cannot rely on them to communicate the sentence that the district court will impose. Defendants' inability to look to the Guidelines for notice underscores why ․ they cannot bring vagueness challenges against the Guidelines.” 802 F.3d at 1194 (quoting Tichenor, 653 F.3d at 365). But this view of the guidelines is at odds with how the Supreme Court views them, and in any event, it is perilous to now deduce this principle from Tichenor. A year after Tichenor was decided, the Supreme Court applied the Ex Post Facto Clause to the Sentencing Guidelines. See Peugh v. United States, __ U.S. __, 133 S. Ct. 2072 (2013). Like the vagueness doctrine, the Ex Post Facto Clause imposes a constitutional requirement of “fair notice.” Weaver v. Graham, 450 U.S. 24, 30, 101 S. Ct. 960, 965 (1981).

Even though the guidelines are now advisory, Peugh reminded us that they “remain the starting point for every sentencing calculation in the federal system.” 133 S. Ct. at 2083. In fact, “[18 U.S.C.] § 3553(a) explicitly directs sentencing courts to consider the Guidelines.” Gall v. United States, 552 U.S. 38, 50 n.6, 128 S. Ct. 586, 596 n.6 (2007). So while judges can choose sentences outside what the guidelines recommend, the Supreme Court has emphasized “the centrality of the Guidelines in the sentencing process.” Molina-Martinez v. United States, 578 U.S. __, __, 136 S. Ct. 1338, 1346 (2016). Molina-Martinez explained that the guidelines are “the starting point for the district court's decision and anchor the court's discretion in selecting an appropriate sentence.” Id. at 1349; see also id. at 1342 (“The Sentencing Guidelines provide the framework for the tens of thousands of federal sentencing proceedings that occur each year.”); Peugh, 133 S. Ct. at 2087 (“District courts must begin their sentencing analysis with the Guidelines ․ and use them to calculate the sentencing range correctly; and those Guidelines will anchor both the district court's discretion and the appellate review process.”).6

The guidelines have this “anchor” effect even when judges depart from them. “Even if the sentencing judge sees a reason to vary from the Guidelines, if the judge uses the sentencing range as the beginning point to explain the decision to deviate from it, then the Guidelines are in a real sense the basis for the sentence.” Peugh, 133 S. Ct. at 2083 (quotation omitted). “In less than one-fifth of cases since 2007 have district courts imposed above- or below-Guidelines sentences absent a Government motion. Moreover, the Sentencing Commission's data indicate that when a Guidelines range moves up or down, offenders' sentences move with it.” Id. at 2084 (citations omitted). Indeed a vague guideline can wreak harm on a defendant even before he is convicted of any crime. Id. at 2085 (plurality opinion) (“[A] defendant charged with an increased punishment for his crime is likely to feel enhanced pressure to plead guilty. This pressure does not disappear simply because the Guidelines range is advisory; the defendant will be aware that the range is intended to, and usually does, exert controlling influence on the sentence that the court will impose.” (citation omitted)).7

Peugh outright rejected the idea the Matchett panel relied on – that the guidelines need not give notice because they are “purely advisory.” Id. at 2087 (majority opinion). The Court wrote: “[i]t is simply not the case that the Sentencing Guidelines are merely a volume that the district court reads with academic interest in the course of sentencing.” Id. Rather, the guidelines are “the Federal Government's authoritative view of the appropriate sentences for specific crimes.” Id. at 2085 (plurality opinion). And they announce “the most recent views of the agency charged by Congress with developing sentencing policy.” Id. at 2087 (majority opinion). For these reasons the guidelines must “give fair warning of their effect and permit individuals to rely on their meaning.” Miller v. Florida, 482 U.S. 423, 430, 107 S. Ct. 2446, 2451 (1987) (quotation omitted) (applying the Ex Post Facto Clause to Florida's advisory sentencing guidelines); see also Peugh, 133 S. Ct. at 2085 (plurality opinion) (“The [Ex Post Facto] Clause ensures that individuals have fair warning of applicable laws.”).

Like the Ex Post Facto Clause, the vagueness doctrine must also apply to the Sentencing Guidelines. Just as the Ex Post Facto Clause ensures “fair warning,” the vagueness doctrine says no law can be “so vague that it fails to give ordinary people fair notice of the conduct it punishes.” Johnson, 135 S. Ct. at 2556. The Matchett panel said Johnson does not apply because “advisory guidelines that inform a sentencing judge's discretion [ ] cannot violate the notice requirement.” 802 F.3d at 1195. Peugh tells us the opposite. Even if Peugh doesn't set out exactly when the vagueness doctrine applies, it shows that the panel's “notice is irrelevant” argument is not enough to shield the guidelines from constitutional scrutiny. And “notice is irrelevant” is the best defense the panel gave for its ruling.

The panel's idea that notice is irrelevant for the Sentencing Guidelines seems to be based on its misreading of Irizarry v. United States, 553 U.S. 708, 128 S. Ct. 2198 (2008). Irizarry held that Rule 32 of the Federal Rules of Criminal Procedure does not require a judge to say in advance of the sentence hearing what sentence she may impose. Id. at 709, 128 S. Ct. at 2200. In contrast, neither Peugh nor Johnson has anything to do with what a judge must say in any single case. Peugh and Johnson tackle the larger question of what notice is due the entire public about what punishment can be expected for a given offense. This notice is required from the “sentencing process long before the district court imposes the sentence.” Molina-Martinez, 136 S. Ct. at 1342. That's why Peugh says that the Guidelines must give notice of what conduct the courts will punish, even if a particular judge is not required to give advance warning about how she will use her discretion in a given case.

The Matchett opinion ignores these lessons from Peugh and Molina-Martinez. Worse, the Matchett panel relied on precedent those cases abrogated. For example, the panel purported to “join” the Sixth Circuit “insofar as we reject Matchett's argument that advisory guidelines can be unconstitutionally vague.” 802 F.3d at 1196. As it happens, the Sixth Circuit has itself considered this same argument and held “that the rationale of Johnson applies equally to the residual clause of the Guidelines.” United States v. Pawlak, __ F.3d. __, 2016 WL 2802723 at *8 (6th Cir. May 13, 2016). The Matchett panel also twice quoted from United States v. Wivell, 893 F.2d 156 (8th Cir. 1990). The Eighth Circuit has recognized that “[t]he reasoning in Wivell that the guidelines cannot be unconstitutionally vague because they do not proscribe conduct is doubtful after Johnson.” United States v. Taylor, 803 F.3d 931, 933 (8th Cir. 2015) (per curiam). And the Seventh Circuit, whose Tichenor opinion the Matchett panel quoted four times, has “proceed[ed] on the assumption that the Supreme Court's reasoning [in Johnson] applies to section 4B1.2.” Ramirez v. United States, 799 F.3d 845, 856 (7th Cir. 2015). Of the other courts cited by the panel, this leaves only the Fifth Circuit, which has also vacated § 4B1.2 sentences because of Johnson. See infra n.1. The Eleventh Circuit is all alone on this.


The Matchett panel got the Sentencing Guidelines wrong in this way. It got the vagueness doctrine wrong as well. It said “[t]he vagueness doctrine applies only to laws that prohibit conduct and fix punishments.” 802 F.3d at 1189. I say federal policy that causes certain conduct to be punished by more years in jail “prohibit[s] conduct and fix[es] punishments.” But even if I am wrong on this, we know that the Supreme Court has long applied the vagueness doctrine to all kinds of things that don't “prohibit conduct and fix punishments.” For example, a half century ago Giaccio v. State of Pennsylvania, 382 U.S. 399, 86 S. Ct. 518 (1966), held unconstitutionally vague a statute that allowed juries to make acquitted defendants pay court costs “without any legally fixed standards.” Id. at 403, 86 S. Ct. at 521. The Supreme Court distinguished the statute, which after all applied only to those acquitted, from laws that “impose[ ] forfeitures, punishments or judgments for costs.” Id. at 404, at 86 S Ct. at 522. This Pennsylvania statute neither prohibited conduct nor fixed punishment, but it was nonetheless unconstitutionally vague.

The panel's reasoning also defies Johnson itself. Johnson reminded us of two ways in which vague laws can violate the Fifth Amendment's guarantee of due process of law: “Our cases establish that the Government violates this guarantee by taking away someone's life, liberty, or property under a criminal law so vague that it fails to give ordinary people fair notice of the conduct it punishes, or so standardless that it invites arbitrary enforcement.” 135 S. Ct. at 2556. The Supreme Court then held that the 13 words of ACCA's “residual clause” were unlawful in both these ways: “We are convinced that the indeterminacy of the wide-ranging inquiry required by the residual clause both denies fair notice to defendants and invites arbitrary enforcement by judges.” Id. at 2557.

The Matchett panel gave no heed to these admonitions against “arbitrary enforcement.” Zero. Instead, the panel addressed only Johnson's “notice” rationale, without ever mentioning the Court's concern about “arbitrary enforcement by judges.” This matters because we have been instructed that the “arbitrary enforcement” concern is “the more important aspect of vagueness doctrine.” Kolendar v. Lawson, 461 U.S. 352, 358, 103 S. Ct. 1855, 1858 (1983). Perhaps reflecting this lesson, every time Johnson told us why the residual clause is not lawful, it underscored the problem that the vague language of the clause led different judges to give similarly situated defendants widely varying sentences.8 But again, the panel made no effort to address this concern about arbitrariness, which the Supreme Court told us is “the more important aspect of vagueness doctrine.”

If the panel had been willing to evaluate how the residual clause in the Sentencing Guidelines leads to “arbitrary enforcement by judges,” then the case would have easily resolved in Mr. Matchett's favor. The Supreme Court has told us that overly vague laws violate our Constitution because they “delegate [ ] basic policy matters to policemen, judges, and juries for resolution on an ad hoc and subjective basis, with the attendant dangers of arbitrary and discriminatory application.” Grayned v. City of Rockford, 408 U.S. 104, 108–09, 92 S. Ct. 2294, 2299 (1972). Or as Johnson put it, “the indeterminacy of the wide-ranging inquiry required by the residual clause” makes the clause unconstitutional because it “invites arbitrary enforcement by judges.” 135 S. Ct. at 2557.

The risks of “discriminatory application” and “arbitrary enforcement” here should be obvious. Two judges who are sentencing defendants with identical records can arrive at different sentences based on each judge's personal sense of what seems like a crime of violence. Judges will certainly try to apply pre-Johnson residual clause opinions correctly when sentencing a person under USSG § 4B1.2.9 But Justice Scalia once said that he worried that, in the end, judges will have little choice but to “simply throw the opinions into the air in frustration, and give free rein to their own feelings as to what offenses should be considered crimes of violence.” Derby v. United States, 131 S. Ct. 2858, 2859 (2011) (Scalia, J., dissenting from denial of certiorari). Of course we expect that judges will not act so ignobly. But “the due process protection against vague regulations does not leave the public at the mercy of noblesse oblige.” FCC v. Fox Television Stations, Inc., __ U.S. __, __, 132 S. Ct. 2307, 2318 (2012) (quotation omitted). Instead it bans any regulation that is “so standardless that it authorizes or encourages seriously discriminatory enforcement.” United States v. Williams, 553 U.S. 285, 304, 128 S. Ct. 1830, 1845 (2008).


Also worrisome, the Matchett panel opinion forces this court and all courts sentencing human beings within the Eleventh Circuit to continue to apply and even add to the body of law that Johnson discredited. The panel recognizes that Johnson “abrogated the previous decisions of the Supreme Court interpreting the residual clause.” 802 F.3d at 1195. But the panel nevertheless instructs courts that they “must still adhere to the reasoning of [these] cases” when interpreting § 4B1.2. Id. at 1195–96. Our court will thus continue to apply cases the Supreme Court derided as “anything but evenhanded, predictable, or consistent.” Johnson, 135 S. Ct. at 2563. Indeed Johnson referenced several § 4B1.2 cases to illustrate how the residual clause “produces more unpredictability and arbitrariness than the Due Process Clause tolerates.” 135 S. Ct. at 2558. The Supreme Court even gave this court an unwanted tip of the hat, when it cited one of our § 4B1.2 opinions to declare that “[t]his Court is not the only one that has had trouble making sense of the residual clause.” Id. at 2559–60 (quotation omitted) (citing United States v. Whitson, 597 F.3d 1218 (11th Cir. 2010)). By my reading, the Supreme Court treated the problem with the residual clause in the statute as identical to the problem with the residual clause in the Sentencing Guidelines.10

Related to that point, the panel warned that applying the vagueness doctrine “would upend our sentencing regime” since “many [guidelines] provisions could be described as vague.” Id. at 1196. But the sky is not really falling. Johnson does not invalidate everything that “could be described as vague.” The Court singled out a far more distinct problem: laws that require judges to apply overly vague standards “to a judicially imagined ‘ordinary case’ of a crime, not to real-world facts.” 135 S. Ct. at 2557; see also id. at 2558 (“It is one thing to apply an imprecise “serious potential risk” standard to real-world facts; it is quite another to apply it to a judge-imagined abstraction.”). In contrast, the guidelines base punishment almost exclusively on a defendant's actual conduct.11 Consider the two provisions that the Matchett panel warned could next be deemed vague. See 802 F.3d at 1196. The first asks if the defendant used “sophisticated means” in committing the actual crime. USSG § 2B1.1(b)(10). The other asks if she played a “minor” role in the actual crime. Id. § 3B1.2(b). Johnson expressly condones standards that assess actual conduct in this way. See 135 S. Ct. at 2561 (“[W]e do not doubt the constitutionality of laws that call for the application of a qualitative standard such as ‘substantial risk’ to real-world conduct.”). I sincerely doubt that resentencing Mr. Matchett and others like him will “upend our sentencing regime.” Every other circuit has applied Johnson to the guidelines without any apparent trouble.


The Matchett opinion is only one of many ways the Eleventh Circuit has uniquely “limited Johnson's reach.” McCall, 2016 WL 3382006, at *2 (Martin, J., concurring). Today marks three months since the Supreme Court's April 18 decision in Welch v. United States, __ U.S. __, 136 S. Ct. 1257 (2016). In those three months, the Eleventh Circuit has ruled on over 1,159 applications for permission to file a successive § 2255 motion based on Johnson. These cases typically come to us as “nothing more than a form filled out by a prisoner.” McCall, 2016 WL 3382006, at *2 (Martin, J., concurring). Because we have so little to go on, “our court has been calling up each prisoner's record” to decide, as though we ourselves are resentencing the prisoner, whether his record would qualify him for an ACCA sentence based on the parts of the statute that Johnson left intact. Id. Having been a part of this process, I can now count at least five ways that this “massive effort to decide the merits of hundreds of habeas cases within 30 days each, all over a span of just a few weeks ․ sets our court apart.” Id. at *3.

First, “[w]e were in the minority of courts that, from the beginning, said prisoners could not benefit from Johnson if they had already filed an earlier § 2255 motion.” Id. That ruling “denied the application of Johnson to potentially hundreds of people based on pro se pleadings and without oral argument or briefing.” In re Franks, 815 F.3d 1281, 1289 (11th Cir. 2016) (Martin, J., dissenting). The government (who, after all, jailed these prisoners in the first place) disagreed with our ruling and “respectfully urge [d] the Court to convene en banc to decide this important issue.” Id. at 1289 n.1.

Second, even after the Welch decision (remarkably issued only nineteen days after the case was argued) abrogated our precedent, prisoners faced another obstacle. Only two months remained until the one-year limitations period for making Johnson claims ran out. In the Eleventh Circuit, over a hundred prisoners tried to file Johnson applications before the Supreme Court decided Welch, most of them in the three months between the Welch certiorari grant and the day Welch was decided. See In re Robinson, __ F.3d __, 2016 WL 1583616, at *2–*5 (11th Cir. Apr. 19, 2016) (Martin, J., concurring) (listing 110 cases and dates). But “unlike all other circuits, the Eleventh Circuit refused to stay applications for successive § 2255 motions pending Welch.”12 A month before the Welch decision, the full Eleventh Circuit refused to vacate the case that Welch eventually abrogated. See In re Rivero, No. 15-13089 (Mar. 16, 2016). A few days later, the full court also vacated a panel opinion that allowed the court to stay Johnson applications “until the Supreme Court decides Welch.” In re Johnson, 810 F.3d 1247, 1253 (11th Cir. 2016). If either of those votes had gone the other way, hundreds of prisoners would have had more time to file their cases, and the judges of this court would have had more time to evaluate them. Instead, inmates had to re-file their applications, and this court had to decide the same cases again, this time in a crush of hundreds of new Johnson cases. Other courts took steps to avoid this problem.

Third, “I am aware of no order from another court of appeals that combs through an applicant's presentence investigation report to decide the merits of his yet-unfiled motion without ever hearing from a lawyer.” McCall, 2016 WL 3382006, at *3 n.7 (Martin, J., concurring). A court of appeals is simply not equipped to construct a new basis for a prisoner's old sentence in this way. At the district court sentence hearing, unlike the pure paper review we are doing here, defense lawyers can point out factual errors in the PSI and otherwise educate the court about why the recommendations in the PSI might not be appropriate. None of that protection exists when prisoners apply to our court for permission to go to the district court to have their sentences reexamined.13 Also, when deciding these cases, our court has over and over again failed to apply the Supreme Court's current interpretation of ACCA.14 This appears to set us apart from our peers too.

Fourth, as far as I know, ours is the only court to force a decision on every one of these cases within 30 days of filing. We do this based on 28 U.S.C. § 2244(b)(3)(D), which says: “The court of appeals shall grant or deny the authorization to file a second or successive application not later than 30 days after the filing of the motion.” Of course we never take lightly the word “shall” in a statute. But others who have considered this statute have concluded that “failure to comply with the thirty-day provision does not deprive this Court of the power to grant or deny a motion under § 2244(b)(3)(A)” because “the provision is hortatory or advisory rather than mandatory.” In re Siggers, 132 F.3d 333, 336 (6th Cir. 1997); see also In re Johnson, 810 F.3d 1247, 1249 (11th Cir. 2016) (“This Court has never decided if this 30-day timeframe is mandatory. All seven of the Courts of Appeals that have decided this question in a published opinion have said it is not.”). The Ninth Circuit recently observed about its experience deciding Johnson applications: “Given the large volume of second or successive applications our court must process each month, it frequently takes us longer—sometimes much longer—than 30 days to rule.” Orona v. United States, __ F.3d __, 2016 WL 3435692, at *2 (9th Cir. June 22, 2016). I can't help but think that if this court had taken the approach taken by others, our work on these cases would be both less frantic and more accurate.

Fifth, another panel of this court recently held that “the federal habeas statute requires us to dismiss a claim that has been presented in a prior application” to file a § 2255 motion. In re Baptiste, __ F.3d __, __, 2016 WL 3752118, at *2 (11th Cir. July 13, 2016). Baptiste was construing another provision of AEDPA, that is 28 U.S.C. § 2244(b)(1), which says any “claim presented in a second or successive habeas corpus application under section 2254 that was presented in a prior application shall be dismissed.” Of course, the § 2255 motions that I am talking about in this dissent are filed by federal prisoners. § 2255 motions are certainly not brought “under section 2254,” which governs petitions filed by state prisoners. But the Baptiste panel ruled that even though § 2244(b)(1) does not mention § 2255 motions, it applies to them anyway, since “it would be odd [ ] if Congress had intended to allow federal prisoners” to do something state prisoners can't do. Id.15 The panel thus held that the court must deny “leave to file a second or successive § 2255 motion presenting the same claims we have already rejected on their merits in a previous application.” Id. We have, of course, yet to see the full effect of Baptiste, but I hope it doesn't work to bar relief for prisoners who ask us to look at their case again if we got it dead wrong the first time.

On the topic of this court's singular approach, I add one more observation. Last month the Supreme Court granted certiorari in the case of a Texas prisoner named Duane Buck. See Buck v. Stephens, No. 15-8049, __ S. Ct. __, 2016 WL 531661 (U.S. June 6, 2016). The Court took the case even though the lower court ruled that Mr. Buck's appeal was so meritless that he couldn't even file it. Mr. Buck's petition for certiorari asked: “did the United States Court of Appeals for the Fifth Circuit impose an improper and unduly burdensome Certificate of Appealability (COA) standard?” Our treatment of applications for successive § 2255 motions may be even more troubling than the issue raised in Buck. Unlike for the denial of a COA, AEDPA provides that “denial of an authorization ․ to file a second or successive application shall not be appealable and shall not be the subject of a petition for rehearing or for a writ of certiorari.” 28 U.S.C. § 2244(b)(3)(E). This means no motion for reconsideration, no motion for en banc review, no appeal, and no petition for certiorari. The decisions we make in these cases are therefore, as a practical matter, not reviewable.16

A month after AEDPA became law, the Supreme Court held that these “new restrictions on successive petitions ․ do not amount to a ‘suspension’ of the writ.” Felker v. Turpin, 518 U.S. 651, 664, 116 S. Ct. 2333, 2340 (1996). Three Justices filed a concurrence warning that “the question whether the statute exceeded Congress's Exceptions Clause power” might need to be revisited “if the courts of appeals adopted divergent interpretations of the gatekeeper standard.” Id. at 667, 116 S. Ct. at 2342 (Souter, J., concurring). I hope someone better equipped than me will take this opportunity to look at whether the divergent views taken by this court require reexamination of this question asked by these Justices so soon after AEDPA was enacted. Twenty years later, I worry that our court's harsh view of our § 2244(b) gatekeeping role brings us perilously close to a suspension of the writ.

Imagine a sentencing guideline that read, “A defendant is a career offender if ‘[p]uddles do not ask for why not? It is cheese! Breath and wind. It is cheese.’ ” Boston Legal, “Word Salad Days” (2006), (last visited Apr. 28, 2016). Now imagine that based on the Guidelines range that that indecipherable language required, a district court sentenced a defendant to twice as much time as it otherwise would have. How could the sentencing court know that the guideline applied? How could the reviewing court know that the correct Guidelines calculation included an enhancement under that guideline? Surely doubling a defendant's sentence based on nonsense would violate due process. But in United States v. Matchett, 802 F.3d 1185 (11th Cir. 2015), we allowed defendants to continue to be sentenced to much more severe sentences than they would otherwise receive, based on the residual clause of the career-offender guideline, a guideline that the Supreme Court has found hardly more scrutable than the hypothetical one above.

No doubt criminal defendants do not have a due-process right to a sentence within a particular Sentencing Guidelines range. But Congress can, and essentially has, required courts to begin the sentencing process by correctly calculating the Guidelines range. The question here is whether, when the Supreme Court strikes language from a statute because it is unconstitutionally vague language and that same language also appears in a guideline, we are constitutionally able to continue to apply that language in the sentencing process that Congress has mandated. The answer, unlike the challenged part of the career-offender guideline, is clear: we are not.

I concur in all but Section I.A of Judge Martin's well-reasoned concurrence. I agree that the Supreme Court's decision in Johnson v. United States, 576 U.S. __, 135 S. Ct. 2551 (2015), holding the Armed Career Criminal Act's (“ACCA”) residual clause unconstitutionally vague renders the exact same language in the Sentencing Guidelines unconstitutional as well.


In Matchett, 802 F.3d 1185, the panel reached the opposite conclusion because it held that the vagueness doctrine does not apply to the Sentencing Guidelines. 802 F.3d at 1193-95. To reach that result, the panel first described the vagueness doctrine as “rest[ing] on [a] lack of notice.” Id. at 1194 (quoting Maynard v. Cartwright, 486 U.S. 356, 361, 108 S. Ct. 1853, 1857 (1988)). Then, the panel construed Irizarry v. United States, 553 U.S. 708, 128 S. Ct. 2198 (2008), as precluding due-process challenges to, essentially, anything having to do with sentencing under the Guidelines, based on the Supreme Court's remark that that “[a]ny expectation subject to due process protection ․ that a criminal defendant would receive a sentence within the presumptively applicable Guidelines range did not survive our decision in United States v. Booker, 543 U.S. 220, 125 S. Ct. 738, 160 L. Ed. 2d 621 (2005), which invalidated the mandatory features of the Guidelines.” Matchett, 802 F.3d at 1194 (quoting Irizarry, 553 U.S. at 713, 128 S. Ct. at 2202). Finally, the panel quoted the Eighth Circuit's decision in United States v. Wivell, 893 F.2d 156, 160 (8th Cir. 1990), for the proposition that “[b]ecause there is no constitutional right to sentencing guidelines ․ the limitations the Guidelines place on a judge's discretion cannot violate a defendant's right to due process by reason of being vague.” Matchett, 802 F.3d at 1194-95 (quoting Wivell, 893 F.2d at 160).


The problem with the first part of the panel's analysis—that the vagueness doctrine “rest[s] on [a] lack of notice”—is that it is incomplete. The vagueness doctrine also protects against arbitrary enforcement by judges. Indeed, in Johnson itself the Supreme Court held that the ACCA equivalent of the 13 words at issue here violated due process because it “both denies fair notice to defendants and invites arbitrary enforcement by judges.” Johnson, 135 S. Ct. at 2557 (emphasis added).


As for the second part of the panel's analysis—that Irizarry precludes due-process challenges to all forms of sentencing error under the Guidelines—I respectfully disagree. In Irizarry, under the advisory Guidelines, a defendant was sentenced above the correctly calculated Guidelines range. 553 U.S. at 712, 128 S. Ct. at 2201. He asserted that his due-process rights had been violated because the sentencing court varied upwards from the Guidelines range without providing him with prior notice. See id. The Supreme Court rejected his argument, explaining that under the advisory Guidelines, “neither the Government nor the defendant may place the same degree of reliance on the type of ‘expectancy’ [of a given sentence] that gave rise to a special need for notice [when the Guidelines were mandatory and the sentencing court departed from them].” Id. 553 U.S. at 713-14, 128 S. Ct at 2202.

Put simply, Irizarry stands for only the proposition that a defendant has no due-process interest in receiving a sentence within the Guidelines range. But Irizarry says nothing about whether a defendant has a due-process right to a correct and fair sentencing process under the Sentencing Guidelines. And while the Supreme Court has not expressly spoken to such a right, the Court's recent decisions strongly indicate that the right exists.

For starters, in Molina-Martinez v. United States, 578 U.S. ___, 136 S. Ct. 1338 (2016), the Supreme Court recently explained that a district court that “improperly calculat[es]” a defendant's Guidelines range makes a “significant procedural error,” id. at 1346 (quoting Gall v. United States, 552 U.S. 38, 51, 128 S. Ct. 586, 597 (2007)) (quotation marks omitted)—so “particularly serious,” id., in fact, that the error generally qualifies in its own right as having “affected the defendant's substantial rights.” Id.1

And that is not surprising, given that the Supreme Court has established that a correct and fair sentencing process necessarily begins with the correct calculation of the Guidelines range. Gall v. United States, 552 U.S. 38, 50 & n.6, 128 S. Ct. 586, 596 & n.6 (2007). Indeed, the Supreme Court has instructed that under 18 U.S.C. § 3553(a), “district courts must begin their analysis with the Guidelines and remain cognizant of them throughout the sentencing process.” Id. at 50 n.6, 128 S. Ct. at 596 n.6 (emphasis added); see Peugh v. United States, 569 U.S. __, 133 S. Ct. 2072, 2083 (2013) (same). The correct Guidelines calculation “anchor[s] both the district court's [sentencing] discretion and the appellate review process.” Peugh, 133 S. Ct. at 2087.

In other words, the Supreme Court has acknowledged that Congress has effectively legislated the requirement that a sentencing court start the sentencing process by first correctly calculating the Guidelines range. That makes § 3553(a) a “statute[ specifying the procedure for] fixing sentences.” See Johnson, 135 S. Ct. at 2556-57 (holding that the vagueness doctrine applies to statutes fixing sentences). So to the extent that, as a part of the statutorily mandated sentencing process, § 3553(a) requires courts in calculating the Guidelines range to use a guideline that is “so standardless that it invites arbitrary enforcement,” the guideline must be struck down. See id. Failure to do so would render the sentencing process that § 3553(a) requires—determining the correct calculation of the Guidelines range—violative of due process because no court could reliably ascertain the correct calculation of the Guidelines range.

That is exactly the problem that the challenged language of the career-offender guideline presents. How can a sentencing court correctly calculate the Guidelines range when it is forced to apply the “hopeless[ly] indetermina[te]” language of the career-offender guideline? Johnson, 135 S. Ct. at 2448. Courts had “trouble making sense” of the very same words when they tried to apply them under the ACCA's residual clause. Id. at 2559-60. The Supreme Court observed that “[n]ine years' experience trying to derive meaning from the residual clause convince[d it] that [it] ha[d] embarked upon a failed enterprise.” Id. at 2560. This “ ‘black hole of confusion and uncertainty’ that frustrates any effort to impart ‘some sense of order and direction,’ ” id. at 2562 (quoting United States v. Vann, 660 F.3d 771, 787 (4th Cir. 2011) (Agee, J., concurring)), does not somehow magically become clearer or more meaningful because the words appear in the guideline, rather than in the ACCA.

Because of this muddle, a sentencing court cannot ascertain whether the challenged part of the career-offender guideline even applies when the guideline is raised, so the court necessarily cannot correctly calculate the Sentencing Guidelines range. As a result, the sentencing court cannot comply with the sentencing process's virtual statutory requirement that the sentencing court first correctly calculate the applicable Guidelines range.

And the confusion only grows on appeal. Determining whether a sentence imposed by a district court was procedurally reasonable requires appellate courts to first ascertain whether the district court correctly calculated the applicable Guideline range. But we are no more skilled in applying “hopeless[ly] indetermina[te]” language than district courts.


Finally, with regard to the third part of the Matchett panel's analysis—that the Sentencing Guidelines cannot be challenged as vague because no constitutional right to sentencing guidelines exists—I again respectfully disagree. True, “legislatures remain free to decide how much discretion in sentencing should be reposed in the judge or jury in noncapital cases.” Lockett v. Ohio, 438 U.S. 586, 603, 98 S. Ct. 2954, 2964 (1978). But legislatures cannot, as Matchett would apparently hold, cabin the sentencing discretion of judges by mandating that they calculate a defendant's sentence using unconstitutionally vague language.

This would be another case entirely if sentencing judges could choose to wholly disregard the unconstitutionally vague career-offender guideline in calculating sentences. They cannot. Instead, district courts must begin the sentencing process by correctly calculating a defendant's Guidelines range. Peugh, 133 S. Ct. at 2083. Under Matchett, that means that Congress has essentially required district courts to apply unconstitutionally vague language in sentencing defendants. But it could not do that. Due process may not require sentencing guidelines, but it does prohibit Congress from requiring judges to apply unconstitutionally vague language in correctly calculating a defendant's sentence under any guidelines it chooses to enact.


At bottom, statutorily, courts are required to begin every sentencing by correctly calculating the Guidelines range. Yet the Supreme Court has recognized that courts cannot reliably know whether the challenged language of the career-offender Guideline applies in any given case. As a result, they cannot possibly know whether a correct calculation of the Guidelines range should or should not include such an enhancement. But in Matchett, we nonetheless required sentencing courts to impose the enhancement and ourselves to uphold it, anyway. Trying to divine meaning from the word salad that is the challenged portion of the career-offender guideline guarantees an arbitrary and unfair sentencing process in violation of due process. For these reasons, I respectfully disagree with our holding in Matchett.

The Supreme Court has told us that it violates the Constitution's guarantee of due process to fix a sentence based on a person's having committed a prior violent felony defined as “involv[ing] conduct that presents a serious potential risk of physical injury to another.” 18 U.S.C. § 924(e)(2)(B)(ii) (containing the so-called “residual clause” of the Armed Career Criminal Act (“ACCA”)); Johnson v. United States, 135 S. Ct. 2551 (2015) (striking the residual clause as unconstitutionally vague). This definition of “violent felony,” the Supreme Court said, is “so shapeless a provision” that any attempt “to derive meaning from” it necessarily will be “a failed enterprise.” Johnson, 135 S. Ct. at 2560.

Since the Supreme Court decided in Johnson that this language is unconstitutionally vague, we have repeatedly misinterpreted and misapplied that decision. As a result of our erroneous application of Johnson, in this Circuit thousands of people remain incarcerated who were sentenced under the very same language. Charles Clayton is one of these people.


Mr. Clayton was sentenced not under the residual clause in the ACCA, but under an identical clause that appears in the career offender sentencing enhancement of the United States Sentencing Guidelines, U.S.S.G. §§ 4B1.1, 4B1.2(a)(2) (amended 2016). The career offender enhancement, just like the ACCA, applies when a defendant sentenced in federal court has been convicted of three violent felonies.1 Just like the ACCA, which lengthens a defendant's maximum sentence of 10 years' imprisonment to a minimum sentence of 15 years, the career offender enhancement has the effect of significantly increasing a defendant's sentence. For Mr. Clayton, it meant that instead of a likely sentence of 120 to 150 months, he faced a likely sentence of at least 360 months. The sentencing judge in Mr. Clayton's case found that he previously had been convicted of a crime that “involve[d] conduct that present[ed] a serious potential risk of physical injury to another” and imposed a 360 month sentence—more than twice what Mr. Clayton likely otherwise would have served. U.S.S.G. § 4B1.2(a)(2).

Mr. Clayton has asked this Court for the opportunity to request relief from the district court because he was subject to a much higher sentence due to 13 words the Supreme Court has held to violate an individual's constitutional rights. But we nonetheless must deny him the opportunity to even bring his claim to the district court's attention because this Court has erected barriers to Mr. Clayton and thousands of others despite the Supreme Court's unambiguous holding in Johnson.


In throwing up these sorts of barriers, this Court consistently got it wrong. For starters, shortly after Johnson was decided, a panel of this Court limited severely the reach of that ruling by making relief unavailable to any inmate who previously had filed a motion to vacate his sentence under 28 U.S.C. § 2255. See In re Rivero, 797 F.3d 986 (11th Cir. 2015). The Supreme Court explained over a decade ago that a new substantive rule of constitutional law is retroactively applicable to cases under review in habeas proceedings. Schriro v. Summerlin, 542 U.S. 348, 351-52 (2004) (applying Teague v. Lane, 489 U.S. 288 (1989)2 ). The Rivero panel acknowledged that in Johnson the Supreme Court announced a new substantive rule of constitutional law. 797 F.3d at 989.

But the panel refused to apply Johnson retroactively for two separate, newly formulated reasons. First, the panel held that Johnson could not be retroactively applicable, even to ACCA cases like Johnson itself, because “Congress could impose the punishment in Johnson if Congress did so with specific, not vague, language.” Id. at 991. And second, the Rivero panel held that for Johnson's rule to be retroactively applicable to the career offender enhancement in the sentencing guidelines, under which Gilberto Rivero had been sentenced, the Supreme Court must also have held specifically that the guidelines are subject to vagueness challenges (which it had not). Id.

The Rivero panel's holdings contradicted what the Supreme Court had already told us about retroactivity principles. And the panel provided no basis in Supreme Court precedent to justify these two newly minted barriers to relief for those sentenced based upon language the Supreme Court had just told us was unconstitutionally vague. As to the first holding, the Supreme Court previously implicitly rejected the idea that the prospect of Congressional intervention could limit the retroactive applicability of a new substantive rule.3 As to the second holding in Rivero, nothing in the Supreme Court's body of retroactivity law so much as hinted at a requirement beyond what Justice O'Connor described in Tyler v. Cain: “[I]f we hold in Case One that a particular type of rule applies retroactively to cases on collateral review and hold in Case Two that a given rule is of that particular type, then it necessarily follows that the given rule applies retroactively to cases on collateral review.” 533 U.S. 656, 668-69 (2001) (O'Connor, J., concurring); see In re Holladay, 331 F.3d 1169, 1172-73 (11th Cir. 2003) (applying Justice O'Connor's test to hold that the Supreme Court had “made” the rule announced in Atkins v. Virginia, 536 U.S. 304 (2002), barring the execution of intellectually disabled persons, retroactively applicable). Although the Rivero panel acknowledged that Justice O'Connor's retroactivity test applied, it failed to employ the test as Justice O'Connor constructed it.

Nonetheless, because of Rivero, from August 2015 until April 2016 we denied relief to every inmate whose Johnson-based request to file a second or successive § 2255 motion we decided. We even continued to deny inmates the opportunity to seek relief after the Supreme Court accepted certiorari in a case in which it would decide the issue of Johnson's retroactivity. See Welch v. United States, 136 S. Ct. 790 (2016) (granting petition for certiorari).4

In Welch, the Supreme Court told us that we were wrong to hold that the rule announced in Johnson did not apply retroactively. 136 S. Ct. 1257, 1268 (2016). Welch was a case from our Circuit. Mr. Welch had filed a first § 2255 motion before Johnson was decided, challenging his ACCA sentence, which was based on the residual clause of that statute. Id. at 1263. The district court denied Mr. Welch relief, and he sought a certificate of appealability (“COA”) from this Court. Id. Even though he notified our Court that Johnson was pending in the Supreme Court and requested that his motion be held pending the Supreme Court's decision, his motion for a COA was denied. Id. “Less than three weeks later,” the Supreme Court observed in overruling us on Johnson's retroactivity, “this Court issued its decision in Johnson.” Id. We had denied Mr. Welch any opportunity for relief knowing that Johnson soon would be decided.5

The Supreme Court in Welch flatly rejected this Court's Rivero holding that Johnson's rule was not retroactive. It noted that it had already rejected the argument that its decisions might not be retroactively applicable if Congress could “enact a new version of the residual clause that imposes the same punishment on the same persons for the same conduct, provided the new statute is precise enough to satisfy due process.” Id. at 1267; see Rivero, 797 F.3d at 991 (relying upon this reasoning). The “clearest example” the Court pointed out, was its prior decision in Bousley, which held a new rule to be retroactive “even though Congress could (and later did) reverse [the rule announced in] Bailey by amending the statute.” Welch, 136 S. Ct. at 1267 (emphasis added); see supra note 3.

Welch was an ACCA case, and it did not speak to the guidelines. But the Supreme Court issued another decision this term that fatally undermines the Rivero panel's alternative holding too. See Montgomery v. Louisiana, 136 S. Ct. 718 (2016). Remember, the Rivero majority acknowledged that the Johnson rule was a new substantive rule of constitutional law. 797 F.3d at 989. In Montgomery, the Supreme Court stated in no uncertain terms that “courts must give retroactive effect to new substantive rules of constitutional law.” 136 S. Ct. at 728 (emphasis added). So although the Rivero panel required that for Johnson's rule to apply retroactively to the guidelines there be a third case holding that the guidelines could be void for vagueness, Montgomery reminded us that the inquiry is simpler. By the Rivero panel's own analysis, the Johnson rule was a new substantive rule of constitutional law. Under Montgomery, that means the rule must be given retroactive effect. Montgomery ends the analysis there.6 In short, we were wrong again.


Even before the Supreme Court could decide Montgomery or Welch, this Court erected yet another barrier to relief for individuals like Mr. Clayton who were sentenced under the very words the Johnson Court struck as unconstitutional. In United States v. Matchett, a panel of this Court held that the rule in Johnson did not apply to individuals sentenced under the advisory guidelines because those guidelines are not subject to the Due Process Clause's vagueness doctrine. 802 F.3d 1185, 1193-94 (11th Cir. 2015). For the reasons articulated in Judge Martin's and Judge Rosenbaum's concurrences in this case, I believe Matchett was wrongly decided. In my view, which I share with my colleagues, Matchett's holding was not grounded in the Constitution, the text of the career offender guideline, or any other solid legal foundation. Rather, the Matchett panel simply decreed that the advisory guidelines, unlike the ACCA, do not “fix punishments” and therefore are not subject to the limitations of due process. Id. at 1195. But in yet another case this term, the Supreme Court underscored that “the Guidelines are not only the starting point for most federal sentencing proceedings but also the lodestar.” Molina-Martinez v. United States, 136 S. Ct. 1338, 1346 (2016). “The Guidelines inform and instruct the district court's determination of an appropriate sentence. In the usual case, then, the systemic function of the selected Guidelines range will affect the sentence.” Id.; see also id. at 1349 (“[The guidelines] serve as the starting point for the district court's decision and anchor the court's discretion in selecting an appropriate sentence.”). By any honest reading, the guidelines fix punishments. As such, in my view, their application must comport with due process.

Matchett's reach in this Circuit is extensive. Now, no person whose advisory sentencing guidelines range was affected by the clause the Supreme Court held to be unconstitutionally standardless in Johnson may obtain relief, no matter how long he has been incarcerated or how diligently he has tried to preserve his claims. Bewilderingly, the Matchett panel erected this barrier even though the United States—the party responsible for the continued incarceration of career offenders—agreed with Mr. Matchett that the residual clause of the career offender guideline was unconstitutionally vague in light of Johnson. See Matchett, 802 F.3d at 1194.


This Court's penchant for deciding these fundamentally important issues in orders on requests for authorization to file—in the absence of any substantive, adversarial briefing—is frustratingly familiar. We have received over 1,800 requests for authorization to file a second or successive § 2255 motion since Welch was decided.7 When an inmate makes such a request, we do not receive briefing from the parties. In nearly all of these cases, we never hear from the government. And at best, we receive a skeletal description of claims from the movant. At least as troublingly, the decisions we make are almost completely insulated from review.8 I believe that in light of the limited time and resources we have to grant or deny authorization and the effective finality of our decision if we deny it, we should avoid making new substantive law in this procedural context.

Unfortunately, not all of my colleagues share my view. In fact, a panel of this Court recently extended Matchett's holding to cover individuals sentenced when the guidelines were mandatory rather than merely advisory. See In re Griffin, No. 16-12012, __ F.3d __, 2016 WL 3002292 (11th Cir. May 25, 2016). I have previously expressed my view of why this decision was deeply flawed. See In re Sapp, No. 16-13338, __ F.3d __, 2016 WL 3648334, at *2-7 (Jordan, Rosenbaum, and Jill Pryor, concurring). This is not the only time since Johnson was decided that we have taken a previous decision of our Court and extended it in the successive § 2255 motion context without any adversarial testing or opportunity for further review. See In re Williams, Nos. 16-13013, 16-13232, __ F.3d __, 2016 WL 3460899, *4 (extending the “concurrent sentence doctrine,” a rule the Supreme Court long ago said offers nothing more than “a rule of judicial convenience,” 9 to the second or successive § 2255 motion context); In re Hires, No. 16-12744, __ F.3d __, 2016 WL 3342668, *4 (11th Cir. June 15, 2016) (extending the holding in Turner v. Warden Coleman FCI (Medium), 709 F.3d 1328 (11th Cir. 2013) , that a conviction under Florida's aggravated assault statue qualifies as a violent felony to the post-Johnson context without analyzing whether it qualifies notwithstanding Johnson and other Supreme Court precedent since Turner that would bear on the issue).

Instead of blazing new trails in the second or successive § 2255 motion context, the only issue we should decide is whether, under our existing precedent, the applicant has made a prima facie showing that his sentence was based on crimes that met the ACCA's definition of “violent felony” before Johnson but no longer do.10


When it comes to Matchett, we soon may be told we are wrong again. On the last day of this year's term, the Supreme Court accepted certiorari in Beckles v. United States, No. 15-8544, 2016 WL 1029080 (U.S. June 27, 2016). Beckles is yet another Johnson case that originated in this Circuit. This time, the petitioner was sentenced as a career offender under the advisory guidelines (just like Mr. Clayton and Mr. Matchett) rather than under the ACCA. So the Supreme Court, in deciding Beckles, the Supreme Court will decide the very issue that Matchett concerns.

If we simply asked whether, on our existing precedent, the applicant has made a prima facie showing that his sentence was based on crimes that met the definition of “violent felony” before Johnson but no longer do, we undoubtedly would be granting authorization to file second or successive § 2255 motions in more cases. At least then these many individuals who may be serving unconstitutional sentences would have a shot at meaningful review, first in the district court and then in this Court on appeal (and maybe even ultimately in the Supreme Court).

I recognize that the number of requests for authorization we have received in the wake of Johnson has been extremely taxing on our Court. We have been inundated with thousands of filings in addition to our regular court work. And I understand that published orders from this Court that categorically foreclose relief to whole groups of individuals, like Matchett and Griffin, may lessen that burden on district courts, too. But such prudential concerns are not reasons to refuse to remedy constitutional violations. As judges we are not sworn to shield district courts; rather, we are sworn to uphold the Constitution and vindicate the individual rights that the Constitution protects.

If the Supreme Court decides Beckles in Mr. Clayton's favor, he may be able to file another request for authorization under § 2255. So too may the hundreds of others who have tried since Johnson, only to be turned away by this Court based upon Matchett. I hope next time around we will avoid the mistakes I have identified. And I hope that, rather than being behind the march of justice, we, as our nation's designated guardians, will be at the front.


1.   See United States v. Soto-Rivera, 811 F.3d 53 (1st Cir. 2016); United States v. Welch, __ F. App'x __, 2016 WL 536656 (2d Cir. Feb. 11, 2016); United States v. Townsend, __ Fed. Appx. __, 2015 WL 9311394 (3d Cir. Dec. 23, 2015); United States v. Frazier, 621 F. App'x 166 (4th Cir. 2015); Order, United States v. Estrada, No. 15-40264 (5th Cir. Oct. 27, 2015); United States v. Pawlak, __ F.3d. __, 2016 WL 2802723 (6th Cir. May 13, 2016); Ramirez v. United States, 799 F.3d 845 (7th Cir. 2015); United States v. Taylor, 803 F.3d 931 (8th Cir. 2015); United States v. Benavides, 617 F. App'x 790 (9th Cir. 2015); United States v. Madrid, 805 F.3d 1204 (10th Cir. 2015); Order, In re Booker, No. 16-3018 (D.C. Cir. June 10, 2016).

2.   See Beckles v. United States, No. 15-8544, 2016 WL 1029080 (U.S. June 27, 2016). In that regard, I note that our court has expanded the ruling in Matchett. While Matchett ruled in a case for which the inmate was sentenced under the advisory guidelines, our court relied on Matchett to hold that prisoners can't even make “a prima facie showing” that Johnson applies to the pre-Booker mandatory guidelines. See In re Griffin, __ F.3d __, 2016 WL 3002293 (11th Cir. May 25, 2016). Three of my colleagues have explained in detail “why [they] believe Griffin is deeply flawed and wrongly decided” even if Matchett is correct. In re Sapp, No. 16-13338-J, 2016 WL 3648334, at *3 (11th Cir. July 7, 2016) (Jordan, Rosenbaum, and Jill Pryor, J.J., concurring). I share their view. I add that Travis Beckles was sentenced after Booker, which means the Supreme Court's ruling in his case might not address the mandatory guidelines issue the Eleventh Circuit decided in Griffin.

3.   See In over 90 percent of these cases, § 4B1.1 “increased the guideline range.” Id. Also, two of the top five districts for § 4B1.1 sentences are in the Eleventh Circuit. See id. This is the latest data published by the Sentencing Commission.

4.   For example, a defendant convicted of being a felon in possession of a firearm who has two felony convictions from earlier in his life normally gets a sentencing range of 15 to 21 months. See id. § 2K2.1(a)(7). If just one of those convictions meets the definition that Johnson said was “nearly impossible to apply consistently,” 135 S. Ct. at 2560, the range becomes 41 to 51 months. See USSG § 2K2.1(a)(4)(A). If both meet that definition, it becomes 63 to 78 months. See id. § 2K2.1(a)(2). This fourfold increase is automatic even if the earlier convictions were themselves punished with as little as a year in state prison. See id. § 4B1.2(a).USSG § 2K2.1 may affect more people than the career offender guideline. The Sentencing Commission's most recent published data shows that “[i]n fiscal year 2014, there were 5,498 offenders convicted under 18 U.S.C. § 922(g), accounting for 7.2% of all offenders sentenced under the guidelines.” See “For each of the past five years, more than half of offenders convicted of violating 18 U.S.C. § 922(g) were sentenced within the [guidelines] range.” Id. And as with § 4B1.1, two of the top five districts for § 922(g) cases are in the Eleventh Circuit. See id. And that's just § 922(g). USSG § 2K2.1 is also used to calculate sentences for violations of 18 U.S.C. §§ 922(a)–(p), (r)–(w), (x)(1), 923, 924(a), (b), (e)–(i), (k)–(o), 2332g, as well as 26 U.S.C. §§ 5685, 5861(a)–(l), 5871. See USSG App. A.

5.   For drug offenders like Mr. Clayton, less than 0.57% percent of defendants sentenced without § 4B1.1 get sentences as long as the lowest end of guideline range for defendants sentenced under § 4B1.1. See

6.   The concept of an “anchor” effect makes sense. If a judge is told a sentencing range, her sentence is likely to be weighted toward that range no matter how far she might be permitted to depart from it. See Timur Kuran & Cass R. Sunstein, Availability Cascades and Risk Regulation, 51 Stan. L. Rev. 683, 705 (1999) (explaining how a number that appears early in a decision-making process “serves as a perceptual ‘anchor’ ” and distorts the ultimate decision even if the decision-maker has wide discretion); see also Stephanos Bibas & Susan Klein, The Sixth Amendment and Criminal Sentencing, 30 Cardozo L. Rev. 775, 779 (2008) (noting that the advisory federal guidelines “provide mental anchors, starting points that influence how judges think about cases and where they wind up”).

7.   Uncertainty about § 4B1.2's meaning also distorts plea bargaining in state courts, where “[p]leas account for nearly 95% of all [felony] convictions.” Padilla v. Kentucky, 559 U.S. 356, 372 & n.13, 130 S. Ct. 1473, 1485 & n.13 (2010). State defendants negotiating plea deals won't likely know how their current conviction will impact a future § 4B1.2 sentence in federal court.

8.   See, e.g., 135 S. Ct. at 2558 (“[T]his Court's repeated attempts and repeated failures to craft a principled and objective standard out of the residual clause confirm its hopeless indeterminacy.”); id. at 2559–60 (“This Court is not the only one that has had trouble making sense of the residual clause. The clause has created numerous splits among the lower federal courts, where it has proved nearly impossible to apply consistently.” (quotation omitted)); id. at 2560 (“Nine years' experience trying to derive meaning from the residual clause convinces us that we have embarked upon a failed enterprise.”); id. at 2562 (“[T]he experience of the federal courts leaves no doubt about the unavoidable uncertainty and arbitrariness of adjudication under the residual clause.”); id. at 2563 (“Decisions under the residual clause have proved to be anything but evenhanded, predictable, or consistent.”).

9.   As detailed in the next section, even though these cases were overruled by Johnson, courts in the Eleventh Circuit are required to keep applying them. See Matchett, 802 F.3d at 1195–96 (“[S]entencing courts interpreting the residual clause of the guidelines must still adhere to the reasoning of cases interpreting the nearly identical language in [ACCA].).

10.   Up until Matchett, this court also recognized that the two “residual clauses are identical” and treated them that way. United States v. Alexander, 609 F.3d 1250, 1253 (11th Cir. 2010). Most notably, we did so on the flip side of the exact issue decided in Matchett. See United States v. Chitwood, 676 F.3d 971, 978 n.3 (11th Cir. 2012) (noting that the Supreme Court's rejection of a vagueness challenge to ACCA's residual clause “appears to foreclose” a vagueness challenge to § 4B1.2's residual clause).

11.   See United States v. Booker, 543 U.S. 220, 250, 125 S. Ct. 738, 759 (2005) (remedial opinion for the Court by Breyer, J.) (“Congress' basic statutory goal—a system that diminishes sentencing disparity—depends for its success upon judicial efforts to determine, and to base punishment upon, the real conduct that underlies the crime of conviction.”). For more on the “real conduct” focus of the Sentencing Guidelines, see Stephen Breyer, The Federal Sentencing Guidelines and the Key Compromises Upon Which They Rest, 17 Hofstra L. Rev. 1, 12 (1988).

12.   Brief of the Federal Public and Community Defenders and the National Association of Federal Defenders as Amici Curiae in Support of Petitioner, at 4 n.3, Jones v. United States, No. 15-8629 (U.S. April 21, 2016). I have not verified this claim, but I see no reason to doubt it.

13.   Prisoners filing applications in our court are confined to a short application form. This form, which can be seen at, does not allow applicants to recreate for us what facts were ultimately found by the sentencing judge, or what legal decisions the judge made. Instead, it gives prisoners exactly two lines to “[s]tate concisely every ground on which you now claim that you are being held unlawfully.” And it warns: “Do not submit separate petitions, motions, briefs, arguments, etc.”

14.   This idea can be illustrated by this court's treatment of the Supreme Court's 2013 ruling in Descamps, which explained the proper method for evaluating a person's prior convictions. This is, of course, the question that comes up for people whose sentences violate Johnson. But in rushing to rule on so many of these cases in such a short period, this court has been erratic about whether and when Descamps applies in this context. The court's first published opinion on this issue held that Descamps did not apply to a Johnson claim because the sentencing judge had cited the Taylor case when imposing sentence. See In re Thomas, __ F.3d __, 2016 WL 3000325, at *2 (11th Cir. May 25, 2016). Weeks later another panel held that Descamps does apply when we can't tell which ACCA definition the sentencing judge had in mind (two ACCA definitions are just halves of one statutory subsection, so judges most often made no distinction). See In re Adams, __ F.3d __, 2016 WL 3269704, at *3 (11th Cir. June 15, 2016). Later that afternoon a third panel ruled that “Descamps cannot serve as a basis” for any Johnson claim. See In re Hires, __ F.3d __, 2744, 2016 WL 3342668, at *5 (11th Cir. June 15, 2016). Two days later, a fourth panel tried to reconcile the “tension” in these cases. Rogers, 2016 WL 3362057, at *2 n.6. That same day another panel held that we can ignore Descamps even for prisoners sentenced after Descamps. See In re Cook, No. 16-12745 (11th Cir. June 17, 2016) (unpublished). Thomas, Adams, Hires, and Rogers are all published opinions, which means they set binding precedent. All of this precedent was established in a very short time period without any of the deliberation or adversarial presentation that goes into a normal appeal, and I'm afraid it shows.

15.   There is nothing “odd” at all about giving different review to prisoners who are contesting convictions imposed by states in state courts compared to prisoners challenging sentences we ourselves imposed in federal court. Comity requires the federal government to respect and defer to the processes put in place by state governments as well as the judgments of their courts. And anyway, state courts are able to fix their own mistakes. Federal courts imposed the sentences of prisoners in federal prisons, so those prisoners must look to us to fix our mistakes. Congress has long developed different standards for people challenging sentences imposed by state courts than for those challenging sentences imposed in federal court.

16.   This problem isn't limited to Johnson claims. Our court recently denied a Johnson application based on the “concurrent sentence doctrine,” which treats an illegal sentence as “harmless” if a prisoner is serving another sentence that is just as long as the illegal one. In re Williams, __ F.3d __, 2016 WL 3460899, at *5 (11th Cir. June 24, 2016). The Supreme Court has long warned that this doctrine is nothing more than a “rule of judicial convenience.” Benton v. Maryland, 395 U.S. 784, 791, 89 S. Ct. 2056, 2061 (1969). And our court had never before “applied harmless error or the concurrent sentence doctrine in the context of an application to file a second or successive § 2255 motion.” Williams, 2016 WL 3460899, at *5. Without any briefing or advocacy on the question, Williams held that this doctrine barred a prisoner from even filing his § 2255 motion. Days later, a split panel used Williams to deny a pro se application for permission to file a § 2255 motion based on Miller v. Alabama, 567 U.S. __, 132 S. Ct. 2455 (2012). See In re Hernandez-Miranda, No. 16-12893 (11th Cir. June 28, 2016) (unpublished). Mr. Hernandez-Miranda was “a juvenile when he joined a conspiracy for which he was later sentenced to life without parole.” Id. at 13 (Martin, J., dissenting). Because of the restrictions on review of our rulings in this context, the panel's order was the beginning and end of Mr. Hernandez-Miranda's Eighth Amendment claim.

1.   If the Guidelines calculation error in Molina-Martinez that resulted in a difference of 7 months' imprisonment on the low end of the Guidelines range constitutes a “significant procedural error,” so too must an error in the application of the career-offender Guideline, which can double and sometimes even triple the otherwise-applicable Guidelines range.

1.   The ACCA enhancement applies when a person convicted of being a felon in possession of a firearm, in violation of 18 U.S.C. § 922(g), previously has been convicted of three violent felonies or serious drug offenses. The career offender enhancement applies when a person convicted of a violent felony or drug offense previously has been convicted of two such offenses. Either way, three convictions are required to impose an enhancement.

2.   Teague set forth a general principle that new rules of law should not be applied retroactively. See Summerlin, 542 U.S. at 351-52. New substantive rules, however, do apply retroactively. Id. at 352.

3.   See Bousley v. United States, 523 U.S. 614 (1998). In Bousley, the Supreme Court declined to apply the general principle that new rules are not retroactive. Id. at 619-21. Kenneth Bousley was convicted in 1990 of “using” a firearm in violation of 18 U.S.C. § 924(c)(1). Id. at 616. After the Eighth Circuit affirmed his conviction, Mr. Bousley filed for collateral relief. Id. at 617. While his appeal from the district court's denial of habeas relief was pending, the Supreme Court held in Bailey v. United States that § 924(c)(1)'s “use” prong required the government to prove “active employment of the firearm.” 516 U.S. 137, 144 (1995). Because Mr. Bousley contended that he merely possessed a firearm during his offense, he argued based on Bailey that his conduct failed to qualify under § 924(c). Bousley, 523 U.S. at 617-18. Amicus, arguing against retroactivity (because the government agreed with Mr. Bousley that he could benefit from Bailey's rule) urged the Supreme Court to apply a Teague bar to Mr. Bousley's claim. The Supreme Court declined, holding that Bailey necessarily was available to Mr. Bousley on collateral review because Bailey announced a new substantive rule. Id. at 620-21; see United States v. Peter, 310 F.3d 709, 711 (11th Cir. 2002) (citing Bousley for the proposition that “[d]ecisions of the Supreme Court construing substantive federal criminal statutes must be given retroactive effect”). As I explained in dissent in Rivero, Congress was in the process of amending § 924(c)(1) to recriminalize the conduct the Court in Bailey held to fall outside the statute's scope. Rivero, 797 F.3d at 999 (Jill Pryor, J., dissenting). Ultimately, the so-called “Bailey Fix Act” passed, but this had no bearing on the Supreme Court's retroactivity decision in Bousley. Id. at 999-1000.

4.   As Judge Martin explains in her concurrence, we were the only Circuit in the nation to have routinely refused to hold in abeyance inmates' applications pending the Welch retroactivity decision.

5.   And, of course, had Mr. Welch sought permission from us to file a second § 2255 motion based on Johnson, we would have denied him based on Rivero.

6.   The Rivero panel's observation that the guidelines must also be subject to a vagueness challenge in order for Mr. Rivero to obtain relief is not wrong (although I disagree with Rivero's suggestion, later born out in Matchett, that the guidelines cannot be unconstitutionally vague). Rather, the observation was wrongly imported into the retroactivity analysis. That inquiry belongs instead in an examination of whether any particular inmate has a meritorious Johnson claim. There is no precedential support for the proposition that these two inquiries somehow are related.

7.   By my rough calculation, approximately one third of inmates making such requests were seeking relief from their guidelines-based sentences.

8.   See 28 U.S.C. § 2244(b)(3)(E) (“The grant or denial of an authorization by a court of appeals to file a second or successive application shall not be appealable and shall not be the subject of a petition for rehearing or for a writ of certiorari.”).

9.   Benton v. Maryland, 395 U.S. 784, 789-91 (1969).

10.   Cf. In re Leonard, Nos. 16-13528, 16-13804, 16-13857, slip op. at 29-30 & n.11 (11th Cir. July 13, 2016) (Martin, J., concurring) (explaining that the question in the context of a request for authorization to file a successive § 2255 motion “should simply be whether [an inmate's] sentence was based on crimes that met ACCA's ‘violent felony’ definition before Johnson but no longer do,” and noting that the answer “should be ‘no’ only if a sentence clearly was based on ‘serious drug offenses' or crimes that we have held are ‘violent felonies' after Johnson” in light of facts the sentencing court found).

Copied to clipboard