Latif v. Obama

Opinion for the Court filed by Circuit Judge BROWN.

Concurring opinion filed by Circuit Judge HENDERSON.

Dissenting opinion filed by Circuit Judge TATEL.

BROWN, Circuit Judge:

The United States appeals the district court’s grant of the writ of habeas corpus to detainee Adnan Farhan Abd A1 Latif. Three errors in the district court’s analysis require us to vacate that decision. First, the court failed to accord an official government record a presumption of regularity. Second, the district court failed to determine Latifs credibility even though the court relied on his declaration to discredit the Government’s key evidence. See Al-Adahi v. Obama, 613 F.3d 1102, 1110 (D.C.Cir.2010). Third, the court’s unduly atomized approach to the evidence is one we have rejected. See id. We remand so the district court can evaluate Latifs credibility as needed in light of the totality of the evidence, including newly available evidence as appropriate.

[Redacted]

[Redacted] In a declaration filed with the district court in 2009, Latif denies ever being part of the Taliban and offers an innocent explanation for his journey. Latif says he left Yemen in 2001 on a quest for medical treatment for head injuries he suffered in a 1994 car accident. He went *748to Pakistan to get help from Ibrahim, a Yemeni he had met at a charitable organization in Yemen. When Latif arrived in Quetta, Ibrahim had already left Pakistan, so Latif followed him to an Islamic studies institute in Kabul, Afghanistan. But once Latif caught up to Ibrahim at the institute, Ibrahim had to leave again and told Latif to wait for him there until they could travel together to Pakistan. After waiting in vain for several weeks, Latif says, he then returned to Pakistan without Ibrahim, fleeing U.S.-supported forces he had been told were advancing from northern Afghanistan.

The district court granted Latif s habeas petition following briefing and a hearing in which Latif declined to testify. Abdah v. Obama (Latif), 2010 WL 3270761, 2010 U.S. Dist. LEXIS 83596 (D.D.C. July 21, 2010). [Redacted]

In a Guantanamo detainee case, we review the district court’s “specific factual determinations” for clear error, and its ultimate grant or denial of habeas de novo. Almerfedi v. Obama, 654 F.3d 1, 5 (D.C.Cir.2011). As in our prior cases, we assume, without deciding, that the district court was correct to hold the Government to the preponderance-of-the-evidence standard. See id., 654 F.3d at 5 n. 4; Al-Bihani v. Obama, 590 F.3d 866, 878 & n. 4 (D.C.Cir.2010); see also Boumediene v. Bush, 553 U.S. 723, 787, 128 S.Ct. 2229, 171 L.Ed.2d 41 (2008) (“The extent of the showing required of the Government in these cases is a matter to be determined.”); Alr-Adahi, 613 F.3d at 1105 (“Although we doubt ... that the Suspension Clause requires the use of the preponderance standard, we will not decide the question in this case.”). To meet its burden, “the government must put forth credible facts demonstrating that the petitioner meets the detention standard, which is then compared to a detainee’s facts and explanation.” Almerfedi, 654 F.3d at 6.

At the heart of the Government’s case is [redacted]

Ordinarily, at this point in our analysis, we would simply review the district court’s comparison of the Government’s evidence with the “detainee’s facts and explanation,” bearing in mind that the ultimate burden is on the Government to establish Latif s detention is legal. Id. We pause here, however, because the district court expressly refused to accord a presumption of regularity to the Government’s evidence, and on appeal the Government continues to assert its Report is entitled to such a presumption.

“The presumption of regularity supports the official acts of public officers and, in the absence of clear evidence to the contrary, courts presume that they have properly discharged their official duties.” Sussman v. U.S. Marshals Serv., 494 F.3d 1106, 1117 (D.C.Cir.2007). The presumption applies to government-produced documents no less than to other official acts. See Riggs Nat’l Corp. v. Comm’r, 295 F.3d 16, 21 (D.C.Cir.2002) (holding that “an official tax receipt” of a foreign government “is entitled to a presumption of regularity”). But Latif (and our dissenting colleague) argue no such presumption can be applied in Guantanamo cases — at least not to [redacted] reports prepared in stressful and chaotic conditions, filtered through interpreters, subject to transcription errors, and heavily redacted for national security purposes.

Since the problems Latif cites are typical [redacted], the rule he proposes would subject all such documents to the he-said/she-said balancing of ordinary evi*749dence. It is impossible to cure the conditions under which these documents were created, so Latifs proposed rule would render the traditional presumption of regularity wholly illusory in this context. We conclude first that intelligence documents of the sort at issue here are entitled to a presumption of regularity, and second that neither internal flaws nor external record evidence rebuts that presumption in this case.

Courts sensibly have anticipated that some sort of presumption is proper in the Guantanamo context, but until now we have not directly addressed the question. The dissent interprets our silence heretofore as disapproval and suggests that a presumption in favor of the Government’s evidence in this case “inappropriately shift[s] the burden” of proof from the Government to the detainee. Dissenting Op. at 783. A Supreme Court plurality said just the opposite, however — -and in a case involving the military detention of an American citizen, no less:

[T]he Constitution would not be offended by a presumption in favor of the Government’s evidence, so long as that presumption remained a rebuttable one and fair opportunity for rebuttal were provided. Thus, once the Government puts forth credible evidence that the habeas petitioner meets the enemy-combatant criteria, the onus could shift to the petitioner to rebut that evidence with more persuasive evidence that he falls outside the criteria.

Hamdi v. Rumsfeld, 542 U.S. 507, 534, 124 S.Ct. 2633, 159 L.Ed.2d 578 (2004).

When the Supreme Court extended the habeas right to non-citizen detainees in 2008, it tasked the lower courts with developing a workable habeas remedy that would give detainees a “meaningful opportunity to demonstrate” the unlawfulness of their detention, Boumediene, 553 U.S. at 779, 128 S.Ct. 2229, yet it left unaddressed the content of the governing law, id. at 798, 128 S.Ct. 2229. Boumediene noted that “common-law habeas corpus was, above all, an adaptable remedy” whose “precise application and scope changed depending upon the circumstances.” Id. at 779, 128 S.Ct. 2229. Our dissenting colleague seems to think Boumediene mandates a skeptical — if not cynical — supervisory role for the courts over the Executive branch’s interactions with its detainees at Guantanamo. Dissenting Op. at 773. In our view, the Boumediene Court envisioned a much more modest judicial role. Aside from a few minimal procedural safeguards, designed to preclude the Government acting as its own judge,1 the Court left the scope of the habeas right to the common-law-like process in which we have been engaged ever since: “[T]he Suspension Clause does not resist innovation in the field of habeas corpus. Certain accommodations can be made to reduce the burden habeas corpus proceedings will place on the military without impermissibly diluting the protections of the writ.” Boumediene, 553 U.S. at 795, 128 S.Ct. 2229.

In that spirit, the district court has operated under a case management order that specifically authorized reliance on evidentiary presumptions. See In re Guantanamo Bay Detainee Litig., 2008 WL *7504858241, at *1-3, 2008 U.S. Dist. LEXIS 97095, at *104 (D.D.C. Nov. 6, 2008) (“The Merits Judge may accord a rebuttable presumption of accuracy and authenticity to any evidence the government presents as justification for the petitioner’s detention if the government establishes that the presumption is necessary to alleviate an undue burden presented by the particular habeas corpus proceeding.”). The Government has frequently invoked this order in urging a presumption that its evidence is accurate, but the district court, with no guidance from us, has been reluctant to grant anything more than a presumption of authenticity. See Benjamin Wittes, Robert M. Chesney & Larkin Reynolds, The Emerging Law of Detention 2.0: Guantanamo Habeas Cases as Lawmaking, at 52-53 nn. 237-43 (May 12, 2011) (citing cases granting a presumption of authenticity but not accuracy), http://www. brookings.edu/papers/2011/05_ guantanamo-wittes.aspx (last visited September 30, 2011). Aside from our silence, there are at least two other reasons why the district court has not applied a presumption of accuracy.

Confusion about the nature of the presumption may account for the district court’s reluctance. In an order applicable to the present case, the district court held, “any evidence presented by the government that has been created and maintained in the ordinary course of business should be afforded a presumption of authenticity,” Dist. Ct. Docket No. 606, but the court rejected the government’s request for a presumption of accuracy “for the reasons stated by Judge Kessler in Ahmed v. Obama, 613 F.Supp.2d 51, 54-55 (D.D.C.2009) and Judge Kollar-Kotelly in Al Mutairi v. United States, [644 F.Supp.2d 78 (D.D.C. July 29, 2009) ].” Id. Those cases misunderstood the nature of the.presumption. In Ahmed and Al Mutairi, the district court assumed the requested presumption would go to the truth of “the facts contained in the Government’s exhibits.” Ahmed, 613 F.Supp.2d at 55. Since “the accuracy of much of the factual material contained in the [Government’s] exhibits [was] hotly contested,” id., quoted in Al Mutairi, 644 F.Supp.2d at 84, and the evidentiary dispute in Ahmed involved allegations that the relevant statements were “obtained by torture,” Ahmed, 613 F.Supp.2d at 55, the court was rightly disinclined to grant them a presumption of truth. But the presumption of regularity does not require a court to accept the truth of a non-government source’s statement.

The confusion stems from the fact that intelligence reports involve two distinct actors' — the non-government source and the government official who summarizes (or transcribes) the source’s statement. The presumption of regularity pertains only to the second: it presumes the government official accurately identified the source and accurately summarized his statement, but it implies nothing about the truth of the underlying non-government source’s statement. There are many conceivable reasons why a government document might accurately record a statement that is itself incredible. A source may be shown to have lied, for example, or he may prove his statement was coerced. The presumption of regularity — to the extent it is not rebutted — requires a court to treat the Government’s record as accurate; it does not compel a determination that the record establishes what it is offered to prove.

Another reason the district court has denied the Government’s motions for a presumption of accuracy may be that such a presumption is often unnecessary or irrelevant. The Government has frequently been able to prove its detention authority without relying on any presumption that *751its records are accurate. And in many cases, detainees do not challenge the Government’s recordkeeping. Instead, they attack the sufficiency of the evidence, or they claim that the Government’s information is unreliable because it resulted from harsh interrogation techniques, multiple levels of hearsay, or unknown sources.

This case presents a different question because Latifs sole challenge is to the accuracy of the Government’s [redacted]. When the detainee’s challenge is to the evidence-gathering process itself, should a presumption of regularity apply to the official government document that results? We think the answer is yes.

To forbid a presumption of regularity in spite of Boumediene’s implicit invitation to innovate, 553 U.S. at 795, 128 S.Ct. 2229, would be particularly counterintuitive, since the field of habeas corpus is already well accustomed to such burden-shifting presumptions. In a state prisoner’s federal habeas proceeding, for example, “a determination of a factual issue made by a State court shall be presumed to be correct,” and “the applicant shall have the burden of rebutting the presumption of correctness by clear and convincing evidence.” 28 U.S.C. § 2254(e)(1); see Al-Bihani, 590 F.3d at 878. And after a state court conviction becomes final, it is subject to a “presumption of regularity,” such that “[i]f that conviction is later used to enhance a [federal] criminal sentence, the defendant generally may not challenge the enhanced sentence through a petition under § 2254 on the ground that the prior conviction was unconstitutionally obtained.” Lackawanna Cty. Disk Att’y v. Coss, 532 U.S. 394, 403-04, 121 S.Ct. 1567, 149 L.Ed.2d 608 (2001); see also Parke v. Raley, 506 U.S. 20, 30, 113 S.Ct. 517, 121 L.Ed.2d 391 (1992) (same for enhancement of a state court sentence).2

Just as principles of vertical comity and federalism justify presumptions in favor of state court judgments in ordinary criminal habeas proceedings, see Sumner v. Mata, 449 U.S. 539, 547, 101 S.Ct. 764, 66 L.Ed.2d 722 (1981), the horizontal separation of powers justifies a presumption in favor of official Executive branch records in Guantanamo habeas proceedings. The district court is uniquely qualified to determine the credibility of hearsay, and the presumption of regularity does not detract from that role. But courts have no special expertise in evaluating the nature and reliability of the Executive branch’s wartime records. For that, it is appropriate to defer to Executive branch expertise. See Boumediene, 553 U.S. at 796-97, 128 S.Ct. 2229 (“In considering both the procedural and substantive standards used to impose detention to prevent acts of terrorism, proper deference must be accorded to the *752political branches.”). Both the Constitution and common sense support judicial modesty when assessing the Executive’s authority to detain prisoners during wartime, for it is an area in which the judiciary has the least competence and the smallest constitutional footprint.

Our dissenting colleague concludes the presumption of regularity should not extend to official intelligence reports because he imagines the presumption of regularity is just a shortcut for crediting the work product of official processes we know to be “transparent, accessible, and often familiar,” Dissenting Op. at 771, and because he thinks we know relatively little about how intelligence reports are created, id. at 771-72. Both premises are false. Courts regularly apply the presumption to government actions and documents that result from processes that are anything but “transparent,” “accessible,” and “familiar.” The presumption of regularity is founded on inter-branch and inter-governmental comity, not our own judicial expertise with the relevant government conduct. In Riggs National, we presumed a foreign government entity’s receipt to be reliable without pretending it was produced by a “familiar” or “transparent” process. Id. at 771; see 295 F.3d at 20-22. Likewise, federal courts need no expertise concerning the procedures of state courts, probation offices, and drug treatment centers to afford their official records a presumption of regularity. See cases cited supra note 2. Thanks to the explanatory declarations [redacted] which we discuss below, see infra at 780, we know far more about the personnel, process, and standards involved in producing intelligence records like the Report than we do about the foreign and state governmental organs whose records we also presume to be reliable, and we have no reason to suspect such documents are fundamentally unreliable.3

Rather than cast doubt on the viability of the presumption of regularity in this context, our only pertinent post-Boumediene discussion of the presumption strongly suggests its continuing viability. In Alr-Bihani, the detainee complained that the district court had “erred by ... presuming the accuracy of the government’s-evidence.” 590 F.3d at 875. Without isolating the components of Al-Bihani’s multifaceted procedural argument — it included attacks on the standard of review, the denial of a full-blown evidentiary hearing, alleged burden-shifting, and the district court’s discovery orders — we said that his “argument clearly demonstrate[d] [his own] error” and that Boumediene’s holding had placed it on “shaky ground.” Id. at 876. Without explicitly confirming that the district court had applied a presumption in favor of the Government’s evidence in that case, we noted that its case management order “reserved the district court’s discretion, when appropriate, to adopt a rebuttable presumption in favor of the accuracy of the government’s evidence.” Id. at 869-70. We implied that the district court had in fact exercised this discretion when we quoted the order with approval in our hearsay analysis. Id. at 880 (“[T]he Court will determine, as to any evidence introduced by the Government, whether a presumption of accuracy and/or authenticity should be accorded.”). Consistent with this order, we noted, the district court had judged the “admissions pre*753sented by the government to be ‘credible and consistent.’ ” Id. Indeed, the district court relied on “certain statements by the petitioner that the Court finds credible and certain classified documents” without entertaining the possibility that the detainee’s statements had been mis-reported. Al Bihani v. Obama, 594 F.Supp.2d 35, 38-39 (D.D.C.2009). We did not distinguish the presumption of regularity from the admission of hearsay evidence generally, but we noted that “had the district court imposed stringent standards of evidence in the first instance, the government may well have been obligated to go beyond Al-Bihani’s interrogation records and into the battlefield to present a case that met its burden,” Al Bihani 590 F.3d at 877-78, and we “disposed of’ “[t]he rest of Al-Bihani’s procedural claims ... without extended discussion,” id. at 881. Although Al-Bihani does not clearly hold the district court may accord government evidence a presumption of regularity, that case is certainly consistent with today’s holding.

Although it was decided under the preBoumediene Detainee Treatment Act of 2005 (DTA), our opinion in Parhat v. Gates, 532 F.3d 834 (D.C.Cir.2008), also lends support to the continuing viability of such a presumption. In Parhat, we noted that the DTA incorporated by reference a “rebuttable presumption that the Government Evidence is genuine and accurate.” Id. at 847 (quotation marks omitted) (quoting Implementation of Combatant Status Review Tribunal Procedures at E-l § G(ll) (July 29, 2004)). We reversed the Tribunal’s decision because the Government’s evidence, despite the presumption in its favor, could not “sustain the determination that Parhat is an enemy combatant.” 532 F.3d at 847. The intelligence consisted of anonymous hearsay in the form of unsupported “bottom-line assertions,” so it was impossible for us to “assess the reliability of the assertions in the documents.” Id. We explained that “[i]f a Tribunal cannot assess the reliability of the government’s evidence, then the ‘rebuttable’ presumption becomes effectively irrebuttable.” Id. Although we found the presumption rebutted in Parhat, we cast no doubt on the propriety of such a presumption in the Guantanamo context. Parhat still “sets the guideposts for our inquiry into the reliability of the [Government’s] evidence in a detainee’s habeas case.” Bensayah v. Obama, 610 F.3d 718, 725 (D.C.Cir.2010) (quoting Barhoumi v. Obama, 609 F.3d 416, 428 (D.C.Cir.2010)). And neither the Supreme Court nor our court has ever rejected the presumption we analyzed in that case.

Our dissenting colleague points to four more recent cases to defend his view that intelligence documents like the Report in this case are undeserving of a presumption of regularity. Dissenting Op. at 774-76 (citing Barhoumi 609 F.3d 416, Bensayah, 610 F.3d 718, Al Ahvi v. Obama, 653 F.3d 11 (D.C.Cir.2011), and Khan v. Obama, 655 F.3d 20 (D.C.Cir.2011)). But we had no occasion to apply such a presumption in any of these cases, and none of them limits our discretion to do so under Boumediene.

In Barhoumi we considered a Government intelligence report containing a translation of a diary. Although we affirmed the district court’s favorable treatment of the Government’s evidence, 609 F.3d at 428-31, we did not apply a presumption of regularity. The reason for that omission is simple. The district court had credited the Government’s evidence without applying a presumption of regularity, and we were reviewing for clear error. See Brief of Respondents-Appellees at 52, Barhoumi 609 F.3d 416, ECF No. 1236093 (observing that “the district court did not presume the accuracy or authentic*754ity of the government’s evidence”). True, the Government’s brief interpreted the criminal cases on which Barhoumi relied as “acknowledging] that, absent ‘unusual circumstances,’ a translation is assumed to be accurate” in “criminal proceedings governed by the Confrontation Clause and the Federal Rules of Evidence.” Id. at 45-46 (quoting United States v. Martinez-Gaytan, 213 F.3d 890, 892 (5th Cir.2000) and United States v. Vidacak, 553 F.3d 344, 352 (4th Cir.2009)). But the Government did not ask us to apply any such presumption to its evidence. Indeed, the Government noted those criminal cases were “clearly distinguishable.” Id. at 45. We agreed. The cases Barhoumi relied on related to the question of admissibility, which we observed was irrelevant in the Guantanamo habeas context since all hearsay is admissible. We rejected the detainee’s contention that deficiencies in the translation rendered it unreliable. See Barhoumi, 609 F.3d at 431. We certainly did not deny the possibility that a presumption of accuracy might apply in the habeas context — we simply were not confronted with that question. Our opinion in Barhoumi therefore cannot bind us to the dissent’s view that the constitutional right to habeas precludes any presumption in favor of an official intelligence report. “Constitutional rights are not defined by inferences from opinions which did not address the question at issue.” Texas v. Cobb, 532 U.S. 162, 169, 121 S.Ct. 1335, 149 L.Ed.2d 321 (2001); see also Lopez v. Monterey Cty., 525 U.S. 266, 281, 119 S.Ct. 693, 142 L.Ed.2d 728 (1999) (“[T]his court is not bound by its prior assumptions.”); cf. Ariz. Christian Sch. Tuition Org. v. Winn, — U.S. -, 131 S.Ct. 1436, 1448-49, 179 L.Ed.2d 523 (2011) (“When a potential jurisdictional defect is neither noted nor discussed in a federal decision, the decision does not stand for the proposition that no defect existed. The Court would risk error if it relied on assumptions that have gone unstated and unexamined.” (citations omitted)); Brecht v. Abrahamson, 507 U.S. 619, 631, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993) (“[S]ince we have never squarely addressed the issue, and have at most assumed the applicability of the Chapman [v. State of California, 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967) ] [‘harmless beyond a reasonable doubt’] standard on habeas, we are free to address the issue on the merits.”).

For the same reason, we cannot extract from Bensayah, Al Alwi, or Khan the dissent’s proposed bar on evidentiary presumptions for intelligence reports. As in Barhoumi, the Government did not request a presumption of regularity in any of these appeals. See Brief of Respondents-Appellees at 38-39, Bensayah, 610 F.3d 718; Brief of Respondents-Appellees at 34-38, Al Alvi, 653 F.3d 11; Brief of Respondents-Appellees at 45-58, Khan, 655 F.3d 20. Thus, the court appropriately refrained from addressing the viability of such a presumption in each of those cases. See Rumber v. District of Columbia, 595 F.3d 1298, 1302 (D.C.Cir.2010) (“[W]e follow our usual practice of declining to reverse the district court based on arguments that the appellant did not raise.”); United States v. Vizcaino, 202 F.3d 345, 348 (D.C.Cir.2000) (“Because [the defendant] failed to preserve the argument for appeal, we review ... at most for plain error.” (emphasis added)). Absent relevant arguments, none of these cases can be read to foreclose a presumption of regularity for government documents in general or intelligence reports in particular.

Apart from its precedential argument, the dissent frets that “in practice” the presumption of regularity will compel courts to rubber-stamp government detentions because it “suggests] that whatever *755the government says must be treated as true.” Dissenting Op. at 779 (quoting Par-hat, 532 F.3d at 849). That fear is unfounded. Again, the presumption of regularity, if not rebutted, only permits a court to conclude that the statements in a government record were actually made; it says nothing about whether those statements are true. And while the presumption applies to government records, it does not apply only to the government’s evidence. If a detainee introduces a government record to support his side of the story — as has been done in the past, see, e.g., Awad, 608 F.3d at 5 (“In support of his petition, Awad introduced into evidence ... additional statements he made to his interrogators”) — he can benefit from the presumption as well. Finally, the presumption likely will never play a larger role in the resolution of a case than it does here (because the reliability of the Report is the central dispute), and even here, the presumption is not dispositive.

A body of judge-made law is not born fully formed, like Athena from the head of Zeus. It grows gradually, developing little by little in response to the facts and circumstances of each new case. Until now, we have not had to decide whether the common-law presumption of regularity applies in Guantanamo habeas proceedings. This case finally forces the issue because Latif challenges only the reliability of the Report, and because the Government persists in its request for a presumption of regularity on appeal.4 We hold that in Guantanamo habeas proceedings a rebut-table presumption of regularity applies to official government records, including intelligence reports like the one at issue here.

Because the Report is entitled to a presumption of regularity, and because the Report, if rehable, proves the lawfulness of Latif s detention, we can only uphold the district court’s grant of habeas if Latif has rebutted the Government’s evidence with more convincing evidence of his own.5

*756Viewed together, both [redacted] and the other evidence he uses to attack its reliability fail to meet this burden.

[Redacted]

[Redacted]6 [Redacted]

[Redacted]

[Redacted]7

[Redacted]

[Redacted]8

[Redacted]9,10

III

The district court issued its decision in this case a week after we published our opinion in Al-Adahi v. Obama, 613 F.3d 1102 (D.C.Cir.2010). We observed that “[o]ne of the oddest things” about that case was that “despite an extensive record and numerous factual disputes, the district court never made any findings about whether Al-Adahi was generally a credible witness or whether his particular explanations for his actions were worthy of belief.” Id. at 1110. The district court’s analysis in this case suffers from the same omission. Because the court relied in part on Latif s declaration in discrediting the Report, see Latif, 2010 U.S. Dist. LEXIS 83596, slip op. at 26 (“[T]he Court cannot credit [the Report] because ... Latif has presented a plausible alternative story.”), the district court was obligated to consider his credibility. Only a credible story could overcome the presumption of regularity to which the Report was entitled. The court’s failure to make a credibility finding is especially puzzling where the inculpatory and exculpatory versions of the detainee’s story overlap so that the factfinder is forced to untangle the detainee’s [redacted] story [redacted] Even doting Uncle Henry managed to evaluate Dorothy’s credibility when she professed that the family and friends gathered around her bed had been with her in Oz. See The Wizard of Oz (MGM 1939) (“Of course we believe you, Dorothy.”). The district court, by contrast, mustered only a guarded finding of plausibility. See Latif, 2010 U.S. Dist. LEXIS 83596, slip op. at 26.

Latif makes two main arguments in defense of the district court’s decision to proceed without an explicit finding of credibility. First, he argues that the court did *757in fact believe his declaration even though its opinion did not use those words. Second, he argues no credibility determination is necessary because the district court relied on the inherent weakness of the Government’s evidence to discredit it. Neither argument has merit.

The closest the district court’s opinion comes to making a credibility determination is in its statements that Latifs story was “plausible” and “not incredible.” Id., slip op. at 26-27. A story may be “plausible” or “not incredible” and yet be very unlikely. Cf. Uthman v. Obama, 687 F.3d 400, 406 (D.C.Cir.2011) (“Uthman’s account ... involves many coincidences that are perhaps possible, but not likely.”). A judgment about credibility, by contrast, measures the truthfulness of the speaker or the likelihood that what he says is true. See Richard Hooker, The Laws of Ecclesiastical Polity bk. II, ch. 4, at 151-52 (George Edelen ed., Harvard Univ. Press 1977) (1594) (“[T]hings are made credible, eyther by the knowne condition and qualitie of the utterer, or by the manifest likelihood of truth which they have in themselves.”). Thus, neither of the district court’s statements is equivalent to a finding that Latifs declaration is more likely true than false. On this, we are all agreed. See Dissenting Op. at 783-84.

By definition, a “plausible” statement is one “seeming reasonable, probable, or truthful”; it may in reality have only “a false appearance of reason or veracity.” Oxford English Dictionary Online, http:// www.oed.com/view/Entry/145466 (definition 4.a) (emphasis added) (last visited June 16, 2011). A plausible explanation does not necessarily compel credence. See Zamanov v. Holder, No. 08-72340, 649 F.3d 969, 974 (9th Cir.2011) (“[Petitioner’s] explanation ... is plausible. However, the record does not compel the finding that the [Immigration Judge’s] unwillingness to believe this explanation ... was erroneous.”). It is when a detainee tells a plausible story that an evaluation of his credibility is most needed. There may be several plausible explanations for Latifs itinerary; it is the district court’s job to decide whether the Government’s explanation is more likely than not. See Al-Adahi, 613 F.3d at 1110 (“Valid empirical proof requires not merely the establishment of possibility, but an estimate of probability.” (quoting David Hackett Fischer, Historians’ Fallacies: Toward a Logic of Historical Thought 53 (1970))).

Likewise, to say Latifs tale is “not incredible” is not to imply its teller ought to be believed. At best, the district court’s statement means a reasonable finder of fact could believe Latifs story, not that he has actually done so. Cf. United States v. Wooden, 420 F.2d 251, 253 (D.C.Cir.1969) (“The appellant’s story was not incredible; indeed, the jury seems to have accepted it, at least in part....”). Different factfinders may come to different conclusions about whether to credit evidence that is “not incredible” as a matter of law.

Other statements in the district court’s opinion confirm that it did not reach a decision on Latifs credibility. For example, the court rejected the Government’s “contention that Latif must be lying,” Latif, 2010 U.S. Dist. LEXIS 83596, slip op. at 27 (emphasis added), while assiduously avoiding any determination that Latif was not lying. The court speculated more than once that the inconsistencies in his statements “may be the result of a misstatement or a mistranslation,” without ever making a finding to that effect. Id., slip op. at 27 (emphasis added); id. (“The smaller inconsistencies ... may be no more than misstatements or mistranslations.” (emphasis added)). Likewise, the *758court found that “Latif did have an injury ... for which he might therefore have sought treatment.” Id., slip op. at 28 (emphasis added); see also id., slip op. at 6 n.4 (citing Latifs “alternative explanation for not having his passport at the time he was seized,” without deciding whether that explanation is more likely than the Government’s incriminating explanation). The district court provided no indication that it actually believed Latifs story and instead noted the story’s “inconsistencies and unanswered questions.” Id., slip op. at 27.

The district court’s decision gives us no reason to believe it would have reached the same result had it not relied on Latifs “plausible” version of the relevant events. The court said it could not “credit” [redacted] partly “because ... Latif has presented a plausible alternative story to explain his travel.” Id., slip op. at 26. Instead of advancing from plausibility to a judgment about Latifs veracity, the court repeated its plausibility finding: [redacted] The district court clearly relied on Latifs alternate account of his trip as one basis for rejecting [redacted]

[redacted]11 Because the district court only found Latifs story “plausible,” not credible, the court merely established the possibility, not the probability, that Latifs story was true. And without a “comparative judgment about the evidence,” there is no finding of fact for this court to review. Ah-Adahi, 613 F.3d at 1110.

By forgoing a determination of credibility for one of plausibility, the district court replaced the necessary factual finding with a legal conclusion that some other reasonable factfinder might believe Latifs story. In other words, the district court took on the role of a reviewing court, assuming in effect that Latif already had been found credible and then applying a deferential standard of review to that imaginary finding. Cf. Awad, 608 F.3d at 7 (“[I]f the district court’s account of the evidence is plausible in light of the record viewed in its entirety, the court of appeals may not reverse it.”). We cannot allow the district court to bypass its factfinding role in favor of an appellate standard of review. Cf. United States v. Anderson, 632 F.3d 1264, 1269-70 (D.C.Cir.2011) (noting that the district court may not apply the appellate court’s standard of review in crafting its own sentence). And since “de novo fact-finding is inconsistent with [an appellate court’s] proper role,” United States v. Brockenborrugh, 575 F.3d 726, 746 (D.C.Cir.2009), we are at an impasse.

In sum, the district court’s failure to resolve the key question of [the lead witness’s] credibility makes it impossible for us to perform our appellate function. “The purpose of an appeal is to review the judgment of the district court, a function we cannot properly perform when we are left to guess at what it is we are reviewing.” We therefore vacate the district court’s order and remand for further proceedings consistent with this opinion.

United States v. Holmes, 387 F.3d 903, 907-08 (D.C.Cir.2004) (quoting United States v. Williams, 951 F.2d 1287, 1290 (D.C.Cir.1991)).

On remand, the district court may consider any relevant, admissible evidence to aid its evaluation of Latifs credibility. If Latif again declines an opportunity to testify, that is another fact bearing on his credibility. Although the district court’s factual findings may be supported

*759by documentary evidence no less than by oral testimony, see Barhoumi, 609 F.3d at 423-24, a civil party’s decision not to testify may support an adverse inference about his credibility, see Mitchell v. United States, 526 U.S. 314, 328, 119 S.Ct. 1307, 143 L.Ed.2d 424 (1999) (“The Fifth Amendment does not forbid adverse inferences against parties to civil actions when they refuse to testify in response to probative evidence against them.”). Latif argues “it would make no sense to require an adverse inference in habeas cases in which the petitioner declines to testify while prohibiting such inferences in criminal cases.” Appellee’s Br. 52. This neglects the crucial point that the rule for criminal cases is based on the Fifth Amendment privilege against self-incrimination. See Mitchell, 526 U.S. at 316, 119 S.Ct. 1307. That privilege has no application outside the criminal context, and a Guantanamo habeas petitioner is not entitled to the same constitutional safeguards as a criminal defendant. Cf. Al-Bihani, 590 F.3d at 879 (“[T]he Confrontation Clause applies only in criminal prosecutions and is not directly relevant to the habeas setting.”). Especially where a detainee’s own self-serving statements comprise the only evidence against the Government’s case, his refusal to testify is relevant to the district court’s credibility determination.12

“[A] court considering a Guantanamo detainee’s habeas petition must view the evidence collectively rather than in isolation.” Salahi v. Obama, 625 F.3d 745, 753 (D.C.Cir.2010). A habeas court’s failure to do so is a legal error that we review de novo, separate and apart from the question of whether the resulting findings of fact are clearly erroneous in themselves. See Al-Adahi, 613 F.3d at 1111 (“[T]he district court clearly erred in its treatment of the evidence and in its view of the law. The court’s conclusion was simply not a permissible view of the evidence. And it reached this conclusion through a series of legal errors.”). Under Alr-Adahi, a detainee is not entitled to habeas just because no single piece of evidence is sufficient by itself to justify his detention. 613 F.3d at 1105-06. It follows that a habeas court may not ignore relevant evidence, for a court cannot view collectively evidence that it has not even considered.

Perhaps because it had already denied the Government’s key evidence a presumption of regularity, the district court committed both errors, explaining away some of the individual contradictions and coincidences in Latifs story one by one, as if each stood alone, and ignoring other probative details altogether. In Alr-Adahi, we reversed the district court’s grant of habeas because the court had failed to consider all the evidence in context. Viewing the evidence as a whole, we concluded the Government had proven the detainee “was more likely than not part of al-Qaida.” Id. at 1111. Although we do not reach an ultimate conclusion on the merits in this case, the district court’s similar treatment of the evidence in this case provides an alternative basis for remand.

The district court’s unduly atomized approach is illustrated by its isolated treatment (or failure to consider) several potentially incriminating inferences that arise from evidence Latif himself offers in support of his petition — namely, (a) striking similarities between Latifs exculpatory *760story and [redacted] (b) the route Latif admits traveling, and (c) contradictions in Latifs exculpatory statements. In addition, the district court improperly declined to consider [redacted] We leave to the district court on remand the unfinished task of weighing this evidence in the aggregate.

What makes Latifs [redacted] story so hard to swallow is not its intrinsic implausibility but its correspondence in so many respects with [redacted] Latifs [redacted] account of what transpired bears a striking resemblance to [redacted]13

[Redacted]

Nor did the district court consider that Latifs admitted route to Afghanistan from his home in Yemen [redacted] We have held that “traveling to Afghanistan along a distinctive path used by al Qaeda members can be probative evidence that the traveler was part of al Qaeda.” Uthman, 637 F.3d at 405 (citing Al Odah, 611 F.3d at 16). At Guantanamo, more than a year after his capture, Latif told his interrogators he flew from Sana’a, Yemen to Karachi, Pakistan in early 2001 with a plane ticket Ibrahim gave him. From there he took a bus to Quetta, Pakistan and a taxi to Kandahar, Afghanistan as Ibrahim had instructed. Then Ibrahim took him by taxi to Kabul, where Latif said he spent five months in the religious study

center.14 This route has been well traveled by al-Qaida and Taliban recruits and by our precedent. See Uthman, 637 F.3d at 405 (noting that Uthman’s route from Sana’a to Karachi by plane, from Karachi to Quetta by bus, from Quetta to a Taliban office by taxi, and from there to Kandahar “is similar to the paths of admitted al Qaeda members”); Al Odah, 611 F.3d at 10,12 (noting that a similar “route used by al Odah was a common travel route for those going to Afghanistan to join the Taliban”). The record in this case is replete with interrogation summaries of other Yemeni detainees who followed the same route to Afghanistan. Instead of focusing on Latifs route, the district court observed that “[n]o other detainee told interrogators that he fled from Afghanistan to Pakistan, from Tora Bora or any other location, with Latif.” Latif, 2010 U.S. Dist. LEXIS 83596, slip op. at 26. That is true. But the court overlooked the implications of Latifs own subsequent admissions about the route he traveled.15 This is relevant evidence, and it should have factored into the district court’s decision. The court’s failure even to consider it is a legal error that compels remand.

[Redacted] This is a welcome step toward the holistic approach to the evidence we called for in AL-Adahi. But as with the other evidence, the district court examined some contradictions in isolation from the rest of the evidence and overlooked others altogether.

*761The court gestured obliquely to what it characterized as “smaller inconsistencies” that it concluded “may be no more than misstatements or mistranslations.” Id. Apparently, the court found it unnecessary to get to the bottom of these contradictions because “even if some details of Latifs story have changed over time, for whatever reason, its fundamentals have remained the same.” Id. [redacted] Applied to Latifs contradictory statements, the district court’s reasoning neglects “the well-settled principle that false exculpatory statements are evidence — often strong evidence — of guilt.” Al-Adahi 613 F.3d at 1107. Thus, even if a given inconsistency in a detainee’s story does not go to the central question of his involvement with the Taliban or al-Qaida, it may be relevant nonetheless to the court’s evaluation of his credibility, which in turn bears on the reliability of the Government’s evidence. Cf. United States v. Philatelic Leasing, Ltd., 601 F.Supp. 1554, 1565 (S.D.N.Y.1985) (citing the principle, “which Wigmore has described as ‘one of the simplest in human experience,’ ” that “when a litigating party resorts to ‘falsehoods or other fraud’ in trying to establish a position, the court may conclude the position to be without merit and that the relevant facts are contrary to those asserted by the party”) (quoting 2 John Henry Wigmore, Evidence § 278, at 133 (1979)).

[Redacted] The court observed, for example, that in Latifs 2009 declaration (in which he claimed to be too disabled to fight) Latif said he “spent three months at the Islamic Jordanian Hospital in Amman, Jordan,” Petitioner’s Decl., ¶ 3, but his own medical records reveal that he was released just five days after admission. The court made no explicit finding about the source of this inconsistency, and it failed to mention that Latif himself testified before the Combatant Status Review Tribunal that he was “treated ... for five days,” ISN 156 CSRT Tr. at 8, a fact that is surely relevant to the credibility of Latifs recent declaration.

[Redacted]

The court failed even to mention other incongruities among the stories Latif has told his interrogators. Latif has said that he stayed with a doctor in Kabul, but also that he stayed in a religious study center there; that Latif was arrested at the Pakistani border fleeing Afghanistan, but also that he was arrested at a hospital in Pakistan; that he paid for his medical treatment, but also that he could not pay; that Ibrahim’s charitable organization is called Jamiat an-Nur, but also that it is called Gameiat al Hekma or, alternatively, Jameiah Islam. Even if some of the inconsistencies in Latifs story “may be,” as the district court suggested about others, “no more than misstatements or mistranslations,” Latif, 2010 U.S. Dist. LEXIS 83596, slip op. at 27, viewed together with the rest of the evidence they undermine the credibility of Latifs declaration. “We do not say that any of these particular pieces of evidence are conclusive, but we do say that they add to the weight of the government’s case against [the detainee] and that the district court clearly erred in tossing them aside.” Al-Adah% 613 F.3d at 1110.

[Redacted]

[Redacted]16

*762E

In a recent case, we held “the location and date of [the detainee’s] capture, together with the company he was keeping, strongly suggest that he was part of al Qaeda.” Uthman> 637 F.3d at 405. The Yemeni detainee in that case was captured in December 2001 with at least five other Yemeni men, two of whom were confessed al-Qaida members, at the Afghan-Pakistani border near Tora Bora, a cave complex in Eastern Afghanistan that was, at that time, the site of a battle between al-Qaida and the United States. Id. Analogous details in the circumstances of Latifs capture should have been weighed in combination with the rest of the Government’s incriminating evidence.

Latif admits that he was captured in “late 2001” after being led across the Afghan border into Pakistan, Appellee’s Br. 7, and he confirmed to his Guantanamo interrogators that an Afghan guide led him across the border. The record contains no direct evidence about Latifs route from Kabul to the Pakistani border. [Redacted] But the district court concluded “the timing of [Latifs] departure from Kabul is not sufficient to create an inference that he was involved in fighting.” Id., slip op. at 27 (emphasis added). This is exactly the formulation we criticized in Alr-Adahi. In that case the district court concluded “Al-Adahi’s attendance at an al-Qaida training camp ‘is not sufficient to carry the Government’s burden of showing that he was a part’ of al-Qaida.” 613 F.3d at 1105 (emphasis added). We cited that statement as an example of the court’s having “wrongly ‘required each piece of the government’s evidence to bear weight without regard to all (or indeed any) other evidence in the case.” Id. at 1105-06. The district court commits exactly the same “fundamental mistake” in this case by considering the time and place of Latifs capture in isolation from the rest of the evidence. Id. at 1106. The question to ask is not whether the circumstances of Latifs capture are sufficient by themselves to prove he was part of the Taliban, but whether, in combination with the rest of the evidence, they make that conclusion more likely than not.

[Redacted]17

The dissent admits the circumstances of Latifs flight from Afghanistan are helpful to the Government’s case, but contends they may not be very helpful since, for all we know, his route was frequented by noncombatants too. Dissenting Op. at 785-86. This bold speculation is beyond our purview as an appellate court, and the district court did not suggest it had so much as considered the possibility. (Indeed, the record contains no evidence to support the dissent’s theory.) At this juncture, all we can say is that the location and timing of Latifs exodus is relevant evidence, and the district court erred by considering his route in isolation and ignoring the similarly situated detainees’ altogether.

To summarize, in addition to viewing Latifs own statements in isolation, the district court ignored the probative value of (1) Latifs familiar, four-leg route to Kabul; (2) Latifs CSRT testimony that he was hospitalized for just five days instead of three months as he now claims; (3) [redacted] One cannot gather from a fair reading of the district court’s opinion that any of these facts informed its conclusion about the Government’s evidence. In light of our application of the presumption of regularity, there can be no question on *763remand but that all of this evidence must be considered — and considered as a whole.

The dissent makes much of the fact that, contrary to the usual practice, we do not assume the court considered all the evidence it failed to mention. Dissenting Op. at 789-90. If that is true, the result flows from the unusual posture of this case. Even in the typical he-said/she-said case— in which two people provide conflicting statements — the court must conduct a close and precise balancing of the evidence to reach a valid result. In detainee cases the difficulties are heightened because it is a he-said/he-said case — the same person provides both the incriminating and exculpatory statements. Thus the Al Adahi formulation becomes critical.

The district court’s failure to address certain relevant evidence leaves us with no confidence in its conclusions about the evidence it did consider. For example, the district court implicitly rejected evidence that Latifs purported benefactor, Ibrahim Al-Alawi, is actually Ibrahim Ba’alawi, known as Abu Khalud, an al-Qaida facilitator. Other detainees have described Ibrahim Ba’alawi in much the same role [redacted]. Several detainees reported meeting Ibrahim Ba’alawi in Taiz, Yemen, near Ibb, [redacted] and being recruited by him to fight jihad. They report that Ba’alawi arranged their travel along the same route Latif took to Afghanistan, lived in Kandahar as Latifs benefactor did, and arranged for their attendance at military training camps. Although noting the similarities between Ibrahim Ba’alawi and the Ibrahim Al-Alawi who appears in Latifs current story [redacted] the district court implicitly concluded they were different men on the basis of exculpatory statements Latif made [redacted] Latif makes much of the fact that Al-Alawi is a different name from Ba’alawi, not just a variant spelling, and at least seven detainees reported their recruiter’s name as Ba’alawi or some variant thereof. But such a minor phonetic mistake could easily result from a translation or transcription error.18 It does not negate altogether the probative value of this link between Latifs current story and a known recruiter [redacted] The district court implied Latifs benefactor was a different person from the known jihadi recruiter, without ever finding that to be so.

Even if the district court had made a clear finding in Latifs favor about Ibrahim’s identity, we could not affirm it on this record. Since the probability of one asserted fact is conditioned upon the likelihood that related facts are true, we cannot uphold the district court’s evaluation of a particular piece of evidence that is susceptible to more than one interpretation when the court has ignored related evidence.

On remand, the district court has an opportunity to evaluate all the evidence as a whole. In the event of another appeal following that evaluation, we would have to decide whether, in light of all the evidence, we are left with “the definite and firm conviction that a mistake has been committed.” Almerfedi, 654 F.3d at 9. In its current posture, this case does not require us to answer that difficult question.19

*764V

Although the district court committed the same errors here as in Alr-Adahi, the evidence before us presents a closer question than we faced in that case and our subsequent reversals. Cf. Almerfedi 654 F.3d 1; Uthman, 637 F.3d at 400. And the Government says it has discovered new evidence [redacted] that neither the district court nor our court has had occasion to consider.

As the dissenters warned and as the amount of ink spilled in this single case attests, Boumediene’s airy suppositions have caused great difficulty for the Executive and the courts. See 553 U.S. at 824-26, 128 S.Ct. 2229 (Roberts, C.J., dissenting); id. at 827-28, 128 S.Ct. 2229 (Scalia, J., dissenting). Luckily, this is a shrinking category of cases. The ranks of Guantanamo detainees will not be replenished. Boumediene fundamentally altered the calculus of war, guaranteeing that the benefit of intelligence that might be gained — even from high-value detainees — is outweighed by the systemic cost of defending detention decisions. Id. at 828, 128 S.Ct. 2229 (Scalia, J., dissenting). While the court in Boumediene expressed sensitivity to such concerns, it did not find them “dispositive.” Id. at 769, 128 S.Ct. 2229. Boumediene’s logic is compelling: take no prisoners. Point taken.

In light of the district court’s expertise as a fact finder and judge of credibility, I am reluctant to reach the merits before the district court has had an opportunity to apply the controlling precedent. But see Concurring Op. at 769 (“[F]urther fact-finding will be a waste of time and judicial resources.”). We therefore vacate and remand the district court’s grant of habeas for further proceedings. On remand the district court must consider the evidence as a whole, bearing in mind that even details insufficiently probative by themselves may tip the balance of probability, that false exculpatory statements may be evidence of guilt, and that in the absence of other clear evidence a detainee’s self-serving account must be credible — not just plausible — to overcome presumptively reliable government evidence.

So ordered.

KAREN LECRAFT HENDERSON, Circuit Judge, concurring in the judgment:

Although I agree with Judge Brown’s analysis and therefore concur in the judgment of remand, I write separately to respond to the dissent and to explain that, in my view, the better course would be to simply reverse the district court’s grant of habeas corpus relief to the detainee Adnan Farhan Abd Al Latif. The dissent attacks Judge Brown’s majority opinion on three grounds. The first two grounds are related: the dissent claims that there is no clear error in the district court’s opinion, Dissenting Op. at 770-71, 779-90 and that we have arrived at the contrary conclusion — finding clear error — only by “undertaking] a wholesale revision of the district court’s careful fact findings,” and “suggesting] [our] own story,” Dissenting Op. at 770, 784; see id. at 784-88. As discussed below, however, the dissent misunderstands the clear error standard of review and its application to this case. The dissent also claims that our use of the *765presumption of regularity “moves the goal posts” and “calls the game in the government’s favor.” Dissenting Op. at 770, 779. As also set forth below, however, the dissent’s high-pitched rhetoric not only ignores the safeguards under which we have already endorsed — albeit not explicitly— the presumption of regularity but also fails to understand how the presumption of regularity in fact aids the reliability inquiry of hearsay evidence. Finally, I believe remand for further factfinding will be a pointless exercise. Assuming he decides to testify, Latif cannot persuasively counter the presumption of regularity. Nor can he overcome the long odds against his exculpatory narrative by testifying, as his declaration already tells his story and any embroidery thereof will only work against him. Accordingly, I concur in the remand judgment only.

This appeal hinges on one question: did the district court correctly find the government’s key piece of evidence unreliable? See Abdah v. Obama (Latif), No. 04-1254, 2010 WL 3270761, at *9, slip op. at 25 (D.D.C. July 21, 2010). “The question whether evidence is sufficiently reliable to credit is one we review for clear error,” Al Alwi v. Obama, 653 F.3d 11, 19 (D.C.Cir.2011), and ordinarily this standard of review creates little controversy.

The clear error standard requires us to reverse a factual finding if “ ‘on the entire evidence’ ” we are “ ‘left with the definite and firm conviction that a mistake has been committed.’” Anderson v. City of Bessemer, 470 U.S. 564, 573, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985) (quoting United States v. U.S. Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 92 L.Ed. 746 (1948)). The dissent first claims that we cannot legitimately find clear error here, relying on our precedent that “[w]here there are two permissible views of the evidence, the factfinder’s choice between them cannot be clearly erroneous,” Awad v. Obama, 608 F.3d 1, 7 (D.C.Cir.2010) (internal quotations omitted), cert. denied, — U.S. -, 131 S.Ct. 1814, 179 L.Ed.2d 773 (2011), and that “[t]he task of resolving discrepancies among the various accounts offered into evidence is quintessentially a matter ... for the district court sitting as the fact-finder,” Al-Madhwani v. Obama, 642 F.3d 1071, 1076 (D.C.Cir.2011) (internal quotations omitted). See Dissenting Op. at 779-80, 784. But the dissent apparently forgets that the quoted passages describe only the starting point for clear error review. Granted, the district court has wide latitude to resolve factual disputes — but only within certain bounds. We must assure ourselves that the district court’s finding is “permissible” or “plausible in light of the record viewed in its entirety.” Anderson, 470 U.S. at 574, 105 S.Ct. 1504. In both Awad and Alr-Madhwani, we examined the evidentiary bases for the district court’s factual findings and, finding them within the range of “permissible” inferences to be drawn from the evidence, concluded that the district court had not clearly erred. See Awad, 608 F.3d at 6-9; Al-Madhwani, 642 F.3d at 1076. But in both Awad and Alr-Madhwani, unlike here, the district court’s permissible inferences were based on the record in its entirety — not on the view that one side’s evidence, standing in isolation, is plausible.

The dissent seems to suggest that if Latif s story “on its own terms[] is not ‘intrinsic[ally] implausible,’ ” then we cannot review the district court’s evaluation of the government’s key piece of evidence or other pieces of evidence. Dissenting Op. at 783, 784-85. It is not enough, however, for the district court to base its factual findings on some evidence in the record. The clear error standard authorizes us to *766reverse a finding, not unless, but “ ‘although there is evidence to support it.’ ” Anderson, 470 U.S. at 573, 105 S.Ct. 1504 (quoting U.S. Gypsum Co., 333 U.S. at 395, 68 S.Ct. 525) (emphasis added); see also Easley v. Cromartie, 532 U.S. 234, 257, 121 S.Ct. 1452, 149 L.Ed.2d 430 (2001) (finding clear error even where “record contains a modicum of evidence offering support for the District Court’s conclusion”). Where the record contains conflicting evidence, then, the clear error standard requires us, as the reviewing court, to assess the comparative weight of the evidence both for and against the district court’s finding. It may be that the evidence relied upon by the district court is insufficiently probative to sustain its finding. See, e.g., Easley, 532 U.S. at 247, 250, 257, 121 S.Ct. 1452 (clear error where statistical evidence “too small to carry significant evidentiary weight,” testimony did not provide “more than minimal support” and other evidence did not “significantly strengthen” district court’s finding). Or the evidence may be outweighed by other, more persuasive evidence. See, e.g., Anderson, 470 U.S. at 575, 105 S.Ct. 1504 (credibility finding clearly erroneous if “[documents or objective evidence ... contradict the witness’ story”); U.S. Gypsum Co., 333 U.S. at 396, 68 S.Ct. 525 (clear error “[w]here ... testimony is in conflict with contemporaneous documents”). The dissent is simply wrong to equate Judge Brown’s careful and complete review of the record evidence — which finds Latifs version both minimally probative, Majority Op. at 761, and decisively outweighed by the government’s evidence, id. at 755-56 — with a “wholesale revision of the district court’s careful fact findings,” Dissenting Op. at 770.

With the clear error framework in mind, there is no difficulty in concluding that the district court clearly erred in failing to credit [redacted]1 See Latif, slip op. at 6-7, 25. As Judge Brown demonstrates, the district court gave insufficient probative weight to the evidence supporting the reliability [redacted] and to the presumption of regularity that we accord a government record, see Majority Op. at 748-56. At the same time, the district court gave undue emphasis both to [redacted] and to Latifs “plausible” alternative explanation for his travels, Latif, 2010 WL 3270761, at *9, slip op. at 26. The second error is especially glaring not only in light of the district court’s failure to make any finding regarding Latifs credibility, see Al-Adahi v. Obama, 613 F.3d 1102, 1110 (D.C.Cir.2010) (by “sp[eaking] only of a possible alternative explanation” for detainee’s actions and failing to “make any finding about whether this alternative was more likely than the government’s explanation,” district court failed to make any “comparative judgment about the evidence [that] is at the heart of the preponderance standard of proof’ (internal quotations omitted)), cert. denied, — U.S. -, 131 S.Ct. 1001, 178 L.Ed.2d 855 (2011), but also in light of the inconsistencies between Latifs alternative explanation — as set forth in his declaration submitted to the district court — and his earlier statements [redacted] see Majority Op. at 760-61.2 After “considering] all of the evidence taken as a whole,” Awad, 608 F.3d at 7,1, like Judge Brown, cannot help but conclude that the district court’s finding regarding the unreliability of [redacted] coupled with its finding regarding the mere plausibility of Latifs story is neither “permissible” nor “plausible in light of the *767record viewed in its entirety,” Anderson, 470 U.S. at 574, 105 S.Ct. 1504.

II.

The dissent also asserts that application of the presumption of regularity to [redacted] “disturbs” the “careful and conscious balance of the important interests at stake” we have struck in past detainee decisions for admitting and assessing the reliability of hearsay evidence. Dissenting Op. at 775. Judge Brown thoroughly disposes of the assertion — laying out in detail that, while we have not heretofore enunciated the presumption of regularity, we have all but done so. See Majority Op. at 752-56. And we most assuredly are not “discarding] the unanimous, hard-earned wisdom” of district courts that have assessed hearsay evidence in detainee cases. Dissenting Op. at 776. To the contrary, sound evidentiary considerations warrant incorporating the presumption of regularity — in the careful manner we expressly do today — into the district court’s overall reliability assessment of these records as we routinely do with others, including the point that the facts supporting the presumption of regularity have significant probative force in their own right, as discussed below.

Moreover, our holding does nothing to disturb the existing framework for hearsay evidence. All hearsay evidence “must be accorded weight only in proportion to its reliability.” Barhoumi v. Obama, 609 F.3d 416, 427 (D.C.Cir.2010). The district court assesses reliability in the first instance, see Parhat v. Gates, 532 F.3d 834, 847-48 (D.C.Cir.2008), and in so doing must consider whatever “indicia of reliability” the hearsay evidence manifests as well as any “ ‘additional information’ ” bearing on the question of reliability. Bensayah v. Obama, 610 F.3d 718, 725-26 (D.C.Cir.2010) (quoting Parhat, 532 F.3d at 849).3 The district court considers a wide range of factors — recognizing that any one of several “hearsay dangers” might render the hearsay unreliable, see Williamson v. United States, 512 U.S. 594, 598, 114 S.Ct. 2431, 129 L.Ed.2d 476 (1994) (“The declarant might be lying; he might have misperceived the events which he relates; he might have faulty memory; his words might be misunderstood or taken out of context by the listener.”). Information “relayed through an interrogator’s account” presents an additional “level of technical hearsay because the interrogator is a third party unavailable for cross examination.” Al-Bihani v. Obama, 590 F.3d 866, 879 (D.C.Cir.2010), cert. denied, — U.S. -, 131 S.Ct. 1814, 179 L.Ed.2d 794 (2011). The presumption of regularity does not come into play with respect to many aspects of hearsay, however; for example, it does not vouch for assertions made about a detainee by a third party nor does it answer the reliability inquiry if the detainee claims he was coerced in making admissions. Rather, the presumption touches on only one dimension of reliability: “it presumes the government official accurately identified the source and accurately summarized his statement, but it implies nothing about the truth of the underlying non-government source’s state*768ment.” Majority Op. at 750. Thus it addresses only the question whether the “interrogator’s account” Al-Bihani, 590 F.Bd at 879, faithfully records the underlying statement. See, e.g., United States v. Smith, 521 F.2d 957, 964-65 (D.C.Cir.1975) (“it is presumed that [the police officer] accurately transcribed and reported [the witness’s] story” but “complaining witnesses] description of the crime, recorded by the police officer in his report, ... does not deserve the presumption of regularity”).

The Federal Rules of Evidence, which carve out exceptions to the general rule against hearsay on the ground that “some kinds of out-of-court statements are less subject to ... hearsay dangers,” Williamson, 512 U.S. at 598, 114 S.Ct. 2431, make certain public records admissible, using “the assumption that a public official will perform his duty properly” as well as “the reliability factors underlying records of regularly conducted activities generally.” Fed.R.Evid. 803(8) advisory committee’s notes (1972 Proposed Rules). Granted, in detainee habeas cases, the Rules do not decide the admissibility of hearsay evidence. Barhoumi, 609 F.3d at 422 (rejecting as “counter to this court’s [precedent]” the claim of error in admission of hearsay evidence “absent a demonstration by the government that they fall within an established hearsay exception in the Federal Rules of Evidence”). But because the presumption of regularity is based on much the same rationale as the public records exception, see United States v. Chem. Found., 272 U.S. 1, 15, 47 S.Ct. 1, 71 L.Ed. 131 (1926) (presumption applies because “courts presume that [public officers] have properly discharged their official duties.”); cf. Legille v. Dann, 544 F.2d 1, 7 n. 39 (D.C.Cir.1976) (presumption of due delivery of the mail and presumption of regularity in government agency’s handling thereof “have a common origin in regularity of action”), the facts supporting the presumption of regularity carry significant probative force in their own right.4 See Legille, 544 F.2d at 9 (“The facts giving rise to the presumption [of procedural regularity] would also have evidentiary force, and as evidence would command the respect normally accorded proof of any fact.”); Webster v. Estelle, 505 F.2d 926, 930 (5th Cir.1974) (“The same special reliability that warrants relaxing the hearsay rule as to [public records] also warrants according them great evidentiary weight.”), cert. denied, 421 U.S. 918, 95 S.Ct. 1581, 43 L.Ed.2d 785 (1975); Stone v. Stone, 136 F.2d 761, 763 (D.C.Cir.1943) (“[T]he basic fact that public officials usually do their duty ... has ... that quality and quantity of probative value to which it is entitled, entirely apart from any presumption; just as is true of any other fact which is based on common experience.”); Alsabri v. Obama, 764 F.Supp.2d 60, 68 (D.D.C.2011) (“The fact that [redacted] were prepared by government agents in the course of their normal intelligence gathering duties provides a degree of support for their reliability.”). The presumption of regularity thus embodies a common-sense judgment about the general reliability of hearsay evidence memorialized in a government record. And the district court’s failure to apply the presumption of regularity is an error going to the heart of the “careful and fine-grained approach to the assessment of reliability,” Dissenting Op. at 776, it is required to undertake.

Nor does the requirement that a challenger offer “clear or specific evidence” to defeat the presumption of regularity, *769Riggs Nat’l Corp. v. Comm’r, 295 F.3d 16, 21 (D.C.Cir.2002), somehow short-circuit the district court’s reliability analysis, as the dissent suggests. Dissenting Op. at 774-75. It is well established that clear error can occur if a district court fails to credit otherwise reliable evidence on the basis of insignificant gaps therein. See, e.g., Almerfedi v. Obama, 654 F.3d 1, 7 (D.C.Cir.2011) (“district court clearly erred in regarding [hearsay evidence] as unreliable” because of “inconsequential” “discrepancy in dates”). Requiring a challenger to produce “clear or specific evidence” — that is, evidence with real probative force — to defeat the presumption of regularity prevents a district court from relying on minor discrepancies to reject a government record. At the same time, it discourages the kind of fly-specking [redacted]5

III.

Based on the considerations outlined above — as well as Judge Brown’s comprehensive opinion — I believe the district court clearly erred in faffing to credit [redacted] Unlike my colleague, however, I also believe remanding the case for further factfinding will be a waste of time and judicial resources. Judge Brown believes remand' — -with the possibility that Latif might choose to testify — is necessary to allow the district court to correctly weigh Latifs credibility. See Majority Op. at 758-59. While I agree that the district court erred in failing to assess Latifs credibility, Majority Op. at 756-59 — for “[a]t no point did the court make any finding about whether [Latifs narrative] was more likely than the government’s explanation,” Al-Adahi v. Obama, 613 F.3d at 1110 — I also believe remand is unnecessary because “the record permits only one resolution of the factual issue,” Pullman-Standard v. Swint, 456 U.S. 273, 292, 102 S.Ct. 1781, 72 L.Ed.2d 66 (1982); see Easley, 532 U.S. at 257, 121 S.Ct. 1452 (finding clear error and reversing because “we do not believe that providing appellees a further opportunity to make their ... arguments in the District Court could change th[e] result”).

The apparent premise behind Judge Brown’s argument for remand is that Latif might offer testimony so compelling that it would shake our confidence [redacted] and overcome any doubt about Latifs credibility. But what testimony could possibly accomplish so much? If Latif were to repeat on the stand the same unpersuasive assertions he made in his declaration— assertions that are inconsistent with his earlier statements [redacted] the district court would have no choice but to disbelieve him. “Credibility involves more than demeanor” and instead “apprehends the over-all evaluation of testimony in light of its rationality or internal consistency and the manner in which it hangs together with other evidence.” United States v. McCoy, 242 F.3d 399, 408 n. 15 (D.C.Cir.) (internal quotations omitted), cert. denied, 534 U.S. 872, 122 S.Ct. 166, 151 L.Ed.2d 114 (2001); see also Anderson, 470 U.S. at 575, 105 S.Ct. 1504 (“[F]actors other than demeanor and inflection go into the decision whether or not to believe a witness. Documents or objective evidence may contradict the witness’ story; or the story itself may be so internally inconsistent or implausible on its face that a reasonable factfinder would not credit it.”). If, on the other hand, Latif were to change his story once again on remand, the very fact that he “made inconsistent statements ... would tend to undermine his credibility.” United States v. Stover, 329 F.3d 859, 867-68 (D.C.Cir.2003), cert. denied, 541 U.S. 1018, 124 S.Ct. 2088, 158 L.Ed.2d 635 *770(2004). Latifs credibility would suffer even if he largely repeated the story in his declaration but also decided to embellish it with additional details — [redacted]—because “[p]rior statements that omit details covered at trial are inconsistent if it would have been ‘natural’ for the witness to include them in the earlier statement.” United States v. Stock, 948 F.2d 1299, 1301 (D.C.Cir.1991). In short, Latif could only dig himself deeper into a hole on remand.6 Because the record can reasonably be viewed in only one way — that is, against him — I would not remand simply to give Latif a shovel but would instead conclude the litigation with the only result the evidence allows: that the government has indeed “shown that Latif is part of A1 Qaeda or the Taliban.” Latif, 2010 WL 3270761, at *1, slip op. at 3.

. Specifically, the Supreme Court held that Guantanamo detainees must have "the means to supplement the record on review," Boumediene, 553 U.S. at 786, 128 S.Ct. 2229, and that the court conducting habeas proceedings must have authority (1) “to assess the sufficiency of the Government’s evidence against the detainee,” id.-, (2) "to admit and consider relevant exculpatory evidence,” id.-, (3) "to make a determination in light of the relevant law and facts,” id. at 787, 128 S.Ct. 2229; and (4) "to formulate and issue appropriate orders for relief, including, if necessary, an order directing the prisoner’s release,” id.

. Even the particular presumption at issue in this case — the presumption that an official government record was accurately produced — applies in ordinary criminal habeas cases. See Hobbs v. Blackburn, 752 F.2d 1079, 1081 (5th Cir.1985) ("Official records, such as this signed [state court guilty plea], are entitled to a presumption of regularity and are accorded great evidentiary weight” in a federal habeas proceeding.); see also Walker v. Maggio, 738 F.2d 714, 717 (5th Cir.1984) ("minute entry of the [state] court” is entitled to a “presumption of regularity”); Thompson v. Estelle, 642 F.2d 996, 998 (5th Cir.1981) ("The district court could properly rely on the regularity of the state court’s documents in preference to Thompson's own self-serving testimony.”); Webster v. Estelle, 505 F.2d 926, 929-30 (5th Cir.1974) (indictment and docket sheet are entitled to presumption of regularity). The same presumption applies to official government records in a probation revocation proceeding, a circumstance like habeas in which liberty is on the line. See United States v. Thomas, 934 F.2d 840, 846 (7th Cir.1991) (probation officer’s report); United States v. Verbeke, 853 F.2d 537, 539 (7th Cir.1988) (treatment center’s report); United States v. McCollum, 677 F.2d 1024, 1026 (4th Cir.1982) (same).

. [Redacted] the Government had no expectation that its intelligence would be used in litigation. Instead, the Government was seeking accurate, actionable intelligence to protect the country from imminent attack. The Government had the strongest incentive to produce accurate reports and no incentive to frame innocent bystanders as Taliban operatives.

. The Government's argument for a presumption of regularity is unambiguous. Observing that "[i]t is well established that there is a strong 'presumption of regularity’ for actions of government officials taken in the course of their official duties," Appellants’ Br. 30 (quoting United States v. Chemical Found., Inc., 272 U.S. 1, 14-15, 47 S.Ct. 1, 71 L.Ed. 131 (1926), the Government argued that the expert descriptions [redacted] “should have been considered in light of the general presumption that government officials are properly carrying out their duties." Id. Developing this argument by analogy, the Government urged that "the factors supporting [the] accuracy” of [redacted] are "even stronger than in the immigration context” in which a border agent "cannot be presumed to be ... other than an accurate recorder” of the alien's statement. Id. at 30-31, 47 S.Ct. 1 (quoting Espinoza v. INS, 45 F.3d 308, 311 (9th Cir.1995)). The dissent's claim that our holding goes "well beyond what the government actually argues in its briefs" is unfounded. Dissenting Op. at 774.

. We need not decide precisely how much more the detainee must show to overcome the presumption of regularity. Depending on the circumstances, courts have required litigants to meet standards ranging from "clear and specific evidence,” Riggs Nat’l, 295 F.3d at 21 (tax), to "clear and convincing evidence,” Riggins v. Norris, 238 F.3d 954, 955 (8th Cir.2001) (habeas); see also United States v. Armstrong, 517 U.S. 456, 464, 116 S.Ct. 1480, 134 L.Ed.2d 687 (1996) (“clear evidence”) (selective prosecution); Dep’t of Labor v. Triplett, 494 U.S. 715, 723, 110 S.Ct. 1428, 108 L.Ed.2d 701 (1990) ("[A]necdotal evidence will not overcome the presumption of regularity”) (effective assistance of counsel); cf. Sussman, 494 F.3d at 1117 (noting a "less stringent standard” applies in at least some FOIA cases (quoting Nat'l Archives & Records Admin, v. Favish, 541 U.S. 157, 174, 124 S.Ct. 1570, 158 L.Ed.2d 319 (2004))). Even if we assume a detainee may overcome the presumption by a mere preponderance of the evidence, Latif cannot meet that standard.

.[Redacted] As we clarified in Al-Bihani, however, an interrogation report involves just one level of hearsay — that of the interrogator. 590 F.3d at 879. Like the diary translated in Barhoumi, and unlike the anonymous hearsay in Parhat, [redacted] are "the underlying reporting on which the government's assertions are founded.” Barhoumi, 609 F.3d at 428. The act of translation "does not affect [redacted] status.” Al-Bihani, 590 F.3d at 879; see Barhoumi, 609 F.3d at 430-31. And, as Parhat and Al-Bihani demonstrate, courts are capable of determining whether official government records that contain hearsay merit the presumption of regularity.

. [Redacted]

. On appeal, Latif retorts that the Government did not put on any witnesses either. Appellee's Br. 51. This misses the point. [Redacted] Latif, by contrast, has every incentive to lie about the purpose of his visit. His failure to testify and subject himself to cross examination therefore undermines his credibility.

. [Redacted]

. Although Latif’s more recent declaration in the district court leaves out some of these details, he does not deny taking this route. Indeed, Latif cites the consistency of his Guantanamo interrogations as evidence that his current story is true. Appellee’s Br. 18-22. Latif’s recent declaration confirms he took a bus to Quetta and a taxi from Quetta to Afghanistan, and then stayed in Kabul before returning to Pakistan.

.The district court did not, as the dissent suggests, "treat[ ] [this] evidence as more akin to traveling along 1-95 than a lonely country road.” Dissenting Op. at 786. The court did not consider it at all.

. We do not ''find []" that this evidence "do[es] in fact implicate” Latif, as the dissent accuses us of doing. Dissenting Op. at 771. Rather, we hold the district court's findings suspect in that the court “failed to take into account” related evidence when it made those findings. Al-Adahi, 613 F.3d at 1108.

. [Redacted]

. Indeed, as the district court acknowledged, the recruiter is identified as Alawi in another detainee's interrogation report. The district court dismissed this evidence, observing that in another case, the district court had discredited this detainee’s statement about an unrelated detail — the timing of another detainee's arrival at a guesthouse — because it conflicted with other detainees' statements. Latif, 2010 U.S. Dist. LEXIS 83596, at *26 n. 10, slip op. at 19 n. 10 (citing Abdah v. Obama, 717 F.Supp.2d 21, 35 (D.D.C.2010)).

. Judge Henderson would reverse the district court’s grant of habeas corpus outright. In her view, “remand is unnecessary because *764'the record permits only one resolution of the factual issue.' ” Concurring Op. at 769 (quoting Pullman-Slandard v. Swint, 456 U.S. 273, 292, 102 S.Ct. 1781, 72 L.Ed.2d 66 (1982)). Because of the legal errors we have both identified, I find it unnecessary to decide that question. Remand is warranted not only when "further fact-finding by the district court is necessary,” but also when it "would be helpful.” Al Alwi, 653 F.3d at 16. This is such a case.

. As Judge Brown explains, [redacted]

. Judge Brown cites a variety of examples— for instance, [redacted]

. Parhat also requires that hearsay evidence “be presented in a form, or with sufficient additional information, that permits the ... court to assess its reliability.” 532 F.3d at 849. As Barhoumi notes, however, the quoted passage has more to do with the form than with the substance of hearsay evidence: "the problem with the intelligence reports at issue in Parhat was that they failed to provide ‘any of the underlying reporting upon which the documents’ bottom-line assertions are founded,’ thus inhibiting our ability to evaluate the reliability of those assertions.” 609 F.3d at 428 (quoting Parhat, 532 F.3d at 846-47). [redacted]

. While the facts surrounding hearsay evidence may not always justify applying the presumption of regularity, it is properly applied here because [redacted]

. [redacted]

. Indeed, even Latifs continued failure to testify would likely work against him. Majority Op. at 759; see Mitchell v. United States, 526 U.S. 314, 328, 119 S.Ct. 1307, 143 L.Ed.2d 424 (1999) (“ ‘[T]he Fifth Amendment does not forbid adverse inferences against parties to civil actions when they refuse to testify in response to probative evidence offered against them.’ ” (quoting Baxter v. Palmigiano, 425 U.S. 308, 318, 96 S.Ct. 1551, 47 L.Ed.2d 810 (1976))).

Related Cases