Law enforcement had a fresh, promising lead for finding Vanessa Armstrong Vixama-a crucial witness. Vixama's former attorney had informed the government that Vixama was with her boyfriend in Delaware. He also had given the government the boyfriend's name and phone number. A simple, routine database search for the boyfriend's address stood a decent chance of leading the prosecution straight to the boyfriend-and likely, to Vixama. But the prosecution did not conduct a database search. Nor did it attempt in any other way to find the boyfriend. Instead, the prosecution, over a span of days before trial, called the boyfriend once and sent him a single text. When the boyfriend did not respond, the prosecution did nothing more before trial to reach him. Even after trial began, the prosecution did nothing more than call the boyfriend one more time. Though no one responded, the government simply stopped looking for Vixama.1
The Sixth Amendment demands more from the government. So does Supreme Court jurisprudence. After all, Delancy's and Smith's Sixth Amendment right to confront Vixama at trial was at stake.
Yet the prosecution asked the district court to admit Vixama's deposition testimony in lieu of her in-person testimony-even though it did not undertake reasonable, *1246routine steps to follow up on its fresh, promising lead for finding Vixama. In support of its request, the government asserted that it had undertaken "reasonable" efforts to find Vixama and had failed, so Vixama was "unavailable" for trial.
Today the Majority Opinion applies a subjective I-know-it-when-I-see-it approach to uphold the government's lackluster efforts as reasonable and deprive Smith and Delancy of their right to the witness's presence at trial. This ruling creates an unconstitutionally low (and unpredictable) bar for what constitutes "reasonable" effort to find a witness. And in so doing, it incorrectly dismisses as surplusage the Sixth Amendment's independent right to the witness's presence at trial. I therefore respectfully dissent from the Majority Opinion's decision to admit Vixama's deposition testimony.2
I divide my discussion into four substantive sections. Section I considers the Confrontation Clause's right to confront the witness. It explains that when other reasonable and promising options remain, the right's "reasonableness" requirement demands more than a couple calls and a text before a witness may be declared unavailable and her preserved testimony presented at trial. Section II applies the proper standard to the facts here, revealing the inadequacy of the government's efforts. Section III examines the errors in the Majority Opinion's analysis. And Section IV concludes that the erroneous admission of Vixama's preserved testimony was not harmless error.
I.
The Sixth Amendment's Confrontation Clause provides that "[i]n all criminal prosecutions, the accused shall enjoy the right ... to be confronted with the witnesses against him." U.S. Const. amend. VI. The right to confront encapsulates two independent rights: the right to cross-examine the witness and the right to the witness's appearance at trial. See United States v. Crawford , 541 U.S. 36, 59, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004) ("Testimonial statements of witnesses absent from trial have been admitted only where the declarant is unavailable, and only where the defendant has had a prior opportunity to cross-examine.").
These rights work in different ways to test the witness's truthfulness. The right to a witness's live testimony protects a defendant's Sixth Amendment rights in ways that a cross-examination conducted outside the crucible of a trial cannot. By ensuring that the witness is in the courtroom and testifying under oath in front of the judge, the jury, and the defendant, the Clause impresses upon the witness the seriousness of the matter and discourages the witness from lying-in a way that cross-examination at a deposition does not. See California v. Green , 399 U.S. 149, 158, 90 S.Ct. 1930, 26 L.Ed.2d 489 (1970) ; see also Barber v. Page , 390 U.S. 719, 721, 88 S.Ct. 1318, 20 L.Ed.2d 255 (1968) (explaining that, unlike cross-examination outside of trial, the presence of the witness at trial compels the witness "to stand face to face with the jury in order that they may look at him, and judge by his demeanor upon the stand and the manner in which he gives his testimony whether he is worthy of belief") (citation and internal quotation marks omitted). Relatedly, when the witness is in the courtroom, the jury can closely scrutinize the witness's demeanor and mannerisms in the judge's and jury's *1247presence to determine the witness's credibility. Ohio v. Roberts , 448 U.S. 56, 63-64, 100 S.Ct. 2531, 65 L.Ed.2d 597 (1980), abrogated in part on other grounds by Crawford , 541 U.S. at 60-69, 124 S.Ct. 1354.
So consequential is the right to confront the witness, the Supreme Court has expressed the view that a failure to allow proper confrontation calls the ultimate integrity of the fact-finding process into question. Roberts , 448 U.S. at 64, 100 S.Ct. 2531. The integrity of the fact-finding process is at stake, of course, because the Confrontation Clause is a procedural protection. See Crawford , 541 U.S. at 61, 124 S.Ct. 1354. So just as we cannot skip a trial, even if we think a defendant is obviously guilty, we cannot skip over the defendant's right to a witness's presence at trial, even if we think the witness's prior recorded testimony is obviously reliable. See id. at 62, 124 S. Ct. 1354.
Of course, the right to a witness's presence at trial is not absolute. But given the significance of a defendant's constitutional right to confront the witness at trial, we do not lightly cast away that right. Before disposing of a defendant's constitutional right to confront a witness, we must be sure that two independent conditions have been met. The defendant must have had an opportunity to cross-examine the witness, and the witness must be "unavailable." See Crawford , 541 U.S. at 59, 124 S.Ct. 1354. So the fact that a thorough cross-examination has occurred is not enough to justify proceeding by recorded testimony, without the witness at trial. Here, only the availability of the witness is at issue.
A witness is "unavailable" when the government cannot secure the witness's presence at trial, despite its good-faith, reasonable efforts. Roberts , 448 U.S. at 74, 100 S.Ct. 2531. So what does that mean in practical terms?
Fortunately, we do not write on a blank slate as to what constitutes reasonableness. The Supreme Court has said that the government need not undertake futile tasks. Id. So if no possibility of procuring a witness exists-the example the Supreme Court used for such a circumstance was when the witness died-the government does not need to do anything. Id. On the other hand, "if there is a possibility, albeit remote," that affirmative steps might produce the witness, then the government may need to take those steps. Id. In determining whether the government must engage in actions falling into this category, the Supreme Court has explained that "the possibility of a refusal is not the equivalent of asking and receiving a rebuff." Id. at 76, 100 S. Ct. 2531 (quoting Barber , 390 U.S. at 724, 88 S.Ct. 1318 ).
In Hardy v. Cross , 565 U.S. 65, 132 S.Ct. 490, 181 L.Ed.2d 468 (2011), the Supreme Court revealed a critical distillation of these principles: under that case, the government must follow up on current, promising leads to find a witness, where reasonable methods to do so exist. In Hardy , the state introduced at trial the recorded testimony of a witness who was missing at the time. Id. at 68, 132 S. Ct. 490. Before the state court declared the witness "unavailable" and allowed the use of her preserved testimony, the prosecution did not contact the witness's boyfriend or any of her other friends in the Chicago area at the time, nor did the prosecution inquire at the cosmetology school where the witness had once been enrolled, concerning the witness's whereabouts. Id. at 70-71, 132 S. Ct. 490. The defendant Cross was convicted.
After exhausting his state appeals, Cross filed a petition for a writ of habeas corpus under 28 U.S.C. § 2254. He argued that the state courts had unreasonably applied Supreme Court precedent concerning the Confrontation Clause's unavailability *1248requirement. Id. at 69, 132 S. Ct. 490. The Supreme Court ultimately disagreed, primarily because, under the Antiterrorism and Effective Death Penalty Act's ("AEDPA") standard of review, it concluded that the state court's decision was "reasonable," so it deferred to that determination. Id. at 72, 132 S. Ct. 490. Of course, we apply a de novo standard of review here on direct appeal, in contrast to the deferential standard that was applicable in Hardy . But Hardy is nonetheless instructive.
True, the Supreme Court held that the state court had acted reasonably in denying Cross's unavailability claim under the Confrontation Clause. But its explanation for why provides the governing principle. Concerning the state's failure to contact the missing witness's boyfriend or any of her friends in the area, the Supreme Court reasoned that none of the victim's "family members or any other persons interviewed by the State provided any reason to believe that any of these individuals had information about [the victim's] whereabouts." Id. at 71, 132 S. Ct. 490. As for the state's failure to make inquiries at the cosmetology school where the witness had been enrolled, the Supreme Court similarly explained that the victim had not attended the school "for some time," so there was "no reason to believe that anyone at the school had better information about [the victim's] location than did the members of her family." Id.
These explanations for denying Hardy 's claim suggest that even when the government's actions are viewed through the highly deferential lens of AEDPA, the government must still follow up on leads if it has "reason to believe" those leads can assist in locating the witness. (And actually, in Hardy , when the witness's mother told the prosecution that the witness could be staying with an ex-boyfriend in Waukegan, Illinois, the prosecution followed up by visiting the Waukegan address and speaking with the ex-boyfriend's mother. Id. at 68, 132 S. Ct. 490.) Here, though, a de novo standard governs. So the government's failure to undertake reasonable steps to follow up on leads that provide "reason to believe" they may succeed in locating a missing witness certainly cannot satisfy the Sixth Amendment's reasonableness standard.
Yet today, the Majority Opinion allows exactly that. To arrive at this mistaken conclusion, the Majority Opinion completely fails to account for Hardy 's reasoning. In fact, it does not even attempt to show that Hardy does not suggest the government must undertake reasonable steps to follow up on promising leads.3 Instead, the Majority Opinion dismisses my analysis of Hardy by simply reciting a lengthy summary of Hardy 's facts-just to make two points I readily agree with: that the government does not need to take every step *1249to secure a witness and that reasonableness depends on the facts of the case. Maj. Op. at 1234-36. But significantly, nothing in these two points contradicts the lesson Hardy 's logic and context teaches: that "reasonable efforts" means the government must take reasonable steps to follow up on a promising lead.
II.
A.
In violation of Hardy 's lesson, the Majority Opinion incorrectly excuses the government's failure to undertake reasonable steps to follow up on leads that provide "reason to believe" they may succeed in finding a witness. To explain why the Majority Opinion's analysis cannot be correct, I first revisit the key facts concerning the government's search for Vixama.
After immigration authorities accidentally let Vixama go in the United States, the case agent obtained Vixama's uncle's address and passed it along to Immigration and Customs Enforcement ("ICE") Enforcement and Removal Operations ("ERO"). Two weeks after it received that information, ERO went to the uncle's address to search for Vixama but came up empty handed. ERO said it "felt like they were getting the runaround" from an occupant of the house, since the occupant told ERO that she was not "sure" if Vixama was living there.
It wasn't until the next month, March, before the case agent followed up with ERO for an update on Vixama. At that point, ERO informed the case agent that it was not going to look for Vixama again. Despite this news, the government again waited another month before acting.
Five days before trial, on April 12-the same day the government advised the court at the calendar call that it intended to introduce Vixama's deposition testimony at trial-the government emailed Vixama's former attorney to ask if he had Vixama's phone number or knew where she was residing. The next day, Vixama's attorney informed the government that Vixama was in Delaware with her boyfriend. In response to this information, the government simply emailed Vixama's former attorney a subpoena for Vixama, which the attorney forwarded to Vixama's boyfriend.
On April 15, the attorney provided the name and phone number of Vixama's boyfriend to the government, advising that the government could "call [Vixama] now at [the boyfriend's] number" and that he "believe[d] she w[ould] cooperate." The government quickly learned that was not to be the case. When the agent called the boyfriend's number on April 15, no one answered, and no one returned the call. The agent followed up with a text, which also was not returned. That one call and one text consisted of the totality of the government's efforts to reach Vixama's boyfriend before trial.
The government knew what this meant. Indeed, the government understood at least as early as the very first day of trial-April 17-that Vixama was not going to appear, since it advised the court that it continued to consider her "unavailable" after having previously indicated on April 12 that it intended to use her deposition testimony.
Even after trial began, the government did nothing more to reach the boyfriend than to try his number one more time. As late as the fourth day of trial, when asked whether the government had run the boyfriend's name through any database or had otherwise made any attempt to find his address, the case agent responded, "Not yet."
B.
The government's efforts to search for Vixama-a witness crucial to the government's case-suffer from serious shortcomings. I begin with the government's *1250efforts during February and March 2017, the period after the government learned that Vixama had been released and before the government's April pretrial efforts. During this time, the government lost nearly two months after ERO's single, unsuccessful February visit to the uncle's residence,4 doing nothing to look for Vixama. It could have followed up with the uncle by sending an agent to the uncle's address to look again for Vixama and to question the uncle. But it did not do so-even though as late as within a week of the trial, the agent thought Vixama was "perhaps" living at her uncle's home in Coral Springs. So the government had at least some reason to believe it might find Vixama with her uncle.
The government's stated reasons for its inaction were insufficient under Supreme Court precedent. According to the case agent, he declined to ask another Homeland Security Investigations ("HSI") agent from an office closer to the uncle's residence to take a second look for Vixama at the uncle's house because he "could have been turned down." But that is just another way of saying those efforts could have worked. And as I have noted, the Supreme *1251Court has held that "the possibility of a refusal is not the equivalent of asking and receiving a rebuff." Barber , 390 U.S. at 724, 88 S.Ct. 1318.
The case agent also said he curtailed his efforts to search for Vixama in part because he knew the government already had what it needed from Vixama: her deposition testimony. But the government's possession of recorded testimony cannot relieve the government of its obligation, under the Sixth Amendment right to the witness's presence, to engage in reasonable efforts to find the witness who offered it. If it did, the Sixth Amendment's unavailability requirement would go the way of the eight-track tape player because the government's mere possession of preserved testimony would always end the government's duties to present the witness in person.
For this reason, it is not surprising that our sister Circuits have described "a good measure of reasonableness [as requiring] the State to make the same sort of effort to locate and secure the witness for trial that it would have made if it did not have the prior testimony available."5 See , e.g. , Cook v. McKune , 323 F.3d 825, 836 (10th Cir. 2003).6 Yet here, the agent conceded he ceased efforts to find Vixama in significant part because the government already had her recorded testimony. That admission confirmed what the record already indicated: the government stopped well short of the efforts it would have undertaken to find Vixama-a witness even the agent deemed "essential," see also infra at Section IV-in the absence of Vixama's preserved testimony.
Nor did the government sufficiently step up its efforts as trial loomed closer. Unlike in Hardy , where nothing indicated that the witness's boyfriend, any of her friends in the Chicago area, or anyone at the cosmetology school had knowledge of her whereabouts, here, the government had good reason to believe that Vixama was with her boyfriend. The government also had the boyfriend's name and phone number. And while it's great that the government was able to secure some assistance from Vixama's former attorney by emailing him, proper follow-up also required performing routine and reasonable law-enforcement tasks pertaining to the name and number of the boyfriend, so the government could secure Vixama.7 Under Hardy , the government *1252should have undertaken these reasonable efforts to follow up on this promising lead.
But it did not. Rather than running a basic, routine, quick, and inexpensive database search of the boyfriend's name to ascertain his address-or, for that matter, trying to obtain the boyfriend's address in any manner-the government did no more than just engage in minimal efforts to twice call and once text the boyfriend. No evidence suggests the government took even five minutes to check for the boyfriend's profile on Facebook, Twitter, or Instagram, or to punch his name into Google to see what those quick searches could dredge up.
Even when it became clear that the two calls and single text were not going to cut it, the government still did nothing more. Instead, as late as the fourth day of trial, the case agent conceded that the government had "[n]ot yet" tried to find the boyfriend's address. With the constitutional rights of two criminal defendants at stake, it is fair to wonder what the government was waiting for. By comparison, teens trying to reach their crushes do more.
And here, Vixama was a crucial witness to the government's case. See infra at Section IV. Winning at trial would have been extremely difficult without her testimony. So we can be sure the government would have engaged in routine, basic, and inexpensive techniques to find Vixama if it had not had her recorded testimony. See Cook , 323 F.3d at 836. Its decision not to do so only further reconfirms the government's failure to satisfy the Sixth Amendment's reasonableness standard.
Yet the Majority Opinion concludes the government's efforts were nonetheless reasonable. In doing so, it relies on five of the reasons the case agent identified for his decision not to engage in further efforts to find the boyfriend: (1) he believed Vixama had received a subpoena (through her boyfriend); (2) he had attempted to contact her boyfriend by phone and text; (3) Vixama was in the country illegally; (4) a criminal action against Vixama was not pending, so the agent could not take her into criminal custody; and (5) ICE was the only agency that could take Vixama into custody before she failed to appear at trial. Maj. Op. at 52-53. In fact, though, the agent also admitted a sixth-that he did not think additional efforts were necessary, since the government already had Vixama's deposition testimony.
None of these reasons-alone or together-suffice to excuse the government from undertaking reasonable and routine efforts to ascertain the boyfriend's address when the government had good reason to believe that Vixama was with the boyfriend.
I have already explained above why the existence of Vixama's deposition testimony could not relieve the government of its obligation to engage in reasonable efforts to find Vixama. See supra at 76-77 & n.6. I have likewise described why the agent's two unreturned calls and single text to the boyfriend did not extinguish the government's responsibility to undertake reasonable efforts to find Vixama. See supra at 77-78. Now I turn to the other four reasons the government set forth.
First, the emailed subpoena forwarded by Vixama's former attorney to her boyfriend did not relieve the government of its *1253responsibility to engage in additional reasonable efforts to find Vixama under the circumstances here. Emailing a subpoena through third parties is not "service" under the Federal Rules of Criminal Procedure. See Fed. R. Crim. P. 17(d). And while I do not argue Vixama did not receive it, the subpoena did not have legal effect, so the agent's reliance solely on it was misplaced, since it became obvious shortly after the subpoena was sent that Vixama did not intend to honor it.
Importantly, though, that the government thought the boyfriend would pass the subpoena to Vixama has tremendous relevance to the reasonableness analysis; it underscores the fact that the government had reason to believe-and did in fact believe-Vixama was with the boyfriend in Delaware, as Vixama's former attorney had advised the government. So the government also had reason to believe that if it found the boyfriend, it would find Vixama. And that required the government to undertake reasonable efforts to try to find the boyfriend.
Nor did Vixama's illegal presence in the country somehow absolve the government of satisfying the constitutional requirement to undertake reasonable efforts to find Vixama. If it did, then the Confrontation Clause's right to the witness's presence at trial would have a hole large enough to drive a Bagger 2938 through whenever the witness was illegally present in the United States at the time of trial. Yet nothing in the Constitution or Confrontation Clause jurisprudence supports the notion that a defendant's Sixth Amendment right to the witness's presence at trial depends upon the witness's immigration status.
The government's excuses that no pending criminal action allowed the agent to take Vixama into custody and that only a bench warrant would enable it to produce Vixama fare no better. Even assuming, arguendo , that no arrest warrant could be obtained,9 that circumstance would not justify the government's failure to produce Vixama. The government knew since February 7, 2017, that Vixama was in the country and had not been deported.
But it did not tell Smith and Delancy that for more than two months. In fact, the government waited until the first day of trial , April 17, to so inform them, failing even to reveal it at the calendar call on April 12. Instead, at the calendar call, without advising the court or Smith and Delancy that Vixama was still in the country, the government told the court that it intended to introduce Vixama's videotaped deposition. The court asked Smith's attorney if he was objecting to the video deposition's *1254admission, and Smith's attorney, not knowing that Vixama was in the country, answered, "Clearly not. We've already had the opportunity to cross-examine."
So when the government finally revealed Vixama's presence in the country on the first day of trial, it was not surprising when Smith and Delancy objected. As Smith's attorney aptly put it, "[T]he whole notion of her being unavailable was because it was presumed that she would be deported and that would make her beyond the jurisdiction of the United States."
Also on the first day of trial, the government told the court that it had sent a subpoena to Vixama through her former attorney to appear on the third day of trial, even though the government expected the trial to take only "[f]our to five days." But almost immediately after telling the court about the subpoena, the government betrayed its doubt that Vixama would comply with it, telling the Court, "At this point we still consider her unavailable." And no wonder, since beginning on April 15, the agent had called and sent a text to the boyfriend, and he had received no response at all.
Of course, nothing prevented the government from subpoenaing Vixama to appear on the first day of trial and obtaining a bench warrant then if she did not comply. At least that way, the government would have had the remaining five trial days (counting April 17) to execute the bench warrant. Instead, however, the government kept Smith and Delancy in the dark and unilaterally chose to limit itself to obtaining a bench warrant only when, by its own calculation, much of the trial would have already concluded. So to the extent that a bench warrant was the only way for the government to compel Vixama's attendance and that there was insufficient time to execute the warrant, the government put itself in that position. It cannot therefore benefit from that self-imposed disadvantage. And if the government had no intention of following up on the bench warrant in the first place, that it even sought one is irrelevant to evaluating the reasonableness of the government's efforts.
Put simply, the government had reason to believe that undertaking routine law-enforcement steps to find Vixama by locating her boyfriend might well succeed. No reason the government offered for failing to take such steps undermines this fact. For this reason, the government's efforts to find Vixama were not "reasonable" under the Sixth Amendment's unavailability standard as the Supreme Court has construed it.
III.
The Majority Opinion protests that I ask too much of the government. It throws a kitchen sink of rationalizations in its attempt to justify the government's failure to conduct routine tasks that it had reason to believe might locate Vixama. Specifically, the Majority Opinion argues that (1) it would be "Monday-morning quarterbacking" to predict whether finding the boyfriend would have led to Vixama, Maj. Op. at 1236-37, 1239-40; (2) the government was "plainly reasonable" in asking ERO for help, id. at 1237-38; (3) it was reasonable for the government to rely on the former attorney to obtain Vixama's presence at trial, rather than engaging in its own efforts to find Vixama, id. at 1238-39; (4) this dissent is internally inconsistent, id. at 1239-40; (5) Vixama was cross-examined, id. at 1240-41; (6) Vixama had great incentive not to be found, id. at 1224, 1230-31; and (7) caselaw supports the conclusion that the government's efforts were reasonable. But even brief consideration of these arguments-both alone and together *1255with each other-reveals the fallacy of them.
(1) he Majority Opinion's "Monday-morning quarterbacking" criticism is riddled with flaws, most notably because it shows the Majority Opinion impermissibly shifted the burden of proof onto Smith and Delancy.
First, the Majority Opinion asserts that this dissent engages in impermissible Monday-morning quarterbacking. In doing so, the Majority Opinion also raises questions about the efficacy of looking for Vixama's boyfriend. Maj. Op. at 1237-37. This argument suffers from four problems.
First, it impermissibly places the burden of proof on Smith and Delancy. As the Majority Opinion correctly notes, the government- not the defendant-bears the burden of proof to show that it acted reasonably. Maj. Op. at 1227 (citing Roberts , 448 U.S. at 74-75, 100 S.Ct. 2531 ). And as the Majority Opinion further correctly acknowledges, although we can always think of more steps the government could have taken, the government can neutralize any intimation that reasonableness required those steps by showing the "great improbability that such efforts would have resulted in locating the witness." Maj. Op. at 1230 (quoting Roberts , 448 U.S. at 75-76, 100 S.Ct. 2531 ). So Smith and Delancy were not required to show that the boyfriend was a promising lead; rather, the government bore the burden of proving that finding the boyfriend was unlikely to lead to Vixama.
Yet the Majority Opinion ignores these rules and improperly places the burden of proof on Smith and Delancy, to show that Vixama would have been found if law enforcement had adequately followed up with the boyfriend. For example, the Majority Opinion criticizes this dissent for "surmis [ing ] that if the government had used databases, it might have found an address for the boyfriend in Delaware, and then it might have found Vixama." Maj. Op. at 1236 (bold added; italics in original).
Here, though, the government, which has the burden of proof, never gave any reason to doubt the efficacy of a database search. Just the opposite. When asked whether it had used databases to search for the boyfriend's address, the government's response was "[n]ot yet," suggesting the potential usefulness of the technique.
Nor did the government give any reason to doubt that Vixama was with her boyfriend. In fact, Vixama's former attorney told the government she was, and the government even sent a subpoena to the boyfriend so Vixama would see it. Only the Majority Opinion manufactures doubt about the likelihood and effectiveness of finding the boyfriend. Then it improperly thrusts the burden of allaying that doubt on Delancy and Smith.
And the burden the Majority Opinion wrongly saddles Smith and Delancy with is also an impossible one for them: law enforcement has unique access to tools and resources that are not available to Smith and Delancy. For that reason, only law enforcement can make the showing the Majority Opinion demands. Smith and Delancy cannot, for example, use law-enforcement databases to search for Vixama's boyfriend's address. So they have no way of removing the doubt the Majority Opinion creates about the efficacy of a database search. Ultimately, the Majority Opinion invents a problem the government never raised and drowns Smith and Delancy in an impossible burden that is not theirs.
Second, the Majority Opinion's argument depends in part on illogical reasoning. The Majority Opinion inexplicably concludes that because "efforts [to find Vixama] had already failed in Florida," they would also necessarily fail in Delaware. Maj. Op. at 1237-38. This is irrational.
*1256First, if Vixama was not in South Florida, she was obviously somewhere else, and that somewhere else could have included Delaware. And second, more significantly, unlike the outdated information concerning whether Vixama was at the uncle's address, the government had current information from Vixama's former attorney that Vixama was in Delaware with the boyfriend. To suggest that one failure necessarily means failure forever-regardless of the different circumstances-is like assuming 7-Eleven won't have Snickers bars just because the library doesn't.
Third, in support of its argument, the Majority Opinion opines that following up with the boyfriend would have been a difficult and involved process. See Maj. Op. at 1238. Not so. No matter how much the Majority Opinion draws out and exaggerates the steps it would have taken for law enforcement to look up the boyfriend's address and follow up with him, the reality is that this is a routine and feasible technique that law enforcement uses all the time-analogous, as the Majority Opinion recognized, to looking up and going to Vixama's uncle's address (which was one of the very first things law enforcement did in this case).
Fourth, as for the concept of "Monday-morning quarterbacking," I agree that a court, with the benefit of hindsight, should not be unreasonably demanding. Maj. Op. at 1236-37. I likewise agree that the "Sixth Amendment does not require the prosecution to exhaust every possible means of producing a witness at trial." Maj. Op. at 1240. But our aversion against Monday-morning quarterbacking does not mean that we never review what the government did. And that the Sixth Amendment does not require the prosecution to take every conceivable step to find a witness is not a talismanic phrase to excuse inadequate efforts. Those two points carry weight only when it is unreasonable to demand more of the government.
Here, the government presented Vixama's recorded testimony in a trial that carried potential maximum sentences of decades in prison for Smith and Delancy. But the government's efforts to follow up on a fresh, promising opportunity to secure the defendants' Sixth Amendment right to the witness's presence at trial essentially consisted of only two calls and a text-even though other reasonable means to find the boyfriend were readily available and inexpensive to undertake. The Sixth Amendment demands more.
(2) That it was reasonable for the prosecution to ask ERO for help does not excuse the prosecution's failure to take other reasonable steps to find Vixama.
Moving on to the Majority Opinion's second justification for why the government's efforts here were reasonable, the Majority Opinion says that the government was "plainly reasonable" in asking ERO for help. Maj. Op. at 1237. I don't disagree. But taking one reasonable step is not enough when that step fails and other reasonable avenues for finding the witness emerge. Here, the case agent's reasons for not doing more once he learned that ERO was not going to look for Vixama-because the government already had her deposition testimony and because he thought he "could have been turned down" had he asked for help-were not reasonable. So the government's request for ERO's assistance did not relieve the government of its obligation through the end of the trial in April to engage in additional efforts to find Vixama when ERO came up short in February.
*1257(3) That the government initially hoped Vixama's former attorney could produce Vixama does not mean that, when that hope dissipated, the government could decline to take reasonable steps to follow up on a promising lead.
As its third rationale, the Majority Opinion asserts that "[b]ecause it appeared that the boyfriend (and Vixama through him) was cooperating with attorney Raben, it was reasonable for the government to rely on attorney Raben to communicate with him rather than to try to track the boyfriend down independently through databases." Maj. Op. at 1239. I agree that the government was right to contact Vixama's former attorney. But when it became clear that the government's earlier communications with the attorney were unlikely by themselves to result in Vixama's presence at trial, and other promising leads for finding Vixama remained, the government did not have the discretion under the Sixth Amendment to refuse to engage in reasonable efforts to locate Vixama just because it had previously contacted her former attorney.
The government knew very quickly that it was not reasonable to assume, based solely on its communications with the attorney, that Vixama would show up for trial; from the very beginning, the boyfriend had not responded to the government. That's why the government noted on the very first day of trial that it anticipated Vixama would be "unavailable"-just two days after the attorney gave the government the boyfriend's phone number. If the government thought the attorney's earlier communications with Vixama would yield her appearance at trial, it obviously would not have made this announcement.
Despite the government's knowledge that Vixama probably would not comply with the subpoena, it did no more before trial than once call and once text Vixama's boyfriend-its current, promising lead. It declined to take basic, inexpensive, and obvious law-enforcement steps like running a database search. And when, just as the government had anticipated, Vixama indeed failed to appear, the government did almost nothing to try to execute the bench warrant it obtained for her.
So while the Majority Opinion is right that it was reasonable for the government to initially rely on Vixama's former attorney, that did not excuse the government's subsequent unreasonable lack of action.
(4) It is logical to conclude that Vixama's boyfriend was such an important lead that, to adequately follow up with him, the government should have done more than just ask someone to forward a subpoena to him.
Fourth, contrary to the Majority Opinion's suggestion, it is not internally inconsistent for this dissent to emphasize the need to adequately follow up with Vixama's boyfriend while also asserting that the government's efforts in sending the subpoena to the boyfriend through Vixama's former attorney were insufficient under the circumstances. Maj. Op. at 1239-40. To be sure, as I have noted, the government was right to try to find Vixama through her former attorney and even to try to send her a subpoena in that way.
But when that effort failed to yield Vixama's cooperation, the government could not just rest on its laurels. Rather, as I have explained, the government believed the boyfriend provided Vixama with the subpoena because it believed Vixama was with the boyfriend. As a result, the government had reason to believe that if it found the boyfriend, it would find Vixama. And that required the government to engage in reasonable efforts to find the boyfriend. See supra at 81.
*1258(5) How robust a previous cross-examination is and what state of mind the parties had during that cross-examination are irrelevant to determining unavailability.
Fifth, while paying lip service to the notion that "[f]or sure, the defendants have not waived their Confrontation Clause claims," the Majority Opinion wrongly dismisses the unavailability requirement's independent importance, stressing that defense counsel "cross-examined Vixama as if she were testifying at trial because everyone assumed this would be her testimony at trial," and that this fact is "relevant and important" in assessing the reasonableness of the government's actions. Maj. Op. at 1240-41. This justification for excusing the government's less-than-reasonable efforts to find Vixama misses the point of the Confrontation Clause's independent right to the witness's presence-a right that is separate from the right to cross-examination. See supra at 1245-47. If robust cross-examination were enough to excuse the government from engaging in reasonable efforts to present a witness at trial, the Supreme Court would not have explained that admission of recorded testimony requires two separate showings: unavailability and cross-examination. See Crawford , 541 U.S. at 59, 124 S.Ct. 1354.
The Majority Opinion's reliance on cross-examination to lower the reasonableness bar of the Sixth Amendment's right to the witness's presence at trial impermissibly conflates these requirements. It fails to recognize that the witness's live testimony at trial serves purposes that cross-examination alone simply cannot. Perhaps for this reason, the Supreme Court has never excused the government from undertaking reasonable efforts to find a missing witness simply because the preserved testimony included thorough cross-examination. Nor does the Majority Opinion point to a single case where any other court has concluded that thorough cross-examination can somehow relieve the government of its obligation under the Sixth Amendment to engage in reasonable efforts to find a witness before it may present that witness's recorded testimony.
(6) That a witness may not want to be found does not relieve the government of its responsibility to take reasonable actions to find the witness.
Sixth, the Majority Opinion repeatedly emphasizes that Vixama "was in hiding, and had a strong incentive not to be found." Maj. Op. at 1224, 1230. But evasiveness is not unavailability, and law enforcement cannot create a self-fulfilling prophecy by abbreviating search efforts just because the witness does not want to be found. See Lynch , 499 F.2d at 1024 ("We are not prepared to equate 'unavailability' with 'evasiveness.' The government failed to establish that [the witness] could not have been located and brought to trial by a reasonably diligent search. Accordingly we hold that the witness was not 'unavailable' ...."). Many witnesses would prefer not to be hauled into court. That's often the very reason why the government must undertake efforts to find them in the first place. So allowing a witness's evasiveness to excuse unreasonable government efforts to find that witness would essentially render meaningless the Confrontation Clause's unavailability requirement.
(7) The Majority Opinion's ruling runs squarely against the caselaw.
Finally, the Majority Opinion suggests that caselaw supports the conclusion that the government's efforts here were "reasonable." Most respectfully, I disagree.
*1259The Majority Opinion's caselaw errors fall into three categories: (1) it does not heed the lessons of Hardy and Roberts ; (2) it does not account for important Sixth Amendment-specific caselaw on "reasonableness;" and (3) it mistakenly concludes that United States v. Tirado-Tirado , 563 F.3d 117 (5th Cir. 2009), bolsters the determination that the government's efforts here were "reasonable." I address each error in turn below.
First, the Majority Opinion repeatedly relies on Hardy for the proposition that "the Sixth Amendment does not require the prosecution to exhaust every avenue of inquiry, no matter how unpromising." Maj. Op. at 1227, 1235-36 (quoting Hardy , 565 U.S. at 71-72, 132 S.Ct. 490 ). I take no issue with Hardy 's statement in that regard. But as I have explained, that is not the beginning and end of Hardy 's significance. As expressed in its reasoning and alluded to in the last phrase of the sentence quoted above, Hardy teaches that the government must undertake reasonable efforts to follow up on leads it does have "reason to believe" might be fruitful in locating a missing witness. See Hardy , 565 U.S. at 71-72, 132 S.Ct. 490 ; see also supra at 1247-49.
Roberts supports this same lesson. There, about a year before the trial, the missing witness's parents were able to reach the witness through information a social worker in San Francisco had provided, since at the time, the social worker was in communication with the parents about a welfare application the witness had filed. Roberts , 448 U.S. at 60, 100 S.Ct. 2531. After that time, though, the witness had called her parents only once and had not been in touch with her siblings. Id. During that last phone call, which occurred about seven or eight months before trial, the witness told her parents that she "was traveling" outside Ohio, but she did not advise them of where she was. Id. The witness's mother attested that she knew of no way to reach the witness, even in case of emergency, and that she did not "know of anybody who knows where she is." Id. (citation and quotation marks omitted).
As for the government's efforts to find the missing witness, it contacted the mother four months before trial and sent five subpoenas over time to the witness at the parents' address. Id. at 75, 100 S. Ct. 2531. The government further noted that the witness's parents had "not been able to locate her for over a year." Id.
Based on these facts, the trial court declared the witness unavailable and admitted the witness's prior testimony. Id. at 59, 100 S. Ct. 2531. The defendant was convicted and appealed, asserting that the use of the prior testimony violated his Sixth Amendment rights. Id.
The Supreme Court agreed with the trial court's decision. Id. at 76, 100 S. Ct. 2531. In reaching this conclusion, significantly, the Supreme Court explained that "the great improbability that [additional] efforts would have resulted in locating the witness, and would have led to her production at trial, neutralizes any intimation that a concept of reasonableness required their execution." Id. Immediately following that statement, the Court reaffirmed the "general rule" that "the possibility of a refusal is not the equivalent of asking and receiving a rebuff." Id . (citation and quotation marks omitted). In other words, under the particular circumstances in Roberts , there was no reason to believe that additional efforts would have located the witness. But when reason to believe additional reasonable efforts may be successful exists, the government must engage in those efforts.
The Majority Opinion completely misses this lesson from Hardy and Roberts . Instead, it inaccurately accuses this dissent of cherry-picking phrases out of context. Compare Maj. Op. at 1233 (accusing this *1260dissent of cherry-picking phrases out of context), with supra at 69-71 (explaining relevant facts in Hardy ), and supra at 95-96 (explaining relevant facts in Roberts ). Then the Majority Opinion single-mindedly focuses on articulating undisputed points, entirely failing to analyze the important principles of the cases. For example, the Majority Opinion notes that in Hardy , the Supreme Court excused the prosecution's failure to reach out to the missing witness's boyfriend or the cosmetology school where she had been enrolled because there was no "reason to believe" those leads would have been fruitful. Maj. Op. at 1235-36 (quoting Hardy , 565 U.S. at 71, 132 S.Ct. 490 ). But remarkably, the Majority Opinion then fails to engage with the fact that, in this case, there was reason to believe that Vixama was with her boyfriend.
Similarly, the Majority Opinion explains that in Roberts , the trail had effectively gone cold for the missing witness: the government's most recent information about the witness's whereabouts was seven-to-eight months old, and the government knew only generally that the witness was, at that time, traveling outside Ohio. Maj. Op. at 1233-34-. That's why the Supreme Court opined that it was "great[ly] improbab[le]" that further government efforts would yield the missing witness.
Again, though, the Majority Opinion fails to register the significance in Smith and Delancy's case that it was not "great[ly] improbab[le]" that additional government efforts would be useful. As I have noted, the government here believed-and had good reason to believe-based on Vixama's former attorney's representations, that Vixama was physically present with the boyfriend in Delaware.
Had the Majority Opinion actually applied the reasoning of Hardy and Roberts here, it necessarily would have reached the conclusion that the government did not undertake reasonable efforts to follow up on the promising lead that was Vixama's boyfriend.10
The Majority Opinion makes its second significant mistake relating to caselaw when it does not account for important Sixth Amendment-specific caselaw on "reasonableness." In explaining the "reasonableness" standard, the Majority Opinion relies heavily on Fourth Amendment jurisprudence. See Maj. Op. at 1228-29 & n.7 (citing United States v. Banks , 540 U.S. 31, 36, 124 S.Ct. 521, 157 L.Ed.2d 343 (2003) ; Ohio v. Robinette , 519 U.S. 33, 39, 117 S.Ct. 417, 136 L.Ed.2d 347 (1996) ; Ker v. California , 374 U.S. 23, 33, 83 S.Ct. 1623, 10 L.Ed.2d 726 (1963) ; United States v. Arvizu , 534 U.S. 266, 273-74, 122 S.Ct. 744, 151 L.Ed.2d 740 (2002) ; Missouri v. McNeely , 569 U.S. 141, 150, 133 S.Ct. 1552, 185 L.Ed.2d 696 (2013) ; Stephens v. DeGiovanni , 852 F.3d 1298, 1315 (11th Cir. 2017) ; Rodriguez v. Farrell , 280 F.3d 1341, 1346-47 (11th Cir. 2002) ).11 And while Fourth Amendment caselaw can be helpful to understanding the meaning of "reasonableness, *1261" Sixth Amendment jurisprudence has put its own unique twist on the "reasonableness" inquiry-a twist that Fourth Amendment jurisprudence does not account for.
For example, significantly, the Supreme Court has explained that, in the Confrontation Clause context, the Framers "were loath to leave too much discretion in judicial hands" and that "open-ended balancing tests" are disfavored because "[v]ague standards are manipulable." Crawford , 541 U.S. at 67-68, 124 S.Ct. 1354. So it is not surprising that Sixth Amendment jurisprudence provides guidelines-like the government need not engage in futile search efforts; it may need to perform actions that have a possibility of finding a witness; and it must perform reasonable tasks that it has reason to believe might be successful-that assist us in determining whether law enforcement's efforts were "reasonable." See , e.g. , Roberts , 448 U.S. at 74, 100 S.Ct. 2531 ; Hardy , 565 U.S. at 71, 132 S.Ct. 490.
To be sure, the Majority Opinion mentions the cases that give us these rules. But it ignores the Sixth Amendment-specific instruction to avoid "open-ended balancing tests." Instead, the Majority Opinion relies on Fourth Amendment law to justify its touchy-feely, I-know-it-when-I-see-it approach to determining whether the government's efforts to find a witness were reasonable. And as I have noted, even when the Majority Opinion pays lip service to the Sixth Amendment-specific rules on reasonableness, it ignores many of the rules most important to this case. As a result, the Majority Opinion fails to recognize the insufficiency of the government's actions here.
Finally, the Majority Opinion makes a third caselaw-related mistake when it asserts that Tirado-Tirado "demonstrates why the government has shown a good-faith, reasonable effort here." See Maj. Op. at 60. In fact, Tirado-Tirado compels the opposite conclusion-that the government's actions here did not satisfy its reasonableness obligation. The Majority Opinion's view to the contrary results from two errors.
First, the Majority Opinion notes that the Tirado-Tirado witness's deposition was taken "only as a precaution" because the government expected the witness to return to the United States to testify at trial. Id. at 61, 100 S. Ct. 2531. But to the extent this discussion suggests that the reason it was wrong in Tirado-Tirado to present the missing witness's testimony was because Tirado-Tirado may have foregone more rigorous cross-examination of the witness in anticipation that the witness would appear again at trial, Tirado-Tirado expressly nixed that idea. Tirado-Tirado explained that the use of the deposition testimony at trial violated the Confrontation Clause "not because Tirado-Tirado did not have a full and fair opportunity to cross examine [the witness]"-indeed, the Fifth Circuit found he did-but because "[the witness] was not 'unavailable' for trial." Tirado-Tirado , 563 F.3d at 125-26. As I have noted, the right to the witness's presence at trial and the right to cross-examination are distinct rights that protect the defendant in different ways. See supra at 66-67. Once again, the Majority Opinion improperly conflates these rights.
Second, the Majority Opinion tries to distinguish Tirado-Tirado , emphasizing that "in Tirado-Tirado , the government 'made no effort' whatsoever to keep in touch with the alien witness" until about eight days before trial. Maj. Op. at 1244 (citing Tirado-Tirado , 563 F.3d at 125 ). But what we have here, the Majority Opinion asserts, "is not a case in which the government 'made absolutely no effort' to locate Vixama and obtain her presence at trial." Id . at 62 These statements are true, *1262as far as they go. But they are misleadingly incomplete, since the Majority Opinion focuses on only the government's efforts in the months before trial and glosses over the comparison of the government's efforts in each case in the several days immediately prior to trial.
In Tirado-Tirado , as the Majority Opinion notes, the government took no action to remain in contact with the witness in Mexico for five months, waiting until eight days before trial to act. Tirado-Tirado , 563 F.3d at 124. And as the Majority Opinion correctly notes, the government here did undertake some action in the two months prior to its last-minute activities five days before trial-it asked ERO to go to the uncle's house. So yes, we have this modest distinction.
But in the eight days before trial, when the government finally began its efforts to obtain the witness's presence at trial in Tirado-Tirado , the government sent the witness a letter with explicit instructions concerning arrangements for the witness to testify and be reimbursed; spoke to the witness's brother about the witness's attendance; obtained the name and telephone number of Tirado-Tirado's common-law wife in the United States; examined the call log of Tirado-Tirado's seized phone for calls to the witness; subpoenaed Western Union for transactions made in Tirado-Tirado's name that could relate to the witness, after discovering in Tirado-Tirado's phone log a call made to Western Union; and reviewed immigration and criminal records for the witness-all in an effort to find the witness. Id. at 120. Even after doing all this, when the government could not come up with the witness, the Tirado-Tirado Court decided that the government had not done enough overall. Id. at 125.
Here, in contrast, the government did little to follow up on the leads it had: it did not revisit the uncle's home, even though the agent believed as late as within a week of trial that Vixama was "perhaps" there; it did not run Vixama's boyfriend's name or number through any database; and it did not attempt to send an agent to find Vixama's boyfriend, even though doing any-or all-of these things to follow up on fresh leads would not have required significant government resources.
Thus, the real distinguishing feature between Tirado-Tirado and this case is how much more overall the government did in Tirado-Tirado to locate a witness outside the country than it did in this case to locate one inside the country-a distinction that especially dooms the sufficiency of the government's efforts here.12
IV.
Since the government's efforts here were unreasonable and Vixama's testimony should not have been admitted, we must consider whether admitting Vixama's deposition testimony was harmless error. Delaware v. Van Arsdall , 475 U.S. 673, 684, 106 S.Ct. 1431, 89 L.Ed.2d 674 (1986). A Confrontation Clause violation is harmless only if we can say beyond a reasonable doubt that the error did not contribute to *1263the verdict. United States v. Gari , 572 F.3d 1352, 1362 (11th Cir. 2009). In contemplating this issue, we consider the importance of the witness's testimony, whether the testimony is cumulative, whether evidence corroborates or contradicts the testimony, the extent of cross-examination permitted, and the overall strength of the prosecution's case. Van Arsdall , 475 U.S. at 684, 106 S.Ct. 1431. We ask whether the average juror would find the prosecution's case less persuasive without the erroneously admitted testimony. Gari , 572 F.3d at 1363.
Here, the admission of the deposition testimony was not harmless. Even the Majority Opinion does not argue that it was.
Indeed, although the Majority Opinion asserts that "there was compelling evidence that the defendants' boat was headed to the United States," Maj. Op. at 58, conspicuously absent from that Opinion is an alternative holding that even if the district court erred in admitting Vixama's testimony, the error was harmless. And for good reason: Vixama's testimony provided the clearest, non-circumstantial evidence that the boat was bound for the United States. Specifically, she testified in her deposition that her mother's friend had arranged for her to be smuggled into Miami for $5,000 and that she understood the boat was going from Freeport directly to Miami.
Aside from the unclear and confusing deposition testimony of Davidson Francois, another material witness, the government presented no other direct evidence that the boat's intended destination was the United States.13 Rather, the other evidence on which the government relied for this necessary element of the crime was entirely circumstantial: that the migrants were told to turn off their cell phones for the trip and to not wave down other vessels once their boat stalled; that none aboard were legally authorized to be in the United States and lacked identification documents; and that Smith's explanation for the boat's track did not make sense. Certainly, these facts are more than sufficient to suggest an illegal operation and even a human-smuggling scheme. But they do not demand or even reasonably support the inference that the boat was United States-bound. Even the case agent agreed in his testimony that Vixama was an "essential" witness. Particularly in view of where the boat was intercepted, I cannot conclude beyond a reasonable doubt that Vixama's testimony did not contribute to the jury's conclusion that the boat was headed for the United States. As a result, the Confrontation Clause violation was not harmless and requires vacatur of the judgment and remand.
V.
To protect Smith's and Delancy's Sixth Amendment right to confront Vixama, the government was required to engage in reasonable efforts to produce Vixama at trial. But the government failed to do so. Among other deficiencies, though the government had good reason to believe that Vixama was physically present with her boyfriend, it did almost nothing to locate the boyfriend-even though it had the boyfriend's phone number and knew he lived in Delaware. Following up on the boyfriend's location would have involved routine, inexpensive law-enforcement techniques such as searching databases or running basic internet searches and sending an agent out to follow up on any resulting leads. Because *1264the government did not engage in these standard and reasonable efforts to follow up on its promising lead, it did not satisfy the Sixth Amendment's unavailability requirement before presenting Vixama's recorded testimony. As a result, Smith's and Delancy's Sixth Amendment right to the witness's presence was violated. And since that was not harmless error on this record, the judgment should be vacated and the case remanded. I therefore respectfully dissent.
Agents have many responsibilities, and I do not mean to suggest that the agent here was lazy. Rather, it appears he did not appreciate the scope of the efforts the Confrontation Clause requires to present a vital witness's live testimony, a circumstance that is understandable, in view of our Circuit's lack of prior reason to opine on the subject. Nonetheless, this circumstance does not make the government's efforts any more reasonable from a constitutional standpoint.
Since the Majority Opinion affirms the district court's admission of Vixama's deposition testimony, the conviction will not be vacated, and the case will not be remanded on that basis. Under those circumstances, I concur in the Majority Opinion's affirmance of the district court's denial of a mistrial on the basis of the prosecutor's closing argument.
The Majority Opinion distorts my discussion of Hardy 's lesson. As I explain, Hardy teaches that the government needs to make reasonable efforts to follow up on promising leads. The Majority Opinion instead claims that I assert a rule that requires the government to take even unreasonable steps to follow up on promising leads. Maj. Op. at 43-44 (claiming that the dissent's rule is "that the government does not make a good-faith, reasonable effort as a matter of law unless it, in effect, pursues each and every lead it has 'reason to believe' might assist in locating a missing witness." (emphasis added)). To be clear, that is not correct. I am not suggesting that the government needs to take unreasonable steps to pursue "each and every" promising lead. Under Hardy , the government must undertake only reasonable efforts-meaning reasonable tasks under the circumstances-to pursue promising leads calculated to find a missing witness. Indicators of reasonableness can include, for example, cost, time requirement, and ease of task, under the circumstances. So where a promising lead exists, performing a free or inexpensive database search for an address, which takes just minutes, is reasonable.
Given the difference in the prosecution's and ERO's interests in finding Vixama, it is, at best, questionable whether we should consider among the prosecution's efforts to present Vixama's testimony at trial ERO's limited efforts to locate Vixama after she was released from custody. I certainly do not suggest that Homeland Security Investigations could not elect to have ERO find Vixama and deport her. And had it done so successfully, there would have been no problem with having Vixama declared unavailable. Nor, as the Majority Opinion incorrectly asserts, do I suggest that "government agencies cannot work together to accomplish a common goal." Maj. Op. at 49-50 n.17. Of course they should. And as I have noted, had they succeeded and had ERO deported Vixama, Vixama would, in fact, have been unavailable for Sixth Amendment purposes. But that's not what happened here. So we must evaluate the efforts of the government to present Vixama at trial. And by definition, Homeland Security Investigations's efforts to have ERO find Vixama were not efforts to obtain Vixama's appearance at trial. In cases where a witness is illegally present in the United States, two agencies have different interests in the witness: the agency prosecuting the case-be it Homeland Security Investigations, the Federal Bureau of Investigation, or any other federal agency-and ERO. The agency prosecuting the case bears the responsibility of producing all necessary government witnesses for trial-even if they are present in the United States illegally. But ERO's mission, in contrast, is to enforce the immigration laws and deport illegal aliens once they have been designated for deportation. The prosecuting agency and ERO generally coordinate their efforts to avoid the premature deportation of a necessary trial witness. But because of the human cost of forcing material witnesses to suffer detention, and because of the financial and physical burdens on the government to detain innocent witnesses, ERO's interest in removal is sometimes prioritized over the prosecuting agency's interest in producing the witness at trial. See United States v. Valenzuela-Bernal , 458 U.S. 858, 865, 102 S.Ct. 3440, 73 L.Ed.2d 1193 (1982) (explaining that human costs and burdens on the government justify prompt deportation of unnecessary witnesses).
If, however, as happened here, ERO is unwilling or unable to perform its removal responsibility, that does not absolve the prosecuting agency of its obligation to produce the witness for trial. The Majority Opinion elides the distinctions between the functions and responsibilities of the prosecuting agency, on the one hand, and ERO, on the other, and, in assessing the reasonableness of the government's efforts to produce the witness at trial, evaluates together ERO's and the prosecuting agency's efforts to fulfill their conflicting functions. But ERO does not engage in any affirmative efforts to make a witness available for trial; just the opposite-it strives to deport a witness designated for deportation. And if the witness has not been deported and is present at the time of trial, the prosecuting agency continues to bear responsibility for producing that witness-regardless of whether the witness is legally or illegally present in the United States.
Under this test, the government's view of the importance of the witness and the seriousness of the crime necessarily modulate the amount of effort the government will undertake to find a given witness. Here, it is clear the government would have tried harder to find Vixama if it had not had her deposition testimony, since Vixama was critical to the case, see infra at Section IV, and the agent admitted that he stopped trying to find her, in significant part, because the government already had her recorded testimony.
See also McCandless v. Vaughn , 172 F.3d 255, 269 (3d Cir. 1999) ("If the prosecution had not had Barth's preliminary hearing testimony and had needed Barth's presence at trial, we are confident that the resources and effort devoted to finding him prior to trial would have been greater than they in fact were. To countenance such a disparity would ill serve the interests protected by the Confrontation Clause."); United States v. Lynch , 499 F.2d 1011, 1023 (D.C. Cir. 1974) ("In the ordinary case, [demonstrating that the government's search was "exercised both in good faith and with reasonable diligence and care"] will require a search equally as vigorous as that which the government would undertake to find a critical witness if it has no preliminary hearing testimony to rely upon in the event of 'unavailability.' ").
The Majority Opinion sets up a false choice between relying on the former attorney to obtain Vixama's appearance and employing basic law-enforcement techniques to follow up on the lead of the boyfriend's name and number. See Maj. Op. at 50 ("Because it appeared that the boyfriend (and Vixama through him) was cooperating with attorney Raben, it was reasonable for the government to rely on attorney Raben to communicate with him rather than to try to track the boyfriend down independently through databases and try to persuade him to help federal law enforcement take Vixama into ICE custody in Delaware and transfer her to Miami for trial." (emphasis added)). Here, though, Hardy 's rule demanded the government do both because both showed promise of obtaining Vixama's appearance, and both were reasonable.
The Bagger 293 is an excavating machine that is 315 feet tall and 738 feet wide. Wayne Grayson, Meet the 31 million-pound bucket wheel excavator. The largest land vehicle ever built , https://www.equipmentworld.com/video-meet-the-the-31-million-pound-bucket-wheel-excavator-the-largest-land-vehicle-ever-built/ (last visited June 29, 2019). It weighs more than 31 million pounds. Id. Despite its size, the vehicle requires only two people to operate it. Id.
While I respect and do not second-guess the government's decision not to charge Vixama with a crime, it is not technically accurate to suggest that the government had no option available to it to obtain a warrant for Vixama's arrest pretrial. HSI could have obtained an arrest warrant for Vixama for entering the United States illegally. See 8 U.S.C. § 1325(a) ("Any alien who (1) enters ... the United States at any time or place other than as designated by immigration officers ... shall, for the first commission of any such offense, be fined under title 18 or imprisoned not more than 6 months, or both ...."). Clearly, it had probable cause to do so. And had it obtained such a warrant and arrested Vixama, it could have exercised prosecutorial discretion to drop the charges and deport her before trial with the consent of Smith and Delancy (in which case, she would have been unavailable) or to have her testify and then drop the charges and deport or release her.
Hardy and Roberts are not the only cases the Majority Opinion summarizes for what seems like little purpose. In a single sentence and a footnote consisting of citations, this dissent relies on precedent from our sister Circuits for the point that other courts have looked, as a measure of reasonableness, to whether the prosecution's efforts match those it would have undertaken had it lacked the missing witness's testimony. See supra at 1227-28 & n.6. In response, the Majority Opinion spends pages laboriously summarizing the cited cases. Maj. Op. at 1241-41. When the dust settles, though, the Majority Opinion does not contest the point of citing the cases in the first place: that a factor courts can consider is whether the government tried as hard to look for the missing witness as it would have if it did not already possess what it needed from that witness.
All of these cases concern the Fourth Amendment.
When a foreign-national witness is happily located outside the United States, the United States generally has no options to obtain that witness's appearance at trial other than a request to appear with a promise to pay for travel. But when a witness is physically present in the United States, the government has various law-enforcement options available to it to produce the witness at trial. For this reason, and as the Majority Opinion recognizes, cases involving foreign-national witnesses happily residing outside the United States-such as United States v. Siddiqui , 235 F.3d 1318 (11th Cir. 2000), discussed by the Majority Opinion at 1228 & n.6-are not helpful in identifying the types of steps the government must engage in to satisfy the unavailability requirement when the witness is physically located here.
At some points in his deposition, Francois seemed to suggest that he was heading to the United States, but Francois's testimony was unclear at best. Significantly, as the Majority Opinion notes, he also testified that he did not personally know where the boat was going. Maj. Op. at 1219-20.