Coleman v. United States

LEE, CIRCUIT JUDGE.

ARGUED FEBRUARY 15, 2023

Appeal from the United States District Court for the Central District of Illinois. No. 4:17-cv-4270 - Sara Darrow, Chief Judge.

Before EASTERBROOK, WOOD, and LEE, Circuit Judges.

LEE, CIRCUIT JUDGE.

In 2014, Frederick Coleman was sentenced to life imprisonment for conspiring to distribute crack cocaine. The district judge based the sentence on 21 U.S.C. § 841(b)(1)(A), which, at the time, mandated a life sentence for a defendant who previously had committed two or more felony drug offenses. After we denied his direct appeal, Coleman filed a pro se motion under 28 U.S.C. § 2255 to vacate his sentence, arguing that his defense counsel had provided ineffective assistance by not informing him of the potential life sentence. Later, he filed a motion to amend his pleading, expanding on his allegations, but, by that time, the limitations period had run. After ruling against Coleman on his original claim, the district court denied the motion to amend, finding that the amendment did not "relate back" to his initial pleading. Because the district court abused its discretion in reaching that conclusion, we reverse and remand for further proceedings as to Coleman's amended ineffective assistance of counsel claim.

I. BACKGROUND

In 2013, Coleman was convicted of conspiring to distribute crack cocaine. See 21 U.S.C. §§ 841(a)(1), (b)(1)(A), 846. Pursuant to the Controlled Substances Act in effect at that time, Coleman was sentenced to the statutorily mandated term of life imprisonment based on having at least two prior convictions for a "felony drug offense." Id. § 841(b)(1)(A) (2012). We affirmed his sentence on direct appeal. United States v. Brown, 822 F.3d 966, 976 (7th Cir. 2016).

The First Step Act of 2018, enacted after Coleman's sentencing, reduced the mandatory minimum sentences from life to 25 years' imprisonment for a defendant having two or more prior convictions for a "serious drug felony" or "serious violent felony." Pub. L. 115-391, § 401(a)(2)(A)(ii), 132 Stat. 5194, 5220 (2018).

Coleman then filed a timely pro se motion to vacate his sentence pursuant to 28 U.S.C. § 2255. Coleman's § 2255 motion asserted, among other things, that his appointed counsel, Anthony Vaupel, was ineffective for having failed to inform him of the government's pretrial 21 U.S.C. § 851 Notice of Enhancement. This notice informed Coleman that, should he be found guilty at trial, the government would seek to enhance his sentence to life imprisonment based on his prior Illinois cocaine-related convictions, which, it believed, qualified as "felony drug offenses" under 21 U.S.C. § 841(b)(1)(A). According to Coleman, had Vaupel shown him this notice, he never would have agreed to go to trial.

To inquire into these allegations, the district court first entered an order that Coleman had waived the attorney-client privilege as to his communications with Vaupel regarding "the government filing a section 851 notice and [Coleman]'s potential life sentence." It also ordered Vaupel to submit an affidavit addressing Coleman's claim.

In his affidavit, Vaupel stated that he had informed Coleman on more than one occasion that he faced, and would receive, a mandatory life sentence if found guilty. Vaupel also attested that he had told Coleman that the mandatory life sentence was due to his prior convictions, and that despite the government's willingness to enter a proffer agreement with Coleman, it was unwilling to waive the § 851 enhancement.

Before the district court ruled on Coleman's § 2255 motion, Coleman moved, again pro se, to amend it pursuant to Federal Rule of Civil Procedure 15(c). His amendment contained more detailed allegations and argued specifically that Vaupel was ineffective by failing to object to the § 851 notice on the grounds that, under the categorical approach espoused in Taylor v. United States, 495 U.S. 575 (1990), Coleman's prior Illinois cocaine convictions did not qualify as "felony drug offenses" under § 841(b)(1)(A) given that Illinois defined "cocaine" more broadly than federal law.

The district court denied Coleman's initial § 2255 motion, as well as his motion to amend. Without reaching the merits of the amended claim, the court determined that Coleman's motion to amend did not relate back to his initial pleading because "the claims rest[ed] on distinct types of attorney misfeasance and [we]re supported by different facts." Coleman v. United States, No. 4:17-cv-04270-SLD-JEH, 2022 WL 673702, at *8 (C.D. Ill. Mar. 7, 2022). And because Coleman's motion to amend was filed more than a year after his conviction became final, the district court found it untimely. Id. at *9; see 28 U.S.C. § 2255(f)(1) (providing a one-year statute of limitations for collateral review). Then, finding that reasonable jurists may disagree on this point, the district court granted Coleman a certificate of appealability on his claim. Id. Coleman appealed, and we recruited counsel.

Coleman's § 2255 motion alleged Vaupel was constitutionally ineffective in three other ways. The district court denied relief, as well as a certificate of appealability, on each of those grounds. See Coleman, 2022 WL 673702, at *5-*7, *9. Our review is therefore limited to Coleman's claim relating to Vaupel's treatment of the government's § 851 notice.

We thank Coleman's counsel, Marc Krickbaum, as well as his colleague Katherine Stallings Bailey of Winston & Strawn LLP, for their advocacy on Coleman's behalf.

II. DISCUSSION

Section 2255 provides a federal prisoner with the means to collaterally attack the propriety of his sentence. Relief under this statute is an "extraordinary remedy," and therefore only available in limited circumstances. Almonacid v. United States, 476 F.3d 518, 521 (7th Cir. 2007). One such circumstance is where defense counsel's representation is so ineffective that it violates the defendant's Sixth Amendment rights. U.S. Const. amend. VI; Strickland v. Washington, 466 U.S. 668, 686 (1984). Coleman argues that his appointed counsel was constitutionally ineffective by failing to object to the § 851 notice. As he sees it, his attorney should have argued that Coleman's prior convictions did not qualify as "felony drug offenses" that could be used to enhance his sentence to life imprisonment under § 841. See generally United States v. Ruth, 966 F.3d 642 (7th Cir. 2020).

Coleman's amended motion was filed prior to our holding in Ruth that an Illinois conviction for possession with intent to distribute cocaine was not a "felony drug offense" under the federal sentencing laws because Illinois's definition of "cocaine" was broader than the federal definition. 966 F.3d at 650.

Before we can reach the merits of Coleman's ineffective assistance of counsel claim, however, we must resolve two preliminary questions. First, we must determine whether appellate review of a district court's disposition of a pleading amendment under the "relation-back" provision of Federal Rule of Civil Procedure 15(c) is de novo or for an abuse of discretion. We must then apply that standard to determine the propriety of the district court's conclusion that Coleman's amendment did not relate back to the filing of his original motion.

A. Standard of Review

When a party seeks to amend its pleading, district courts "should freely give leave when justice so requires." Fed.R.Civ.P. 15(a)(2). "The Supreme Court has interpreted this rule to require a district court to allow amendment unless there is a good reason-futility, undue delay, undue prejudice, or bad faith-for denying leave to amend." Life Plans, Inc. v. Sec. Life of Denver Ins. Co., 800 F.3d 343, 357-58 (7th Cir. 2015) (citing Foman v. Davis, 371 U.S. 178, 182 (1962)). This "liberal standard," id. at 357, is limited, however, by any applicable statute of limitations. See Mayle v. Felix, 545 U.S. 644, 655 (2005).

Where a party seeks to amend a pleading after the statute of limitations has run, he must turn to the relation-back provision of Rule 15(c), which allows such amendment if it "asserts a claim or defense that arose out of the conduct, transaction, or occurrence set out-or attempted to be set out-in the original pleading." Fed.R.Civ.P. 15(c)(1)(B).

It is well-settled that review of a district court's disposition of a motion to amend under Rule 15(a) is for an abuse of dis-cretion. See, e.g., Foman, 371 U.S. at 182 ("Of course, the grant or denial of an opportunity to amend [under Rule 15(a)] is within the discretion of the District Court[.]"); Runnion ex rel. Runnion v. Girl Scouts of Greater Chi. &Nw. Ind., 786 F.3d 510, 524 (7th Cir. 2015) (applying deferential review but noting that, where a district court's denial is futility-based, our review "includes de novo review of the legal basis for the futility"). But we have been less clear on the appropriate standard of review when a district court decides whether a proposed amendment "relates back" to a pleading under Rule 15(c). Indeed, over six years ago, we made a mental note to "clarify the correct standard in a future case when the matter is properly before the court." Mulvania v. Sheriff of Rock Island Cnty., No. 16-1711, 2017 WL 2726577, at *2 (7th Cir. May 16, 2017) (denying petition for rehearing en banc). Compare Arreola v. Godinez, 546 F.3d 788, 796 (7th Cir. 2008) (applying abuse of discretion review), O'Brien v. Ind. Dep't of Corr. ex rel. Turner, 495 F.3d 505, 507 (7th Cir. 2007) (same), and Bibbs v. Sheriff of Cook Cnty., 618 Fed.Appx. 847, 851 (7th Cir. 2015) (same), with Delgado-Brunet v. Clark, 93 F.3d 339, 342 (7th Cir. 1996) (applying de novo review). With this issue squarely before us, we make clear that a district court's disposition of a motion to amend a pleading that turns on the "relation-back" provision of Rule 15(c) is reviewed for an abuse of discretion.

The same is true for amendments made under Rule 15(b), which governs amendments during and after trial. See Aldridge v. Forest River, Inc., 635 F.3d 870, 875 (7th Cir. 2011).

We also observe a split of authority in our sister circuits on this question. Compare United States v. Ciampi, 419 F.3d 20, 23 (1st Cir. 2005) (applying abuse of discretion review in § 2255 context), United States v. Pittman, 209 F.3d 314, 316 (4th Cir. 2000) (same), Mandacina v. United States, 328 F.3d 995, 1000 (8th Cir. 2003) (same), Davenport v. United States, 217 F.3d 1341, 1343 n.4 (11th Cir. 2000) (same), with Slayton v. Am. Exp. Co., 460 F.3d 215, 227-28 (2d Cir. 2006) (applying de novo review), United States v. Santarelli, 929 F.3d 95, 100 (3d Cir. 2019) (same in § 2255 context), Durand v. Hanover Ins. Grp., Inc., 806 F.3d 367, 374 (6th Cir. 2015) (applying de novo review), United States v. Marulanda, 226 Fed.Appx. 709, 710 (9th Cir. 2007) (same in § 2255 context), United States v. Roe, 913 F.3d 1285, 1298 (10th Cir. 2019) (same), and Anza Tech., Inc. v. Mushkin, Inc., 934 F.3d 1359, 1367 (Fed. Cir. 2019) (applying de novo review). The Fifth Circuit declined to resolve this issue in the § 2255 context in United States v. Alaniz, 5 F.4th 632 (5th Cir. 2021), but noted that "[its] cases tend to apply the abuse of discretion standard." Id. at 635 n.2.

As an initial matter, the fact that this issue comes to us via a § 2255 proceeding is of no consequence. Habeas corpus cases under § 2255 are civil cases generally governed by the Federal Rules of Civil Procedure. Conley v. United States, 5 F.4th 781, 794 (7th Cir. 2021) (citing Banister v. Davis, 140 S.Ct. 1698, 1705 (2020)); see Beason v. Marske, 926 F.3d 932, 938 (7th Cir. 2019) (discussing application of Rule 15's relation-back provision to a § 2255 motion); Fed.R.Civ.P. 81(a)(4). See generally Mayle, 545 U.S. at 655 (applying Rule 15(c) to § 2244 habeas corpus proceedings). Accordingly, our holding today applies generally to the relation-back provision of Rule 15(c); it is not limited to habeas proceedings.

Determining whether a proposed amendment asserts a claim arising from the same conduct, transaction, or occurrence involves more than a facial comparison of the original and amended pleadings. In making this assessment, district courts also must consider whether, under the particular circumstances of the litigation, the opposing party has been put on notice as to the claim raised in the proposed amendment. See Supreme Auto Transp., LLC v. Arcelor Mittal USA, Inc., 902 F.3d 735, 741 (7th Cir. 2018) ("The central inquiry under Rule 15(c) is whether the original complaint 'gave the defendant enough notice of the nature and scope of the plaintiff's claim that he shouldn't have been surprised by the amplification of the allegations of the original complaint in the amended one.'") (quoting Santamarina v. Sears, Roebuck &Co., 466 F.3d 570, 573 (7th Cir. 2006)); see also Krupski v. Costa Cro-ciere S. p. A., 560 U.S. 538, 550 (2010) (explaining that the purpose of relation back is "to balance the interests of the defendant protected by the statute of limitations with the preference expressed in the Federal Rules of Civil Procedure in general, and Rule 15 in particular, for resolving disputes on their merits").

In some cases, it may be clear from the face of the proposed amendment that the original pleading put the defendant on notice of a later-added claim. See, e.g., Tiller v. Atl. Coast Line R.R. Co., 323 U.S. 574, 581 (1945) (finding that a new legal theory related back where defendant was on notice "from the beginning that [plaintiff] was trying to enforce a claim against it" based on the same facts alleged in the complaint). In others, it may be clear from the face of the proposed amendment that the opposing party was not on notice of such a claim. See, e.g., Mayle, 545 U.S. at 648-49 (no relation back where petitioner initially claimed that admission of videotaped testimony from a prosecution witness violated the Sixth Amendment's Confrontation Clause, but his amended claim asserted that the admission of his own pretrial testimony violated the Fifth Amendment right against self-incrimination).

But there will be those cases in the middle, where the district court's familiarity with and proximity to the parties and proceedings will help it decide whether a defendant would (or reasonably should) be "surprised by the amplification" of the plaintiff's originally asserted claims. Supreme Auto, 902 F.3d at 741. Such decisions involving questions that "immerse courts in case-specific factual issues" we typically review under the abuse-of-discretion standard. U.S. Bank Nat'l Ass'n ex rel. CWCapital Asset Mgmt. LLC v. Vill. at Lakeridge, LLC, 138 S.Ct. 960, 967 (2018).

We find our analysis in Supreme Auto instructive. There, the district court determined that the plaintiffs' amended claims did not relate back to the initial complaint because the transactions at issue "were wholly distinct" from those of the amended complaint. 902 F.3d at 741. We agreed. But we based our conclusion on more than a facial comparison of the two pleadings; we also looked to the parties' representations of the scope of the dispute throughout the course of litigation. Id. at 741-42. Viewing the dispute through this lens, we concluded that at no time prior to the plaintiffs' proposed amendment could the defendant have reasonably expected to defend against the "wholly distinct" newly alleged conduct. See id. ("What's more, Supreme Auto gave no indication during the first seven years of litigation that its suit included any more products other than [those alleged in the initial complaint]."). Although we did not clarify in Supreme Auto the standard under which we reviewed the district court's decision, when reviewing mixed questions of fact and law, the standard turns on "which 'judicial actor is better positioned' to make the decision." U.S. Bank, 138 S.Ct. at 967 (quoting Miller v. Fenton, 474 U.S. 104, 114 (1985)). Because the district court in Supreme Auto had been steeped in the circumstances of that case for ten years before it reached us, we had no doubt the district court was better positioned to determine whether the defendant was or should have been on notice of the plaintiffs' amended claims.

In sum, because the application of Rule 15(c)'s "relation-back" provision requires a district court to consider whether a nonmoving party would have been on notice of the moving party's amended claims based on all the relevant circumstances of a particular case, our review of a district court's application of that rule is for an abuse of discretion.

We are not persuaded by Coleman's reliance on Krupski to argue that the "mandatory nature" of the Rule 15(c) inquiry signals de novo review. 560 U.S. at 553. Krupski merely holds that, once a district court finds that all the requirements of Rule 15(c) are satisfied, the district court has no discretion to deny the amendment. See id. It does not hold, as Coleman suggests, that the district court lacks discretion to determine whether those requirements are satisfied in the first place.

B. Relation Back

Even under the deferential standard we have settled on today, it is not difficult for us to conclude that, in this case, the district court abused its discretion in denying Coleman's motion to amend.

As already noted, "amendments relate back to the date of the original pleading when the claim asserted in the amended plea 'arose out of the conduct, transaction, or occurrence set forth or attempted to be set forth in the original pleading.'" Mayle, 545 U.S. at 656 (quoting Fed.R.Civ.P. 15(c)(2)). Even where an amendment "invoke[s] a legal theory not suggested by the original complaint and relie[s] on facts not originally asserted," relation back is in order "[s]o long as the original and amended petitions state claims that are tied to a common core of operative facts." Id. at 659, 664. Indeed, "[e]ven 'significant' changes to a complaint ... can relate back so long as the defendant had fair notice of the substance of the new allegations from the outset." Supreme Auto, 902 F.3d at 741.

Construing Coleman's original pro se § 2255 petition liberally, as we must, see Erickson v. Pardus, 551 U.S. 89, 94 (2007), we conclude that the government was reasonably on notice of the claims contained within Coleman's amended petition. Those claims are substantively similar in time and type to those asserted in his original motion: his attorney's failure to provide legal advice as to the contents and impact of the § 851 notice.

Coleman's original § 2255 motion advanced an ineffective assistance of counsel claim based on his attorney's failure to "inform" him of the government's § 851 notice that the government would use Coleman's prior convictions to enhance his sentence to life imprisonment if he lost at trial. His amended petition sought to expand on this claim to allege that his attorney failed to adequately "research, investigate, or inform" Coleman whether those prior convictions could be used to support a sentencing enhancement under § 851 at all. Comparing these two claims, the district court acknowledged that, "broadly speaking," both the original and amended claims challenged his attorney's conduct as to the § 851 notice. Coleman, 2022 WL 673702, at *8. Nevertheless, the district court concluded that the two claims were "supported by different facts," and therefore the amendment did not relate back. Id. In the district court's view, "[w]hile the original petition contained a claim centered on an alleged failure to inform Coleman of the § 851 notice prior to Coleman's decision to go to trial, the new claims center on substantive failures of counsel throughout case preparation, client counseling, plea negotiations, and at sentencing." Id.

This is too crabbed a view of Coleman's claims, particularly given his pro se status at the time. The fact that an amended claim may involve some different facts than those originally alleged does not necessarily mean that that claim is not tied to the original claim via a "common core of operative facts." Mayle, 545 U.S. at 664 (emphasis added). Coleman's success on either of his theories would require the district court to inquire into his attorney's treatment of the § 851 notice and the adequacy of advice he provided to Coleman as to the impact the notice may have had on the case. This commonality is enough to unite the claims.

We also fail to see how the government could have reasonably been surprised by Coleman's amended claim. The government no doubt encounters countless pro se filings such as this and knows that they must be construed liberally. Coleman's allegations that his attorney had inadequately counseled him regarding the § 851 notice and the convictions listed therein were enough to put the government on notice that Coleman might challenge the use of those convictions for the enhancement. Furthermore, the district court alerted the parties that, by filing his original motion, Coleman had waived the attorney-client privilege as to "any and all communications between [Coleman] and Mr. Vaupel relating to ... the government filing a section 851 notice and [Coleman]'s potential life sentence." Text Order of 1/5/18 (emphasis added). This order was sufficiently broad to uncover the facts underlying Coleman's amended claims. Accordingly, on balance, any prejudice the government may suffer by allowing the amended complaint to relate back to the original filing is outweighed by "the preference expressed in the Federal Rules of Civil Procedure in general, and Rule 15 in particular, for resolving disputes on their merits." Krupski, 560 U.S. at 550.

We therefore find that the district court abused its discretion in determining that Coleman's amended § 2255 pleading did not relate back to his original motion. Finding that Coleman's amendment was timely under the provisions of Rule 15(c), we turn to the merits of his ineffective assistance of counsel claim.

C. Ineffective Assistance of Counsel

Under Strickland, Coleman is not entitled to collateral relief on his claim unless he can establish that (1) his attorney's "performance was deficient" and (2) "the deficient performance prejudiced the defense." 466 U.S. at 687.

1. Performance

We begin by considering whether defense counsel's failure to challenge the use of Coleman's prior Illinois cocaine convictions in the § 851 notice to enhance Coleman's sentence was constitutionally deficient. As we have noted elsewhere, "[s]uch an argument, novel then, would succeed today." Harris v. United States, 13 F.4th 623, 625 (7th Cir. 2021); see Ruth, 966 F.3d at 650 (applying the categorical approach established in Taylor and holding that Illinois cocaine convictions are not "felony drug offenses" that can be used to enhance a sentence under § 841(b)(1)(C)). But, to succeed on this first prong, Coleman must show that his counsel's performance "fell below an objective standard of reasonableness." Strickland, 466 U.S. at 688.

We have long recognized the general principle that "[t]he Sixth Amendment does not require counsel to forecast changes or advances in the law." Lilly v. Gilmore, 988 F.2d 783, 786 (7th Cir. 1993); Strickland, 466 U.S. at 689. "Rather, the reasonableness of counsel's performance must be assessed 'in the context of the law' at the time." Harris, 13 F.4th at 629 (quoting Lilly, 988 F.2d at 786). Nevertheless, "there are some circumstances where [defense counsel] may be obliged to make, or at least to evaluate, an argument that is sufficiently foreshadowed in existing case law." Bridges v. United States, 991 F.3d 793, 804 (7th Cir. 2021) (citations omitted).

In Harris, we determined that defense counsel should have known about a possible categorical challenge to the use of Illinois cocaine convictions as predicate offenses-the same that succeeded in Ruth-at least as early as 2017. 13 F.4th at 629 ("By 2017, ... the categorical approach was well-established.") (citing Bridges, 991 F.3d at 803). And the groundwork for such an argument was, at the very least, foreshadowed by numerous decisions issued before 2014 (the year Coleman was sentenced) that applied the categorical approach to predicate offenses in other contexts. See, e.g., Descamps v. United States, 570 U.S. 254, 264-65 (2013) (concluding that a California burglary conviction could not serve as a predicate offense under the Armed Career Criminal Act ("ACCA") because the state statute was broader than generic burglary); United States v. Tucker, 703 F.3d 205, 209 (3d Cir. 2012) (applying categorical approach of Taylor to determine if prior convictions qualify as "serious drug offenses" under ACCA); United States v. Bynum, 669 F.3d 880, 885 (8th Cir. 2012) (same); Ruiz-Vidal v. Gonzales, 473 F.3d 1072, 1078 (9th Cir. 2007), abrogated on other grounds by Kwong v. Holder, 671 F.3d 872 (9th Cir. 2011) (holding that California's definition of "controlled substance" was broader than the federal definition because California "punishe[d] the possession of optical and geometrical isomers" of methamphetamine while federal law "punishe[d] the possession of optical isomers alone"); Samas v. United States, No. 3:10-CV-422 (JCH), 2011 WL 221866, at *4 (D. Conn. Jan. 20, 2011) (concluding that § 2255 petitioner's state conviction was not a "felony drug offense" because two analogs of fentanyl criminalized in the Connecticut narcotics statute were not "narcotic drug[s]" as defined by federal law). These cases demonstrate that, although the categorical approach had not yet been applied to prior convictions like Coleman's by 2014, the framework for making such a challenge had been established.

What is more, in White v. United States, 8 F.4th 547, 557 (7th Cir. 2021), we determined (albeit under different circumstances) that "a challenge to cocaine delivery predicate offenses was neither novel .. nor foreclosed" in as early as 2013. And we recognized that the "basis and authority" for Ruth had been established as early as 1990, when "the Supreme Court first laid out the categorical approach in [Taylor] and when the relevant portions of the Illinois and federal drug statutes had taken their current form." White, 8 F.4th at 556. Given that there was no adverse precedent foreclosing the viability of such a challenge in 2013, we held that the petitioner had procedurally defaulted his argument by failing to timely raise it in his direct appeal. See id. Similarly, here, we are unpersuaded by the government's argument that application of the categorical approach to cocaine isomers was too novel in 2014 to have been recognized by competent defense counsel. As we have "repeatedly" made clear, "comparing statutory definitions is part of competent representation," and application of the categorical approach to drug isomers is "not complex." Harris, 13 F.4th at 630.

White concerned a collateral challenge to the constitutionality of the petitioner's sentence. See id. at 551. It did not address whether defense counsel was constitutionally ineffective for failing to raise such a challenge.

Indeed, in Harris, "the government admit[ted that] counsel only had to compare the plain language of the statutes" to determine if the predicate Indiana cocaine offenses were "felony drug offenses." 13 F.4th at 630.

Accordingly, we hold that it would have been objectively unreasonable for Coleman's defense counsel to have not even considered a categorical challenge to the government's reliance on prior Illinois cocaine convictions to enhance Coleman's sentence. This is particularly true given that an enhanced sentence here would have resulted in a mandatory life sentence. Of course, if counsel did consider the argument but had credible strategic reasons for not raising it, that would be a different question. See id. at 631. But because the claim alleged in Coleman's amended § 2255 motion, if proven, would entitle him to relief, we find that an evidentiary hearing on this issue is necessary. See Kafo v. United States, 467 F.3d 1063, 1067 (7th Cir. 2006). On remand, the district court should determine whether Vaupel considered the possibility of a categorical challenge to Coleman's predicate offenses. And, if he did consider it, what reasons he had for not raising it.

2. Prejudice

To succeed on his claim, Coleman also must establish that he was prejudiced by defense counsel's purportedly deficient performance. "[P]rejudice occurs when there is a 'reasonable probability' that 'the end result of the criminal process would have been ... a sentence of less prison time.'" Harris, 13 F.4th at 629 (quoting Missouri v. Frye, 566 U.S. 134, 147 (2012)). Coleman has met his burden.

We have held that, in cases where a court relies on an incorrectly calculated Guidelines range to sentence a defendant, a reasonable probability of prejudice is presumed absent unusual circumstances. Bridges, 991 F.3d at 808-09 (citing Molina-Martinez v. United States, 578 U.S. 189, 200-01 (2016)); see United States v. Wylie, 991 F.3d 861, 864 (7th Cir. 2021) (same); see also United States v. Coby, 65 F.4th 707, 713-14 (4th Cir. 2023) (same). This principle comes from Molina-Martinez, where the Supreme Court observed that "a defendant who has shown that the district court mistakenly deemed applicable an incorrect, higher Guidelines range has demonstrated a reasonable probability of a different outcome." 578 U.S. at 200. But the Court was cautious, observing that "[t]here may be instances when, despite [the error], a reasonable probability of prejudice does not exist," such as when the record shows that the district court would have given the same sentence in any event. 578 U.S. at 200. Only in these "unusual circumstances," the Court held, was the defendant "required to show more" to meet his burden. Id. at 201. We see no reason why a court's purportedly erroneous application of a statutorily mandated minimum sentence, where it is stripped of all sentencing discretion, should be treated differently.

The allegations of prejudice in Coleman's § 2255 petition, which went unchallenged both in the district court and on appeal, are sufficient to support application of the presumption here. As Coleman observes, absent the district court's application of the § 851 enhancement, he would not have been subject to a mandatory life sentence and the district court would have had discretion to sentence him after full and fair consideration of the sentencing factors of 18 U.S.C. § 3553(a). To be sure, as Coleman acknowledges, at the sentencing hearing, the district court noted that, even if it had the discretion to impose a lower sentence, it would have "seriously con-sider[ed]" imposing life imprisonment. But therein lies the rub-the district court had no discretion whatsoever. Even if it would have considered a life sentence (in the counterfactual scenario), the record falls short of indicating that this is what the court actually would have done. Given the significant stakes of such a sentence, combined with the district court's lack of sentencing discretion, we are comfortable concluding that Coleman sufficiently alleged that he was prejudiced by his counsel's purportedly deficient performance.

Indeed, the entirety of the government's position, here and below, was that the argument was too novel in 2014 to succeed-an argument we have already rejected. A bald speculation that an argument would have been unsuccessful does not preclude a finding of prejudice.

III. CONCLUSION

For these reasons, we REVERSE the district court's determination that Coleman's amended § 2255 motion did not relate back to the original motion and REMAND the case to the district court to hold an evidentiary hearing consistent with this opinion.

EASTERBROOK, Circuit Judge, dissenting. I agree with my colleagues about the standard of appellate review and with their conclusion that the district court should have entertained the petition as amended. But I do not think that a remand is necessary.

Coleman was sentenced in 2014, long before United States v. Ruth, 966 F.3d 642 (7th Cir. 2020), held that the way Illinois law defines "cocaine" means that cocaine convictions in Illinois do not count toward federal recidivist sentencing. (Illinois defines as cocaine the products of coca leaves, including optical, positional, and geometric isomers; federal law mentions only optical and geometric isomers.) Not until 2018 did any court of appeals reach the conclusion that Ruth adopted in 2020. See Lorenzo v. Sessions, 902 F.3d 930, 935-36 (9th Cir. 2018). As of 2014 neither any court of appeals nor any federal district court had held that a state-law reference to positional isomers disqualifies a cocaine conviction for federal purposes. The tools that led to Ruth existed-the "categorical approach" of Taylor v. United States, 495 U.S. 575 (1990), and its successors, plus the texts of the state and federal statutes-but tools exist to make thousands of legal inquiries and arguments. Most inquiries come up dry, and most novel arguments flop. That's why lawyers usually limit their research to lines that have support in judicial decisions. In 2014 an argument about positional isomers had none.

Ruth's novelty is a reason why we held it non-retroactive. See White v. United States, 8 F.4th 547, 556-57 (7th Cir. 2021). And in Harris v. United States, 13 F.4th 623 (7th Cir. 2021), we rejected an effort to make Ruth retroactive indirectly, by treating as ineffective assistance a lawyer's failure to make a Ruthlike argument before Ruth was issued. Harris observes that counsel are not expected to be prescient and that legal work should not be condemned with the benefit of hindsight.

Yet that is exactly what my colleagues do today. They observe (as Ruth, White, and Harris all state) that the tools to make a Ruth-like argument existed in 2014 and hold that counsel therefore should have at least done the research to find the ingredients of such an argument and contemplated the possibility of advancing it. This is hard to reconcile with Harris and harder to reconcile with the principles established in Strickland v. Washington, 466 U.S. 668 (1984), the genesis of modern ineffective-assistance law.

"The proper measure of attorney performance remains simply reasonableness under prevailing professional norms." Strickland, 466 U.S. at 688. Did "professional norms" in 2014 require lawyers to investigate a potential legal argument with zero support in decided cases? I don't think so. Attorneys representing criminal defendants have dozens of potentially important tasks, and like physicians they must engage in triage. The important issues, those with the largest potential returns, must be seen to-and then counsel must move to the defense of other clients. In determining what is most likely to be productive, a reasonable attorney uses as cues what has been productive. The number of potential lines of investigation and argument that have never succeeded in any court is large, as is the cost of investigating all of them, but the anticipated return is small. Sometimes a novel argument will prevail-that's what happened in Ruth-but only with the aid of hindsight can a court say that it was so likely to prevail that any reasonable lawyer would have pursued it.

Judicial scrutiny of counsel's performance must be highly deferential. It is all too tempting for a

defendant to second-guess counsel's assistance after conviction or adverse sentence, and it is all too easy for a court, examining counsel's defense after it has proved unsuccessful, to conclude that a particular act or omission of counsel was unreasonable. A fair assessment of attorney performance requires that every effort be made to eliminate the distorting effects of hindsight, to reconstruct the circumstances of counsel's challenged conduct, and to evaluate the conduct from counsel's perspective at the time.

Strickland, 466 U.S. at 689 (internal citation omitted). Only through hindsight could a court say that, in 2014, every reasonable lawyer would have investigated the possibility that a state law's unusual reference to positional isomers of cocaine would eliminate the use of a cocaine conviction in a federal recidivist prosecution.

One may be tempted to reply that Taylor itself is support for the sort of argument that prevailed in Ruth. In a general sense, that's true. But only in a general sense. If we define support at a high level of generality, then every lawyer must investigate every argument, for every argument has "support" in a general sense. Taylor is the basis of hundreds if not thousands of arguments that one or another state law is a mismatch for one or another federal statute. (Just look at the number of decisions that the Supreme Court and the courts of appeals have rendered since 1990 applying the "categorical" and "modified categorical" approaches. A search by computer turns up 197 for the Seventh Circuit alone, and that search was limited to decisions that cite Taylor rather than one of its successors.) What a lawyer needs to know is whether a particular "categorical" argument has reasonable prospects- and in 2014 competent lawyers would not have rated the positional-isomer argument high on that list.

This discussion of "support" is parallel to the question whether a particular legal rule has been clearly established. The Justices say often that, to overcome qualified immunity, the rule must be established concretely, as applied to a situation, rather than at a high level of generality. See, e.g., White v. Pauly, 580 U.S. 73 (2017); Mullenix v. Luna, 577 U.S. 7 (2015). The same proposition holds for the law of ineffective assistance. Taylor supplies support at a high level of generality for many inquiries into whether a state law is a match for a particular federal law, but it takes a decision such as Ruth to apply this principle so concretely that all reasonable lawyers must try to take advantage.

What fraction of criminal defense lawyers in 2014 investigated the possibility that a state statute's mention of positional isomers would prevent use of a state conviction in recidivist sentencing? I'll hazard a guess that less than 1% did so (for if even 1% made such an argument, starting soon after Taylor, the holding of Ruth would have come decades earlier). Can it really be that 99% of all criminal defense lawyers are in the bottom 10% of the profession? If "prevailing professional norms" govern, then a lawyer who does or omits what almost every other lawyer at the time does or omits satisfies the constitutional standard. That's what Strickland tells us. Only the distorting effect of hindsight could lead to a contrary conclusion.