PRECEDENTIAL

UNITED STATES COURT OF APPEALS FOR THE THIRD CIRCUIT

No. 19-1631

UNITED STATES OF AMERICA, v. DAVID PAYO, Appellant

Appeal from the United States District Court for the Western District of Pennsylvania (District Court No. 2:17-cr-00211-001) District Judge: Honorable Mark R. Hornak

Argued on January 22, 2025

Before: HARDIMAN, McKEE, and AMBRO, Circuit Judges (Opinion filed: April 28, 2025) Laura S. Irwin Matthew S. McHale [ARGUED] OFFICE OF UNITED STATES ATTORNEY 700 Grant Street Suite 4000 Pittsburgh, PA 15219 Counsel for Appellee

Renee Pietropaolo [ARGUED] OFFICE OF FEDERAL PUBLIC DEFENDER 1001 Liberty Avenue Suite 1500 Pittsburgh, PA 15222 Counsel for Appellant

___________

OPINION OF THE COURT ___________

AMBRO, Circuit Judge David Payo pled guilty to committing two robberies in 2017. The District Court applied a career-offender enhance- ment to his sentence based on three of his previous robbery convictions: one federal conviction from 2001, one Pennsylva- nia conviction from 2008, and one Pennsylvania conviction from 2010. Payo argued that the enhancement was unwar- ranted because the 2008 and 2010 convictions did not involve crimes of violence. The District Court disagreed, holding that both did—the former based on a state-court docket sheet the Government had not produced, the latter based on an argument the Government had not advanced. We disagree with both

2 decisions, so we vacate Payo’s sentence and remand for further proceedings.

I. BACKGROUND Payo was indicted for two counts of bank robbery and one count of Hobbs Act robbery. He pled guilty to the two bank-robbery counts, although his plea agreement allowed him to appeal any career-offender sentencing enhancement. The Probation Office determined under the Sentencing Guidelines that Payo was a career offender with a total offense level of 29 and a criminal history category of VI. According to the Guidelines, a defendant is a career offender subject to a sentencing enhancement if he or she, among other things, has “at least two prior felony convictions of … a crime of vio- lence.” U.S. Sent’g Guidelines Manual § 4B1.1(a) (U.S. Sent’g Comm’n Nov. 2016). A prior conviction is a crime of violence if the underlying offense was “punishable by imprisonment for a term exceeding one year,” and either (1) “has as an element the use, attempted use, or threatened use of physical force against the person of another,” or (2) “is murder, voluntary manslaughter, kidnapping, aggravated assault, a forcible sex offense, robbery, arson, extortion, or the use or unlawful pos- session of a firearm … or explosive material.” U.S.S.G. § 4B1.2(a). We often call the first prong the force (or elements) clause and the second the enumerated-offenses clause. The Probation Office based Payo’s career-offender sta- tus on a 2001 federal conviction for robbing two banks; a 2008 Pennsylvania conviction for robbing two stores; and a 2010 Pennsylvania conviction for robbing another bank. At the time of Payo’s 2008 and 2010 convictions, Pennsylvania’s robbery statute provided in relevant part that:

3 A person is guilty of robbery if, in the course of committing a theft, he: (i) inflicts serious bodily injury upon an- other; (ii) threatens another with or intentionally puts him in fear of immediate serious bod- ily injury; [or] … (iv) inflicts bodily injury upon another or threatens another with or intentionally puts him in fear of immediate bodily in- jury ….

(a)(1). To determine Payo’s sentence, the District Court or- dered the parties to submit sentencing memorandums. Payo submitted one, but the Government did not. After reviewing Payo’s submission, the District Court acknowledged that he had “raise[d] a number of legal issues” about “whether the ‘Ca- reer Offender’ provisions … appl[ied].” App. 40 (Dkt. No. 56). It was “stumped as to why the United States would not want to weigh in as to such important matters” and “provide[d] [it] with a final opportunity to do so.” App. 40 (Dkt. No. 56). This time, the Government responded. It argued that the career-offender enhancement applied to Payo based on his 2001 and 2008 convictions. In its view, Payo’s two-count con- viction for bank robbery under

(a) counted as two crimes of violence. For good measure, it also claimed, re- lying solely on the presentence report, that his 2008 Pennsyl- vania conviction for robbery was under § 3701(a)(1)(ii) and thus involved a crime of violence. The District Court held a status conference to discuss the parties’ sentencing submissions, during which it explained

4 that it could not determine, based on the presentence report alone, which subsection of § 3701(a)(1) had supported Payo’s 2008 conviction. As we explain below, the answer to that ques- tion matters a great deal. Subsection (ii)—the one the Govern- ment claimed Payo was convicted under—is a crime of vio- lence. But other subsections might not be. To answer this ques- tion, the District Court intended to deploy something called the modified categorical approach, which would involve looking at a narrow category of reliable materials known as Shepard documents to determine which subsection in fact supported Payo’s conviction. See infra Section III.A.1. After the hearing, the District Court ordered the Government to produce Shepard documents shedding light on the 2008 conviction. Over Payo’s objection, the Government produced sev- eral documents. It now claimed that the documents showed “Payo was convicted of two counts of 18 Pa. [Cons. Stat. §] 3701(a)(1)(i) or (ii),” and “submit[ted] that either of these convictions would qualify as a crime of violence.” App. 176 (emphases added). The Government also argued in its cover memorandum, apparently for the first time, that Payo’s 2010 conviction under § 3701(a)(1)(iv) involved a crime of violence under the force clause. It said nothing about the enumerated-offenses clause. After reviewing the Government’s documents, the Dis- trict Court concluded that Payo’s two-count federal conviction constituted only a single crime of violence. And it also found that the Government’s Shepard documents had not established whether Payo’s 2008 conviction was under subsection (i) or (ii). To resolve the matter, it took judicial notice of a state-court docket sheet indicating that the conviction was under the latter. It rested its conclusion on the docket sheet even though it was not generated by the Pennsylvania convicting court and

5 included a prominent disclaimer that its contents may be inac- curate. The Court then turned to Payo’s 2010 conviction. It de- clined to adopt the Government’s argument that subsection (iv) qualified as a crime of violence under the force clause. It in- stead held that subsection qualified under the enumerated-of- fenses clause because its elements categorically matched those of the generic definition of robbery. In the end, the Court agreed with the Probation Office’s Guidelines calculation: Payo’s offense level was 29, his crimi- nal history category was VI, and his Guidelines range was be- tween 151 and 188 months. It sentenced him to two concurrent terms of 127 months’ incarceration and three years’ supervised release. Payo timely appealed.

II. JURISDICTION AND STANDARD OF REVIEW The District Court had jurisdiction under

. We have jurisdiction under

because this is an appeal from a final decision of a district court and under

(a) because this is an appeal of a sen- tence imposed under the Sentencing Reform Act of 1984. We review anew the District Court’s legal conclusions, including its determination that a conviction constitutes a “crime of violence” under the Guidelines. United States v. Hen- derson,

, 210 n.1 (3d Cir. 2023) (quoting United States v. Chapman,

(3d Cir. 2017)). We re- view any factual findings for clear error. United States v. Bent- ley,

(3d Cir. 2022). And we review determi- nations that the Government waived or forfeited legal argu- ments for abuse of discretion. Kars 4 Kids Inc. v. Am. Can!,

, 219 n.9 (3d Cir. 2021).

6 III. THE DISTRICT COURT EXCUSED FORFEITURES AND CONSIDERED A NON-SHEPARD DOCUMENT. Before us, Payo focuses on three arguments, though each contains sub-arguments. First, he contends that the Dis- trict Court violated party-presentation and separation-of-power principles by resurrecting arguments that the Government had forfeited. In his view, the parties, not the courts, present the issues and facts, so when a district court excuses one side’s for- feitures without explanation, it deprives the parties of their right to frame the case. And when the Federal Government is the party whose forfeitures are excused, Payo claims a district court also interferes with the choices of a coordinate branch. Second, Payo argues that the Court abused its discretion by re- lying on the state-court docket sheet to determine that his 2008 conviction was under § 3701(a)(1)(ii) rather than (i). He be- lieves that it should have confined its analysis to the Shepard documents the parties submitted and that it strayed when it considered the docket sheet. And third, Payo claims that neither subsection (i) nor (iv) categorically matches generic robbery, and thus neither qualifies as a career-offender crime.

A. Did the District Court Impermissibly Excuse any Government Forfeitures? Payo argues that the District Court violated party- presentation and separation-of-powers principles in three ways. 1. The Government forfeited its right to introduce Shepard documents, and the District Court improperly excused that forfeiture when it afforded the Government another opportunity to do so before sentencing. 2. The Court improperly relieved the Government of its burden to prove that the sentencing enhancement should

7 apply by relying on a document—the state-court docket sheet—that the Government never submitted. 3. The Court advanced, and then accepted, an argument the Government never made: that Payo’s 2010 convic- tion under § 3701(a)(1)(iv) satisfied the enumerated-of- fenses clause. According to Payo, the Government ar- gued only that Pennsylvania robbery satisfied the force clause—an argument the District Court declined to adopt. Though Payo’s first two arguments are wrong, his third is right.

1. The District Court Could Afford the Government Another Opportunity to Produce Shepard Documents Before Sentencing. In the ordinary case, a court deciding whether a prior conviction qualifies as a crime of violence may look only at the elements of the offense that supported the prior conviction. We call this mode of analysis the categorical approach. In cases like this one, however—when several independently sufficient elements could have supported the conviction—the parties may introduce a narrow set of reliable materials solely to help the court identify the elements that in fact sustained the con- viction. See Shepard v. United States,

(2005) (typical materials include the “charging document, written plea agreement, transcript of plea colloquy, and any explicit factual finding by the trial judge to which the defendant assented”). Because this is a variation on the standard categorical ap- proach, courts call it the modified categorical approach and the limited set of materials Shepard documents. “The Supreme Court has repeatedly stressed that there is a ‘demand for cer- tainty’ in determining whether a defendant was convicted of a

8 qualifying offense.” United States v. Gandy,

(11th Cir. 2019) (quoting Mathis v. United States,

(2016)). To that end, Shepard documents must have the “conclusive significance of a prior judicial record,” Shepard,

, and must “speak plainly” in establish- ing the elements of the underlying offense, Mathis,

. The Government “bears the burden of proving by a pre- ponderance of the evidence that a sentencing enhancement ap- plies.” United States v. Napolitan,

(3d Cir. 2014). According to Payo, the “[G]overnment waived its right to present evidence and argument to satisfy its burden” by re- peatedly declining to submit Shepard documents and choosing instead to rest on the presentence report. Appellant’s Br. 15–16. The Court’s decision to give the Government another chance before the sentencing hearing, he believes, was an abuse of dis- cretion. “We enforce waiver and forfeiture against criminal de- fendants and the [G]overnment equally.” United States v. Dow- dell,

(3d Cir. 2023) (Hardiman, J.). “The pol- icy supporting waiver and forfeiture is the ‘party presentation principle,’ which applies ‘in both civil and criminal cases, in the first instance and on appeal.’”

at 140–41 (quoting Greenlaw v. United States,

(2008)). “[C]at- egorically excus[ing] waivers or forfeitures … would violate the party presentation principle,” which “ensures that courts decide only those issues argued by interested and motivated litigants.” Id. at 145. “And when, as in this case, the [G]overn- ment is a party, categorically excusing forfeiture would raise separation of powers concerns” too, because at sentencing “the [G]overnment—not the Court—bears the burden of proving” that an enhancement applies. Id. at 146.

9 Payo cites no cases holding that a sentencing court abuses its discretion, or violates any of the other principles he invokes, by affording a party additional opportunities to build the record before sentencing. The cases he does cite—United States v. Sineneng-Smith,

(2020), and Dowdell— are not on point. In Sineneng-Smith, the Ninth Circuit appointed, sua sponte, amici to make a constitutional argument that the parties never raised.

. The Supreme Court held that this “takeover of the appeal” was a “radical transformation of th[e] case [that went] well beyond the pale.”

at 379–80. By con- trast, affording the Government another opportunity to support an argument it had made all along—before the sentencing hear- ing—is not a “radical transformation.” And in Dowdell, the Government forfeited a legal argu- ment against suppressing evidence from a traffic stop, yet it argued that the district court should have considered it anyway.

. We disagreed. “Had the [d]istrict [c]ourt inter- vened [t]here by excusing the Government’s forfeiture and ap- plying [its] own novel legal theory, it would have undermined the judiciary’s neutrality and encroached upon the executive branch’s prosecutorial prerogative to argue its case.”

. Again, affording the parties another opportunity to flesh out the record before resolving a crucial factual question is different from and more ordinary than advancing and accepting a legal argument that neither party made.

2. A District Court Can Judicially Notice Documents the Government Never Produced. Payo next argues that even if the District Court’s deci- sion to afford the Government another opportunity to submit Shepard documents was within its discretion, its decision to

10 seek out those documents independently was not. The Govern- ment disputes the premise. In its view, the Court did not con- duct its own investigation. It merely judicially noticed a docket sheet—something it may do even if neither party submitted it. The Government has the better argument. To be sure, Payo is correct that the Government bears the ultimate burden of proving that the career-offender enhancement applies. Na- politan,

. But just because it bears that burden does not mean that the District Court cannot consider judicially noticeable material. Although the Federal Rules of Evidence do not apply to sentencing matters, Fed. R. Evid. 1101(d)(3), Rule 201(c)(1) illustrates the general principle that a court may “take judicial notice on its own,” even though one party almost always has the ultimate burden of proof. Whether the District Court could, however, consider this docket sheet when employing the modified categorical ap- proach to determine whether a sentencing enhancement applies is a different question. As we later explain, it could not because this docket sheet was not a Shepard document. See infra Sec- tion III.B.1.b.

3. The Government Forfeited Its Argument That Payo’s 2010 Conviction Under § 3701(a)(1)(iv) Satisfied the Enumerated-Offenses Clause. Payo also claims that the District Court erred by advanc- ing and accepting an argument for the sentencing enhancement that the Government never made. In its sentencing memoran- dum, the Government never claimed that subsection (iv) was a crime of violence. It instead rested its sentencing argument on Payo’s prior federal conviction and 2008 Pennsylvania convic- tion. It was not until its Shepard submission that the Govern- ment finally asserted that subsection (iv) was a crime of

11 violence. Even then, it limited its argument to the force clause. The District Court implicitly acknowledged as much in its sen- tencing opinion when it noted that “only the Defendant has of- fered a generic version of robbery.” App. 18. The closest the Government came was when it cited United States v. Ball for the proposition that “[t]he First Circuit has held that 18 [Pa. Cons. Stat. §] 3701(a)(1)(iv) is a crime of violence.” App. 176 (citing

(1st Cir. 2017)). The District Court declined to hold that subsection (iv) was a crime of violence under the force clause. And it reasoned that Ball was not on point because its analysis focused on the Armed Career Criminal Act’s (ACCA’s) now-defunct residual clause. See Johnson v. United States,

(2015) (holding residual clause unconstitutionally vague). The Court thus “decline[d] the Government’s invitation to treat Ball as informative as to Mr. Payo’s case.” App. 18. The Court nonetheless proceeded to analyze subsec- tion (iv) under the enumerated-offenses clause, and concluded that it categorically matched the generic definition of robbery. Payo contends that by failing to argue that subsection (iv) sat- isfied the enumerated-offenses clause before sentencing, the Government forfeited that argument. And by not only excusing that forfeiture but also raising and adopting a legal argument that the Government never advanced to apply a sentencing en- hancement, the District Court violated party-presentation and separation-of-powers principles. We agree. In Dowdell, we reasoned that if the district court had “intervened … by excusing the Government’s forfeiture and applying [its] own … legal theory, it would have undermined the judiciary’s neutrality and encroached upon the executive branch’s prosecutorial prerogative to argue its case.”

. Yet the District Court here did just that. It mined our

12 cases for a generic definition of robbery that the Government did not provide and concluded that subsection (iv) was a cate- gorical match based on arguments the Government never made. Despite its insistence to the contrary, the Government’s unelaborated citation to Ball was no enumerated-offenses ar- gument. It cited that case only for the broad proposition that subsection (iv) was a crime of violence. It did not even mention the enumerated-offenses clause, let alone explain why subsec- tion (iv) qualifies under it. “[S]imply citing a case in the Dis- trict Court is not sufficient to raise all arguments that might flow from it.” United States v. Dupree,

(3d Cir. 2010) (Hardiman, J.). Our concurring colleague claims that the “record shows that Dowdell does not apply” here because the “Government argued that Payo had been ‘convicted of two crimes’”—includ- ing § 3701(a)(1)(iv)—“that categorically include the use of force or the threatened use of force.” Conc. Op. 4 (quoting App. 177). But this language, which comes from the Govern- ment’s cover memorandum for its Shepard documents, shows only that it had argued subsection (iv) was a crime of violence under the force clause. No one disputes that. This language does not show that the Government argued that subsection (iv) was a crime of violence under the enumerated-offenses clause. Our colleague also claims that the Government’s ge- neric citation to Ball was enough to raise an enumerated-of- fenses clause argument because that case also discussed ge- neric robbery and the District Court “took … ‘the opportunity to consider the argument.’” Conc. Op. 5 n.1. He believes that Dupree is distinguishable because “[t]here, the [G]overnment cited a case to make one argument in the district court and a different argument on appeal.” Id. (citing Dupree,

at 730–31). That is an accurate statement of the procedural pos- ture in Dupree, but it is unclear why this distinction matters. Unless the standard for preserving an issue for a district court’s consideration is substantially different from the standard for preserving an issue for appeal, then an unexplained citation to a case, without more, is not enough to preserve all arguments that flow from it. Dowdell illustrates the point. The district court there held that the Government had forfeited a potentially winning argument, even though it had cited cases theoretically implicating that argument. We agreed that the Government had forfeited its argument before the district court. Not only did we quote Dupree for the proposition that case citations without elaboration cannot preserve arguments not explicitly made, Dowdell,

, but we also held that the dis- trict court would have abused its discretion had it excused the Government’s forfeiture and considered the argument any- way,

. It is no answer to claim, as our colleague does, that the District Court ultimately reached the enumerated-offenses is- sue anyway. The question is not, as the concurrence frames it, whether the Government as appellee forfeited the argument be- fore us; it is whether the Government as prosecutor forfeited the argument before the District Court. The issue we must de- cide is whether the District Court abused its discretion in im- posing a sentencing enhancement by relying on an argument the Government did not make. * * * Payo’s 2010 conviction cannot support his sentencing enhancement because the Government forfeited the argument that § 3701(a)(1)(iv) is a crime of violence under the enumer- ated-offenses clause and concedes that it is not one under the

14 force clause. We next address whether his 2008 conviction can do the job.

B. Was Payo’s 2008 Conviction Under § 3701(a)(1)(i) or (ii)? The District Court concluded that the 2008 conviction was under § 3701(a)(1)(ii), which all agree is a crime of vio- lence, rather than § 3701(a)(1)(i), which might not be. It rested its conclusion on an online criminal docket sheet it found on its own. Payo argues that the District Court erred in doing so because the docket sheet is not only unreliable but also in ten- sion with the Shepard documents the parties actually produced. The Government claims that there is ample evidence for us to affirm the District Court’s finding. First, it asserts that the Court could consider the docket sheet because it was a judi- cially noticeable Shepard document. Second, it contends that we may consider the plea colloquy from Payo’s 2008 convic- tion—no doubt a Shepard document—which it offers for the first time on appeal. And third, it argues that we may consider Payo’s presentence report. We address each argument in turn.

1. The District Court Erred in Considering the State-Court Docket Sheet. Payo claims that even if the District Court could inde- pendently look at materials that the Government never submit- ted, those materials should have been limited to Shepard doc- uments, and the state-court docket sheet was not one. The Gov- ernment makes two counterarguments. First, it responds that the District Court could look beyond Shepard documents to de- termine the subsection under which Payo was convicted be- cause determining the bare fact of a conviction is analytically different from determining the nature of that conviction. In its

15 view, the modified categorical approach is concerned only with the latter, so Shepard’s restrictions do not limit the District Court in the way Payo claims. Second, the Government’s fallback is that even if the District Court’s analysis were so constrained, the docket sheet is reliable enough to qualify as a Shepard document. Neither argument is persuasive. Payo is correct that the District Court’s inquiry was limited to Shepard documents and that the docket sheet is not such a document.

a. Courts conducting a modified categorical analysis are confined to Shepard documents. Our precedent lays out a three-step framework for de- termining whether to apply a career-offender enhancement. First, we ask if the relevant statute is divisible. United States v. Ramos,

(3d Cir. 2018). Second, if it is, we ask whether “the limited set of extra-statutory materials that we may consult under the modified categorical approach estab- lish[es] with certainty which subsection” of the divisible stat- ute “provided the basis for [the] conviction.”

And third, if it does, we ask whether that specific offense “categorically qualif[ies] as a predicate crime of violence under the Guide- lines.”

Payo acknowledges that § 3701(a)(1) is divisible. United States v. Peppers,

(3d Cir. 2018); Ap- pellant’s Br. 31 (noting that the question “posed here” is “which subsection of a divisible statute [Payo’s] prior convic- tion rested” on). He contends that the District Court could thus look to only a “limited set of extra-statutory materials” to de- termine “which subsection … provided the basis for [his] con- viction.” Ramos,

; see also Appellant’s Br. 28–29. In his view, the docket sheet is not sufficiently reli- able to qualify as a Shepard document, and so the District Court erred by relying on it.

16 The Government responds that the District Court could rely on the docket sheet, even if it were not a Shepard docu- ment, because there is “an analytical distinction between rec- ords used to determine the bare factual question of a defend- ant’s prior conviction” and “the legal question of what ele- ments comprise the offense of conviction.” Appellee’s Br. 25. According to the Government, determining which subsection of § 3701(a)(1) Payo was convicted under involves the former question, so “Shepard’s strictures do not fully apply.” Appel- lee’s Br. 25–26. The Government acknowledges that we have not expressly recognized this distinction. So it instead directs us to the Fourth Circuit’s decision in United States v. Washing- ton, which distinguished between two questions: “what records may the [G]overnment use in determining [a defendant’s] of- fense of conviction, and what records may the [G]overnment use in determining whether the nature of that conviction qual- ifies it as an ACCA predicate.” United States v. Washington,

(4th Cir. 2011). The Government is correct that these inquiries can be analytically distinct, but whether they are depends on the ob- ject of the modified categorical approach. For example, in Washington the defendant was convicted under a Maryland statute that made it illegal “[t]o manufacture, distribute, or dis- pense, or to possess a controlled dangerous substance in suffi- cient quantity to reasonably indicate … an intent to manufac- ture, distribute, or dispense, a controlled dangerous substance.”

at 407–08 (quoting Md. Code. Ann. art. 27 § 286(a)(1)). Whether a sentence enhancement under ACCA applied did not turn on whether he was convicted under that statute; it turned on which drug was involved in the offense. So the sentencing court was not limited to Shepard documents in conducting the former inquiry, but it was in conducting the latter.

17 Here, determining whether Payo was convicted under § 3701(a)(1)(ii) is not, as in Washington, logically prior to the modified categorical analysis; it is itself the object of the anal- ysis. Recall that once we determine the statute is divisible, “we apply the ‘modified categorical approach’ … to identify the subsection under which [the defendant] was convicted.” Hen- derson,

(emphasis added). The Government’s argument would have us consider non-Shepard documents to determine which subsection of a divisible statute a defendant was convicted under, even when we are purporting to use the modified categorical approach to answer precisely that ques- tion. That does not make sense. If we acknowledge that we are “apply[ing] the ‘modified categorical approach’ … to identify the subsection under which” Payo was convicted, then we can- not say, as the Government asserts, that we are also allowed to consider non-Shepard materials to make this same determina- tion.

If we could, the modified categorical approach, and thus our cases applying it to § 3701(a)(1), would contribute nothing. When a district court is conducting the modified cate- gorical approach, rather than some predicate sentencing fact- finding as in Washington, “the ‘judicially noticeable docu- ments’ that are appropriate for consideration are materials such as the ‘charging document, written plea agreement, transcript of plea colloquy, and any explicit factual finding by the trial judge to which the defendant assented’”—in other words, Shepard documents. Reina-Rodriguez v. United States,

(9th Cir. 2011) (quoting Shepard,

). Allowing the Government—or the District Court on its own initiative—to rely on non-Shepard materials to determine the elements that supported a defendant’s conviction would welcome the very “evil [the Supreme Court] sought to

18 prevent—inquiries into the underlying facts that would essen- tially turn the sentencing hearings into mini-trials.”

b. The state-court docket sheet is not a Shepard document. If the District Court was properly confined to Shepard documents in determining whether Payo was convicted under § 3701(a)(1)(ii), and it relied on a docket sheet in reaching that conclusion, then we must next determine whether that docket sheet is a Shepard document. What documents a sentencing court “may rely on to determine the nature of a prior convic- tion” is a “question[] of law, which we review de novo.” United States v. Howard,

(3d Cir. 2010) (citations omitted). Payo argues that the docket sheet here is not sufficiently reliable to qualify as part of the “specific set of extra-statutory documents” the sentencing court may use in conducting the modified categorical analysis. Ramos,

. He con- tends that “[c]omputerized docket sheets obtained from an online repository of data that is inputted by unknown individ- uals from various agencies at unknown times are not conclu- sive records made or used in adjudicating guilt.” Appellant’s Br. 29. He also notes that “each page of the computerized docket sheets” that the District Court relied on contained the following disclaimer: Neither the courts of the Unified Judicial System of the Commonwealth of Pennsylvania nor the Administrative Office of Pennsylvania Courts assume any liability for inaccurate or delayed data, errors or omissions on these reports. Docket Sheet information should not be used in place of a criminal history background check which can

19 only be provided by the Pennsylvania State Po- lice. Moreover an employer who does not com- ply with the provisions of the Criminal History Record Information Act may be subject to civil liability as set forth in 18 Pa. C.S. Section 9183.

Appellant’s Br. 30 (quoting App. 26–33). Shepard documents are more limited than those docu- ments the Government can introduce to prove other sentencing facts, like the fact of a prior conviction. “The Supreme Court has repeatedly stressed that there is a ‘demand for certainty’ in determining whether a defendant was convicted of a qualifying offense.” Gandy,

(quoting Mathis,

). That demand for certainty limits us to documents that have the “conclusive significance of a prior judicial record.” Shepard,

. Those documents must “speak plainly” in establishing the elements of the underlying offense. Mathis,

. Several other circuits have categorically held that docket sheets are not Shepard documents. See, e.g., United States v. Leaverton,

(10th Cir. 2018) (ob- serving that “Shepard documents are limited to conclusive rec- ords made or used in adjudicating guilt”; “docket sheets fall far short of this standard” and “do not qualify as Shepard docu- ments” (internal quotation marks omitted)). But we need not address whether a docket sheet can ever be a Shepard docu- ment because this one clearly is not. It was prepared not by the convicting court but by some other entity and then made avail- able through the Unified Judicial System of Pennsylvania Web Portal. Each page of the docket sheet also prominently declares that “[n]either the courts of the Unified Judicial System of the Commonwealth of Pennsylvania nor the Administrative Office of Pennsylvania Courts assume any liability for inaccurate or

20 delayed data, errors or omissions on these reports.” App. 26–33. The Government characterizes this warning as “boilerplate,” aimed only at “protecting the courts from civil liability.” Appellee’s Br. 27. We are more inclined to read the warning to mean what it says: the docket sheet may be inaccu- rate. Because the docket sheet does not “approach[] the cer- tainty of the record of conviction,” it is not a Shepard document and the District Court erred by relying on it. Shepard,

. The Government resists this conclusion. It points to our decisions in Howard and United States v. Henderson,

(3d Cir. 2016), which in its view held that docket sheets like this one are Shepard documents. But both cases are distin- guishable. Howard involved neither the modified categorical approach nor Shepard documents. The defendant there argued that the Government could not prove the bare fact of a prior felony conviction with “an incomplete certified conviction rec- ord and [municipal court] Docket entries.” Howard,

. In the defendant’s view, those documents lacked “suf- ficient indicia of reliability to support [their] probable accu- racy,”

at 271 (quoting United States v. Miele,

(3d Cir. 1993))—the ordinary reliability standard for con- sidering information at sentencing, see United States v. Free- man,

(3d Cir. 2014) (discussing indicia-of- reliability standard). We disagreed that the docket sheet failed to clear that lower reliability threshold. We held that the sen- tencing court could rely on uncertified docket entries to apply the career-offender enhancement because those “docket entries [we]re the type of judicial records that are permissible for sen- tencing courts to use to establish past convictions for sentenc- ing purposes.” Howard,

. In other words, the Government in Howard was trying to establish a more garden- variety sentencing fact—the kind the Fourth Circuit in

21 Washington held was not subject to Shepard’s heightened reli- ability requirement. See supra at 17–18. Howard thus stands merely for the proposition that a state-court docket entry is re- liable enough to prove the fact of a prior felony conviction. Although it cites Shepard, it is not a Shepard case. Henderson is off point too. That case, unlike Howard, did involve a modified categorical analysis, but not a docket sheet. The document at issue there was a form “completed by the Clerk of Courts in the Court of Common Pleas for Alle- gheny Pennsylvania” and “used for reporting … violations of the Controlled Substance, Drug, Device and Cosmetic Act.” Henderson,

n.10. Henderson cited Howard for the proposition that “we have deemed ‘other reliable judicial records’ to be sufficient, including incomplete certified convic- tion records and docket entries because we have found that ‘both … are records of the convicting court.’”

(quot- ing Howard, 599 F.3d at 272–73) (internal citation omitted) (omission in original). But as explained, we did not hold that state-court docket sheets were Shepard documents in Howard. The point in Henderson was only that the modified categorical approach is not limited to “‘[a]ctual’ conviction documents”; it can include other sufficiently reliable “records of the convict- ing court.” Id.

2. We Cannot Consider Shepard Documents Introduced on Appeal. The Government next argues that it can establish Payo’s 2008 conviction was for violating § 3701(a)(1)(ii) even with- out the state-court docket sheet. “To eliminate any doubt,” it “has now obtained the plea-colloquy transcript, which further confirms that subsection (ii) is ‘the only logical choice’ for [the] statute of conviction.” Appellee’s Br. 18. According to the Government, the plea colloquy shows that Payo “threatened

22 each victim by pulling back a piece of clothing to show what appeared to be a gun tucked into his waistband.” Id. at 19. On its reading, “[t]here was no factual basis presented for Payo to admit to causing serious bodily injury,” which is an element of § 3701(a)(1)(i). Id. The Government insists that it “has ob- tained this transcript on appeal not to sandbag, but in a good- faith effort to assure itself and the Court of the accuracy of the underlying convictions.” Id. at 20. It cites our unpublished de- cision in United States v. Volek for the principle that “we are able to take judicial notice of the Shepard documents submitted by the [G]overnment for the first time on appeal.”

, 126 (3d Cir. 2019). It is hard to harmonize Volek with our earlier published decisions holding that the Government ordinarily must defend a sentencing enhancement using the record it produced before the sentencing court. In United States v. Dickler, we held that when “the [G]overnment has the burden of production and per- suasion as it does on issues like enhancement of the offense level under [the Guidelines], … its case should ordinarily have to stand or fall on the record it makes the first time around.”

(3d Cir. 1995). “It should not normally be af- forded ‘a second bite at the apple.’”

(internal citation omit- ted). To be sure, we noted that there is “no constitutional or statutory impediment to the district court’s providing the [G]overnment with an additional opportunity to present evi- dence on remand.”

But it may do so only “if [the Govern- ment] has tendered a persuasive reason why fairness so re- quires.”

In United States v. Rowe, we reiterated “our distaste for ‘a second bite at the apple.’”

(3d Cir. 2019) (internal citation omitted). We acknowledged that Dick- ler “recognized a limited exception,” but held that the Govern- ment did not provide a good reason for its failure, so we va- cated the sentence, remanded the case for resentencing, and

23 specified that “the Government may not introduce new evi- dence to prove drug quantity.” Id.; see also In re Indian Palms Assocs.,

, 205 (3d Cir. 1995) (“Judicial notice may be taken at any stage of the proceeding, including on appeal, as long as it is not unfair to a party to do so ….” (emphasis added) (quotation marks and citations omitted)). We will not permit the Government here to introduce new Shepard documents on appeal. It bore the burden at sen- tencing of proving that the career-offender enhancement ap- plied to Payo. Napolitan,

. That burden included showing that he committed two predicate offenses. To prove that his 2008 conviction was a crime of violence, and thus a predicate offense, the Government introduced some Shepard documents, but it repeatedly chose not to produce the plea col- loquy. Although Volek states that we can review Shepard doc- uments introduced on appeal, we believe that decision clashes with our general rule—reaffirmed several times in our pub- lished opinions—that the Government must, absent a persua- sive reason, defend its sentencing enhancements based on the record it created before the District Court. See, e.g., Rowe,

; Dickler,

.

3. Payo’s Presentence Report Is Not a Shepard Document. Leaving no stone unturned, the Government argues that the docket sheet and plea colloquy are “not the only bas[e]s for determining the subsection of conviction here.” Appellee’s Br. 16. In its view, “[t]he [presentence report] also supports the [D]istrict [C]ourt’s finding that the convictions were under subsection (ii)” because “Payo’s robberies involved no actual bodily injury, … only threats of serious bodily injury.”

(ci- tation omitted).

24 A presentence report “can be a Shepard document when the defendant does not object to a factual account of a crime therein.” United States v. Doe,

(3d Cir. 2015). The Government acknowledges that Payo objected to the factual descriptions of his prior conviction in the presen- tence report, but insists that those were “boilerplate blanket ob- jections.” Appellee’s Br. 16. Such objections are purportedly “ineffective to nullify the [presentence report’s] findings.”

According to the Government, Payo’s objection would be ef- fective only if he “include[d] evidence to cast ‘real doubt’ on the [presentence report’s] reliability.”

at 17 (quoting United States v. Meherg,

(7th Cir. 2013)). “[O]bjec- tions to the [presentence report that] are merely rhetorical and unsupported by countervailing proof” are not enough.

(quoting United States v. Cyr,

(1st Cir. 2003)). The Government is conflating different ways a presen- tence report might be used. It is correct that a court ordinarily may consider a presentence report in determining sentencing facts. For example, the First Circuit in Cyr held that the district court had not erred by using the presentence report to establish drug quantity for sentencing purposes.

. The de- fendant could not stop the court from considering a presentence report to establish sentencing facts merely by registering a blanket objection. As we have explained, however, establishing ordinary sentencing facts is different from determining which of several alternative elements in a divisible statute supported a defend- ant’s conviction. Under the modified categorical approach, “the universe of records the [G]overnment may use in proving the nature of a prior conviction” is limited to Shepard docu- ments. Washington,

. Those must have the “con- clusive significance of a prior judicial record.” Shepard, 544

25 U.S. at 25. A presentence report is not ordinarily a Shepard document because it is often itself based on non-Shepard doc- uments, like police reports. See United States v. Serrano-Mer- cado,

(1st Cir. 2015) (Lipez, J., concurring) (“[P]olice reports are a typical source of the facts reported in a [presentence report] ….”); see also Shepard,

(holding courts may not rely on police reports when conducting a modified categorical analysis). In Doe, we held that a presentence report may qualify as a Shepard document if, but only if, the defendant registered no objection to its contents.

. The logic under- lying that rule is that a presentence report has conclusive sig- nificance when the defendant accepts that its factual contents are accurate. But when a defendant does not concede that the factual contents of the presentence report are accurate, then that assumption no longer holds. The Government’s argument that a defendant must make an effective objection with evi- dence that casts doubt on the presentence report is right for or- dinary sentencing facts, but gets it backwards for the modified categorical analysis. The default presumption is that a presen- tence report is not a Shepard document. A contrary rule would allow the Government to smuggle information based on non- Shepard materials into the modified categorical analysis, and leave the defendant with the evidentiary burden of disproving the Government’s claims. Because Payo objected to the factual content of the presentence report, it is not a Shepard document, and the District Court was right not to rely on it.

26 IV. WHAT IS THE APPROPRIATE REMEDY?

A. The District Court Shall Determine Whether Payo’s 2008 Conviction Was Under § 3701(a)(1)(i) or (ii). We need not reach Payo’s third and final argument that subsections (i) and (iv) do not categorically match the generic offense of robbery because he succeeds on his first two argu- ments. To recap: the District Court should not have concluded that Payo’s 2010 conviction under § 3701(a)(1)(iv) satisfied the enumerated-offenses clause. The Government never made that argument before the District Court, and, in failing to do so, forfeited it. By excusing the Government’s forfeiture without explanation and advancing a legal argument the Government did not make, the Court violated party-presentation and sepa- ration-of-powers principles. That leaves Payo’s 2008 conviction. The District Court concluded, relying on a state-court docket sheet, that the con- viction was under § 3701(a)(1)(ii), which is a crime of violence under the force clause—the argument the Government made all along. But as we explained, that docket sheet was not a Shepard document. The District Court should not have relied on it. And we cannot, as the Government requests, affirm the District Court’s conclusion by looking at the 2008 plea collo- quy, which is untimely, or the presentence report, which is not a Shepard document because of Payo’s objection. The sole remaining question is factual: Based on the Shepard documents the Government submitted to the District Court, was Payo’s 2008 conviction under subsection (i) or (ii)? If subsection (ii), then the enhancement stands because the par- ties do not dispute that it satisfies the force clause. If subsec- tion (i), then the enhancement was improper, at least here. That

27 is because on appeal the Government concedes that subsection (i) cannot satisfy the force clause. And it already forfeited in the District Court the argument that subsection (i) satisfies the enumerated-offenses clause.1 Because this residual question is essentially factual, we will remand to the District Court, in deference to its superior familiarity with the details of this case, to conduct the inquiry a second time. It should, however, confine its review to the Shepard documents the Government compiled the first time around. Because the Government did not provide a good rea- son for its failure to submit the docket sheet or plea colloquy at sentencing, we will not afford it “a second bite at the apple.” Dickler,

(citation omitted). If the District Court concludes that the remaining mate- rials are inconclusive, then it must assume that the 2008 con- viction was based on “the least culpable conduct hypothetically necessary to sustain a conviction under the statute.” United States v. Harris,

(3d Cir. 2023) (citation omit- ted). Here, that would be § 3701(a)(1)(i). As the Government concedes, a defendant can violate subsection (i) with a mens rea of recklessness. By contrast, § 3701(a)(1)(ii) requires a more culpable mens rea: the defendant must “threaten[]

1 Our concurring colleague again claims that the Government preserved this argument by claiming that its Shepard docu- ments “‘reveal[ed] … Payo was convicted of two counts of 18 Pa. C.S. [§] 3701(a)(1)(i) or (ii)’ and ‘submit[ted] that either of these convictions would qualify as a crime of violence.’” Conc. Op. 3 (quoting App. 176). No one disputes that the Govern- ment argued subsection (i) was a crime of violence. The prob- lem is that it never argued subsection (i) satisfied the enumer- ated-offenses clause.

28 another with or intentionally put[] him in fear of immediate se- rious bodily injury.”

(a)(1)(ii). Finally, we request that the District Court act expedi- tiously on remand to ensure it resolves Payo’s career-offender status before he completes his sentence. Nearly six years elapsed between when Payo was sentenced and when we heard argument, and at the time of our writing his projected release date is November 17, 2026.

B. The District Court May Consider Payo’s Personal History. On remand, if the District Court ultimately determines that Payo is not a career offender, he may be eligible for release immediately upon resentencing. In that event, the District Court will have to determine what, if any, conditions to include in any term of supervised release it might impose. In fashioning the appropriate conditions of that release, the Court might con- sider how Payo’s criminality is rooted in his lifelong struggles with mental illness, drug addiction, and chronic pain. For most of his life, Payo has experienced chronic back pain, drug addiction, and severe mental health issues. He has been diagnosed with anxiety disorder, bipolar depression, post- traumatic stress disorder, and schizoaffective disorder. He has a history of suicidal ideation and has several times received in- patient mental health treatment. In 2017, Payo committed the robberies that supported the conviction before us now. A year earlier, he underwent spi- nal fusion surgery and was committed twice to inpatient mental health treatment. Unable to manage his pain, he told an emer- gency room nurse that he “want[ed] to die” and planned to “jump from the fourth floor building or use a razor blade to slit his throat.” App. 113 (internal quotation marks omitted). These

29 pain and mental-health struggles helped fuel Payo’s addiction as he resorted to drugs to self-medicate. And after he was ar- rested for the 2017 robberies, Payo told the interviewing of- ficer, “I planned on having you guys shoot me.” App. 115 (in- ternal quotation marks omitted). These misfortunes do not excuse Payo’s conduct. But they do provide context. If he is ever to reenter society success- fully and remain crime-free, then he must address his mental- health and drug-use struggles. To that end, the District Court may want to consider including conditions of release targeting the underlying causes of Payo’s significant history of criminal- ity.2

2 We also note that Payo is 60 years old. He was 53 years old when he was sentenced. The United States Sentencing Com- mission has published research suggesting an inverse relation- ship between age and criminality. See U.S. Sent’g Comm’n, Older Offenders in the Federal System 41–44 (July 2022); U.S. Sent’g Comm’n, The Effects of Aging on Recidivism Among Federal Offenders 22 (Dec. 2017). Older offenders are less likely to recidivate and, when they do, are much more likely to commit less serious offenses then their younger counterparts. See Older Offenders in the Federal System 41–44; The Effects of Aging 22. Payo’s age alone greatly reduces the risk he pre- sents to society if he were to be released. During an eight-year follow-up period of all individuals released in 2010 from fed- eral custody, “[t]he recidivism rate of older offenders (21.3%) was less than half that of offenders under the age of 50 (53.4%).” Older Offenders in the Federal System 42–43. And during an eight-year follow-up period of all individuals re- leased in 2005 from federal custody, the recidivism rate for in- dividuals released at age 60 years or older was 16.4%; it was

30 V. CONCLUSION The parties, not the courts, are “responsible for advanc- ing the facts and arguments entitling them to relief” in our criminal legal system. Dowdell,

(quoting Greenlaw,

). In a laudable effort to reach the right answer, the District Court failed to hold the Government to the arguments and factual record it chose to develop. Instead, it resuscitated the Government’s forfeited argument that the 2010 conviction was a crime of violence under the enumerated- offenses clause and purported to take judicial notice of a non- Shepard document. We disagree with those decisions. The only remaining live issue is whether Payo’s 2008 conviction was under § 3701(a)(1)(i) or (ii). We thus vacate Payo’s sentence and remand for the District Court to answer that question as expeditiously as possible using only the Shep- ard documents produced by the Government the first time around. If it can prove that Payo’s 2008 conviction was under subsection (ii), the sentence enhancement stands; otherwise, it does not.

64.8% for those who were released at an age younger than 30 years old. The Effects of Aging 2–3, 22.

31 HARDIMAN, Circuit Judge, concurring in the judgment.

I agree with my colleagues that we must vacate David Payo’s sentence and remand for the District Court to resentence him. The resentencing hearing should be de novo.

This appeal comes to us in an unusual procedural posture. Payo was sentenced on March 7, 2019. The parties disputed whether he was a “career offender” within the meaning of U.S.S.G. § 4B1.1. To answer that question, the parties focused mostly on whether Payo’s two-count conviction for Pennsylvania robbery in violation of

(a)(1)(i) or (ii) involved the use of force.

After Payo was sentenced, his appeal was stayed for over four years pending the resolution of other cases involving categorical approach issues. The law changed during that time. First, the Supreme Court held that a criminal offense that requires only a mens rea of recklessness does not satisfy the elements clause in the Armed Career Criminal Act’s definition of “violent felony.” See Borden v. United States,

(2021) (plurality opinion). Second, we determined that

(a)(1)(ii) qualifies as a “crime of violence” under U.S.S.G. § 4B1.2(a)’s elements clause. See United States v. Henderson,

(3d Cir. 2023), cert. denied,

(2024).

The decisions just mentioned caused the parties to shift their focus on appeal. In the District Court, the Government had argued that Payo pleaded guilty to violating § 3701(a)(1)(i) or (ii) and that both subsections qualify as a “crime of violence” under U.S.S.G. § 4B1.2(a)’s elements clause. Meanwhile, Payo had contended that it was unclear whether he pleaded guilty to violating § 3701(a)(1)(i) or (ii), and he had

1 tactically argued that under Shepard the District Court could not consider a state court sentencing order that indicated he had pleaded guilty to violating § 3701(a)(1)(i). The Government now concedes that § 3701(a)(1)(i) cannot qualify as a crime of violence under U.S.S.G. § 4B1.2(a)’s elements clause because it requires only a mens rea of recklessness. So the Government argues that the Shepard documents show Payo pleaded guilty to violating § 3701(a)(1)(ii) and that § 3701(a)(1)(i) is an enumerated offense. For his part, Payo now argues that the Shepard documents show he pleaded guilty to violating § 3701(a)(1)(i), which he contends is not an enumerated offense.

In view of these legal and tactical changes, the District Court deserves the opportunity to resentence Payo without constraint, subject to the now-controlling precedents. The majority requires the opposite, severely circumscribing the District Court’s ability to get this case correct based on the facts and the law. In my view, the majority makes three fundamental errors.

First, the majority holds that the District Court erred by relying on the docket face sheet because it was not a Shepard document. I agree that the District Court erred by relying exclusively on the docket face sheet in view of the apparent conflict between the docket face sheet and the sentencing order. So I would remand for the District Court to reexamine the docket face sheet, the state court sentencing order, and any other appropriate Shepard documents to determine whether Payo was convicted under

(a)(1)(i) or (ii).

Second, the majority holds that the Government forfeited the argument that § 3701(a)(1)(i) satisfies the

2 enumerated offenses clause. In doing so, the majority turns forfeiture on its head by invoking the doctrine against the appellee in this procedural posture. But as we have explained, “forfeiture has greater consequences for appellants than for appellees.” Montemuro v. Jim Thorpe Area Sch. Dist.,

(3d Cir. 2024). The Government was “not required to raise all possible alternative grounds for affirmance to avoid [forfeiting] those grounds.” Eichorn v. AT&T Corp.,

, 657–58 (3d Cir. 2007). And “[w]e may affirm a district court for any reason supported by the record.” Brightwell v. Lehman,

(3d Cir. 2011). So the Government may defend the judgment below on a new ground. See, e.g., Los Rovell Dahda v. United States,

, 449–50 (2018) (affirming based on “an argument that the Government did not make below” that was “closely related to the arguments the Government did make below”). Considering alternative grounds for affirmance is especially appropriate where, as here, there has been a significant intervening change in the law. See Borden,

; see also United States v. Quailes,

, 218 n.4 (3d Cir. 2025) (“[W]e may reach forfeited arguments that relate to an intervening change in controlling case law that occurs while appeal is pending.”).

The Government also raised this issue in the District Court, so it is especially mistaken to preclude it from renewing this argument on remand. When it submitted the Shepard documents, the Government argued that they “reveal[ed] . . . Payo was convicted of two counts of 18 Pa. C.S. 3701(a)(1)(i) or (ii)” and “submit[ted] that either of these convictions would qualify as a crime of violence.” App. 176. Because the Government raised this issue, I would allow the District Court to consider it on remand and would not restrict the sources it

3 may consider in determining whether Payo’s conviction was for § 3701(a)(1)(i) or (ii).

Third, the majority holds that because the Government forfeited its argument that

(a)(1)(iv) is a crime of violence under U.S.S.G. § 4B1.2(a)’s enumerated offenses clause, the District Court abused its discretion by accepting it. Citing dicta in United States v. Dowdell,

(3d Cir. 2023), the majority concludes that the District Court “violated party-presentation and separation-of-powers principles” by reaching a forfeited argument. Maj. Op. Section III-A-3. Once again, the majority flips forfeiture on its head by invoking the doctrine against the appellee on an argument that the District Court actually reached.

The District Court record shows that Dowdell does not apply to this case. The Government argued that Payo had been “convicted of two crimes that categorically include the use of force or the threatened use of force.” App. 177. Those two crimes included

(a)(1)(i) or (ii) and

§ 3701(a)(1)(iv). App. 176. As for the latter, the Government said only that “[t]he First Circuit has held that 18 Pa. C.S. 3701(a)(1)(iv) is a crime of violence.” Id. (citing United States v. Ball,

(1st Cir. 2017)). But in an opinion following that written submission, the District Court recognized that Ball discussed whether § 3701(a)(1)(iv) substantially corresponds with generic robbery. United States v. Payo,

, at *4 (W.D. Pa. Jan. 16, 2019).

In sum, the Government’s sparse filing apprised the District Court of its argument.1 Although the District Court

1 The majority also says that “simply citing a case in the District Court is not sufficient to raise all arguments that might

4 could have required the Government to further develop its argument, its decision not to do so was far from an abuse of discretion. See Dowdell, 70 F.4th at 140–42. And its conclusion that § 3701(a)(1)(iv) was an enumerated offense is a required judicial function, not a violation of the separation of powers.

The idiosyncrasies of this case require a full and fair hearing in the District Court so the sentencing judge can ascertain, once and for all, whether Payo is a career offender. With respect, I concur only in the judgment.

flow from it.” Maj. Op. Section III-A-3 (quoting United States v. Dupree,

(3d Cir. 2010)) (cleaned up). That language from Dupree is taken out of context. There, the government cited a case to make one argument in the district court and a different argument on appeal. Dupree, 617 F.3d at 730–31. We held the argument first raised on appeal was forfeited because the district court was not given “the opportunity to consider” it.

. Here, by contrast, the District Court had—and took—“the opportunity to consider the argument” when the Government cited Ball. Id.; Payo,

, at *4–6.

5