Opinion
No. 21-6056
03-21-2023
UNITED STATES of America, Plaintiff-Appellee, v. Deron Edwards ROBINSON, Defendant-Appellant.
ON BRIEF: Benjamin H. Perry, LAW OFFICE OF BENJAMIN H. PERRY, Nashville, Tennessee, for Appellant. Robert E. McGuire, UNITED STATES ATTORNEY'S OFFICE, Nashville, Tennessee, for Appellee.
Appeal from the United States District Court for the Middle District of Tennessee at Nashville. No. 3:05-cr-00196-2—William Lynn Campbell, Jr., District Judge. ON BRIEF: Benjamin H. Perry, LAW OFFICE OF BENJAMIN H. PERRY, Nashville, Tennessee, for Appellant. Robert E. McGuire, UNITED STATES ATTORNEY'S OFFICE, Nashville, Tennessee, for Appellee. Before: McKEAGUE, WHITE, and MURPHY, Circuit Judges.
OPINION
MURPHY, Circuit Judge.
This case raises two questions about supervised release. The first question concerns the exclusionary rule. That rule sometimes bars the government from using evidence at a criminal trial if the police obtained the evidence in violation of the Fourth Amendment. Does the rule also bar illegally obtained evidence from a hearing at which a court decides whether to revoke a defendant's supervised release and send the defendant back to prison? We answer "no" because the Supreme Court has held that the rule does not apply in the analogous parole setting. See Pa. Bd. of Prob. & Parole v. Scott, 524 U.S. 357, 359, 118 S.Ct. 2014, 141 L.Ed.2d 344 (1998).
The second question concerns the right to a jury trial. Courts have long held that defendants do not have a right to a jury at supervised-release hearings. See United States v. Johnson, 356 F. App'x 785, 790-92 (6th Cir. 2009). Yet the Supreme Court recently found that this jury-trial guarantee applied to a unique provision—18 U.S.C. § 3583(k)—that imposed a minimum 5-year prison term on a defendant who committed specified federal crimes while on supervised release. United States v. Haymond, — U.S. —, 139 S. Ct. 2369, 2378-79, 204 L.Ed.2d 897 (2019) (plurality opinion); id. at 2386 (Breyer, J., concurring in the judgment). Does Haymond also render unconstitutional a separate provision—18 U.S.C. § 3583(g)—that requires a court to impose a prison term of unspecified length if a defendant has engaged in certain conduct (such as possessing a gun) while on supervised release? We again answer "no" because the narrow logic of the controlling opinion in Haymond does not cover § 3583(g). These two answers lead us to affirm the district court's judgment.
I
In 2005, police officers spotted DeRon Robinson as he left a suspected stash house for illegal drugs. Robinson had a warrant out for his arrest. After officers arrested him, they uncovered a firearm and drugs in his car. Robinson pleaded guilty to possessing a firearm as a felon, in violation of 18 U.S.C. § 922(g)(1), and possessing drugs with the intent to distribute them, in violation of 21 U.S.C. § 841(a)(1). The district court originally sentenced Robinson to 130 months' imprisonment, but a retroactive change in the Sentencing Guidelines led it to reduce his sentence to 92 months. Robinson finished his prison time and started a 4-year term of supervised release in 2017.
On February 9, 2021, a few months before the expiration of this supervised-release term, Robinson was driving late at night outside Nashville in Hendersonville, Tennessee. A Hendersonville officer stopped Robinson purportedly for a tinted-window infraction. According to Robinson, however, the dash-cam video on the police cruiser caught the officer stating that he had stopped the car after running its license plate because "he had a '50/50 shot' of pulling over a criminal[.]" Order, R.203, PageID 826; Mot., R.196, PageID 783. (We do not know what this video shows because the parties failed to include it in our record on appeal.) Upon questioning from the officer, Robinson noted that he had almost completed his supervised-release term. The officer gave Robinson a written warning for the tinted-window violation. He then asked for consent to search the car, a request that Robinson apparently evaded. The officer allegedly told Robinson that a refusal to consent might violate the conditions of his supervised release. The officer added that he did not want to see Robinson run afoul of those conditions. Robinson responded: "I already know because[ ] I got, I have, I got a gun in the glove compartment man." Pet., R.179, PageID 745. With this admission in hand, the officer searched the car and found a loaded handgun, marijuana, cocaine, and prescription drugs.
The Hendersonville police arrested Robinson on firearm and drug charges. Yet neither state nor federal prosecutors indicted him. According to the district court, they likely declined to prosecute because of the "obvious Fourth Amendment violations." Tr., R.213, PageID 869-70. A probation officer did, however, petition the court to revoke Robinson's supervised release because his conduct had violated the conditions on that release.
In the ensuing supervised-release proceedings, Robinson moved to suppress the firearm and drugs found in his car on the ground that the search violated the Fourth Amendment. The government conceded the Fourth Amendment violation but argued that this violation did not require the court to exclude the evidence from Robinson's revocation hearing. The district court agreed that the exclusionary rule does not apply in the supervised-release context and denied Robinson's motion.
Robinson separately moved for a jury trial over whether he had violated the conditions of his supervised release. He argued that Haymond clarified the Sixth Amendment's scope and gave him a right to a jury in this context. The district court disagreed, citing out-of-circuit decisions to distinguish Haymond.
At Robinson's revocation hearing, the district court relied on the officer's dash-cam video to find as a fact that Robinson had possessed the firearm and drugs in violation of his supervised-release conditions. The court revoked Robinson's supervised release and sentenced him to another 28 months' imprisonment.
II
Federal law allows district courts to require defendants to serve a term of "supervised release" after they complete their prison terms. 18 U.S.C. § 3583(a). Courts usually impose many conditions on released prisoners, including, for example, that they refrain from committing further crimes and possessing drugs. See id. § 3583(d). When a probation officer believes that a defendant has violated a condition of supervised release, the officer may ask a district court to revoke supervised release and order the defendant back to prison. See Fed. R. Crim. P. 32.1(a). A district court will then hold a revocation hearing to decide whether the defendant engaged in conduct that violated the supervised-release condition and, if so, the appropriate punishment. See 18 U.S.C. § 3583(e)(3); Fed. R. Crim. P. 32.1(b)(2).
On appeal, Robinson asks us to incorporate two criminal-trial protections into supervised-release revocation hearings. He argues that the exclusionary rule and the right to a jury trial should apply at these hearings, which courts hold to determine whether defendants have violated their conditions of supervised release. Reviewing these legal arguments de novo, we disagree with Robinson on both fronts. See United States v. Roberge, 565 F.3d 1005, 1010, 1012 (6th Cir. 2009).
A. Exclusionary Rule
The Fourth Amendment restricts the government's ability to ferret out crime, including by banning "unreasonable searches and seizures[.]" U.S. Const. amend. IV. In Robinson's case, the government concedes that the Hendersonville officer violated this provision because, after issuing a tinted-window warning, the officer extended the stop without reasonable suspicion that Robinson had committed other crimes. See Rodriguez v. United States, 575 U.S. 348, 354, 135 S.Ct. 1609, 191 L.Ed.2d 492 (2015). But this case does not concern the amendment's restrictions; it concerns the amendment's remedies.
The Fourth Amendment's text does not identify the relief that courts must (or may) provide when a police officer violates one of its limits on government investigations. See Davis v. United States, 564 U.S. 229, 236, 131 S.Ct. 2419, 180 L.Ed.2d 285 (2011). The Supreme Court has nevertheless adopted the "exclusionary rule" to deter the police from exceeding these limits. See id. at 236-37, 131 S.Ct. 2419. This rule sometimes compels courts to exclude incriminating evidence from a trial if the police violated the amendment when obtaining it. See United States v. Reed, 993 F.3d 441, 450 (6th Cir. 2021).
The Supreme Court's use of the rule has changed over time. The Court once treated the rule as a constitutionally compelled remedy for all Fourth Amendment violations. See United States v. Baker, 976 F.3d 636, 646 (6th Cir. 2020). But the Court has come to recognize that the rule represents a judge-made remedy that it may adjust in common-law fashion. See Davis, 564 U.S. at 238, 131 S.Ct. 2419. It has also come to recognize that the rule imposes an "extreme" penalty, Herring v. United States, 555 U.S. 135, 140, 129 S.Ct. 695, 172 L.Ed.2d 496 (2009) (citation omitted), with "heavy costs," Davis, 564 U.S. at 237, 131 S.Ct. 2419. Nowadays, therefore, the Court does not extend the rule to all Fourth Amendment violations. See Hudson v. Michigan, 547 U.S. 586, 591-92, 126 S.Ct. 2159, 165 L.Ed.2d 56 (2006). Rather, the Court reserves the rule for cases in which its benefits (in deterring "future Fourth Amendment violations") exceed its costs (in interfering with a trial's truth-finding function and freeing wrongdoers). Davis, 564 U.S. at 236-37, 131 S.Ct. 2419; see also Herring, 555 U.S. at 140-42, 129 S.Ct. 695; Scott, 524 U.S. at 362-63, 118 S.Ct. 2014.
The Supreme Court's current utilitarian approach to the exclusionary rule leads us to the proper question here: If we extended the rule to the supervised-release context, would its upsides of deterring Fourth Amendment violations outweigh its downsides of potentially allowing a defendant to violate a supervised-release condition without consequence? The Court's resolution of this cost-benefit inquiry in Scott goes a long way toward deciding it in this case too. Scott held that the exclusionary rule does not compel a state to exclude illegally obtained evidence from a parole-violation hearing at which the state seeks to revoke a defendant's parole. 524 U.S. at 364, 118 S.Ct. 2014. The Court's weighing of the rule's benefits and costs in the parole context largely carries over to this related supervised-release context. See id. at 364-69, 118 S.Ct. 2014.
Start with the exclusionary rule's benefits. The rule's expansion to supervised release would produce only modest additional deterrence of Fourth Amendment violations. See id. at 367-69, 118 S.Ct. 2014. Just as in the parole context, officers will often not even know that a subject is on supervised release. See id. at 367, 118 S.Ct. 2014. Usually, such an officer will be investigating crime with an eye toward a criminal trial, not a supervised-release proceeding. See id. Because the exclusionary rule would apply at trial, the officer already has a strong incentive to adhere to the Fourth Amendment's requirements. See id. And the "remote possibility" that the government might also use uncovered evidence in supervised-release proceedings would have "little, if any, effect on the officer's incentives." Id.
Even when an officer knows that a suspect is on supervised release, extending the rule to supervised-release proceedings would not increase the officer's incentive to follow the Fourth Amendment by much. See id. at 368, 118 S.Ct. 2014. Again as in the parole context, a typical officer's primary focus will remain on "obtaining convictions," not supervised-release revocations. Id. The two proceedings are not one and the same. A supervised-release revocation usually triggers a shorter sentence, the maximum term of which generally depends on the defendant's original crime rather than the new offense. See 18 U.S.C. § 3583(e)(3), (g)(1). Robinson's case proves the point. The court sentenced him to 28 months in prison. If the government had instead indicted him for the gun and drug crimes, he likely would have faced a much longer term. 18 U.S.C. § 924(a)(8); 21 U.S.C. § 841(b). According to the district court, however, the government declined to indict him due to the risk that a court would exclude the key evidence. This real-world effect shows that the exclusionary rule would incentivize the police to follow the Fourth Amendment even if it does not cover supervised-release proceedings. See Scott, 524 U.S. at 368, 118 S.Ct. 2014.
Turn to the exclusionary rule's costs. The rule would undermine the purpose of supervised release in the same way that it would undermine the purpose of parole. See id. at 365, 118 S.Ct. 2014. The two programs serve the same function. See Johnson v. United States, 529 U.S. 694, 710-11, 120 S.Ct. 1795, 146 L.Ed.2d 727 (2000). By requiring released prisoners to follow the conditions that a district court finds necessary for their successful reentry into society, supervised release seeks to help these "releasees" during this transition and ensure that they do not endanger the community. See id. at 708-10, 120 S.Ct. 1795. Yet the exclusionary rule would reduce a court's ability to hold releasees accountable for violations of the conditions and so reduce the releasees' incentives to follow them. See Scott, 524 U.S. at 365, 118 S.Ct. 2014.
If anything, the exclusionary rule could have more harmful effects in this supervised-release context than in the parole context. When switching from parole to supervised release, Congress changed from a system that applied to all releasees to one in which courts target "supervision to those releasees who need[ ] it most." Johnson, 529 U.S. at 709, 120 S.Ct. 1795. Because supervised release applies only to the defendants most in need of supervision, the federal government has an even more "overwhelming interest" in ensuring that they follow their supervised-release requirements. Scott, 524 U.S. at 365, 118 S.Ct. 2014 (quoting Morrissey v. Brewer, 408 U.S. 471, 483, 92 S.Ct. 2593, 33 L.Ed.2d 484 (1972)). If courts cannot adequately monitor this group of releasees due to the exclusionary rule, the rule's application might "disadvantage" these releasees by making it more likely that the courts would impose longer initial prison terms instead in order to ensure public safety. Cf. id. at 367. Scott made a similar point when it noted that the use of the exclusionary rule in the parole context could make "the hearing body . . . less tolerant of marginal deviant behavior and feel more pressure to reincarcerate than to continue nonpunitive rehabilitation." Id. (quoting Gagnon v. Scarpelli, 411 U.S. 778, 788, 93 S.Ct. 1756, 36 L.Ed.2d 656 (1973)).
This balance of costs and benefits also follows from the nature of the prison term that a court imposes after it revokes a defendant's supervised release. That prison term does not represent punishment for the conduct that violated a supervised-release condition. See Johnson, 529 U.S. at 700, 120 S.Ct. 1795. It represents "part of the penalty for the initial offense[.]" Id. So a supervised-release hearing resembles a sentencing hearing because both seek to choose the proper sentence for a crime that a jury has already found beyond a reasonable doubt. Yet we (and most courts) have held that the exclusionary rule does not generally bar a judge from considering illegally obtained evidence when choosing a prison term at sentencing. See United States v. Jenkins, 4 F.3d 1338, 1344-45 (6th Cir. 1993); see also United States v. Acosta, 303 F.3d 78, 84-85 (1st Cir. 2002) (collecting cases). It is difficult to see how courts could coherently hold that the exclusionary rule should apply at a hearing to modify a sentence when they do not believe that it should apply at the hearing to pick the original sentence. See United States v. Phillips, 914 F.3d 557, 559-60 (7th Cir. 2019).
The Supreme Court's consistent track record of rejecting the exclusionary rule in all contexts other than a criminal trial further undergirds our cost-benefit analysis. United States v. Hill, 946 F.3d 1239, 1241 (11th Cir. 2020) (per curiam). Do the rule's benefits exceed its costs when a prosecutor seeks to present illegally obtained evidence to a grand jury? No. United States v. Calandra, 414 U.S. 338, 351-52, 94 S.Ct. 613, 38 L.Ed.2d 561 (1974). How about when the government's tax lawyers use this evidence to seek overdue taxes in a civil case? Also no. United States v. Janis, 428 U.S. 433, 458-60, 96 S.Ct. 3021, 49 L.Ed.2d 1046 (1976). Or when the government's immigration lawyers use the evidence to seek an immigrant's deportation? No, for a third time. INS v. Lopez-Mendoza, 468 U.S. 1032, 1050-51, 104 S.Ct. 3479, 82 L.Ed.2d 778 (1984). We fail to see why courts should treat supervised release differently.
Lastly, seven other circuit courts have now held that the exclusionary rule does not apply in supervised-release proceedings. See United States v. Hightower, 950 F.3d 33, 36-38 (2d Cir. 2020) (per curiam); Hill, 946 F.3d at 1241-42; Phillips, 914 F.3d at 558-60; United States v. Charles, 531 F.3d 637, 640 (8th Cir. 2008); United States v. Hebert, 201 F.3d 1103, 1104 (9th Cir. 2000) (per curiam); United States v. Armstrong, 187 F.3d 392, 393-95 (4th Cir. 1999); United States v. Montez, 952 F.2d 854, 857-59 (5th Cir. 1992). We have likewise refused to apply the rule in those proceedings, albeit in lightly reasoned unpublished decisions. See United States v. Williams, 858 F. App'x 827, 830 (6th Cir. 2021); United States v. Alexander, 1997 WL 592807, at *1 (6th Cir. Sept. 24, 1997) (order); United States v. Blackshear, 1993 WL 288297, at *4-5 (6th Cir. July 29, 1993) (per curiam). We now officially join this unanimous circuit precedent.
Robinson responds with both a broad argument and a narrow one. As a broad matter, he asserts that the exclusionary rule should apply at all supervised-release revocation hearings. Robinson asks us to distinguish Scott (and create a circuit split) because he says that a hearing to revoke a defendant's supervised release is "drastically" different from a hearing to revoke a defendant's parole. Appellant's Br. 15. He accurately identifies distinctions between the two hearings. An agency might oversee a parole hearing, Scott, 524 U.S. at 366, 118 S.Ct. 2014, whereas a judge oversees a supervised-release hearing, 18 U.S.C. § 3583(e)(3). And Scott described a parole hearing as "nonadversarial," 524 U.S. at 366, 118 S.Ct. 2014, whereas a supervised-release hearing pits the government against a counseled defendant, see Fed. R. Crim. P. 32.1(b)(2). Yet these factual distinctions do not make a legal difference. Indeed, we have already called supervised release and parole "analogous for constitutional purposes[.]" Johnson, 356 F. App'x at 791; see also Johnson, 529 U.S. at 710-11, 120 S.Ct. 1795. And even if Scott addressed only nonadversarial proceedings, the Court has refused to apply the exclusionary rule in adversarial contexts too. See Janis, 428 U.S. at 459-60, 96 S.Ct. 3021.
Citing the dissent in Scott, Robinson next argues that the government often chooses to proceed with a revocation hearing over new criminal charges because it can prove a supervised-release violation to a court more easily than it can prove a criminal violation to a jury. See Scott, 524 U.S. at 378-79, 118 S.Ct. 2014 (Souter, J., dissenting). Because the two types of proceedings are "substitutes," this argument goes, the exclusionary rule would not provide proper incentives unless it applied in both places. As we have noted, however, the two are not substitutes. A supervised-release revocation generally will trigger a shorter sentence. See 18 U.S.C. § 3583(e)(3). That is because courts generally treat a supervised-release revocation as a sanction for breach of the court's trust, not for the new crime. Johnson v. United States, 179 F. App'x 246, 248 (6th Cir. 2006); Haymond, 139 S. Ct. at 2386 (Breyer, J., concurring in the judgment); U.S.S.G. ch. 7, pt. A, intro. 3(b). So a releasee's new crimes often lead to both fresh charges and a supervised-release revocation. See, e.g., United States v. Massey, 758 F. App'x 455, 457-59 (6th Cir. 2018). Regardless, the parolee in Scott could have made the same "substitute" argument for the government's use of a parole-violation hearing over a criminal trial. But Scott held that any added deterrence benefits did not justify extending the exclusionary rule to that hearing. 524 U.S. at 367-69, 118 S.Ct. 2014. If this argument did not suffice in that case, it cannot suffice in this one. See Phillips, 914 F.3d at 559.
As a narrow matter, Robinson alternatively asks us to carve out a "harassment" exception that would allow courts to apply the exclusionary rule if an officer harasses a defendant because of the defendant's supervised-release status. Although we have discussed the possibility of this exception, we have never found that it applies. See Alexander, 1997 WL 592807, at *1; Blackshear, 1993 WL 288297, at *4; see also United States v. Farmer, 512 F.2d 160, 162 (6th Cir. 1975). Scott also rejected a request to create a "special" exclusionary "rule for those situations in which the officer performing the search knows that the subject of his search is a parolee." 524 U.S. at 367-68, 118 S.Ct. 2014; cf. Hightower, 950 F.3d at 38 n.2. Yet we need not resolve whether this exception survives Scott. Even if we could apply the exclusionary rule when a defendant makes a sufficient showing of harassment, Robinson has failed to prove such harassment here. Apart from the alleged Fourth Amendment violation itself, the district court saw no evidence of harassment by the otherwise "courteous" Hendersonville officer and no evidence that this officer knew that Robinson was on supervised release when the officer stopped his vehicle. Tr., R.203, PageID 826-27. Robinson does not attempt to show that these findings of fact were clearly erroneous. Cf. Montez, 952 F.2d at 859. We thus have no basis to overturn the court's rejection of the exclusionary rule.
B. Jury Trial
The Fifth and Sixth Amendments give an "accused" both the right to "an impartial jury" and the right to "due process of law" before the government may deprive the accused of liberty. U.S. Const. amend. V-VI. When these two rights apply, they require a jury to find beyond a reasonable doubt any fact that is necessary to increase a defendant's punishment above the statutory maximum sentence that would otherwise apply without that fact. See Apprendi v. New Jersey, 530 U.S. 466, 490, 120 S.Ct. 2348, 147 L.Ed.2d 435 (2000). They also require a jury to find beyond a reasonable doubt any fact that is necessary to increase a defendant's mandatory-minimum punishment above the statutory minimum sentence that would otherwise apply without that fact. See Alleyne v. United States, 570 U.S. 99, 103, 133 S.Ct. 2151, 186 L.Ed.2d 314 (2013).
In the supervised-release statute, however, Congress has adopted a system that does not provide these procedural protections for the hearing at which a court determines whether to revoke a defendant's supervised release and return the defendant to prison. 18 U.S.C. § 3583(e)(3). The statute allows a district court—not a jury—to find as a factual matter that the defendant has violated a condition of supervised release. Id. And the statute orders the district court to make this finding under a less demanding standard (preponderance of the evidence) than the standard (beyond a reasonable doubt) that due process would demand at a criminal trial. Id.
Do these procedures conflict with the Fifth and Sixth Amendments? As a general matter, circuit courts have long rejected constitutional attacks on § 3583's processes. The Supreme Court has held that a parole-violation hearing does not count as an independent "criminal prosecution" for purposes of the Fifth and Sixth Amendments because reimprisonment after a parole revocation counts as punishment for the original crime that the jury found the parolee committed beyond a reasonable doubt. See Morrissey, 408 U.S. at 480, 487-90, 92 S.Ct. 2593. And reimprisonment for a supervised-release violation is "analogous" to reimprisonment for parole, Johnson, 356 F. App'x at 791, because it too qualifies "as part of the penalty for the initial offense" that led to the initial sentence, Johnson, 529 U.S. at 700, 120 S.Ct. 1795. So revocation hearings in the supervised-release context likewise generally are not "criminal prosecution[s]" that require a jury or a reasonable-doubt burden of proof. See Johnson, 356 F. App'x at 790-91 (citing cases); see also United States v. Melton, 782 F.3d 306, 309 (6th Cir. 2015); United States v. Kirby, 418 F.3d 621, 627-28 (6th Cir. 2005).
As a specific matter, however, the Supreme Court's Haymond decision carved out an exception to this precedent. In that case, Andre Haymond committed a child-pornography offense that came with a prescribed punishment of 0 to 10 years in prison. 139 S. Ct. at 2373 (plurality opinion). A court sentenced Haymond to 38 months in prison and 10 years on supervised release. Id. He served his prison time. Id. at 2374. In subsequent supervised-release proceedings, a district court found that he committed additional child-pornography crimes that triggered a unique supervised-release provision: 18 U.S.C. § 3583(k). Id. Under § 3583(k), when a court finds that certain defendants have committed one of the crimes on a list of federal sex-related offenses, the court must revoke their supervised release and impose a punishment between 5 years and life imprisonment. See id. The court thus imposed an additional 5-year prison term on Haymond. Id.
In a 4-1-4 decision, the Supreme Court held that the Fifth and Sixth Amendments required a jury to find beyond a reasonable doubt that Haymond committed the crimes listed in § 3583(k). A 4-Justice plurality reasoned that this subsection's 5-year mandatory minimum triggered these two amendments because Haymond would never have faced any mandatory-minimum sentence except for the court's later finding that he committed the new crimes. Id. at 2378. With that judicial fact-finding, his mandatory minimum jumped from no prison time to 5 years' time. Id. Because this fact "increased 'the legally prescribed range of allowable sentences,' " the plurality reasoned, a jury had to find it beyond a reasonable doubt. Id. (quoting Alleyne, 570 U.S. at 115, 133 S.Ct. 2151). In response to the dissent's charge that the plurality opinion would require that "all supervised release proceedings comport with Apprendi," the plurality also stressed that its "decision [was] limited to § 3583(k)—an unusual provision enacted little more than a decade ago—and the Alleyne problem raised by its 5-year mandatory minimum term of imprisonment." Id. at 2383.
Justice Breyer concurred in the outcome for different reasons. He rejected the plurality's choice to expand Apprendi and its progeny wholesale into this supervised-release context. Id. at 2385 (Breyer, J., concurring in the judgment). Yet he still found § 3583(k) unconstitutional based on several factors that made it more like a punishment for a new crime than a revocation of supervised release. Id. at 2386. To begin with, § 3583(k) applied only when the defendant committed specifically identified crimes from the U.S. Code. Id. Next, § 3583(k) took away a district court's discretion to determine whether a supervised-release violation warranted additional prison time and the length of that time. Id. Last, § 3583(k) restricted this discretion "in a particular manner" by imposing a 5-year mandatory minimum for the defendant's new federal crimes. Id.
Citing Haymond, Robinson argues that the Fifth and Sixth Amendments required a jury to find beyond a reasonable doubt that he had a gun and drugs while driving in Hendersonville—the facts that led the district court to revoke his supervised release and reimprison him for 28 months. Although Robinson's case does not implicate § 3583(k), he argues that Haymond's logic extends to a supervised-release provision that his case does implicate: 18 U.S.C. § 3583(g). Subsection 3583(g) provides that a court "shall" revoke supervised release and "require" a defendant to serve more prison time if the defendant commits one of four types of supervised-release violations, including the possession of a controlled substance or a firearm. Id. According to Robinson, § 3583(g) has the same constitutional problem as § 3583(k).
We disagree. At the outset, though, it is not clear how the Haymond plurality's Apprendi-based logic would apply here. The plurality did not "express a view" on whether its logic extended to § 3583(g). 139 S. Ct. at 2382 n.7 (plurality opinion). Indeed, the plurality "emphasized [that its] decision [was] limited to § 3583(k)." Id. at 2383. Nonetheless, we can see arguments both ways. On the one hand, this case does not appear to raise a problem under Apprendi itself, which prohibits a court from using a judge-found fact to increase a defendant's sentence above the statutory maximum that would apply based on the jury's findings alone. 530 U.S. at 490, 120 S.Ct. 2348. The combination of Robinson's initial sentence (92 months) and his supervised-release revocation sentence (28 months) added up 120 months. Yet this total number did not exceed the maximum that Robinson could have received for either his original felon-in-possession conviction (120 months) or his original drug conviction (480 months)—let alone for both of them. 18 U.S.C. § 924(a)(2) (2006); 21 U.S.C. § 841(b)(1)(B) (2006). Even the Haymond plurality opined that a combined sentence like Robinson's that fell below the statutory maximum for the original conviction would not raise an Apprendi problem. 139 S. Ct. at 2384 (plurality opinion).
On the other hand, the correct result under the plurality's reasoning gets murkier if we consider Alleyne, which prohibits a court from finding a fact that increases a defendant's mandatory-minimum sentence beyond the statutory minimum that would apply based on the jury's conviction alone. 570 U.S. at 103, 133 S.Ct. 2151. Robinson's original drug conviction included a 5-year mandatory minimum. See 21 U.S.C. § 841(b)(1)(B) (2006). And § 3583(g) required only a mandatory "term of imprisonment" without specifying a length, so a single day might suffice. See United States v. Seighman, 966 F.3d 237, 244 (3d Cir. 2020). Robinson's original minimum (5 years) thus exceeded his minimum under § 3583(g) (1 day). This fact distinguishes Haymond because the defendant there faced no mandatory minimum under his original conviction and a 5-year minimum under § 3583(k). 139 S. Ct. at 2378 (plurality opinion). But perhaps the issue is not as simple as comparing the length of the original minimum with that of the later one. After all, Robinson had already served his 5-year minimum when the court decided to revoke his supervised release. In that sense, § 3583(g) imposed a mandatory minimum on top of (not in lieu of) the original minimum. A court seeing the question in this light might treat § 3583(g) as enhancing "the legally prescribed range of allowable sentences" because it changed Robinson's mandatory minimum from 5 years to 5 years and 1 day. Haymond, 139 S. Ct. at 2378 (plurality opinion) (quoting Alleyne, 570 U.S. at 115, 133 S.Ct. 2151).
Given these complexities, we opt not to decide how the plurality's Apprendi-based logic applies in this case. We instead hold that Justice Breyer issued the Haymond opinion that binds us here because it took the narrowest path to the judgment finding § 3583(k) unconstitutional. See Marks v. United States, 430 U.S. 188, 193, 97 S.Ct. 990, 51 L.Ed.2d 260 (1977). By doing so, we join the other courts to consider this question. See, e.g., United States v. Moore, 22 F.4th 1258, 1268 (11th Cir. 2022); United States v. Henderson, 998 F.3d 1071, 1076 (9th Cir. 2021); United States v. Salazar, 987 F.3d 1248, 1259 (10th Cir. 2021); United States v. Garner, 969 F.3d 550, 552 (5th Cir. 2020); Seighman, 966 F.3d at 242; United States v. Coston, 964 F.3d 289, 295-96 (4th Cir. 2020); United States v. Doka, 955 F.3d 290, 295-96 (2d Cir. 2020). Critically, moreover, Justice Breyer refused to "transplant" Apprendi and Alleyne to the supervised-release context. Haymond, 139 S. Ct. at 2385 (Breyer, J., concurring in the judgment). Rather, he invoked fact-intensive elements—"considered in combination"—to hold that the Constitution required a jury to find beyond a reasonable doubt that a defendant on supervised release committed a crime listed in § 3583(k). Id. at 2386.
When examined through the lens of Justice Breyer's narrower analysis, § 3583(g) does not contain the constitutional defects that plagued § 3583(k). Justice Breyer in Haymond pragmatically asked whether, considering the totality of the circumstances, § 3583(k) acted more like a "punishment for a new crime" (which would trigger the jury-trial right) or a traditional supervised-release sanction that merely sent a defendant to prison as a result of the original crime (which would not). Id. In this sense, Justice Breyer's concurrence resembles his dissents in the Apprendi line of cases—which have readily conceded that Congress might violate the Constitution if it allows a judge to find such a significant "sentencing fact" that this fact acts as the "tail" that "[wags] the dog of the substantive offense" reserved for the jury. Blakely v. Washington, 542 U.S. 296, 344, 124 S.Ct. 2531, 159 L.Ed.2d 403 (2004) (Breyer, J., dissenting) (citation omitted).
The same pragmatic assessment here, however, leaves no doubt that § 3583(g) looks more like a traditional supervised-release sanction than a penalty for a new crime. Most notably, § 3583(g) depends on the defendant's original crime in a way that § 3583(k) does not. It provides that any prison term may not "exceed the maximum term of imprisonment authorized under" § 3583(e)(3). That paragraph, in turn, states that "a defendant whose term is revoked under this paragraph may not be required to serve on any such revocation more than 5 years in prison . . . [for] a class A felony, more than 3 years in prison [for] a class B felony, more than 2 years in prison [for] a class C or D felony, or more than one year in any other case[.]" 18 U.S.C. § 3583(e)(3). It thus ties the maximum term to "the severity of the original crime of conviction, not the conduct that results in revocation." Haymond, 139 S. Ct. at 2386 (Breyer, J., concurring in the judgment). The more serious the original crime, the longer the maximum post-revocation prison term. Subsection 3583(k), by contrast, mandates the same sentencing range for everyone—no matter the severity of their original crimes. Its punishment thus focuses on the new crime in a more significant way than the punishment in § 3583(g), which looks more like "part of the penalty for the initial offense," Johnson, 529 U.S. at 700, 120 S.Ct. 1795.
The differences in allowable punishments across the two subsections confirm this point. By incorporating § 3583(e)'s sentencing caps, § 3583(g) allows a court to impose only "significantly restrained" sanctions. U.S.S.G. ch. 7, pt. A, intro. 3(b). Even if the defendant's original crime was a Class A felony, the court may sentence the defendant to no more than 5 years. This limited sentencing power reinforces that the sentence represents a sanction for a defendant's failure "to abide by the conditions of the court-ordered supervision," not a "punishment for any new criminal conduct[.]" Id. Subsection 3583(k), by comparison, states that a court should impose a prison term "without regard" to § 3583(e)(3)'s "exception" containing its sentencing caps. 18 U.S.C. § 3583(k). By requiring a mandatory-minimum sentence of 5 years (and by permitting a life sentence), § 3583(k) objectively appears designed to prosecute and punish the defendant for the new crime (but without the burdens of a jury trial and a reasonable-doubt standard of proof).
The two subsections also generally reach different types of violations. The four paragraphs requiring the revocation of supervised release in § 3583(g) focus primarily on whether defendants have violated their conditions of supervised release, not other federal offenses in the U.S. Code. The first paragraph asks if the defendant possessed drugs "in violation of the [supervised-release] condition" imposed by another subsection; the second asks if the defendant possessed a firearm in violation of federal law or "a condition of supervised release"; the third asks if the defendant refused to engage in drug testing "imposed as a condition of supervised release"; and the fourth asks if the defendant failed that testing. Id. § 3583(g)(1)-(4). Only a portion of one of these paragraphs turns on a violation of federal law (rather than a condition of supervised release). And two of the supervised-release violations (refusing or failing a drug test) are not even standalone federal offenses. In this way, too, § 3583(g) stands in contrast to § 3583(k). Congress wrote the latter provision to ask whether the defendant had committed a new federal offense, not a violation of a supervised-release condition. It thus applies "only" to conduct that qualifies as an independent federal crime. Haymond, 139 S. Ct. at 2386 (Breyer, J., concurring in the judgment).
For these reasons, the other circuit courts that have considered § 3583(g)'s constitutionality have all agreed that Haymond does not affect it. See United States v. Richards, 52 F.4th 879, 883-85 (9th Cir. 2022); United States v. Vickers, 2022 WL 2048486, at *3-4 (11th Cir. June 7, 2022) (per curiam); Garner, 969 F.3d at 552-53; Seighman, 966 F.3d at 243-44; Coston, 964 F.3d at 294-96; United States v. Ewing, 829 F. App'x 325, 330 (10th Cir. 2020); see also United States v. Wilson, 939 F.3d 929, 932-33 (8th Cir. 2019). We now join these courts. (Although Richards, Vickers, Seighman, Coston, and Ewing arose on plain-error review, we find their reasoning persuasive nonetheless.)
Robinson's responses do not change things. He points out that § 3583(g) (like § 3583(k)) makes supervised-release revocation mandatory if a defendant commits one of the supervised-release violations listed in the subsection. Yet Robinson identifies nothing to suggest that mandatory revocation alone makes an ensuing prison term more like a punishment for a new crime than a sanction for the original one. Some states, for example, have operated parole schemes that compelled the relevant administrative body to revoke parole if the defendant engaged in certain conduct, and we did not suggest that this nondiscretionary system transformed parole revocation into punishment for a new crime. See Sneed v. Donahue, 993 F.2d 1239, 1243-44 (6th Cir. 1993). Even the old federal parole system indicated that a parolee who possessed an illegal drug "shall have his parole revoked" for an undefined term. 18 U.S.C. § 4214(f) (1976) (emphasis added). For his part, moreover, Justice Breyer went out of his way in Haymond to emphasize that § 3583(k) eliminated a district court's discretion not just over whether to revoke supervised release but also over "how long" the defendant must serve in prison. 139 S. Ct. at 2386 (Breyer, J., concurring in the judgment). Conversely, § 3583(g) reserves the court's flexibility over the prison term's length. So Robinson's 28-month sentence arose from judicial discretion, not congressional command.
Robinson next argues that his specific supervised-release violations that triggered § 3583(g)—firearm and drug possession—also happen to qualify as independent felonies. Yet Justice Breyer looked to how § 3583(k) operated as a whole, not to specific applications. Id. Subsection 3583(k) explicitly incorporated a list of federal offenses. Id. Subsection 3583(g) does not. And Justice Breyer found that § 3583(k) violated the Constitution only by considering all his cited factors "in combination[.]" Id. Subsection 3583(g) lacks many of these factors.
Robinson lastly asserts that "experience" has shown that district courts broadly impose supervised release as new punishment for new conduct, not as an additional sanction for the initial offense. Appellant's Br. 23-24. But all the opinions in Haymond rejected this view by reiterating that supervised-release revocations are "part of the penalty for the initial offense." 139 S. Ct. at 2380 (plurality opinion) (quoting Johnson, 529 U.S. at 700, 120 S.Ct. 1795); id. at 2386 (Breyer, J., concurring in the judgment); id. at 2391 (Alito, J., dissenting). We must follow this binding authority until the Supreme Court instructs us otherwise. See State Oil Co. v. Khan, 522 U.S. 3, 20, 118 S.Ct. 275, 139 L.Ed.2d 199 (1997).
We affirm.