THAPAR, District Judge:

The wire-fraud statute, 18 U.S.C. § 1343 does not enact as federal law the Ninth Commandment given to Moses on Sinai.1 For § 1343 forbids only schemes to defraud, not schemes to do other wicked things, e.g., schemes to lie, trick, or otherwise deceive. The difference, of course, is that deceiving does not always involve harming another person; defrauding does. That a defendant merely "induce[d] [the victim] to enter into [a] transaction" that he otherwise would have avoided is therefore "insufficient" to show wire fraud. See United States v. Starr, 816 F.2d 94, 98 (2d Cir.1987).

Here, the defendants feared that the jury might convict them of wire fraud based on "fraudulent inducements" alone. Hence they asked the district court to give the jurors the following instruction: that they must acquit if they found that the defendants had tricked the victims into entering a transaction but nevertheless gave the victims exactly what they asked for and charged them exactly what they agreed to pay. The district court refused to give that instruction, and the jury ultimately convicted the defendants of wire fraud and other crimes, most of which were predicated on the wire-fraud convictions. The question presented in this appeal is whether the district court abused its discretion when it refused to give the requested instruction.

I.

A.

During the defendants' trial, the parties disagreed about much of what happened in the clubs the defendants owned. But they did agree on one thing: that the defendants had tricked men to come into the defendants' clubs. The government presented evidence that the defendants had hired Eastern European women — known as "Bar Girls" or "B-girls" — to pose as tourists, locate visiting businessmen, and lure them into the defendants' bars and nightclubs. [DE 1121 at 39]. And the defendants did not seriously dispute that evidence: they admitted that they knew the B-girls concealed their relationship with the clubs to persuade the men to go to the clubs. Indeed, the defendants testified that they believed this scheme was a perfectly legitimate business model. [Id. at 59-60, 99].

The parties' stories diverged, however, as to what happened after the men entered the clubs. In the government's story, the defendants' scheme began with the B-Girls' lies but went far beyond that. Once inside the clubs, employees would pour vodka in the men's beer to get them drunker, misrepresent the prices of drinks, hide menus, cover up prices, and even forge the men's signatures on credit-card receipts. [Id. at 40-42].

The defendants' story, on the other hand, began and ended with the B-girls. Yes, they admitted they knew the B-girls were posing as tourists to get the men to come to the clubs with them. From there, though, they proceeded to mount what one might call the Casablanca defense, arguing that they were "shocked, shocked" to learn that fraud was taking place within their South-Beach versions of Rick's Café Américain.2 As for the swindling going on inside the clubs — the lying about prices, the forging of signatures, and so on — the defendants said that they knew nothing about it. Instead, the defendants testified, they were merely investors in the clubs — or in charge of the credit-card transactions — but were not involved in the day-to-day workings of the clubs. [See id. at 60, 69-70, 86, 103]. In the defendants' story, none of these allegedly swindled men were truly victims: they knowingly entered the clubs, bought bottles of liquor, and drank them with their female companions. Thus, in the defendants' view, these men got what they paid for — nothing more, nothing less. [See id. at 62, 100-102].

B.

In addition to the factual dispute about what happened after the men came into the clubs, the parties also disagreed about the legal significance of the lies that the B-Girls used to get the men to come into the clubs in the first place. In the government's view, the jury could convict the defendants of wire fraud based on those lies alone. [See DE 1121 at 39-40; R. 1154 at 63]. The defendants argued just the opposite — that "just because they have [used promoters to persuade men to come back to the respective establishments] does not constitute fraud with regard to the wire fraud or conspiracy to commit these frauds." [DE 1152 at 285].

At the close of evidence, the defendants asked for a jury instruction to support this theory. Specifically, they asked the court to instruct the jury that "[f]ailure to disclose the financial arrangement between the B-girls and the Bar, in and of itself, is not sufficient to convict a defendant of any offense[.]" [DE 921 at 1]. The court denied that theory-of-the-defense instruction, however, because the court did not believe it was "an accurate statement of the law." [DE 1152 at 285, 289].

During closing argument, the government argued exactly what the defendants had expected it would argue: that the B-girls' concealment of their bar-affiliation to the men were material misrepresentations sufficient to constitute fraud. [DE 1154 at 63 ("The first lie was by the girls to get them to come to the clubs by not telling them that they work for the clubs and got a percentage and this was material. This was important because, as even the defendant's own witness told you, had they known that these women worked for the clubs they likely wouldn't have even gone.")]. When defense counsel stood up to make their closing arguments, they did so in front of a jury that had just heard that the B-girls' lies were material and had never received an instruction to the contrary. Perhaps for this reason, defense counsel focused their efforts elsewhere, arguing that there was not enough evidence to connect the defendants to the other fraudulent activities at the clubs. Specifically, the defendants argued that they were not involved in the alleged fraud that took place within the clubs, [id. at 147-48, 183, 191, 214-15], and that they did not believe they were doing anything illegal. [Id. at 147-48, 170-71, 201-02]. Following closing arguments, the jury convicted the defendants on several counts, including multiple counts of wire fraud and money laundering. [D.E. 954; D.E. 956; D.E. 957]. This appeal followed.

II.

We review for an abuse of discretion a district court's refusal to give a requested jury instruction. United States v. Dohan, 508 F.3d 989, 993 (11th Cir. 2007).

A.

To show that the district court abused its discretion when it refused to give a proposed jury instruction, a defendant must first show that the requested instruction was a correct statement of the law. United States v. Eckhardt, 466 F.3d 938, 947-48 (11th Cir.2006). "We review the legal correctness of a [requested] jury instruction de novo." Fid. Interior Const., Inc. v. Se. Carpenters Reg'l Council of United Bhd. of Carpenters & Joiners of Am., 675 F.3d 1250, 1259 (11th Cir.2012).

1.

The law in question here is the wire-fraud statute, which makes criminal any "scheme or artifice to defraud."3 18 U.S.C. § 1343. The statute itself, however, does not explain what constitutes such a scheme or artifice. United States v. Bradley, 644 F.3d 1213, 1240 (11th Cir.2011). Thus, the meaning of the phrase "scheme to defraud" has been "judicially defined." United States v. Pendergraft, 297 F.3d 1198, 1208 (11th Cir.2002). And that definition is a broad one, "broad[er] ... than the common law definition of fraud." Id. It is a "reflection of moral uprightness, of fundamental honesty, fair play and right dealing in the general and business life of members of society." Gregory v. United States, 253 F.2d 104, 109 (5th Cir.1958).

"[D]espite its breadth," however, "the judicial definition" of a "scheme to defraud" has some limits. Bradley, 644 F.3d at 1240. The most important limit is obvious from the statute itself: the scheme must be a scheme to defraud rather than to do something other than defraud. For Congress could have made criminal any "scheme" simpliciter, but chose not to do so. The first question presented in this case, then, is what the word "defraud" means.

To answer that question, we turn first to the dictionaries. For "[t]he ordinary-meaning rule is the most fundamental semantic rule of interpretation." Antonin Scalia & Bryan A. Garner, Reading Law 69 (2012). And "to determine the common usage or ordinary meaning of a term, [we] often turn to dictionary definitions for guidance." CBS Inc. v. PrimeTime 24 Joint Venture, 245 F.3d 1217, 1223 (11th Cir.2001); see also Stein v. Paradigm Mirasol, LLC, 586 F.3d 849, 854 (11th Cir. 2009) (noting that "[a] term that is undefined in a statute carries its ordinary meaning" and turning first to the dictionary to determine that meaning). Black's defines the word "defraud" as "[t]o cause injury or loss to (a person or organization) by deceit." Black's Law Dictionary 516 (10th ed. 2014.). Webster's says much the same. Webster's Third New International Dictionary 593 (2002) (defining the word "defraud" as "to take or withhold from (one) some possession, right, or interest by calculated misstatement or perversion of truth, trickery, or other deception").

These definitions make clear that there is a difference between deceiving and defrauding: to defraud, one must intend to use deception to cause some injury; but one can deceive without intending to harm at all. See Black's at 492 (defining the word "deception" as "[t]he act of deliberately causing someone to believe that something is true when the actor knows it to be false"); Webster's at 585 (defining the word "deception" as "the act of deceiving, cheating, hoodwinking, misleading, or deluding"). Thus, deceiving is a necessary condition of defrauding but not a sufficient one. Put another way, one who defrauds always deceives, but one can deceive without defrauding.

For this reason, the law in the Eleventh Circuit makes clear that a defendant "schemes to defraud" only if he schemes to "depriv[e] [someone] of something of value by trick, deceit, chicane, or overreaching." Bradley, 644 F.3d at 1240. But if a defendant does not intend to harm the victim — "to obtain, by deceptive means, something to which [the defendant] is not entitled" — then he has not intended to defraud the victim. Id.

From that conclusion, a corollary follows: a schemer who tricks someone to enter into a transaction has not "schemed to defraud" so long as he does not intend to harm the person he intends to trick. And this is so even if the transaction would not have occurred but for the trick. For if there is no intent to harm, there can only be a scheme to deceive, but not one to defraud.

Consider the following two scenarios. In the first, a man wants to exchange a dollar into four quarters without going to the bank. He calls his neighbor on his cell phone and says that his child is very ill. His neighbor runs over, and when she arrives he asks her to make change for him. She agrees; the quarters pass to the man; the dollar passes to the woman; and they part ways. She later learns that the child was just fine all along. The second scenario is identical to the first, except that instead of giving the woman a true dollar, he gives her a counterfeit one.

The first scenario is not wire fraud; the second one is.4 Although the transaction would not have occurred but-for the lie in the first scenario — the woman would have remained home except for the phony sickness — the man nevertheless did not intend to "depriv[e] [the woman] of something of value by trick, deceit, [and so on]." Bradley, 644 F.3d at 1240. But in the second scenario he did intend to do so.

More specifically, the difference between the scenarios is that, in the first scenario, the man did not lie about the nature of the bargain: he promised to give the woman a true dollar in exchange for the quarters, and he did just that. In the second, he lied about the nature of the bargain: he promised to give her a true dollar but gave her a fake one instead.

Now imagine another, more common scenario: a young woman asks a rich businessman to buy her a drink at Bob's Bar. The businessman buys the drink, and afterwards the young woman decides to leave. Did the man get what he bargained for? Yes. He received his drink, and he had the opportunity to buy a young woman a drink. Does it change things if the woman is Bob's sister and he paid her to recruit customers? No; regardless of Bob's relationship with the woman, the businessman got exactly what he bargained for. If, on the other hand, Bob promised to pour the man a glass of Pappy Van Winkle5 but gave him a slug of Old Crow6 instead, well, that would be fraud. Why? Because the misrepresentation goes to the value of the bargain.

Thus, a "scheme to defraud," as that phrase is used in the wire-fraud statute, refers only to those schemes in which a defendant lies about the nature of the bargain itself. That lie can take two primary forms: the defendant might lie about the price (e.g., if he promises that a good costs $10 when it in fact costs $20) or he might lie about the characteristics of the good (e.g., if he promises that a gemstone is a diamond when it is in fact a cubic zirconium). In each case, the defendant has lied about the nature of the bargain and thus in both cases the defendant has committed wire fraud. But if a defendant lies about something else — e.g., if he says that he is the long-lost cousin of a prospective buyer — then he has not lied about the nature of the bargain, has not "schemed to defraud," and cannot be convicted of wire fraud on the basis of that lie alone.

The Second Circuit has interpreted the wire-fraud statute in precisely this way. Their cases have "drawn a fine line between schemes that do no more than cause their victims to enter into transactions that they would otherwise avoid — which do not violate the mail or wire fraud statutes — and schemes that depend for their completion on a misrepresentation of an essential element of the bargain — which do violate the mail and wire fraud statutes." United States v. Shellef, 507 F.3d 82, 108 (2d Cir. 2007); see also United States v. Starr, 816 F.2d 94, 98 (2d Cir.1987) ("Misrepresentations amounting only to a deceit are insufficient to maintain a mail or wire fraud prosecution. Instead, the deceit must be coupled with a contemplated harm to the victim [that] affect[s] the very nature of the bargain itself. Such harm is apparent where there exists a discrepancy between benefits reasonably anticipated because of the misleading representations and the actual benefits which the defendant delivered, or intended to deliver.") (internal quotation marks omitted); United States v. Regent Office Supply Co., 421 F.2d 1174, 1182 (2d Cir.1970) ("[W]e conclude that the defendants intended to deceive their customers but they did not intend to defraud them, because the falsity of their representations was not shown to be capable of affecting the customer's understanding of the bargain nor of influencing his assessment of the value of the bargain to him, and thus no injury was shown to flow from the deception.").

Moreover, the Second Circuit's interpretation of the wire-fraud statute is not a parochial interpretation of an ambiguous provision of federal law. Their interpretation follows as a matter of logic from Congress's decision to use the phrase "scheme to defraud" rather than "scheme" or "scheme to deceive." We therefore adopt that interpretation as our own. A jury cannot convict a defendant of wire fraud, then, based on "misrepresentations amounting only to a deceit." Shellef, 507 F.3d at 108. Thus, even if a defendant lies, and even if the victim made a purchase because of that lie, a wire-fraud case must end in an acquittal if the jury nevertheless believes that the alleged victims "received exactly what they paid for." Id.

2.

Here, the defendants asked the district court to instruct the jury that "[f]ailure to disclose the financial arrangement between the B-girls and the Bar, in and of itself, is not sufficient to convict a defendant of any offense[.]" R. 921 at 1; see also R. 1152 at 285 ("So [Pavlenko's] theory of defense is just because [the defendants used the B-girls to lure men into the clubs without disclosing their financial arrangement] does not constitute fraud with regard to the wire fraud or conspiracy to commit these frauds."). If the jury believed that the defendants had deceived the victims only about the B-girls' pay arrangement, then the jury would likewise believe that the defendants' misrepresentations "amount[ed] only to deceit" and that the victims "received exactly what they paid for." Shellef, 507 F.3d at 108. Thus, "failure to disclose the financial arrangement between the B-girls and the Bar" was not "in and of itself" sufficient to convict the defendants of wire fraud. The defendants' requested instruction therefore seems like a correct statement of the law.

The government responds in a few ways. First, it argues that the Second Circuit cases have not "been applied in this Circuit." United States Br. at 43. True, but neither have we rejected the Second Circuit's interpretation; we simply have never considered the question they answered. And the "law" at issue is the proper interpretation of the federal statute itself, not some common-law creature unique to the Second Circuit. After all, the defendants' argument is not that the Eleventh Circuit is bound by the opinions of the Second Circuit. The argument is that the Second Circuit's interpretation of the statute is the right interpretation. Thus, the defendants argue, their proposed jury instruction was a correct statement of "the law," i.e., the statute. We have already explained why we agree.7

Second, the government argues that the proposed instruction "was incorrect legally, because it proposed to tell the jury that [the] defendants' conduct was outside the reach of the wire fraud statute." United States Br. at 54. No, it did not. It proposed only to tell the jury what portion of the defendants' conduct could be criminal under the statute and what portion could not.

Third, the government argues that "Pavlenko's instruction was incorrect in its portrayal of the charges" because the government "never described the [defendants'] scheme" as one "to have attractive women induce patrons to purchase and consume alcoholic beverages under the illusion that the patron may later persuade the women to have sexual relations." United States Br. at 55. Fair enough. But the question at hand is a purely legal one: whether the defendants' proposed instruction was a correct statement of the law. And the government's characterization of the defendants' scheme has no bearing on the answer to that question.

Finally, the government argues that the instruction "was incorrect in its depiction of the facts" because "[t]he evidence showed convincingly that the conspirators... intoxicated patrons without their knowledge or knowing consent, forcing liquor onto patrons, adulterating beverages, and physically pouring it down victims' throats in some instances." United States Br. at 55-56. Again, fair enough. But, again, totally irrelevant. What "the evidence showed" — even "showed convincingly" — simply has no bearing on whether the defendants' proposed instruction was a correct statement of the law.

In sum, the defendants asked the district court to tell the jurors that they could convict only if they found that the defendants had schemed to lie about the quality or price of the goods sold to the victims. And § 1343, which we interpret de novo, requires the jury to make just such a finding before convicting a defendant of wire fraud. Thus, the proposed instruction was a correct statement of the law.

B.

To show an abuse of discretion, however, a defendant must show more than just that the requested instruction was a correct statement of the law. He must also show that the instruction dealt with a sufficiently important point raised during trial. Eckhardt, 466 F.3d at 947-48.

Here, the defendants' case theory seemed to go as follows: yes, we knew that the B-girls were tricking the victims into coming to the bar by posing as tourists, but we nevertheless did not commit wire fraud because we gave the victims exactly what they ordered — i.e., absurdly expensive drinks at the bar — and thus any lies about the B-girls' employment status did not misrepresent the value of the bargain.8 The government's position was that, although there was evidence that the defendants did far more than just trick the victims into coming to the bar, the jury could convict based on that trickery alone.

Given the positions that the government and defendants took during trial, the defendants' proposed instruction — that "[f]ailure to disclose the financial arrangement between the B-girls and the Bar, in and of itself, is not sufficient to convict a defendant of any offense" — certainly "dealt with some point in the trial so important that failure to give the requested instruction seriously impaired the defendant's ability to conduct his defense." Eckhardt, 466 F.3d at 947-48. After all, if the jurors believed that they could convict based only on the B-girls' failure "to disclose the financial arrangement between the B-girls and the Bar," then the defense's theory would have collapsed entirely.

C.

Finally, to show that the district court abused its discretion by refusing to give a proposed instruction, a defendant must show that the proposed instruction "was not substantially covered by a charge actually given." Eckhardt, 466 F.3d at 948. Here, the district court gave the jury an instruction as to the elements of the offense that in relevant part provided:

R. 1154 at 25-26 (emphasis added). Nowhere in that instruction — or anywhere else — did the district court tell the jurors that "[f]ailure to disclose the financial arrangement between the B-girls and the Bar, in and of itself, is not sufficient to convict a defendant of any offense[.]" Nor did the district court say anything like that. Thus, it would seem that the district court's instructions did not "substantially cover" the defendants' requested instruction.

In response, the government argues that the district court's "good-faith instruction" was enough. Specifically the government points out that the court told the jury that:

R. 1154 at 37-38. That instruction, the government argues, "substantially covered" the defendants' requested one.

The government's argument simply misunderstands the defendants' case theory. They did not argue that they lacked the specific intent to deceive the victims; indeed they admitted that they fervently hoped to do just that. The defendants instead argued that they had intended to deceive the victims in only one way — by tricking them into coming to the bars — and that such a deception was not wire fraud. Put another way, the defendants did not dispute that they lacked specific-enough intent; they argued that what they specifically intended to do was not a crime. The district court's good-faith instruction concerned only what it meant for the defendants to have a specific intent to deceive. It said nothing about what kind of deception could constitute wire fraud. Thus, the good-faith instruction did not "substantially cover" the defendants' requested instruction.

One might also argue (though the government does not) that the district court's definition of "scheme to defraud" substantially covered the proposed instruction. This is a more difficult argument to answer. After all, the court told the jurors that they could convict only if they found that the defendants intended to "deceive or cheat someone out of money or property." And if someone is lured to a bar under false pretenses but nevertheless gets precisely what he pays for, he has hardly been "deceive[d] or cheat[ed] out of money or property." Thus, under the given instructions, it would have been hard for the jury to convict if they found only that the defendants lured the victims into the bar under false pretenses. To the contrary, to convict under the given instructions, the jury would have needed to find — as a matter of logic — that the defendants cheated the victims out of money or property, e.g., by running up fake bills on the victims' credit cards and charging absurd drink prices that the menus nowhere advertised.

The question before us, however, is not whether the proposed instruction was "logically entailed" by the given instruction, but whether it was "substantially covered"; and those are meaningfully different concepts. After all, the average juror is not Mr. Spock. If he were, then a trial-court judge's job would be much easier. He could instruct the jury in broad strokes — instructing only as to the bare elements of the crime, perhaps — and be confident that the jury would deduce all of the finer-grained implications that must logically follow. As it stands, however, the vast majority of American juries are composed exclusively of humans. And humans, unlike Vulcans, sometimes need a bit more guidance as to exactly what the court's instructions logically entail.

For that reason, this case is distinguishable from other Eleventh Circuit cases holding that a given instruction substantially covered a requested one — even the two cases that are most difficult for the defendants, Martinelli and Hill. The difference between those cases on the one hand, and this case on the other, is the size of the logical leap that a juror would need to make to get from the instruction the court gave to the instruction the defendant requested.

In Martinelli, the court instructed the jury that the defendant was guilty of money laundering only if he knew that the money was derived from felonious activities. The defendant in turn asked the court to instruct the jury that the defendant was not guilty if he thought the money-producing business was a legitimate one. United States v. Martinelli, 454 F.3d 1300, 1315-16 (11th Cir.2006). Thus, to get from the given instruction to the requested one, the jury needed to infer only one thing: that "legitimate businesses" do not engage in "felonious activities." We have no doubt that a juror can bridge that logical gap all on her own.

Similarly, in Hill, the court instructed the jury that the defendant was guilty of credit-application fraud only if he made false statements to the bank knowingly and willfully. The defendant asked the court to instruct the jury that he was not guilty if he believed the statements were true. United States v. Hill, 643 F.3d 807, 852-54 (11th Cir.2011). Thus, to get from the given instruction to the requested one, the jury needed to infer only one thing: that a person cannot lie "knowingly and willfully" if he speaks what is in his view the truth. That inference, too, hardly requires Holmesian feats of deduction.9

In contrast, the court here instructed the jury that the defendants were guilty of wire fraud only if they intended to "deceive or cheat someone out of money or property." The defendants asked the court to instruct the jury that they were not guilty if they merely used lies to lure the victims into an otherwise-legitimate transaction. To get from the given instruction to the requested one, the jury needed to make the following inference: that a person is not "deceived or cheated out of money or property" if he gets exactly what he paid for even though he is deceived into paying in the first place. Suffice it to say that, if a juror is going to make that kind of logical inference, he is probably going to need some guidance from the district court.

Indeed, the government still refuses to make that inference without some judicial "guidance." The prosecution repeatedly argued below that the B-Girls' lies about their employment status was enough to convict the defendants of wire fraud. See, e.g., R. 663 at 3-4; R. 1121 at 39-40; R. 1154 at 63. And even before this court, the government argues that "[t]here is no reason that deceiving victims about the girls' relationship to the clubs would not be enough to sustain a wire-fraud conviction[.]" United States Br. at 39. Although the government's argument might be a reason to hold that the defendants' requested instruction was not a "correct statement of a law" — an argument that we rejected above — it also shows that the given instructions did not "substantially cover" the requested one. After all, if even the government's attorneys did not believe that the court was instructing the jury that "deceiving victims about the girls' relationship to the clubs would not be enough to sustain a wire-fraud conviction," then why would a jury composed exclusively of lay people?

In sum, a district court does not need to tell a jury that proceeds from a felony are not proceeds from a legitimate business. Nor does a court need to tell a jury that a person does not lie willfully if he thinks he is telling the truth. The jurors will figure those things out on their own. But a court does need to tell a jury that a scheme to trick patrons to come into a bar — without more — is not wire fraud. That is not the sort of thing that we can expect jurors to simply infer. The district court's elements-of-the-crime instruction therefore did not "substantially cover" the defendants' requested instruction.

D.

We recognize that the district court presided over a complex criminal trial that took many months to complete. And the court did an admirable job in overseeing those proceedings. Indeed, we commend the court, whose evidentiary and other legal rulings were, in our view, nearly flawless. Nevertheless, the district court refused to give a jury instruction that was a correct statement of the law, was critical to the defense's case theory, and was not substantially covered by other instructions. Thus, as a matter of law, the district court abused its discretion by refusing to give that instruction. The question, then becomes whether this error requires reversal or whether it was instead harmless.

III.

A.

This would have been an easy case to decide during the Kennedy administration. Under the Supreme Court's NASA-era jurisprudence, the rules were clear with respect to botched jury instructions. If a trial judge gave two instructions — and one turned out to be improper — then the appellate court would reverse the conviction so long as "it [wa]s impossible to say under which [instruction] the conviction was obtained." Stromberg v. California, 283 U.S. 359, 368, 51 S.Ct. 532, 75 L.Ed. 1117 (1931); see also Yates v. United States, 354 U.S. 298, 311-12, 77 S.Ct. 1064, 1 L.Ed.2d 1356 (1975). Under that test, the defendants here would surely be entitled to a new trial. After all, the district court provided the jury with multiple paths by which they could arrive at a guilty verdict. One of those paths — as we have explained above — was an impermissible one. It is of course "impossible to say" which path the jury proceeded down. And thus it is likewise "impossible to say" whether the jury would have nevertheless arrived at the same destination — a conviction — if the trial court had restricted the jury to the proper route.

Things are harder now. In 1967, the Supreme Court announced its decision in Chapman v. California, 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967), which held that even constitutional errors might, in certain situations, still be harmless ones. "In a series of post-Chapman cases," the Court "concluded that various forms of [jury-instruction] error[s] are not structural" — i.e., they do not automatically require reversal, no questions asked — but are instead "subject to harmless-error review." Hedgpeth v. Pulido, 555 U.S. 57, 60, 129 S.Ct. 530, 172 L.Ed.2d 388 (2008). And although those cases "did not arise in the context of a jury instructed on multiple theories of guilt, one of which is improper," the Court made clear that the same harmless-error standard would apply in cases, like the one here, in which a jury could have convicted based on an erroneous instruction. Id. at 61, 129 S.Ct. 530.

That standard was framed most clearly — and most recently — in Neder v. United States, 527 U.S. 1, 119 S.Ct. 1827, 144 L.Ed.2d 35 (1999). There, the government indicted a man named Ellis Neder for tax fraud. "According to the Government, Neder failed to report more than $1 million in income in 1985 and more than $4 million in income for 1986, both amounts reflecting profits Neder obtained from [various] fraudulent[ly] [obtained] real estate loans." Id. at 6, 119 S.Ct. 1827. He was therefore indicted for two counts of filing a false income tax return, in violation of 26 U.S.C. § 7206(1). Id. To convict Neder of violating that section of the Internal Revenue Code, the government needed to prove that he had willfully "ma[de] and subscrib[ed]" a "return, statement, or other document, which contains or is verified by a written declaration that it is made under the penalties of perjury, and which he does not believe to be true and correct as to every material matter." 26 U.S.C. § 7206(1) (emphasis added).

Encouraged by then-existing circuit precedent, the district court glossed over this "every material matter" language and instructed the jury that it "need not consider the materiality of any false statements even though that language [was] used in the indictment." Neder, 527 U.S. at 6, 119 S.Ct. 1827 (internal quotation marks omitted). "The question of materiality, the [district] court instructed, `is not a question for the jury to decide.'" Id. By the time the case arrived at the Supreme Court, the parties agreed that the district court had made a mistake when it gave that instruction. Id. at 8, 119 S.Ct. 1827. After all, the statute did not forbid a taxpayer from making any false statements on his tax return; it forbade him only from making material false statements.

The question before the Court was what to do about the error. "Unlike such defects as the complete deprivation of counsel or trial before a biased judge," the Court held, "an instruction that omits an element of the offense does not necessarily render a criminal trial fundamentally unfair[.]" Id. at 9, 119 S.Ct. 1827. The Court then imported the Chapman standard and thus framed the harmless-error standard — at a high level of generality — as follows: the question is whether "it appears beyond a reasonable doubt that the error complained of did not contribute to the verdict obtained." Id. at 15, 119 S.Ct. 1827 (internal quotation marks omitted).

In addition to this general rule — framed, as it were, as a wide-angle shot — the Neder Court also zoomed in to give additional guidance as to exactly when an error would be harmless beyond a reasonable doubt. Specifically, the Court held that the question was "whether the record contains evidence that could rationally lead to a contrary finding with respect to the omitted element," id. at 19, 119 S.Ct. 1827 — i.e., whether a properly instructed jury could have "reasonably f[ound]" the defendant not guilty, id. at 16, 119 S.Ct. 1827. If not, the Court held, then a jury-instruction error is harmless. Answering this question as to Neder himself, the Court held that "no jury could reasonably find that Neder's failure to support [some $5 million] of income on his tax returns was not `a material matter.'" Id. And thus the Court concluded that it was "beyond cavil" that "the error `did not contribute to the verdict obtained.'" Id. at 17, 119 S.Ct. 1827 (quoting Chapman, 386 U.S. at 24, 87 S.Ct. 824).

The general question presented in this case is therefore as follows: whether "it appears beyond a reasonable doubt that the [faulty jury instruction] did not contribute to the verdict obtained." See id. at 15, 119 S.Ct. 1827 (internal quotation marks omitted). The more specific question is this: does "the record contain[] evidence that could rationally lead" a jury to find that the defendants lacked the intent to defraud? See id. at 19, 119 S.Ct. 1827. If so, the defendants are entitled to a new trial. If not, the failure to give the requested instruction was harmless. Id.

B.

1.

Here, the record contains sufficient evidence such that, had the requested instruction been given, a rational jury could find that the defendants lacked the intent to defraud on almost all of the wire-fraud counts. At trial, the defendants each testified that they were aware that the B-girls concealed their relationship to the clubs. See D.E. 1147 at 187 (Pavlenko testimony); D.E. 1149 at 95 (Feldman testimony); D.E. 1151 at 34, 38, 88, 213 (Takhalov testimony).10 However, they each also testified that they did not intend to defraud any customers and that they were not aware of any (other) illicit activity by the B-girls. See, e.g., D.E. 1146 at 119-20 (Pavlenko testimony); D.E. 1147 at 31-32 (same); D.E. 1149 at 44-47, 109, 181, 184, 223-24 (Feldman testimony); D.E. 1151 at 38-39, 143, 184 (Takhalov testimony). Had the requested instruction been given, a rational jury certainly could have chosen to believe the defendants' testimony that they did not intend to defraud customers. And in fact, the record suggests the jury did believe the defendants, at least in part — the jury acquitted the defendants on the majority of the counts in the indictment. D.E. 954; D.E. 956; D.E. 957; see United States v. Thomspon, 473 F.3d 1137, 1142 (11th Cir.2006) ("The jury gets to make any credibility choices, and we will assume they made them all in the way that supports the verdict."). Regardless of the verdict, though, "[i]t is not for us to decide which witnesses we find more believable — credibility choices lie within the province of the jury." United States v. Johnson, 713 F.2d 654, 661 (11th Cir.1983) (internal quotation marks omitted). As such, the defendants' testimony provided sufficient evidence for a rational jury to find that the defendants lacked the intent to defraud.

The government makes five arguments in response.11 First, the government explains that the defendants' fraud scheme had many components, and the government never told the jury to convict the defendants solely on their failure to disclose the B-girls' relationship to the clubs. True. But regardless of whether the government told the jury convict on this basis, the jury still could have convicted the defendants on the wire fraud counts solely on this basis. To convict the defendants of wire fraud, the jury had to find, in part, that (1) the defendants "knowingly devised or participated in a scheme to defraud or to obtain money by false pretenses, representations or promises," and (2) these "false pretenses, representations or promises were about a material fact." D.E. 1154 at 25 (jury instructions). In the government's closing, the government argued that the concealment of the B-girls' relationship to the clubs was a "material" misrepresentation. Id. at 63. As such, even if the jury did not believe that the defendants made any other material misrepresentations, the jury still could have convicted the defendants because the government proved that the defendants made false representations about a material fact: the B-girls' relationship to the clubs.

Second, the government argues that it presented "overwhelming evidence" that the defendants were guilty of wire fraud beyond their failure to disclose the B-girls' relationship. While this may be correct, it does not change the outcome here. The question is not whether the jury could still have convicted the defendants if the instruction had been given. The question is whether the jury could have acquitted them. And the evidence against the defendants here was not so overwhelming that an acquittal would have been irrational.

Third, the government asserts that "if the jury was truly under the `mistaken' impression that simply concealing the girls' relationship to the clubs was sufficient to convict, there would have been no acquittals whatsoever on the substantive fraud counts." United States Supp. Br. at 11. It is true that the jury acquitted the defendants on some of the substantive fraud counts. This does not mean that the jury did not have the `mistaken impression' that concealing the girls' relationship to the club was, on its own, fraud. Maybe the jury did not believe the government proved the defendants' involvement in those counts, or maybe the jury reached inconsistent verdicts. Regardless, the Court will not read into the split verdict to determine whether or not the jury believed the concealment of the B-girls' relationship was sufficient to prove fraud. See United States v. Powell, 469 U.S. 57, 64-65, 105 S.Ct. 471, 83 L.Ed.2d 461 (1984) ("[W]here truly inconsistent verdicts have been reached, the most that can be said ... is that the verdict shows that either in the acquittal or the conviction the jury did not speak their real conclusions, but that does not show that they were not convinced of the defendant's guilt." (internal quotations omitted)).

Fourth, the government states that the defendants did not argue their proposed defense theory — that the concealment of the B-girls' relationship did not constitute fraud — in closing. Thus, the government claims, we cannot find that the jury convicted the defendants solely on the concealment of the B-girls' relationship to the clubs. But the government does not point the Court to a single case that requires a defendant to raise a defense theory in closing arguments after that the trial court rejected that theory. And, even if such a case existed, the government's logic is flawed. Imagine that the defendants had argued in closing that the concealment of the B-girls' relationship did not constitute fraud. Without an instruction supporting the defendants' theory, the jury was not required to believe this theory. Instead, the jury could believe what the government argued in its closing: that the concealment was material and the defendants acted with the intent to deceive or cheat the victims. Thus, even if the defendants had argued their theory in closing, the jury still could have convicted the defendants of fraud based on the concealment of the B-girls' relationship.

Finally, the government argues that "the jury convicted Pavlenko on a substantive wire fraud count that was not tied to the interaction of a customer and a B-girl, further ameliorating any doubt that the jury may have thought the only deception here was failing to disclose the girls' relationship to the clubs." United States Supp. Br. at 11. Specifically, the government points to Count 21 of the indictment, which charged that Pavlenko carried out the fraud scheme by sending an email to American Express regarding a picture of Bolaris, one of the victims, with a B-girl. Id. In his email to American Express, Pavlenko implied he did not know who the girl in the photo was, when in fact she worked for the bar. See United States Ex. 223. So, the government argues, the instruction at issue here would not have altered the verdict on Count 21.

Pavlenko responds that Count 21 of the indictment was tied to the interaction of a customer and a B-girl. The jury instructions stated that the substantive wire fraud counts, including Count 21, did not require the wire communication to be false. Instead, the government could prove those counts if the wire communication contained a true statement, as long as that communication was in furtherance of the fraud scheme. R. 1154 at 26. This jury instruction meant that the jury did not have to find that Pavlenko's email to American Express contained a lie; instead, the jury just had to find that the email was sent in furtherance of the general scheme to defraud customers. So, the argument seems to go, the jury could have found that this email was in furtherance of the scheme to defraud the customers without finding that the email contained a material misstatement with the intent to defraud American Express. But the question this court must focus on is: would the jury have convicted if properly instructed? In other words, what Pavlenko is arguing is that the jury might have convicted under an improper theory. But that is not the question. The question is whether a reasonable jury could have acquitted when instructed under the proper theory. The answer is no. Why? Here, the email to American Express was clearly false. Indeed, Pavlenko himself admitted that he "didn't want to lie to [American Express], but [he] didn't want them to know she was an employee." See D.E. 1219 at 30. And, for Count 21, this lie was both the focus of the indictment and the government's theory to the jury. As such, had the jury been properly instructed, it still would have convicted. Thus, the error was harmless as to Count 21.12

The government, however, has not presented the court with sufficient evidence showing the trial court's error was harmless as to the remaining wire-fraud counts and conspiracy-to-commit-wire-fraud counts.13 And it is the government's burden to demonstrate that the error was harmless. United States v. Robison, 505 F.3d 1208, 1222-23 (11th Cir.2007). The defendants' convictions with respect to those counts are therefore reversed.

2.

The defendants also argue that their convictions on the money-laundering counts, Count 29 and Count 39, should be reversed because the missing instruction tainted these counts as well. [Pavlenko Initial Brief at 24.] Count 29 charged all three defendants with conspiracy to violate two provisions of the money-laundering statute: money laundering by concealment and money laundering by international promotion. D.E. 953 at 20-21.14 Count 39 also charged Takhalov with conspiracy to commit money laundering by concealment. Id. at 32-33. Although money laundering by concealment and money laundering by international promotion involve different elements, both offenses criminalize transactions that promote an underlying criminal activity. 18 U.S.C. § 1956(a). But to prove either offense, the government does not have to prove that the defendant committed the underlying criminal activity. Id.; see also United States v. Martinelli, 454 F.3d 1300, 1311 (11th Cir.2006). Instead, for money laundering by concealment, the government has to prove that the defendant knew: (1) that the proceeds of the transaction involved proceeds of an unlawful activity, and (2) that the transaction was designed to conceal some aspect of those proceeds. 18 U.S.C. § 1956(a)(1). And for money laundering by international promotion, the government has to prove that the defendant engaged in the transaction "with the intent to promote the carrying on of [a] specified unlawful activity." Id. § 1956(a)(2)(A).

Here, the government charged two specified unlawful activities: wire fraud, in violation of 18 U.S.C. § 1343, and misuse of a visa document, in violation of 18 U.S.C. § 1546. D.E. 953 at 21, 33.15 The first problem for the government is that it did not provide us any evidence to show that the jury would have convicted the defendants of money laundering based on misuse of a visa document. Chapman, 386 U.S. at 23-24, 87 S.Ct. 824 (Harmless-error analysis puts "the burden on the beneficiary of the error either to prove that there was no injury or to suffer a reversal of [the] erroneously obtained judgment."); United States v. Fern, 155 F.3d 1318, 1327 (11th Cir.1998) ("Here, the government's burden, more precisely stated, is to show that the guilty verdict against [the defendant] was `surely unattributable' to the incorrect jury instruction[.]"). And the Court's review of the record does not reveal sufficient evidence for such a finding. In fact, in the government's opening statement — a roadmap of the government's case — the government did not mention anything about money laundering related to misuse of a visa document. See D.E. 1121. Instead, the government expressly tied the money-laundering counts only to wire fraud, stating that the defendants engaged in money laundering "because the wire fraud was being concealed." Id. at 51. This brings us to the second problem: we cannot be sure that a properly instructed jury would have found that the specified unlawful activity of wire fraud actually occurred. See supra Part III.B.1. And, without that, we cannot conclude beyond a reasonable doubt that the jury would have convicted the defendants of money laundering. See United States v. Neder, 197 F.3d 1122, 1129 (11th Cir.1999) ("[T]he government must show that the evidence [of the proper theory of guilt] is so overwhelming ... that no rational jury, properly instructed ..., could have acquitted [the defendants] on that count."). As such, the defendants' convictions for conspiracy to commit money laundering, Count 29 and Count 39, must be reversed.

IV.

The remaining conviction against Takhalov, Count 38 for conspiracy to defraud the United States Department of Homeland Security, remains because it was unrelated to wire fraud. D.E. 953 at 30-32. Finding no other errors that would justify reversal on Counts 21 and 38, we reverse on all but Counts 21 and 38, and remand to the district court for further proceedings consistent with this opinion.