KETHLEDGE, J., delivered the opinion of the court in which GUY, J., joined. STRANCH; J., joined the opinion in part and the result in part. STRANCH, J. (pp. 893-97), delivered a separate opinion joining in Parts II.B and III of the majority opinion and concurring in the judgment only with respect to Part II.A.
OPINION
KETHLEDGE, Circuit Judge.In Fourth Amendment cases the Supreme Court has long recognized a distinction between the content of a communication and the information necessary to convey it. Content, per this distinction, is *884protected under the Fourth Amendment, but routing information is not. Here, Timothy Carpenter and Timothy Sanders were convicted of nine armed robberies in violation of the Hobbs Act. The government’s evidence at trial included business records from the defendants’ wireless carriers, showing that each man used his cellphone within a half-mile to two miles of several robberies during the times the robberies occurred. The defendants argue that the government’s collection of those records constituted a warrantless search in violation of the Fourth Amendment. In making that argument, however, the defendants elide both the distinction described above and the difference between GPS tracking and the far less precise locational information that the government obtained here. We reject the defendants’ Fourth Amendment argument along with numerous others, and affirm the district court’s judgment.
I.
In April 2011, police arrested four men suspected of committing a string of armed robberies at Radio Shacks and T-Mobile stores in and around Detroit. One of the men confessed that the group had robbed nine different stores in Michigan and Ohio between December 2010 and March 2011, supported by a shifting ensemble of 15 other men who served as getaway drivers and lookouts. The robber who confessed to the crimes gave the FBI his own cellphone number and the numbers of other participants; the FBI then reviewed his call records to identify still more numbers that he had called around the time of the robberies.
In May and June 2011, the FBI applied for three orders from magistrate judges to obtain “transactional records” from various wireless carriers for 16 different phone numbers. As part of those applications, the FBI recited that these records included “[a]ll subscriber information, toll records and call detail records including listed and unlisted numbers dialed or otherwise transmitted to and from [the] target telephones from December 1, 2010 to pres-entí,]” as well as “cell site information for the target telephones at call origination and at call termination for incoming and outgoing calls[.]” The FBI also stated that these records would “provide evidence that Timothy Sanders, Timothy Carpenter and other known and unknown individuals” had violated the Hobbs Act, 18 U.S.C. § 1951. The magistrates granted the applications pursuant to the Stored Communications Act, under which the government may require the disclosure of certain telecommunications records when “specific and articulable facts show[ ] that there are reasonable grounds to believe that the contents of a wire or electronic communication, or the records or other information sought, are relevant and material to an ongoing criminal investigation.” 18 U.S.C. § 2703(d).
The government later charged Carpenter with six counts, and Sanders with two, of aiding and abetting robbery that affected interstate commerce, in violation of the Hobbs Act, and aiding and abetting the use or carriage of a firearm during a federal crime of violence. See 18 U.S.C. §§ 924(c), 1951(a). Before trial, Carpenter and Sanders moved to suppress the government’s cell-site evidence on Fourth Amendment grounds, arguing that the records could be seized only with a warrant supported by probable cause. The district court denied the motion.
At trial, seven accomplices testified that Carpenter organized most of the robberies and often supplied the guns. They also testified that Carpenter and his half-brother Sanders had served as lookouts during the robberies. According to these wit*885nesses, Carpenter ’ typically waited in a stolen car across the street from the targeted store. At his signal, the robbers entered the store, brandished their guns, herded customers and employees to the back, and ordered the employees to fill the robbers’ bags with new smartphones. After each robbery, the team met nearby to dispose of the guns and getaway vehicle and to sell the stolen phones.'
FBI agent Christopher Hess offered expert testimony regarding the cell-site data provided by Carpenter’s and Sanders’s wireless carriers, MetroPCS' and T-Mobile. Hess explained that cellphones work by establishing a radio connection with nearby cell towers (or “cell sites”); that phones are constantly searching for the strongest signal from those towers; and that individual towers project different signals in each direction or “sector,” so that a cellphone located on the north side of a cell tower will use a different signal than a cellphone located on the south side of the same tower. Hess said that cell towers are typically spaced widely in rural areas, where a tower’s coverage might reach as far as 20 miles. In an urban area like Detroit, however, each cell site covers “typically anywhere from a half-mile to two miles.” He testified that wireless carriers typically log and store certain call-detail records of their customers’ calls, including the date, time, and length of each call; the phone numbers engaged on the call; and the cell sites where the call began and ended.
With the cell-site data provided by Carpenter’s and Sanders’s wireless carriers, Hess created maps ■ showing that Carpenter’s and Sanders’s phones were within a half-mile to two miles of the location of each of the robberies around the time the robberies happened. Hess used Me-troPCS call-detail records, for example, to show that Carpenter was within that proximity of a Detroit Radio Shack that was robbed around 10:35 a.m. bn December 13, 2010. Specifically, MetroPCS records showed that at 10:24 a.m. Carpenter’s phone received a call that lasted about four minutes. At the start and end of the call, Carpenter’s phone drew its signal from MetroPCS tower 173, sectors 1 and 2, located southwest of the store and whose signals point north-northeast. After the robbery, Carpenter placed an eight-minute call originating at tower 145, sector 3, located northeast of the ■ store, its signal pointing southwest; when the call ended, Carpenter’s phone was receiving its signal from tower 164, sector 1, 'alongside Interstate 94, north of the Radio Shack. See Carpenter App’x at 11. Hess provided similar analysis concerning the locations of Carpenter’s and Sanders’s phones at the time of a December 18, 2010 robbery in Detroit;- a March 4, 2011 robbery in Warren,-Ohio; and an April 5, 2011 robbery in Detroit. See Carpenter App’x át 12-15.
The jury convicted Carpenter and Sanders on all of the Hobbs Act counts and convicted Carpenter on all but one of the § 924(c) gun counts. Carpenter’s convictions on the § 924(c) counts subjected him to four mandatory-minimum prison sentences of 25 years, each to be served consecutively, leaving him with a Sentencing Guidelines range of 1,395 to 1,428 months’ imprisonment. The district court sentenced Carpenter to 1,395 months’ imprisonment and Sanders to 170 months’ imprisonment. Carpenter and Sanders now appeal their convictions and sentences.
II.
A.
Carpenter and Sanders challenge the district court’s denial of their-motion to exclude their cell-site data from the evidence at trial. Those data themselves *886took the form of business records created and maintained by the defendants’ wireless carriers: when the defendants made or received calls with their cellphones, the phones sent a signal to the nearest cell-tower for the duration of the call; the providers then made records, for billing and other business purposes, showing which towers each defendant’s phone had signaled during each call The government thereafter collected those records, and hence these cell-site data, for a range of dates (127 days of records for Carpenter, 88 days for Sanders) encompassing the robberies at issue here. The government did so pursuant to a court order issued under . the Stored Communications Act, which .required the government to shpw “reasonable grounds” for believing that the records were “relevant and material to an ongoing , investigation.”,. 18 U.S.C. § 2703(d).. Carpenter,and Sanders argue that the Fourth Amendment instead required the government to obtain a. search warrant, pursuant to a showing of probable cause, before collecting the data. The district court rejected that argument, holding that the government’s collection of cell-site records created and maintained by defendants’ wireless carriers was not a search under the Fourth Amendment. We review the district court’s decision de novo. See United States v. Lee, 793 F.3d 680, 684 (6th Cir.2015).
The Fourth Amendment protects “[t]he right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures[.]” U.S. Const. amend. IV. “[F]or most of our history the Fourth Amendment was understood to embody a particular concern for government trespass upon the areas (‘persons, houses, papers, and effects’) it enumerates.” United States v. Jones, — U.S. -, 132 S.Ct. 945, 950, 181 L.Ed.2d 911 (2012). Government trespasses upon those areas normally count as a search; Id. In Katz v. United States, 389 U.S. 347, 88 S.Ct. 507, 19 L.Ed.2d 576 (1967), however, the Supreme Court moved beyond a property-based understanding of the Fourth Amendment, to protect certain expectations of privacy as well. To fall within these protections, an expectation of privacy must satisfy “a twofold requirement”: first, the .person asserting it must “have exhibited an actual (subjective) expectation of privacy”; and second, that expectation must “be one that society is prepared to recognize as ‘reasonable.’” Katz v. United States, 389 U.S. 347, 361, 88 S.Ct. 507, 19 L.Ed.2d 576 (1967) (Harlan, J., concurring). When an expectation of privacy meets both of these requirements, government action that “invade[s]” the expectation normally counts as a search. Smith v. Maryland, 442 U.S. 735, 740, 99 S.Ct. 2577, 61 L.Ed.2d 220 (1979).
This' case involves an asserted privacy interest' in information related to personal communications. 'As to that kind of information, the federal courts have long recognized a core distinction: although the content of personal communications is private, the information necessary to get those communications from point A to point B is not. For example, in Ex parte Jackson, 96 U.S. 727, 733, 24 L.Ed. 877 (1878), the Court held that postal inspectors needed a search warrant to open letters and packages, but that the “outward form and weight” of those mailings — including, of course, the recipient’s name and physical address — was not constitutionally protected. Id. That was true even though that information could sometimes bring embarrassment: “In a small village, for instance, a young gentleman may not altogether desire that all the loungers around the store which contains the Post-office shall be joking about the fair object of his affections.” *887Our Letters, N.Y. Times, Dec. 12, 1872, at 4.
In the twentieth century, the telephone call joined the letter as a standard form of communication. The law eventually followed, recognizing that police cannot eavesdrop on a phone call — even a phone call placed from a public phone booth — without a warrant. See Katz, 389 U.S. at 352-55, 88 S.Ct. 507. But again the Supreme Court distinguished between a communication’s content and the information .necessary to send it. In Katz, the Court held that “[t]he .Government’s activities in electronically listening to and recording the petitioner’s icords ” was a search under the Fourth Amendment. Id. at 353, 88 S.Ct. 507 (emphasis added). But in Smith, the Court held that the police’s installation of a pen register — a device that tracked the phone numbers a person dialed from his home phone — was not a search because the caller could not reasonably expect those numbers to remain private. “Although [the caller’s] conduct may have been calculated to. keep the contents of his conversation private, his conduct was not and could not have been calculated to preserve the privacy of the number he dialed.” Smith, 442 U.S. at 743, 99 S.Ct. 2577 (emphasis in original).
Today, the same distinction applies-to internet communications. The Fourth Amendment protects the content of the modern-day letter,, the email. See United States v. Warshak, 631 F.3d 266, 288 (6th Cir.2010). But courts ■ have not (yet, at least) extended those protections to the internet analogue to envelope markings, namely the metadata used to route internet communications, like sender and recipient addresses on an email, or IP addresses. See, e.g., United States v. Christie, 624 F.3d 558, 574 (3d Cir.2010); United States v. Perrine, 518 F.3d 1196, 1204-05 (10th Cir.2008); United States v. Forrester, 512 F.3d 500, 510 (9th Cir.2008).
The business records here fall on the unprotected side of this line. Those records say nothing about the -content of any calls.- Instéad the records include routing information, which the wifeless providers gathered in the ordinary course of business. Carriers necessarily track their customers’ phones across different cell-site sectors to connect and maintain their customers’ calls. And carriers keep records of these data to. find weak spots in their network árid to determine whether roaming charges apply, among other purposes. Thus, the cell-site data — like mailing addresses, phone numbers, and IP addresses — are information that facilitate personal communications, rather than part of the content of those communications themselves. The government’s collection of business records' containing these data therefore is not a search.
The Supreme Court’s decision in Smith confirms the point. At the outset, the Court observed that Smith could not claim that “his ‘property’ was invaded” by the State’s actions, which meant he could not claim any property-based protection under the Fourth Amendment. And as to privacy, the Court hewed precisely to the content-focused distinction that we make here. 442 U.S. at 741, 99 S.Ct. 2577. The Court emphasized (literally) that the State’s pen register did “not acquire the contents of communications.” Id. (emphasis in original). Instead, the Court observed, the phone numbers acquired by the State had been dialed “as a means of establishing communication.” Id. Moreover, the Court, pointedly refused to adopt anything like a “least-sophisticated phone user” (to paraphrase the Fair Debt Collection-Practices. Act) standard in determining whether phone- users know that they convey that information to the phone company: “All *888telephone users realize that they must ‘convey’ phone numbers to the telephone company, since it is through telephone company switching equipment that their calls are completed.” Id. at 742, 99 S.Ct. 2577. The Court likewise charged “telephone users” with knowledge that “the phone company has facilities for recording” numerical information and that “the phone company does in fact record this information for a variety of legitimate business purposes.” Id. at 743, 99 S.Ct. 2577. Thus, the Court held, Smith “voluntarily conveyed numerical information to the telephone company and ‘exposed’ that information to its equipment in the ordinary course of business.” 442 U.S. at 744, 99 S.Ct. 2577. Hence the numerical information was not protected under the Fourth Amendment.
The same things are true as to the locational information here. The defendants of course lack any property interest in cell-site records created and maintained by their wireless carriers. More to the point, when the government obtained those records, it did “not acquire the contents of communications.” Id. at 741, 99 S.Ct. 2577. Instead, the defendants’ cellphones signaled the nearest cell towers — thereby giving rise to the data obtained by the government here — solely “as a means of establishing communication.” Id. Moreover, any cellphone user who has seen her phone’s signal strength fluctuate must know that, when she places or receives a call, her phone “exposes” its location to the nearest cell tower and thus to the company that operates the tower. Accord United States v. Davis, 785 F.3d 498, 511 (11th Cir.2015) (en banc); In re Application for Historical Cell Site Data, 724 F.3d 600, 614 (5th Cir.2013). And any cellphone user who has paid “roaming” (ie., out-of-network) charges — or even cellphone users who have not — should know that wireless carriers have “facilities for recording” locational information and that “the phone company does in fact record this information for a variety of legitimate business purposes.” Id. at 743, 99 S.Ct. 2577. Thus, for the same reasons that Smith had no expectation of privacy in the numerical information at issue there, the defendants have no such expectation in the locational information here. On this point, Smith is binding precedent.
The defendants and their amicus, the American Civil Liberties Union, argue that Jones liberates us to hold otherwise. In Jones, five Justices (though not the Court in its majority opinion) agreed that “longer term GPS monitoring in government investigations of most offenses impinges on expectations of privacy.” 132 S.Ct. at 964 (Alito, J., concurring in the judgment); id. at 955 (Sotomayor, J., concurring) (same). But there are at least two problems with the defendants’ argument as made here. The first is that the government action in this case is very different from the government action in Jones. That distinction matters: in applying Katz, “it is important to begin by specifying precisely the nature of the state activity that is challenged.” Smith, 442 U.S. at 741, 99 S.Ct. 2577 (emphasis added). Whether a defendant had a legitimate expectation of privacy in certain information depends in part on what the government did to get it. A phone conversation is private when overheard by means of a wiretap; but that same conversation is unprotected if an agent is forced to overhear it while seated on a Delta flight. Similarly, information that is not particularly sensitive — say, the color of a suspect’s vehicle — might be protected if government agents broke into the suspect’s garage to get it. Yet information that is highly sensitive — say, all of a suspect’s credit-charges over a three-month period — is not protected if the government *889gets that information through business records obtained per a subpoena. See United States v. Phibbs, 999 F.2d 1053, 1077-78 (6th Cir.1993).
This case involves business records obtained from a third party, which can only diminish the defendants’ expectation of privacy in the information those records contain. See United States v. Miller, 425 U.S. 435, 443, 96 S.Ct. 1619, 48 L.Ed.2d 71 (1976); Phibbs, 999 F.2d at 1077-78. Jones, in contrast, lands near the other end of the spectrum: there, government agents secretly attached a GPS device to the underside of Jones’s vehicle and then monitored his movements continuously for four weeks. That sort of government intrusion presents one set of Fourth Amendment questions; government collection of business records presents another. And the question presented here, as shown above, is answered by Smith.
The second problem with the defendants’ reliance on-Jones is that — unlike Jones — this is riot a GPS-tracking case. GPS devices are accurate within about 50 feet, which is accurate enough to show that the target is located within an individual building. Data with that kind of accuracy might tell a story of trips to “the strip club, the criminal defense attorney, the by-the-hour motel, the union meeting, the mosque, synagogue or church, the gay bar and on and on[.]” Jones, 132 S.Ct. at 956 (Sotomayor, J., concurring) (internal quotation marks omitted). But here the cell-site data cannot tell that story. Instead, per the undisputed testimony at trial, the data could do no better than locate the defendants’ cellphones within a 120- (or sometimes 60-) degree radial wedge extending between one-half mile and two miles in length. Which is to say the locational data here are accurate within a 3.5 million square-foot to 100 million square-foot area — as much as-12,500 times less accurate than the GPS data in Jones. And cell phone locational data are even less precise in suburban and rural settings. Areas of this scale might encompass bridal stores and Bass Pro Shops, gay bars and straight ones, a Methodist church and the local mosque. The ACLU responds that so-called “femtocells” can provide service (and thus identify a phone’s location) within areas as small as ten meters. But. our task is to decide this case, not hypothetical ones; and in this case there are no femto-cells to be found. The defendants’ argument is without merit.
The defendants similarly rely on Riley v. California, — U.S.-, 134 S.Ct. 2473, 2485, 189 L.Ed.2d 430 (2014), where the Court held the government may not access a smartphone’s internal data — or, one might say, its contents — without a warrant. But the'Court’s rationale was that smartphones typically store vast amounts of information about their users — vastly more, of course, than whether the user happens'to be located within a two-mile radial wedge. Riley only illustrates the core distinction we make here.
Some other points bear mention. One is that Congress has specifically legislated on the .question before us today, and in doing so has struck the balance reflected in the. Stored Communications Act. The Act stakes out a middle ground between full Fourth Amendment protection and no protection at all, requiring that the government show “reasonable grounds” but not “probable cause” to obtain the cell-site data at issue here. See 18 U.S.C. §. 2703(d). The defendants and’ the ACLU effectively ask us to declare that balance unconstitutional. There .is considerable irony in that request. The Katz standard asks whether the defendants’ asserted expectation of privacy “is ‘one that society is prepared to recognize as reasonable[.]’” Smith, 442 U.S. At 740, 99 S.Ct. 2577 *890(quoting Katz, 389 U.S. at 361, 88 S.Ct. 507). Here, one might say that society itself—in the form of its elected representatives in Congress—has already struck a balance that it thinks reasonable. That is not to say that courts should defer to Congress’s judgment on constitutional questions. But when the question itself turns on society’s views, and society has in a meaningful way already expressed them, judges should bring a certain humility to the task of deciding whether those views are reasonable—lest judges “confuse their own expectations of privacy,” Jones, 132 S.Ct. at 962 (Alito, J., concurring), with those that every reasonable person must hold.
A second point is related. Constitutional judgments typically rest in part on a set of empirical assumptions. When those assumptions concern subjects that judges know well—say, traffic , stops— courts are well-equipped to make judgments , that strike a reasonable . balance among the conipeting interests at stake. See Kerr, The Fourth Amendment and New Technologies: Constitutional Myths and the Case For Caution, 102 Mich. L.Rev. 801, 863 (2004). But sometimes new ' technologies—say, the latest iterations of smartphones or social media— evolve at rates more common to superbugs than to large mammals. In those situations judges are less good at evaluating the empirical • assumptions that underlie their constitutional judgments. Indeed the answers- to those empirical questions , might change as quickly as the technology itself does. With regard to the Katz test in particular, for example, “[djramatic technological change may lead to periods in which popular expectations are in flux and may ultimately produce significant changes in popular attitudes.” Jones, 132 S.Ct. at 962 (Alito, J., concurring); ' Congress is usually better equipped than courts are to answer the empirical questions that such technologies present. Thus, “[wjhen technologies are new and their impact remains uncertain, statutory rules governing law enforcement powers will tend to be more sophisticated, comprehensive, forward-thinking, and flexible than rules created by the judicial branch.” Kerr, 102 Mich. L.Rev. at 859-60. These concerns favor leaving undisturbed the Congressional judgment here.
In sum, we hold that the government’s collection of business records containing cell-site data was not a search under the Fourth Amendment.
B.
Sanders argues that the district court should have suppressed the government’s cell-site evidence for another reason, namely that (in his view) the government’s applications to obtain the cell-site records failed to show “reasonable grounds” for believing that the records were “relevant and material to an ongoing criminal investigation.” 18 U.S.C. § 2703(d). There are several problems with that argument, but -the simplest is that suppression of evidence is not among the remedies available under the Stored Communications Act. Quite the contrary: the statute identifies a handful of civil remedies, including “damages” and “equitable or declaratory relief,” .18 U.S.C. § 2707(b), and provides that those “are the only judicial remedies and sanctions for nonconstitutional violations of this chapter.” 18 U.S.C. § 2708; see United States v. Guerrero, 768 F.3d 351, 358 (5th Cir.2014). The relief that Sanders seeks is therefore unavailable under the Act. See United States v. Abdi, 463 F.3d 547, 556 (6th Cir.2006),
III.
A.
Carpenter argues that the district court erred when it denied Carpenter’s *891motion for acquittal for lack of venue over counts seven and eight.. Those counts charged Carpenter with aiding and abetting a Hobbs Act robbery in Warren, Ohio, and with aiding and abetting the use of a firearm in connection with that robbery. See 18 U.S.C. §§ 924(c), 1951(a). We review the district court’s decision de novo. See United States v. Kuehne, 547 F.3d 667, 677 (6th Cir.2008).
Carpenter was prosecuted in the Eastern District of Michigan. Venue was proper there so long as a rational trier of fact, viewing the evidence in the light most favorable to the government, could find by a preponderance of the evidence that any of Carpenter’s accessorial acts, or the underlying crime itself, occurred in the Eastern District of Michigan. Relatedly, “[w]here venue is appropriate for the underlying crime of violence, so too it is for the § 924(c)(1) offense.” United States v. Rodriguez-Moreno, 526 U.S. 275, 282, 119 S.Ct. 1239, 143 L.Ed.2d 388 (1999).
Here, Carpenter’s accomplices testified that, while in the Eastern District of Michigan, Carpenter recruited the robbers for the Warren robbery, described for them the robbery’s general scheme, and from there drove them to Ohio. Two of these witnesses also testified that, while in Michigan, Carpenter made arrangements to have another accomplice supply the robbers with a gun when they got to Warren. A reasonable trier of fact could credit this testimony, and conclude that much of Carpenter’s conduct in abetting both the Warren robbery and the use of a firearm during it took place in the Eastern District of Michigan. The district court correctly denied Carpenter’s motion for acquittal on counts seven and eight.
B.
Carpenter argues that the district court should have allowed him to use a report prepared by FBI Special Agent Vicente Ruiz to refresh the memory of government witness Adriane Foster on cross-examination. We review that evi-dentiary' ruling for an abuse of discretion; See United States v. Morales, 687 F.3d 697, 701 (6th Cir.2012).
At trial, Carpenter’s counsel cross-examined Foster — an accomplice of Carpenter — about Foster’s past statements to Agent Ruiz. Foster testified that he told Ruiz that Carpenter had provided Foster with advance information about the robbery in Warren. According to Ruiz’s written summary of the interview, however, Foster told Ruiz that Sanders, not Carpenter, had provided Foster with advance information about the robbery. Carpenter’s counsel sought to introduce Ruiz’s report to “refresh [Foster’s] memory” of the interview.
A document may be used to refresh a witness’s memory only after his memory has been “exhausted.” Rush v. Ill. Cent. R.R. Co., 399 F.3d 705, 716 (6th Cir.2005). Here, Foster seemed to have no trouble remembering his conversation with Agent Ruiz. Foster repeatedly testified that he did remember telling Ruiz that Timothy Carpenter — not Timothy Sanders — had told him about the plans for the Warren robbery. Carpenter’s counsel then asked Foster whether he remembered “saying something different” to Ruiz. Foster said that he did not. That answer did not show that Foster’s memory needed refreshing; it showed that Foster disagreed with Carpenter about what Foster had told Ruiz. What Carpenter actually sought to do with the report was not refresh Ruiz’s memory, but impeach his testimony. The district court did not abuse its discretion in ruling that Carpenter could not use the report for that purpose.
*892To the same end, Carpenter argues that the district court should have allowed him to introduce Ruiz’s report as extrinsic evidence of a prior inconsistent statement under Federal Rule of Evidence 613(b). But an FBI agent’s written summary of an interview with a declarant cannot be used to impeach the declarant’s later testimony unless the declarant has attested to the report’s accuracy. See United States v. Barile, 286 F.3d 749, 757 (4th Cir.2002); United States v. Schoenborn, 4 F.3d 1424, 1429 & n. 3 (7th Cir.1993). Foster has not done that here; to the contrary, Foster testified that Ruiz’s report would have been wrong if it said that Sanders rather than Carpenter had told him about the plans for the Warren robbery. The district court thus correctly barred the report’s introduction at trial.
C.
1.
Carpenter’s remaining argument is that his 1,395-month sentence is so disproportionate to his crimes as to violate the Eighth Amendment’s prohibition on cruel and unusual punishment. He also argues that his mandatory-minimum sentences for his § 924(c) convictions violate the constitutional separation of powers. We consider both issues de novo. See United States v. Kelsor, 665 F.3d 684, 701 (6th Cir.2011).
“[O]nly an extreme disparity between crime and sentence offends the Eighth Amendment.” United States v. Odeneal, 517 F.3d 406, 414 (6th Cir.2008). In Solem v. Helm, 463 U.S. 277, 103 S.Ct. 3001, 77 L.Ed.2d 637 (1983), the Supreme Court held that the Eighth Amendment prohibited a state court from sentencing to life imprisonment without parole a recidivist criminal who wrote a bad check for $100. But in Ewing v. California, 538 U.S. 11, 123 S.Ct. 1179, 155 L.Ed.2d 108 (2003), the Supreme Court rejected the Eighth Amendment claim of a defendant who was sentenced to 25 years to life for stealing several golf clubs. 538 U.S. at 28-30,123 S.Ct. 1179.
Carpenter has a long criminal history. In this case, as the district court observed, Carpenter organized and led several “very violent” robberies that put his victims “in extreme danger[.]” Meanwhile, in other armed-robbery cases, we have already held that sentences even longer than Carpenter’s were constitutionally permissible. See United States v. Clark, 634 F.3d 874, 877-78 (6th Cir.2011) (2,269 months); United States v. Watkins, 509 F.3d 277, 282 (6th Cir.2007) (1,772 months). Carpenter’s sentence does not violate the Eighth Amendment.
Nor do his mandatory-minimum sentences violate the constitutional separation of powers. “Congress has the power to define criminal punishments without giving the courts any sentencing discretion.” Chapman v. United States, 500 U.S. 453, 467, 111 S.Ct. 1919, 114 L.Ed.2d 524 (1991). Carpenter acknowledges that we have “flatly rejected” his argument in other cases. Carpenter Br. at 54; see, e.g., United States v. Cecil, 615 F.3d 678, 696 (6th Cir.2010). This case is no different.
2.
Sanders challenges • his sentence on non-constitutional grounds, arguing that the district court misapplied the Sentencing Guidelines and that his sentence is “greater than necessary” to accomplish the remedial objectives of incarceration. See 18 U.S.C. § 3553. We review for clear error the district court’s factual findings in support of Sanders’s sentence, and review the sentence itself for an abuse of discre*893tion. See United States v. Randolph, 794 F.3d 602, 614 (6th Cir.2015).
Sanders argues first that the district court incorrectly applied sentencing enhancements for brandishing or possessing a firearm in furtherance of robbery, and for physically restraining a person in furtherance of a robbery. See U.S.S.G. §§ 2B3.1(b)(2)(C), (b)(4)(B). Sanders himself need not have committed those acts in order for the enhancements to apply; rather, he need only have known it was “reasonably probable” that a co-participant would commit them.’ 'See United States v. Woods, 604 F.3d 286, 291 (6th Cir.2010).
That standard is met here. During the January 7, 2011 robbery, Sanders’s accomplice Juston Young returned to the getaway car with gun in hand. Thus, when Sanders teamed up.with Young and others for another robbery on March 4, Sanders could have easily foreseen that Young would brandish • a firearm in the course of the crime — as in fact Young did. The district court did not clearly err in finding that the firearm enhancement applied to Sanders.
Nor did the court err in finding that Sanders could foresee that Young would physically restrain someone during the March 4 robbery. As a general matter, an accomplice to robbery should foresee that robbery likely entails physical restraint or worse. See U.S. Sentencing Guidelines Manual § 1B1.3 cmt.' n; 2 (2012). And Sanders knew specifically that the plan for that robbery was for the robbers to move customers and employees to the back of the store. The physical-restraint enhancement was therefore proper. That leaves an enhancement • for brandishing a firearm during the January 7 robbery. But that enhancement had no effect on Sanders’s Guidelines range: the offense, level for the March 4 robbery was higher than the offense level for the January 7 robbery, even with the brandishing enhancement; and the offense level for the March 4 robbery, not the January 7 one, thus determined his total offense level under the Guidelines. Any error as to the brandishing enhancement for the January 7 robbery was therefore harmless. See United States v. Jeross, 521 F.3d 562, 574-76 (6th Cir.2008).
Finally, the district court acted within its discretion in sentencing Sanders to 170 months’ imprisonment, which fell squarely within his Guidelines range of 151 to 188 months. Within-Guidelines sentences are presumptively reasonable in this circuit. See United States v. Kamper, 748 F.3d .728, 739-40 (6th Cir.2014). Moreover, the district court considered and rejected the arguments that Sanders raised at his sentencing hearing, and otherwise properly determined that the sentence was appropriate in light of 18 U.S.C. § 3553(a). The court did not abuse its discretion.
The judgments in both cases are affirmed.