OPINION

SUTTON, Circuit Judge.

Calvin Reid, a man in his late forties, took J.H., a girl in her early teens, across state lines to have sex with her. We affirm his resulting conviction under the Mann Act and his sentence.

I.

J.H. has known Calvin Reid, an acquaintance of her family, since early childhood. In 2011, Reid (then forty-eight) began to send J.H. (then thirteen) flirtatious, then suggestive, then amatory text messages and phone calls. He led her to think of him as her boyfriend, and he eventually had sex with her. At first these sexual encounters happened in Memphis, Tennessee, where Reid and J.H. lived. But in November 2011, Reid took J.H. to a hotel across the Mississippi border to have sex with her there.

A few weeks later, J.H. told Reid that she wanted to run away from home. Reid encouraged her to leave, telling her he would go with her. One day, he picked up J.H. on her way to school and, Humbert Humbert-like, headed west, having sex along the way. Once they reached Las Vegas, J.H. left Reid and called her family, who had been searching for her. A friend of the family retrieved J.H. from Nevada and brought her home.

The United States charged Reid with violating the Mann Act, which prohibits knowingly transporting a minor in interstate commerce with intent that the minor engage in illegal sexual activity. 18 U.S.C. § 2423(a). The jury convicted Reid of two counts of the crime, one for the trip to Mississippi, the other for the trip to Nevada. The district court sentenced Reid to 198 months of imprisonment.

II.

On appeal, Reid targets four errors: in the selection of the jury that convicted him; in the admission of evidence at his trial; in the application of an enhancement at his sentencing; and in the calculation of his criminal history score.

Jury selection. In a felony case like this one, the Federal Rules of Criminal Procedure give defendants ten peremptory challenges, what amounts to ten (largely) unrestricted opportunities to strike potential jurors. See Fed.R.Crim.P. 24(b)(2). Reid claims that the trial court inadvertently gave him only nine. The mistake arose after Reid challenged Kevin Britt for cause. The court excused Britt, but the judge inadvertently recorded the removal as peremptory. Reid learned of the judge's mistake some time later, when the court informed him that he had used nine peremptory challenges and turned aside his claim that he had used only eight.

Reid is right about one thing: The trial court erred in classifying the Britt challenge as peremptory. When Britt came up, Reid made a "[m]otion for cause." R. 109 at 161. The court repeated, "[m]otion for cause." Id. After Britt answered a few more questions, the prosecutor said, "No objection to challenge for cause," and the court excused the juror. Id. at 164. In ruling that Reid challenged Britt peremptorily, the trial court thus erred.

The government agrees that the trial court misclassified Reid's challenge. Yet it initially maintains that the court nonetheless respected Criminal Rule 24. How can that be? The rule guarantees a defendant ten peremptory challenges. When a judge counts a for-cause challenge as peremptory, the defendant gets only nine peremptory challenges. Nine is fewer than ten.

The government's argument rests on the premise that, despite the trial court's mistake, Reid got what the rules are designed to give him: "a fair and impartial jury." United States Br. at 23. The premise overlooks another purpose, some would say the main purpose, of peremptory challenges: giving the defendant a jury that he considers fair and impartial. Worse, the argument assumes that a court follows the law so long as its actions comport with the law's purpose. Laws sometimes tell judges not only where to go but also how to get there, and the path deserves as much respect as the destination. Respect for the path marked by Rule 24 requires giving the defendant ten peremptory challenges.

United States v. Martinez-Salazar does not say otherwise. 528 U.S. 304, 120 S.Ct. 774, 145 L.Ed.2d 792 (2000). Martinez-Salazar held that no violation of Criminal Rule 24 occurs when a defendant uses a peremptory challenge to cure a trial court's erroneous denial of a challenge for cause. Id. at 315, 120 S.Ct. 774. In that situation, the defendant does not "lose" a peremptory challenge; he merely chooses to use one curatively. Id. at 315-16, 120 S.Ct. 774. (By the way, he has another choice: He could allow the juror to sit, await conviction, and raise a juror bias claim on appeal. Id. at 315, 120 S.Ct. 774.) Here, by contrast, the defendant did lose a peremptory challenge. He never decided to use a peremptory challenge against Britt, but the trial court said he used one anyway.

All of this, however, does not require reversal of Reid's conviction. The trial court's mistake caused Reid no harm, see Fed.R.Crim.P. 52, a reality confirmed by the fact that Reid did not exhaust his stock of peremptory challenges. At the end of jury selection, he had one or two peremptory challenges left — one by the court's tally, two by the right tally.

Not just math, but common sense too, proves the harmlessness of this error. Taking away a peremptory challenge often hurts the defendant, because it leads to the seating of a juror whom the defendant would have struck with the wrongfully withheld peremptory. Yet when a defendant does not use all of his peremptory challenges, we know that the defendant was satisfied with everyone on his jury. Had he not been satisfied with someone, he would have used the last challenge. Far from forcing an objectionable juror on Reid, this mistake had only one effect: Reid had one challenge in his pocket rather than two. That does not count as harm. A defendant cannot cash in extra peremptory challenges at the end of a trial or take them home like leftovers from a restaurant. Court after court agrees. It makes no difference whether a defendant gets nine or ten peremptories if he wants to use only eight. See, e.g., The Anarchists' Case, 123 U.S. 131, 168, 8 S.Ct. 21, 22, 31 L.Ed. 80 (1887); Hopt v. People, 120 U.S. 430, 436, 7 S.Ct. 614, 30 L.Ed. 708 (1887); United States v. Torres, 960 F.2d 226, 228 (1st Cir.1992) (Breyer, C.J.); United States v. Hardy, 941 F.2d 893, 897 (9th Cir.1991). No authority runs the other way.

A miscount, to be sure, might in some situations affect the defendant's jury selection strategy. A defendant who believes he has only one challenge left might for instance exercise more caution than a defendant who knows he has two. But that speculative possibility does not suffice to reverse the conviction, because this type of effect on jury selection strategy does not amount to an impairment of "substantial rights," Fed.R.Crim.P. 52(a). For one thing, the Supreme Court has taken the position that a defendant in Reid's position suffers "no injury." Hopt, 120 U.S. at 436, 7 S.Ct. 614 (emphasis added). The Court indeed "entertain[s] no doubt" that "no injury is done [to this kind of] defendant." Anarchists, 123 U.S. at 168 (emphasis added). The Court's categorical position makes it difficult to maintain that defendants like Reid suffer some kind of injury after all.

For another, the Supreme Court has let a jury's verdict stand even though a juror inaccurately answered a question during voir dire and even though the inaccuracy affected the litigant's exercise of peremptory challenges. McDonough Power Equip., Inc. v. Greenwood, 464 U.S. 548, 104 S.Ct. 845, 78 L.Ed.2d 663 (1984). The Court reasoned that, although the inaccuracy in some sense harmed the litigant, the harm was too meager to justify a fresh trial. "A trial represents an important investment of private and social resources," and in circumstances like these "it ill serves the important end of finality to wipe the slate clean simply to recreate the peremptory challenge process." Id. at 555, 104 S.Ct. 845. Although Greenwood involved a civil trial, its reasoning applies in this setting too. The civil and criminal harmless error rules after all spring from the same statute, use more or less the same language, and in general require courts to apply the same standard. See O'Neal v. McAninch, 513 U.S. 432, 441, 115 S.Ct. 992, 130 L.Ed.2d 947 (1995). Here too, a mistake might affect how a defendant thinks about exercising peremptory challenges. And here too, the mistake causes too slender and too speculative a disadvantage to justify a reversal.

Trying to sidestep these points, Reid claims that prejudice has nothing to do with it. He is doubly wrong. In the first place, in pushing the point, he relies on cases that apply automatic reversal to some constitutional mistakes. See Rivera v. Illinois, 556 U.S. 148, 161, 129 S.Ct. 1446, 173 L.Ed.2d 320 (2009). But taking away one of a defendant's peremptory challenges violates the Criminal Rules, not the Constitution. Id. at 157, 129 S.Ct. 1446. In the second place, some constitutional errors trigger automatic reversal only if, by infecting "the entire trial," they produce "consequences that are necessarily unquantifiable and indeterminate." United States v. Gonzalez-Lopez, 548 U.S. 140, 150, 126 S.Ct. 2557, 165 L.Ed.2d 409 (2006). Yet the consequences of the mistake here are neither unquantifiable nor indeterminate; they are essentially zero.

United States v. McFerron addressed a different situation. 163 F.3d 952 (6th Cir. 1998). It automatically reversed a trial court that improperly invoked Batson v. Kentucky, 476 U.S. 79, 106 S.Ct. 1712, 90 L.Ed.2d 69 (1986), to seat prospective jurors whom the defendant wanted to remove. 163 F.3d at 955-56. There, however, someone wrongly ended up on the jury. Because we had no way to assess the effects of a wrongfully seated juror's participation — in view of the secret deliberation of juries — automatic reversal made sense. Here, by contrast, nobody wrongly ended up on the jury.

Reid also seeks refuge in Swain v. Alabama, which says that "denial or impairment of the right [to use peremptory challenges] is reversible error without a showing of prejudice." 380 U.S. 202, 219, 85 S.Ct. 824, 13 L.Ed.2d 759 (1965). But, regrettably for Reid, the Supreme Court has since "disavowed this statement." Rivera, 556 U.S. at 160, 129 S.Ct. 1446.

Admission of evidence. Evidence Rule 404(b) restricts the use of propensity evidence — evidence of other crimes used to show that the defendant has a proclivity to commit crimes, as opposed to evidence used to show that the defendant committed this crime. Observing that the government charged him only with taking the victim to other States to have sex with her there, Reid claims that the trial court erred by admitting evidence of his in-State sexual encounters with J.H.

Under Rule 404(b), evidence of other crimes may not come in to show that the defendant acted "on a particular occasion" as he acted before, but it may come in "for another purpose, such as proving... intent." That exception defeats Reid's claim, because the trial court admitted the evidence to prove his intent, and permissibly so. The government charged Reid with transporting a minor to another State with intent to engage in criminal sexual activity. 18 U.S.C. § 2423(a). Evidence that Reid had a long-running sexual relationship with the victim helps prove this element of the crime. It blunts the possibility that Reid first took J.H. to another State and only later decided to have sex with her, and it bolsters the possibility that he had sex on his mind all along. More, the trial court admitted this evidence only to prove intent. It warned the jury that it could consider in-State sex "only on the issue of whether the defendant had the necessary specific intent," not "to decide whether the defendant carried out the acts involved in the crimes." R. 86 at 112.

To be sure, evidence of other crimes, even if admissible under Rule 404(b), must stay out under Rule 403 if "its probative value is substantially outweighed by a danger of ... unfair prejudice." But the district court did not abuse its discretion by ruling that the balance in this case favored letting the evidence in.

Sentencing enhancement. The Sentencing Guidelines advise courts to increase the sentence of a defendant who "unduly influenced a minor to engage in prohibited sexual conduct." U.S.S.G. § 2G1.3(b)(2)(B). A criminal's influence over a minor counts as "undue" if it "compromise[s] the voluntariness of the minor's behavior." Id. cmt. n.3(B). Reid's influence over J.H. passes this test with ease. A rebuttable presumption of undue influence arises when the criminal is at least ten years older than the minor. Id. Reid is thirty-five years older than J.H. Reid in addition manipulated J.H. He started their relationship, led her to think of him as her boyfriend, and encouraged her to run away from her family. To top it off, he took her hundreds of miles away from her home with no way to get back. All of that qualifies as compromising the voluntariness of the victim's behavior.

It makes no difference that J.H. "was not handcuffed to a bed" or "kidnapped off the street." R. 106 at 70-71. The undue-influence enhancement "is not limited to force, fraud, or coercion." United States v. Willoughby, 742 F.3d 229, 241 (6th Cir. 2014). It also reaches "manipulating" and "preying upon" a vulnerable victim, id. — just what we have here.

Criminal history score. The Sentencing Guidelines set forth a formula to convert a defendant's prior crimes into a criminal history score (which in turn plays a lead role when the court fixes the defendant's sentence). See U.S.S.G. § 4A1.2. In this case the district court plugged three of Reid's prior crimes into this formula: cocaine possession, larceny and escape from prison. Reid accepts the district court's consideration of cocaine possession but says that it should not have counted larceny and escape.

The dispute turns on U.S.S.G. § 4A1.2(e)(1), which says that a crime counts toward the criminal history score only if it resulted in the defendant's imprisonment "during any part" of the fifteen years preceding the start of the present offense. Under that rule, larceny and escape count only if they resulted in Reid's imprisonment at some moment after November 11, 1996. The trial court concluded that they did, and its finding must stand unless clearly wrong.

No clear error occurred. Both sides agree that Michigan's records of Reid's numerous convictions, paroles and sentences "[are] not as complete as we would like," R. 106 at 51, but even so two pieces of evidence support the district court's decision. First, Reid received a Certificate of Termination of his cocaine possession sentence in December 1996. The certificate says that, although Reid's cocaine possession sentence had ended, his larceny and escape sentences had not. That shows Reid's imprisonment for larceny and escape continued after November 11, 1996. Second, other records say that the larceny and escape sentences both expired, and that Reid as a result went free, on September 29, 1999. That too shows that Reid's imprisonment for larceny and escape continued after November 11, 1996.

Reid has no evidence that his imprisonment for larceny and escape stopped before the cutoff date. He instead attacks the evidence on the other side, homing in on an alleged contradiction between the prison records (which say that the larceny and escape sentences expired on September 29, 1999) and the court records (which say the larceny and escape sentences ran consecutively). How, he asks, could consecutive sentences expire on the same day? The answer, we suspect, lies in the Corrections Department's policy: "If a prisoner is serving ... consecutive sentences, none of the sentences which are part of the consecutive string shall be terminated until all sentences in that consecutive string have been served." Mich. Dep't of Corr., Policy Directive 03.01.135; see also Lickfeldt v. Dep't of Corr., 247 Mich.App. 299, 636 N.W.2d 272 (2001). Besides, the alleged inconsistency casts doubt on at most one piece of evidence — prison records showing Reid's discharge date. It leaves unscathed another part of the evidence — the Certificate of Termination showing that Reid's imprisonment for larceny and escape continued after the cutoff date. This kind of disparity does not make a district court's decision clearly erroneous. It takes more than a "conflict in the [evidence]" to overturn a factual finding. Harrison v. Monumental Life Ins. Co., 333 F.3d 717, 722 (6th Cir.2003).

For these reasons, we affirm.