Full opinion text
Opinion for the Court filed PER CURIAM. PER CURIAM: The Hostage Taking Act, 18 U.S.C. '§ 1203, prescribes criminal penalties for foreign nationals who abduct American citizens. In this case, nationals of the Republic of Trinidad and Tobago abducted wealthy individuals, held them captive in the island’s mountainous forests, and extorted ransoms from terrified family and friends. The scheme proved quite profitable — at least until they kidnapped an American citizen and ran headlong into the Hostage Taking Act. The conspirators were extradited to the United States, tried, and convicted of violating the Act. But does that statute apply if, as defendants allege, the victim secured his United States citizenship through fraud? The district court held that it does, and for the reasons set forth in this opinion, we agree. Rejecting all of the conspirators’ other challenges, we affirm their convictions in all respects. I. By early 2005, defendants Wayne Pierre, Ricardo De Four, and Zion Clarke had perfected their hostage-taking protocol and regularly extorted six-figure ransoms (Trinidadian dollars). Looking to up the ante, the three enlarged their organization to include defendants Kevon Demerieux, Kevin Nixon, Christopher Sealey, and Anderson Straker, and set their sights on Trinidad-native Balram Maharaj, whom they believed had amassed a fortune in the United States. Although naturalized as an American citizen in 1995, Maharaj frequently visited his children in Trinidad. Defendants, assisted by a host of unindicted co-conspirators, planned to abduct Maharaj during one of those visits. On the night of April 6, 2005, defendants executed their plan. Sealey and Nixon, armed with handguns, dragged Maharaj from Samaan Tree Bar in Aranguez, Trinidad and forced him into a getaway car while De Four drove ahead in a separate vehicle to clear the way. Sealey and Nixon delivered Maharaj to an isolated camp deep within the forest where Clarke and Demerieux guarded him. A nightmare ensued. The two guards tied Maharaj to a post and gave him little food and water. Suffering from severe diabetes, hypertension, and tuberculosis, Maharaj pleaded for medication. Clarke and Demerieux ignored his pleas while their co-conspirator, Winston Gittens, used Maharaj’s worsening health as leverage to demand three million Trinidadian dollars from his family. Bound and gagged, Maharaj. repeatedly refused defendants’ attempts to record a “proof of life” video, even when Straker threatened to harm Maharaj’s son. After six days in captivity, Maharaj slipped into a diabetic coma and died. Well aware that they had killed a United States citizen, defendants voted to conceal their crime. “No body, no evidence, no case,” proclaimed Pierre. Using a machete and their bare hands, Clarke, Demerieux, and Pierre removed Maharaj’s internal organs and dismembered his body. They packed the remains in Styrofoam containers and buried them in the woods. As with most buried secrets, however, defendants’ misdeeds .eventually surfaced. In late 2005, the Trinidad and Tobago Police Service began an investigation of defendants’ hostage-taking ring. Assisted by the FBI, Trinidadian police ultimately uncovered evidence of Maharaj’s death. The United States extradited defendants and charged them with conspiracy and hostage-taking resulting in death in violation of the Hostage Taking Act. The facts and circumstances surrounding the kidnapping are largely undisputed — indeed, five of the seven defendants confessed. Defendants primarily argue that Maharaj misrepresented key facts on his immigration applications, thus negating his United States citizenship — an essential element of a Hostage Taking Act prosecution. The district court rejected this argument, as well as numerous other objections. After a ten-week trial, the jury convicted defendants of all charges, and the district court sentenced them to life imprisonment without the possibility of release. Defendants now appeal their convictions on numerous grounds.' We address the arguments pertaining to Maharaj’s citizenship in Part II and then consider defendants’ other arguments in Parts III through XI. II. Enacted to fulfill the United States’ obligations under the International Convention Against the Taking of Hostages, the Hostage Taking Act, 18 U.S.C. § 1203, makes extraterritorial hostage-taking a criminal offense when the victim is a United States national. On appeal, it is undisputed that at the time of his death Balram Maharaj possessed an authentic certificate of naturalization. Before trial, however, defendants uncovered evidence they claim demonstrates that Maharaj obtained his naturalization through fraud. According to this evidence, Maharaj, formerly Aladdin Barlow John, first entered the United States in 1967 as a non-immigrant transit en route to Canada. Following a short visit there, Maharaj returned to the United States, briefly settling in New York before enlisting in the Army in 1968. Ha deserted seven months later. In order to avoid prosecution for desertion, Maharaj completed a clemency program and was ultimately discharged from the Army as undesirable. In April 1986, the Immigration and Naturalization Service (INS), having discovered that Maharaj had overstayed his 1967 transitory permit, ordered him to leave the United States. But instead of leaving, he petitioned INS for permanent-resident alien status, also known as a green card. Asked on his green card application whether he had ever been convicted of a crime involving moral turpitude, Maharaj checked “no” even though, defendants claim, he had once pleaded guilty to petty larceny and was on probation. Maharaj also checked “no” when asked whether he had ever suffered an “attack of insanity,” narcotic drug addiction, or chronic alcoholism, even though his ex-wife would years later attest in an unrelated proceeding that he had previously spent time in a mental-health facility, attempted suicide, and chronically abused alcohol and prescription drugs. INS granted Maharaj’s petition. After maintaining permanent-resident status for five years, Maharaj applied for full naturalization in 1994. On the application, Maharaj checked “no” when asked whether he had ever been ordered deported. And when asked whether he had ever knowingly committed a crime for which he had not been arrested, he cheeked “no” even though his ex-wife had testified that he once physically assaulted and raped her. INS granted Maharaj citizenship in 1995. The question whether Maharaj’s misrepresentations negated his citizenship and, in turn, defendants’ guilt, was the most contested issue throughout the proceedings in the district court. Defendants first raised the issue in a motion to dismiss the indictment, arguing that because conviction under the Hostage Taking Act requires U.S. citizenship and because Maharaj’s frau,d negated his citizenship, the district court lacked jurisdiction. The district court disagreed. Citing a long and unbroken line of Supreme Court precedent, see, e.g., United States v. Zucca, 351 U.S. 91, 95 & n. 8, 76 S.Ct. 671, 100 L.Ed. 964 (1956); see also Bindczyck v. Finucane, 342 U.S. 76, 83, 72 S.Ct. 130, 96 L.Ed. 100 (1951), the district court held that 8 U.S.C. § 1451, which permits the United States Attorney to institute denaturalization proceedings in a federal district court, is the exclusive procedure for voiding the citizenship of a person naturalized due to fraud. United States v. Clarke, 628 F.Supp.2d 1, 9 (D.D.C.2009). Citizenship, the court held, remains valid until a district court, acting upon a United States Attorney’s section 1451 motion, determines that naturalization was “procured by concealment of a material fact or fraud.” Id. at 6 (quoting 8 U.S.C. § 1451(a)). Given that no district court had ever made such a finding as to Maharaj, the court denied the motion. Clarke, 628 F.Supp.2d at 10. The district court also granted the government’s motion in limine to exclude from trial any evidence regarding Maharaj’s alleged fraud. Id. at 13. Conviction under the Hostage Taking Act, the court held, requires the government to prove that the victim acquired citizenship by birth or naturalization. Id. at 13. Evidence disputing whether the victim should have been naturalized or the circumstances surrounding naturalization is irrelevant. Id. The court therefore rejected defendants’ argument that they had a Sixth Amendment right to present evidence regarding Maharaj’s alleged fraud to the jury. Id. at 14. “[T]he jury,” the district court concluded, “may not decide the validity of Maharaj’s citizenship.” Id. at 13. At trial, the government offered Maharaj’s certificate of naturalization and two passports as evidence of his citizenship. An employee of the U.S. Citizenship and Immigration Service testified that INS issued the certificate of naturalization in 1995. A State Department Fraud Program Manager testified that Maharaj’s two passports, issued in 1995 and 2000, were authentic. On cross examination, however, defense counsel pointed out that Maharaj’s 1995 passport was unsigned and, as the witness conceded, invalid. See 22 C.F.R. § 51.4(a) (“A passport book is valid only when signed by the bearer in the space designated for signature[.]”). The district court struck the 1995 passport from evidence and ruled that challenges to the authenticity of the 2000 passport, issued solely on the basis of the stricken 1995 passport, were fair game. In sum, then, the district court permitted defendants to argue that the citizenship documents were inauthentic, i.e., forged or counterfeit, but barred them from collaterally attacking INS’s decision to naturalize Maharaj. In its final instructions, the district court reminded the jury that the victim’s citizenship was an essential element of the crime that the government must prove beyond a reasonable doubt. The 2000 passport and the naturalization certificate, the court instructed, may be considered as evidence of citizenship. In closing arguments, defense counsel pressed this point, claiming that the invalidity of the 1995 passport called into question the authenticity of the 2000 passport, as well as the naturalization certificate. After the jury returned a guilty verdict, defendants, reiterating their earlier arguments, moved for a judgment of acquittal, or alternatively a new trial. Relying on its prior reasoning, the district court denied the motion. United States v. Clarke, 767 F.Supp.2d 12, 65 (D.D.C.2011). The criminal trial was not the only forum in which defendants challenged Maharaj’s citizenship. Prior to trial, defendants submitted affidavits to the United States Attorney for the District of Columbia detailing Maharaj’s alleged fraud and requesting initiation of section 1451 proceedings. When the U.S. Attorney took no action, defendants petitioned the district court for a writ of mandamus requiring the U.S. Attorney to initiate denaturalization proceedings. The' government moved to dismiss pursuant to Federal Rules of Civil Procedure 12(b)(1) for lack of jurisdiction and 12(b)(6) for failure to state a claim. The district court granted the Rule 12(b)(1) motion, finding that defendants lacked constitutional standing because they failed to demonstrate that their requested remedy — initiation of section 1451 proceedings — would redress their claimed injury, i.e., denial of the right to present evidence in the criminal trial. It is “wholly speculative,” the court concluded, whether the U.S. Attorney could meet the high burden of proof necessary to denaturalize Maharaj, “especially in light of evidentiary problems that may arise so long after Maharaj’s death and his inability to defend himself.” Clarke v. Holder, 767 F.Supp.2d 106, 109 (D.D.C.2011). Alternatively, the district court found that even if defendants could establish standing, they failed to meet the threshold requirements for mandamus relief because the U.S. Attorney had no clear duty to seek posthumous denaturalization of Maharaj. Id. at 112— 13. On appeal, defendants argue that the district court lacked jurisdiction under the Hostage Taking Act and erred in excluding evidence contesting the validity of Maharaj’s naturalization. Defendants also appeal the district court’s denial of their petition for a writ of mandamus. We address each issue in turn. The Criminal Conviction Defendants argue that “[s]ubject matter jurisdiction ... hinges entirely on Mr. Maharaj’s citizenship.” Def. Br. 12. This, however, misapprehends the nature of federal court jurisdiction in criminal cases. A claim that an element of the offense is unsatisfied — that the victim was not a United States citizen, for example— goes only to a defendant’s guilt or innocence. In other words, jurisdiction hinges not on the merits, but rather on the court’s constitutional or statutory power to adjudicate the case. Lamar v. United States, 240 U.S. 60, 64, 36 S.Ct. 255, 60 L.Ed. 526 (1916) (“Jurisdiction is a matter of power, and covers wrong as well as right decisions.”). Under 18 U.S.C. § 3231, federal district courts possess statutory authority over “all offenses against the laws of the United States.” Because violation of the Hostage Taking Act is an offense against the laws of the United States, our jurisdictional inquiry ends and we turn to the merits of defendants’ appeal. United States v. Fahnbulleh, 752 F.3d 470, 476 (D.C.Cir.2014) (“If an indictment or information alleges the violation of a crime set out in Title 18 or in one of the other statutes defining federal crimes, that is the end of the jurisdictional inquiry.”) (internal quotation marks omitted). Defendants make two arguments. They first claim that the district court’s exclusion of evidence regarding Maharaj’s alleged fraud violated their constitutional right to present a complete defense. They also argue that the district court removed the citizenship question from the jury, thus relieving the government of its burden to prove Maharaj’s citizenship beyond a reasonable doubt. With respect to both arguments, our starting point is the text of the Hostage Taking Act. Section 1203(b)(1)(A) criminalizes hostage-taking that occurs outside the United States if “the person seized or detained is a national of the United States.” A “national of the United States” is, in turn, defined by reference to the Immigration and Nationality Act as “a citizen of the United States.” 18 U.S.C. § 1203(c); 8 U.S.C. § 1101(a)(22). By its plain language, then, section 1203 broadly protects United States citizens. The statute imposes no restriction on this protection. It does not, for example, exclude citizens who, in retrospect, are unworthy of the honor. Nor does it exclude persons whose citizenship might at some later time be invalidated. In other words, section 1203 protects victims according to their status at the time of the hostage-taking. True, section 1203 is written in the present tense — the statute applies if “the person seized or detained is a national of the United States.” But that clause appears in a criminal statute that requires examination of past events — whether the victim was seized or detained. See Cullen v. Pinholster, 563 U.S. 170, 131 S.Ct. 1388, 1398, 179 L.Ed.2d 557 (2011) (use of backward-looking language such as “resulted in” and “involved” in federal habeas statute, 28 U.S.C. § 2254(d), requires examination of the state-court decision at the time it was made). A more familiar statute, the Armed Career Criminal Act, provides a helpful parallel. The ACCA imposes a sentence enhancement if the defendant “has three previous convictions ... for ... serious drug offense[s]” for which “a maximum term of imprisonment of ten years or more is prescribed by law.” 18 U.S.C. § 924(e) (emphasis added). As the Supreme Court has explained, “[t]he plain text of ACCA” therefore “requires the court to determine whether a ‘previous conviction’ was for a serious drug offense.” McNeill v. United States, 563 U.S. 816, 131 S.Ct. 2218, 2221-22, 180 L.Ed.2d 35 (2011) (emphases added). To answer that “backward-looking question,” the Court held that the sentencing court must “consult the law that applied at the time of that conviction.” Id. at 2222. So too here. Determining whether an American citizen was seized or detained under the Hostage Taking Act requires examination of the victim’s status at the time of the abduction. This focus on status at the time of the crime is hardly unusual. For instance, convictions under the federal statute that bars felons from possessing firearms rest solely upon the fact of the prior felony. Lewis v. United States, 445 U.S. 55, 62-65, 100 S.Ct. 915, 63 L.Ed.2d 198 (1980). The validity of the prior conviction is irrelevant even when that conviction is patently unconstitutional. Id.; see also Custis v. United States, 511 U.S. 485, 493-97, 114 S.Ct. 1732, 128 L.Ed.2d 517 (1994) (defendants in federal sentencing proceedings may not, with a narrow exception for certain convictions obtained in violation of the right to counsel, challenge the validity of prior state convictions used to enhance sentences under the Armed Career Criminal Act). Likewise, defendants accused of providing material support to designated foreign terrorist organizations in violation of 18 U.S.C. § 2339B may not challenge the validity of the designation. United States v. Hammoud, 381 F.3d 316, 331 (4th Cir.2004) (en banc) (“Congress has provided that the fact of an organization’s designation as an FTO is an element of § 2339B, but the validity of the designation is not.”), rev’d on other grounds, 543 U.S. 1097, 125 S.Ct. 1051, 160 L.Ed.2d 997 (2005); see also United States v. Mandel, 914 F.2d 1215, 1222 (9th Cir.1990) (defendants charged with illegally exporting items on the Secretary of Commerce’s Commodity Control List may not challenge the validity of the Secretary’s designation). Finally, our interpretation of the Hostage Taking Act reinforces its purpose. When President Reagan proposed the bill that ultimately became the Act, he declared that it would “send a strong and vigorous message to friend and foe alike that the United States will not tolerate terrorist activity against its eitizens[.]” President’s Message to the Congress Transmitting Proposed Legislation to Combat International Terrorism, Pub. Papers, Admin, of Ronald Reagan 3-4 (Apr. 26, 1984). This “strong and vigorous message” would be severely diluted if foreign nationals could target American citizens for abduction and then avoid prosecution in the United States by impugning the victim’s character. This is especially true where, as here, defendants targeted the victim not only because he was an American, but also because he had assets in the United States. Permitting them to escape prosecution by arguing that Maharaj was undeserving of United States citizenship would weaken the protection Congress intended to extend to Americans abroad. For all of these reasons, the district court properly excluded evidence of Maharaj’s alleged fraud as irrelevant. Congress has vested sole naturalization authority in the Attorney General, 8 U.S.C. § 1421(a), and a certificate of naturalization represents conclusive evidence of the Attorney General’s determination, Tutun v. United States, 270 U.S. 568, 577, 46 S.Ct. 425, 70 L.Ed. 738 (1926); 8 U.S.C. § 1443(e). As explained above, whether the Attorney General, acting through INS, should have issued a certificate to Maharaj — as opposed to whether the certificate was itself authentic — is irrelevant under the Hostage Taking Act. The Mandamus Proceedings This brings us to defendants’ appeal of the district court’s denial of their petition for a writ of mandamus requiring the United States Attorney to initiate posthumous denaturalization proceedings against Maharaj under 8 U.S.C. § 1451. Recall that the district court denied the petition on the grounds that defendants lacked standing and, alternatively, that they failed to meet the requirements for mandamus relief. Clarke v. Holder, 767 F.Supp.2d 106, 116 (D.D.C.2011). Because we agree with the former ruling, we need not address the latter. In order to have Article III standing to bring their mandamus action, defendants must prove that they suffered (1) an “injury in fact” that is (2) “fairly ... traceable to the challenged action,” and that is (3) likely to be “redressed by a favorable decision.” Lujan v. Defenders of Wildlife, 504 U.S. 555, 560-61, 112 S.Ct. 2130, 119 L.Ed.2d 351 (1992) (internal quotation marks omitted). The district court found that defendants 'failed to establish the third element, redressability, ie., “a substantial likelihood that the relief requested will redress the injury claimed.” Duke Power Co. v. Carolina Envtl. Study Grp., Inc., 438 U.S. 59, 75 n. 20, 98 S.Ct. 2620, 57 L.Ed.2d 595 (1978) (internal quotation marks omitted). Reviewing defendants’ evidence, the district court thought it “wholly speculative” whether the government could meet section 1451’s burden of proving fraud by “clear, unequivocal, and convincing evidence.” Clarke, 767 F.Supp.2d at 109, 112 (quoting Fedorenko v. United States, 449 U.S. 490, 505, 101 S.Ct. 737, 66 L.Ed.2d 686 (1981)). Given this, defendants’ alleged “injury — their hostage taking convictions — is not redress-able by an order directing [the government] to initiate [a section 1451] proceeding.” Clarke, 767 F.Supp.2d at 109. Defendants challenge the district court’s conclusion on two grounds. First, they claim that their “injury is the lack of opportunity to present a defense,” which is “necessarily ... redressed if the Government is compelled to bring a § 1451 hearing against Mr. Maharaj.” Def. Mandamus Br. 9. Second, they argue that the district court’s evidentiary rulings in the criminal trial, together with its dismissal of their mandamus petition, ensnare them in a catch-22: defendants “could not show the citizenship evidence at trial because the district court had ruled that a Government-initiated § 1451 proceeding was the exclusive avenue for presenting such evidence, but they could not present the evidence in a § 1451 proceeding because the district court would not allow them to challenge the Government’s- inaction.” Id. at 8. Both arguments are foreclosed by the conclusion we reached in the previous section, i.e., that conviction under the Hostage Taking Act depends upon the victim’s citizenship at the time of the crime. Balram Maharaj possessed American citizenship when defendants abducted him in 2005. Whatever happens now is irrelevant. In other words, even if the government were to strip Maharaj of his United States citizenship under section 1451, defendants’ convictions would stand because Maharaj possessed a valid naturalization certifícate at the time of the crime. Defendants’ alleged injury is therefore incapable of redress not because the outcome of the section 1451 proceeding is too speculative (as the district court found), but because the outcome of those proceedings could not possibly affect defendants’ right to present a full defense to a charge under the Hostage Taking Act. We reach this conclusion despite section 1451’s relation-back provision, which provides that denaturalization on account of fraud “shall be effective as of the original date of the [naturalization] order and certificate.” 8 U.S.C. § 1451(a). Based on this provision and pre-section 1451 case law espousing the relation-back principle, defendants argue that denaturalization on account of Maharaj’s fraud would retroactively void his naturalization, meaning that he never possessed citizenship and thus was not a national under the Hostage Taking Act. Def. Mandamus Br. 11 n. 8. The Supreme Court, however, has rejected mechanical application of section 1451’s relation-back principle. Costello v. Immigration and Naturalization Service, 376 U.S. 120, 130, 84 S.Ct. 580, 11 L.Ed.2d 559 (1964). In Costello, the Court examined section 1451’s language and legislative history to determine whether the provision applied to the “general deportation provisions” of the Immigration and Nationality Act. Id. at 129, 84 S.Ct. 580. The Court concluded that Congress intended the relation-back provision simply to codify preexisting case law that retroactively voided fraudulently-acquired naturalization for the purpose of determining derivative citizenship, ie., citizenship conveyed to children through the naturalization of one or both parents. Id. Calling relation-back a “legal fiction,” the Court refused, absent express congressional command, to extend that fiction to require deportation of a, denaturalized individual on the theory that crimes committed prior to his denaturalization rendered him an alien despite possession of a then-valid naturalization certificate. Id. at 130, 132, 84 S.Ct. 580. Likewise, nothing in the text or legislative history of section 1451' — or the Hostage Taking Act — suggests that Congress intended the relation-back doctrine to apply in a criminal prosecution. Indeed, as explained above, applying the doctrine in this context would weaken the purpose of the Hostage Taking Act. See supra at 585. We therefore decline, as did the Supreme Court in Costello, to extend the legal fiction of relation-back into the realm of criminal law. See Costello, 376 U.S. at 130, 84 S.Ct. 580. III. Objections to Introduction of Evidence of Other Crimes Defendants challenge the district court’s admission, over their repeated objections, of evidence of three other, uncharged hostage takings that occurred within four months of the Maharaj hostage taking. We conclude that the district court did not err in admitting that evidence. It was relevant under Rule 404(b) as background showing how the conspiracy formed and certain defendants’ intent. Given the care the district court took to limit and focus the evidentiary presentation, its prejudicial potential did not outweigh its probative value under Rule 403. A. Before trial, the government submitted a> notice of its intent to introduce evidence that some of the defendants participated in uncharged hostage takings. The district court admitted the evidence as relevant to issues other than the defendants’.bad character — namely “the background of the conspiracy and how the relationships between the participants developed, as well as defendants’ motive, intent, knowledge, preparation, and plan.” United States v. Straker, 567 F.Supp.2d 174, 178-79 (D.D.C. 2008). The evidence was strongly probative, in the district court’s view, and thus its value was not substantially outweighed by the “fairly low” danger of unfair prejudice. Id. at 179. The district court restricted the government’s presentation of the “other crimes” evidence in order to avoid “unnecessary presentation of cumulative evidence and to minimize the danger of unfair prejudice.” Id. The district court precluded the government from introducing any evidence concerning the Gopaul hostage taking — an offense in which the hostage takers apparently killed the victim after they received an unsatisfactory ransom offer — because that evidence presented “the most likely case for some degree of unfair prejudice.” Id. Additionally, the court permitted the government to introduce only three of the four other uncharged hostage takings the government identified, id.; see also J.A. 961-62, and limited the testimony on those to three hours each. J.A. 3171-72. The government accordingly introduced at trial evidence that defendants Pierre and De Four participated in three of the uncharged hostage takings, and that defendant Clarke participated in two of them. The evidence came in through the testimony of the four cooperating co-conspirators and six additional witnesses, including the victim of one hostage taking and Trinidadian law enforcement officials involved in investigating the other hostage takings. The district court gave limiting instructions to the jury periodically during the trial, informing it that the evidence of other hostage takings was admissible only against the specific defendants the jury found were involved in them and explaining the purposes for which the evidence could, and could not, be considered. See Clarke, 767 F.Supp.2d at 27-28. B. On appeal, defendants argue that the district court’s admission of evidence of the three uncharged hostage takings violated Rule 404(b) because it was not admitted for any of the valid purposes enumerated in that Rule, but impermissibly to show their bad characters and propensity to commit the charged crimes. The evidence’s probative value was substantially outweighed by its unfair prejudicial effect, they assert, and was presented in a confusing and prejudicial manner, so should have been excluded under Rule 403. The evidentiary limitation in Rule 404(b) implements the fundamental tenet of our criminal justice system that defendants may be convicted only for violating the law, not for being bad people. United States v. Sutton, 801 F.2d 1346, 1360 (D.C.Cir.1986). The Rule prohibits the admission of evidence of a crime, wrong, or other bad act “when offered for the purpose of proving that a defendant acted in conformity with his character, but allows admission so long as the evidence is offered for any other relevant purpose.” United States v. Lawson, 410 F.3d 735, 741 (D.C.Cir.2005); United States v. Bowie, 232 F.3d 923, 930 (D.C.Cir.2000); see also Fed.R.Evid. 404(b)(1). Relevant, non-propensity purposes include “proving motive, opportunity, intent, preparation, plan, knowledge, identity, absence of mistake, or lack of accident.” Fed.R.Evid. 404(b)(2). Even if a court determines that the prosecution’s other-crimes evidence is relevant to an issue apart from propensity, the evidence may nonetheless be excluded under Rule 403 if its “probative value is substantially outweighed by a danger of ... unfair prejudice, confusing the issues, misleading the jury, undue delay, wasting time, or needlessly presenting cumulative evidence.” Fed.R.Evid. 403; Bowie, 232 F.3d at 930. Even if it is concededly relevant, unduly prejudicial evidence may be excluded to prevent jurors from impermissibly relying on biases, dislikes, or the emotional impact of the evidence, for example by drawing on assumptions about a defendant’s bad character, rather than proof of the criminal conduct charged. The Rule’s requirement that the danger of unfair prejudice substantially outweigh probative value calls on us, in close cases, to lean towards admitting evidence. United States v. Douglas, 482 F.3d 591, 600 (D.C.Cir.2007); United States v. Manner, 887 F.2d 317, 322 (D.C.Cir.1989). We review a district court’s admission of other-crimes evidence for abuse of discretion, United States v. Mathis, 216 F.3d 18, 25-26 (D.C.Cir.2000), according “substantial deference” to the district court, Lawson, 410 F.3d at 741; see also United States v. Long, 328 F.3d 655, 660 (D.C.Cir.2003). We review Rule 403 balancing decisions “only for grave abuse.” Douglas, 482 F.3d at 596 (internal quotation marks omitted). 1. The district court did not abuse its discretion in concluding that evidence of particular defendants’ involvement in uncharged hostage takings was relevant to both how those defendants started to .work together as kidnappers, and their motive and intent to kidnap wealthy civilians to extort ransom money. “In a conspiracy prosecution, the government is usually allowed considerable leeway in offering evidence of other offenses” to, for example “inform the jury of the background of the conspiracy charged” or “help explain to the jury how the illegal relationship between the participants in the crime developed.” Mathis, 216 F.3d at 26 (internal quotation marks omitted); see Manner, 887 F.2d at 322. Evidence that defendants jointly engaged in other criminal activity can be relevant to shed light on how the “relationship of mutual trust” developed between those individuals. See United States v. Escobar-de Jesús, 187 F.3d 148, 169 (1st Cir.1999) (collecting cases). The district court .admitted the other-crimes evidence as tending to show that a criminal relationship formed between Pierre, De Four, Clarke, and the cooperating co-conspirators during other, uncharged hostage takings. That prior criminal relationship helped to explain how Pierre, De Four, and Clarke knew they could rely on one another during the Maharaj hostage taking. The district court did not impermissibly admit evidence of the uncharged hostage takings merely to allow the government to provide the jury with general background information that completed the prosecutor’s narrative, see Bowie, 232 F.3d at 929, but rather admitted it as tending to establish how the defendants in this case formed the Maharaj hostage-taking conspiracy. The uncharged hostage takings were also relevant to establish the defendants’ state of mind. Information showing that Pierre, De Four, and Clarke had worked closely before on very similar hostage takings helped to dispel any doubt as to whether they knowingly and intentionally joined together to carry out these crimes in order to extract significant ransoms. Intent, knowledge, and motive are “well-established non-propensity purposes for admitting evidence of prior crimes or acts.” Bowie, 232 F.3d at 930; see also Fed.R.Evid. 404(b)(2). As we have previously observed, evidence relevant to intent and motive “is particularly probative where the government has alleged conspiracy.” Mathis, 216 F.3d at 26 (internal quotation marks omitted). To prove conspiracy to commit hostage taking, the government was required to establish that the conspiracy was knowingly formed and that defendants willfully participated in the plan to commit it with the intent to further some purpose of the conspiracy. See United States v. Yunis, 924 F.2d 1086, 1096 (D.C.Cir.1991). Potential juror doubt about whether any of these three defendants was somehow mistakenly swept up into activities he did not know were part of a criminal conspiracy is powerfully undermined by the evidence of similar criminal teamwork with some of the same people, both before and after the Maharaj hostage taking. Any questions about motive also -tended to be put to rest by evidence that the conspirators successfully obtained ransoms in the other, uncharged hostage takings. The district court thus permissibly held that the other-crimes evidence was relevant for non-propensity purposes. 2. That conclusion does not end our inquiry. Even where other-crimes evidence is relevant for a non-propensity purpose, it nevertheless is inadmissible under Rule 403 if the potential for prejudice from introducing the evidence outweighs its probative value. See Douglas, 482 F.3d at 600. We see no reason, however, to disturb the district court’s carefully reasoned Rule 403 determination. See Straker, 567 F.Supp.2d at 179. The district judge identified the strong probative value of the evidence for the purposes we have just discussed: to show the defendants’ willingness to trust one another and work together to kidnap civilians as a means to extort ransom money. The danger of unfair prejudice was minimal because the other-crimes evidence added “‘no emotional or other pejorative emphasis not already introduced by the evidence’” of the crime charged in this case. Id. (quoting Lawson, 410 F.3d at 742). Indeed, the facts of this kidnapping are significantly more damning because, unlike the uncharged other crimes the judge allowed the prosecution to establish, this one went awry. After defendants held their victim, an American citizen, hostage for seven days without necessary diabetes medication^ he died. Several defendants then dismembered his body with a machete and packed the pieces in two large coolers in an effort to conceal their crime. The district court effectively barred cumulative evidentiary presentations and used safeguards to minimize any potential prejudice from the admission of the other-crimes evidence. In addition to limiting the number of other crimes about which the prosecution could introduce evidence, and strictly rationing the trial time allowed for those evidentiary presentations, Straker, 567 F.Supp.2d at 179; see also J.A. 961-62; J.A. 3171-72, the court paid careful attention to the nature of the testimony that was introduced and prevented the government from soliciting testimony about particularly prejudicial details. See J.A. 3462-66; J.A. 3686-88. The court excluded all evidence of two uncharged hostage takings, including the one that raised the greatest risk of unfair prejudice. As already noted, the evidence in that kidnapping suggested that the captors intentionally killed their hostage in response to an insufficient ransom offer, see Straker, 567 F.Supp.2d at 179 — a response arguably even more brutal than the deprivation of life-sustaining medication that led to the predictable demise of the victim in this case. Defendants further argue that the uncharged hostage-takings evidence was “wholly unnecessary” to the government’s case,' such that any minimal probative value was substantially outweighed by the prejudice it caused, in violation of Rule 403. Defendants find unpersuasive the theories discussed above, regarding the probativeness of the other-crimes evidence to questions of relationship, motive, and intent; they contend that the government had ample, less prejudicial ways to make the same points. Defendants contend that the evidence was unnecessary to show their relationships, given other evidence that the defendants formed friendships in boyhood or during military service. That evidence, however, does not speak to their repeated experience of trusting one another to carry through with felonious conspiracies without revealing their activities to law enforcement. Defendants also contend that the evidence of the uncharged hostage takings was unnecessary to show their intent or motive because they did not raise innocent-motive defenses. That argument ignores the government’s burden, regardless of the nature of the defense, to prove beyond a reasonable doubt that defendants knowingly and intentionally joined the conspiracy to kidnap Maharaj. See Douglas, 482 F.3d at 597; Old Chief v. United States, 519 U.S. 172, 187-88, 117 S.Ct. 644, 136 L.Ed.2d 574 (1997). In sum, the district judge’s careful sorting of the other-crimes evidence and the limitations he placed on how much other-crimes evidence the prosecution could use successfully allowed the evidence which was most probative while avoiding unfair prejudice. The district court did not abuse its discretion by refusing to exclude the evidence of the uncharged hostage takings under Rule 403. 3. Lastly, we turn to defendants’ argument that the government presented evidence of the uncharged hostage takings in such a “disorganized and confusing fashion” that they were prejudiced. Def. Br. 56 (citing United States v. Sampol, 636 F.2d 621, 645 (D.C.Cir.1980); United States v. Foskey, 636 F.2d 517, 524 n. 6 (D.C.Cir.1980)). As we discuss more fully, infra, in connection with the post-trial severance motion, the district court concluded that the jury was able to correlate the evidence with each defendant against whom it was properly introduced, and to avoid spillover consideration of evidence against defendants to whom it did not relate. See Clarke, 767 F.Supp.2d at 26-27. Defendants have failed to persuade us that the district court abused its discretion in so concluding. The government presented evidence about the other hostage takings, in part through testimony of cooperating co-conspirators who spoke about some of the uncharged offenses before describing the charged offense in greater detail. Because the cooperating co-conspirators generally testified about the hostage takings in chronological order and each hostage taking involved a different victim, the jury was provided with clear guideposts with which to differentiate and compartmentalize each event. The district court repeatedly and carefully instructed the jury as to which defendants were involved in which of the other crimes, and cautioned the jurors to consider evidence only against those specific defendants, thereby protecting all of the defendants against any potential confusion stemming from the other-crimes evidence. See Long, 328 F.3d at 662 (“[Limiting instructions ordinarily suffice to protect the defendant’s interests.”) (citing Spencer v. Texas, 385 U.S. 554, 561, 87 S.Ct. 648, 17 L.Ed.2d 606 (1967)). The court gave limiting instructions concerning the evidence of other hostage takings six times throughout the course of the trial: after opening statements, the first time that the other-crimes evidence was introduced by the government, at several points during the trial, and as part of the final jury instructions. In the various instructions, the court cautioned the jury that the evidence of the uncharged hostage takings was admissible for only limited purposes: informing the jurors of the background of the conspiracy, helping them decide whether there were relationships between the co-conspirators, and aiding their determinations as to whether the defendants had motive, intent, knowledge, or a plan to commit the Maharaj hostage taking. The court also made clear that the other-crimes evidence was admissible only against the particular defendants the jurors found were involved in those other crimes, alternating between specifically naming the defendants involved in the other hostage takings and generally referring to those defendants. Juries are presumed to follow instructions that caution them to draw only permissible inferences from Rule 404(b) evidence. See United States v. Brown, 597 F.3d 399, 405-06 (D.C.Cir.2010). The district court’s detailed instructions' guided the jurors to compartmentalize and properly consider the evidence of uncharged hostage takings. Defendants have identified a handful of examples where the district judge expressed some confusion regarding the presentation of evidence of the uncharged hostage takings. In several instances, the district court identified potential confusion only to reinforce to the prosecution the importance of making clear to the jury which hostage taking the witness was discussing. Several references to confusion were made during bench conferences out of the jury’s hearing. The record portions on which defendants rely demonstrate not that the evidence was presented in a misleading fashion, but rather that the district court took great care to reduce the potential for confusion. The isolated examples of confusion identified by defendants do not show that the other-crimes evidence was so unclear or misleading that the jury was unable to follow the district court’s limiting instructions. Defendants claim the government’s closing argument intensified confusion about the other-crimes evidence by stating that the same “crew” or “organization” carried out all of the hostage takings, and that Pierre was the “godfather” of the crew. Contrary to defendants’ characterization, however, no prosecutor argued that Pierre’s crew committed all of the hostage takings. The government’s closing did refer to the “crew” or group of individuals involved in the Maharaj hostage taking. The government also argued that Pierre was involved in all of the hostage takings, and indeed was in charge during each of those hostage takings. The evidence presented at trial supported each of those points. What the government did not assert was that Pierre led the same crew in each instance, and the trial evidence and judge’s instructions protected against any such conclusion. For all of these reasons, we conclude that the district court acted within its sound discretion in admitting at trial evidence about three uncharged hostage takings. IV. Confrontation Clause Challenge to Use of Codefendants’ Statements Defendants next contend that the district court violated their Sixth Amendment rights to confront the witnesses against them when it admitted into evidence redacted confessions made by their fellow defendants that inculpated them in the Maharaj hostage taking. We conclude that the admission of the redacted confessions did not violate the Confrontation Clause, with the exception of the violation acknowledged by the government, which, in view of the independent and overwhelming evidence in support of the conviction, was harmless. A. The conspiracy was alleged to have involved at least a dozen men, seven of whom were tried jointly in this case. After they were arrested, five of the defendants gave statements to law enforcement officials confessing their own participation in the Maharaj hostage taking. Before trial, the government filed a notice of its intent to introduce at trial the out-of-court statements made by codefendants Clarke, Demerieux, and Sealey. Defendants objected on Sixth Amendment grounds to the admission of the statements and also moved to sever their trials. The district court concluded that separate trials would not be necessary because each statement could be adequately redacted and other safeguards used in order fully to protect the non-declarant defendants’ Sixth Amendment confrontation rights. The district court ordered the government to redact the statements to remove references identifying defendants other than the declarant whenever possible. When full redaction was not possible, the court instructed the government to replace a particular name with a neutral term, sufficient to “avoid creating an inevitable association with a particular defendant or defendants when the statement is viewed together with other evidence.” J.A. 2079. The court also provided the government with a detailed set of guidelines, recounted below, specifying the types of neutral terms that would be acceptable and those that would not, along with other safeguards of defendants’ confrontation rights. B. Defendants contend that the use of neutral-pronoun redactions was inadequate, and that the Confrontation Clause instead demands full redaction of codefendants’ confessions to eliminate any reference to fellow defendants that jurors might infer to be references to non-declarant defendants. See Richardson v. Marsh, 481 U.S. 200, 107 S.Ct. 1702, 95 L.Ed.2d 176 (1987). Alternatively, redactions Used here were inadequate because, defendants claim, when a redacted confession was considered alongside the other evidence presented at trial, it inevitably pointed an inculpatory finger at a particular defendant, contrary to Gray v. Maryland, 523 U.S. 185, 118 S.Ct. 1151, 140 L.Ed.2d 294 (1998). The government responds that the Confrontation Clause does not require removal of all references to defendants, anonymized as the remaining references were, and that full redaction would have substantially diminished the value of the statements against the declarants themselves. Full redactions were not always practicable in this case, the government contends, because the declarants were charged with conspiracy. Eliminating from their statements all references to their co-conspirators jointly on trial would have “deprived the government of powerful conspiracy evidence” that it was entitled to use. Gov’t Br. 94., We review de novo the district court’s legal conclusions under the Confrontation Clause, United States v. Wilson, 605 F.3d 985, 1003 (D.C.Cir.2010), and subject to harmless-error analysis any legal errors it may have made, United States v. Moore, 651 F.3d 30, 69 (D.C.Cir.2011). 1. The Confrontation Clause of the Sixth Amendment provides a criminal defendant with the right “tn be confronted with the witnesses against him,” including the right to cross-examine those witnesses. U.S. Const. Amend. VI; see Pointer v. Texas, 380 U.S. 400, 404, 85 S.Ct. 1065, 13 L.Ed.2d 923 (1965). Use of a defendant’s own confession against him raises no confrontation issues. The admission of a codefendant’s confession implicating another defendant, however, poses special risks to the defendant’s confrontation rights. When the declarant expressly implicates another defendant yet renders himself unavailable for cross examination by asserting his Fifth Amendment right not to testify, use of his statement violates the defendant’s right to confront his accuser. See Bruton v. United States, 391 U.S. 123, 137, 88 S.Ct. 1620, 20 L.Ed.2d 476 (1968). But at least when (1) the jury is instructed to consider the confession against the declarant only, and (2) redactions are made such that the statement, together with other trial evidence, neither expressly identifies defendants nor creates any inevitable association between them and the criminal activity the statement describes, there is no Sixth Amendment violation. See Richardson, 481 U.S. 200, 211, 107 S.Ct. 1702; United States v. Washington, 952 F.2d 1402, 1406-07 (D.C.Cir.1991). The framework for analyzing limitations on use of codefendants’ statements is established by a trilogy of Supreme Court Confrontation Clause cases: Bruton, 391 U.S. 123, 88 S.Ct. 1620, Richardson, 481 U.S. 200, 107 S.Ct. 1702, and Gray, 523 U.S. 185, 118 S.Ct. 1151. The district judge who presided over defendants’ joint trial in Bruton admitted into evidence a non-testifying codefendant’s confession incriminating the defendant. 391 U.S. at 124, 88 S.Ct. 1620. The Supreme Court found a Sixth Amendment violation, overruling a prior decision sustaining a conviction in similar circumstances, because that precedent placed unwarranted confidence in the efficacy of limiting jury instructions. Id. at 126, 88 S.Ct. 1620 (overruling Delli Paoli v. United States, 352 U.S. 232, 77 S.Ct. 294, 1 L.Ed.2d 278 (1957)). The Bruton Court acknowledged that it is unrealistic to expect a jury to rely on a statement when deciding the guilt of the confessing codefendant, yet ignore the same statement when considering the guilt of the defendant it mentions as an accomplice. Id. at 131, 135-36, 88 S.Ct. 1620. Justice Stewart summed up the inadequacy of limiting jury instructions in such settings: “A basic premise of the Confrontation Clause ... is that certain kinds of hearsay are at once so damaging, so suspect, and yet so difficult to discount, that jurors cannot be trusted to give such evidence the minimal weight it logically deserves, whatever instructions the trial judge might give.” Id. at 138, 88 S.Ct. 1620 (Stewart, J., concurring) (internal citation omitted). The Court has since established that non-testifying codefendants’ statements may be introduced at joint trials if sufficient redactions can be made and adequate jury instructions given to protect the rights of codefendants. In Richardson, the codefendant’s confession was fully redacted to eliminate all references to the defendant. 481 U.S. at 203, 107 S.Ct. 1702. The confession only implicated the defendant when it was considered alongside her own testimony, placing her at the scene of a critical conversation the confession described. Id. at 205-06, 208, 107 S.Ct. 1702. The Court found no Confrontation Clause violation in Richardson because the statement was “not incriminating on its face,” but became potentially incriminating “only when linked with evidence introduced later at trial.” Id. at 208, 107 S.Ct. 1702. Even then, the inference was not obvious, such that the limiting jury instructions sufficed to guard against the remaining risk of “inferential incrimination.” Id. In contrast to the “[sjpecific” and “vivid” incriminating statement at issue in Bruton that created an “overwhelming probability” that jurors would fail to heed limiting instructions, id. at 208-09, 107 S.Ct. 1702, the redacted statement in Richardson made any incriminating implication sufficiently indirect that jury instructions could be counted on to “dissuad[e] the jury from entering onto the path of inference in the first place.” Id. at 208, 107 S.Ct. 1702. The adequacy under the Confrontation Clause of redacting a non-testifying codefendant’s statement depends on how effectively the redaction eliminates the statement’s accusatory implication. Evaluations of such effectiveness are necessarily contextual. In Gray v. Maryland, the prosecution only crudely redacted a codefendant’s statement by whiting out the names of Gray and the other alleged perpetrator (who had since died), leaving blank spaces separated by commas. 523 U.S. at 188, 118 S.Ct. 1151. The police witness reading to the jury from the confession said “deleted” or “deletion” each time he encountered a blank. Id. Such redaction did little, if anything, to cure the prejudice to the defendant. The Court determined that the “blank space in an obviously redacted confession ... points directly to the defendant, and it accuses the defendant in a manner similar to [the non-testifying codefendant’s] use of Bruton’s name or to a testifying codefendant’s accusatory finger,” and thus requires the same result as in Bruton. Id. at 194, 118 S.Ct. 1151. Even though the redacted confession in Gray never named the defendant on trial, it called the jurors’ attention to his codefendants’ inculpation of him with sufficient clarity that no limiting jury instruction could suffice. The difference in outcomes in Gray and Richardson depended “in significant part upon the kind of, not the simple fact of, inference.” Id. at 196, 118 S.Ct. 1151. Gray “involve[d] inferences that a jury ordinarily could make immediately, even were the confession the very first item introduced at trial,” id., whereas the inferences in Richardson were attenuated. The Supreme Court has not yet determined the permissibility under the Sixth Amendment of the type of redaction at issue here, which eliminated names and identifying references to specific defendants (without signaling that changes had been made), but left intact some of the statements’ descriptions of people doing things to advance the crimes with which the defendants were charged. Indeed, the Court in Richardson was careful to note that it “expressed] no opinion on the admissibility of a confession in which the defendant’s name has been replaced with a symbol or neutral pronoun.” 481 U.S. at 211 n. 5, 107 S.Ct. 1702. The redactions in this case fall somewhere between the full redaction that Richardson sustained, and the obviously inculpatory blank spaces and deletions that Gray held to be insufficient. The Court has, however, hinted how redactions might effectively be used in cases involving several perpetrators: In disapproving the obvious redactions in Gray, the Court noted that the incriminating references to “Me, deleted, deleted, and a few other guys” could have been changed to “Me and a few other guys.” 523 U.S. at 196, 118 S.Ct. 1151. Our circuit has infrequently considered the. kind of neutral-pronoun redactions approved by the district court in this case. Evaluation of the potential inculpatory implications of a non-testifying codefendant’s redacted confession is necessarily contextual. We have held that putatively anonymized references to a defendant in a codefendant’s statement violated Bruton where the statement still called attention to the declarant’s accusation against the defendant. See Serio v. United States, 401 F.2d 989, 990 (D.C.Cir.1968) (per curiam). Elsewhere, we found neutral-pronoun redaction constitutionally adequate where, describing a transaction in which several people were involved, a statement was redacted to replace the defendant’s name •with neutral pronouns that, in context, did not inevitably refer to the defendant. See Washington, 952 F.2d at 1406; see also United States v. Applewhite, 72 F.3d 140, 145 (D.C.Cir.1995). Serio, a case we decided immediately after Bruton and that, like Bruton, involved just two alleged perpetrators, held that the admission of a codefendant’s confession in which the defendant’s name was replaced with the phrase “another man” violated the defendant’s confrontation right because of the “well-nigh inevitable association of [the defendant] as the ‘other man’ referred to in [his eodefendant’s] confession.” 401 F.2d at 989-90. After Richardson but before Gray, however, we sustained in Washington the use of nonobvious, neutral-pronoun redaction together with limiting jury instructions in circumstances in which the redacted statement could have referred to several individuals other than the defendant. 952 F.2d at 1406. In that context, the neutrally redacted statement created no “inevitable association” between the defendant and the inculpatory conduct the statement describes. Id. (citing Serio, 401 F.2d at 990). Washington questioned whether Serio’s “inevitable association” standard might be more protective of defendants’ rights than the Sixth Amendment requires, but left that matter undecided because that challenge failed even under Serio. Id. Here, too, we see no need to consider whether Serio is overprotective, because that standard was satisfied here. In sum, at least when “all references to the defendant in a codefendant’s statement are replaced with indefinite pronouns or other general terms, the Confrontation Clause. is not violated by the redacted statement’s admission if, when viewed together with other evidence, the statement does not create an inevitable association with the defendant, and a proper limiting instruction is given.” Washington, 952 F.2d at 1406-07. In such circumstances, provided that the jury is instructed not to consider the codefendant’s statements as evidence against anyone but the declarant himself, as happened here, Bruton is not violated. 2. In their Bruton challenge, defendants first argue that the district court erred by failing to require the government to redact fully the codefendants’ confessions to eliminate even anonymized references to other perpetrators. They assert that neutral-pronoun redactions do not adequately protect defendants’ confrontation rights because they leave in place inevitable associations with the defendants. In their view, only full redaction would suffice. Defendants further contend that, even if the Sixth Amendment permits jurors to hear anonymized references to defendants’ criminal activity, the redactions here were inadequate. The sheer number of redactions, combined with grammatical errors in redaction, they claim made it clear to the jury in this case that the confessions were altered, and thus impermissibly pointed the finger at them. We do not accept the defendants’ claim that anything short of full redaction violates their confrontation rights. The prosecution made full redactions in several places where it could do so without creating unacceptable confusion or distortion. But, as Washington makes clear, the Confrontation Clause does not always mandate full redactions. Carefully made neutral-pronoun redactions can avoid the defect of elisions so crude that they “obviously refer directly to someone, often obviously the defendant.” . Gray, 523 U.S. at 196, 118 S.Ct. 1151. The critical question is whether the redactions adequately conceal the fact that the declarant identified the defendant in particular — a fact that, if known, would make it unlikely that a jury would be able to follow a limiting instruction. See id. Viewing the text of the statements as a whole and in the context of the facts and evidence in the case, we disagree that the redactions made it obvious that the statements referred to specific defendants. The neutral-pronoun redactions here were a far cry from those in Gray, where the method of redaction only strengthened the inference that the Confrontation Clause required be attenuated. Despite the need for frequent redactions in this case, each resulting statement resembled a confession that a defendant might have made if he were trying to avoid identifying his co-conspirators. A defendant endeavoring not to point the finger at his confederates would need repeatedly to rely on the kinds of vague references, such as “them” and “the other guy,” that these redacted statements use. In fact, even before his statement was redacted, Clarke referred to