OPINION OF THE COURT
Lek Berishaj, an ethnic Albanian from Montenegro, petitions for review of a decision of the Board of Immigration Appeals (BIA), which affirmed without opinion the decision of an immigration judge (IJ) denying him asylum and relief under the Convention Against Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment (CAT). Under our case-law, see Dia v. Ashcroft, 353 F.3d 228 (3d Cir.2003) (en banc), it is the IJ’s decision that we review, no mean task here because the IJ’s opinion is cursory, thinly reasoned, and discusses the case without any reference to the governing legal standards. Nonetheless, we understand the IJ to have concluded that Berishaj’s testimony regarding past persecution was not credible; *317that, even taking Berishaj’s testimony as true, country conditions in the Federal Republic of Yugoslavia (which embraced Montenegro at the time of the IJ’s decision) had changed such that Berishaj could no longer have a well-founded fear of future persecution; and that Berishaj’s CAT claim failed because there was no objective evidence that a return to Montenegro would expose him to torture.
Reviewing the IJ’s decision under the “substantial evidence” standard, see id. at 247-50, we conclude that the IJ’s rejection of Berishaj’s asylum claim cannot stand. First, the IJ’s adverse credibility determination has no basis in the record. Second, the IJ misapplied the law in concluding that changed conditions in Montenegro have obviated any persecution claim that Berishaj might once have had. In such a posture, the burden of showing changed country conditions is on the government, see 8 C.F.R. 208.13(b)(l)(ii), and we hold that the government must rebut the alien’s well founded fear of future persecution with specific evidence, which it did not produce. We will therefore grant the petition for review of the decision insofar as it rejected Berishaj’s asylum claim, and his related claim for withholding of removal. We leave it to the Agency to make a proper determination in the first instance of the merits of those claims. With respect to Berishaj’s CAT claim, the IJ’s decision passes muster (though barely), and we will deny the petition for review of the IJ’s CAT decision.
As we will explain in greater detail, we .think this case to be a particularly apt example of a disturbing trend we often encounter in petitions for review of the BIA. In many cases in which country conditions are at issue, the administrative records are grossly out-of-date, requiring us to engage in the rather artificial exercise of ruling on situations that existed several years in the past, but do not exist today. Here, we work from an administrative record in which the most recent country conditions report is over four years out-of-date. While SEC v. Chenery Corp., 318 U.S. 80, 63 S.Ct. 454, 87 L.Ed. 626 (1943), and the constraints of process-based review of administrative decision making prevent us from supplementing a grossly out-of-date administrative record, they do not command blindness to the emerging pattern of stale records. Considering the rapid, frequent political changes in countries from which asylum and CAT applicants usually come, and the potentially dire consequences of sending such an applicant back to his country of origin to face possible persecution or torture on the basis of such a stale report, we call on Congress, the Department of Justice, the Department of Homeland Security, and the BIA to improve the structure and operation of the system, so that all may have the confidence that the ultimate disposition of a removal case bears a meaningful connection to the merits of the petitioner’s elaim(s) in light of contemporary world affairs.
I. The Administrative Record and the IJ’s Decision
As will become clear, the IJ’s credibility determination rested on his rejection of a fairly narrow slice of Berishaj’s testimony. But we will discuss Berishaj’s testimony in full, because his claims depend on aspects of it beyond the specific testimony on which the IJ based his adverse credibility determination. CAT claims and questions of changed country conditions are, for the most part, evaluated with reference to documentary evidence of contemporary country conditions; questions of corroboration are evaluated with reference to documentary evidence of past conditions. We will therefore address the documentary materi*318als in the record with a focus on both past and contemporary events. To set the context for Berishaj’s testimony, we set forth in the margin a capsule chronology of events in the Balkans from 1991 to 20011
A. Berishaj’s Testimony and Corroborating Affidavit
Berishaj is an ethnic Albanian who spent his youth in Montenegro, at the time part of Yugoslavia.2 In the summer of 1991, he went to Kosovo, a neighboring province of Yugoslavia, to attend a university that conducted classes in his native Albanian tongue. (At that time, no university in Montenegro conducted classes in Albanian.) Serb forces had taken control in Kosovo in 1990, and had officially closed the university, but it continued to function underground, with classes held in private homes in Pristina, a major city in Kosovo. Berishaj’s uncle, Palok, with whom Beri-shaj lived at the time, was one of the leading organizers of the illegal university, and he recruited Berishaj to find private homes in which to hold classes. Because of his activities, Palok was arrested in 1991, and again in 1994, when he was detained and beaten for several days. Berishaj was arrested in Pristina in the spring of 1992 for his assistance to the illegal university; he was beaten with a rifle butt and detained overnight. Not trusting the Serbian doctors at the hospital, Berishaj was treated by an Albanian doctor practicing illegally, and returned to Montenegro a few weeks later.
Shortly after returning to Montenegro, Berishaj was inducted into the army; he was sent to serve in Serbia, where he spent eleven months. He was easily recognizable as an ethnic Albanian, among a predominantly Serbian army-“Berishaj” is a well-known Albanian name; indeed, one Sali Berishaj was the former president of Albania. In the army, Berishaj served as a tank gunman following a three-month training period in which he learned to operate the tank gun from instructions in Serbian. In the tank crew of three or four, Berishaj was the lowest in rank, taking orders from Serbs in charge of the tank. Berishaj did not, in these eleven months, go to war in Bosnia. Berishaj attributed this to the Serbian officers not trusting Albanians enough to send them to war. Berishaj’s duties mostly consisted of *319cleaning the tank gun and guarding the tank. He was beaten at the direction of Serbian officers for singing songs in Albanian, and he stopped speaking Albanian publicly, relying on the Serbian he learned while in the army. After completing eleven months of military service, Berishaj was discharged. He returned to Montenegro, then to Kosovo briefly to take university examinations, and then back to his parents’ home in Montenegro.
In December 1993, four policemen - apparently military police-came to his parents’ house at midnight and took him to fight in Bosnia. He served again as a gunman, and was ordered to destroy buildings, houses, and shoot at the army and at Muslim civilians in Bosnia. Berishaj explained that he had no choice but to shoot civilians: “I would either, you know, shoot or [the Serbians] would kill me.” Ethnic Albanians in the army were not trusted to shoot without being under Serbian control. For example, Berishaj explained, “When we were in the tank, you know, using the gun, we would have somebody behind us [a Serbian] with an automatic gun.... Their function was that if somebody does not obey the order to shoot with a gun, they would kill him.” Berishaj spent two months in the army this time, and escaped during the night in February 1994. He returned to his parents’ home in Montenegro.
Fearing that he would be arrested and returned to the army, Berishaj crossed illegally from Montenegro into. Albania, where he spent the next fourteen months in hiding at his cousin’s home. He sought, but was unable to obtain, legal status in Albania; as a result, the Albanian authorities learned of him. His cousin, fearing the Albanian police, convinced Berishaj to return to Montenegro in April 1995. Upon returning to his parents’ home in Montenegro, Berishaj learned that he was wanted by the police, and he went to reside with his sister, who lived in another village several miles from his parents’ home. The five months with his sister were spent mostly indoors, as were the following months, which he spent with an uncle in yet another village.
Berishaj ultimately was located by the Montenegrin police in September 1996, and detained for two days. They asked him why he deserted from the army, and why he did' not finish his studies at the official university (i.e., the Serbian-run university); it was clear that the police knew he had participated in the illegal university. Berishaj was released from custody apparently when an uncle fabricated a story about Berishaj needing to visit an ill family member and posted bail for him. Berishaj returned illegally to Albania, where he resided until February 1997, when he was smuggled to Belgrade, and from there to France, then Brazil, then the United States.
Since being in the United States, Beri-shaj has had limited contact with family members in Montenegro, fearing that his family would be coerced by tjie police into revealing his whereabouts. In January 2001, Berishaj spoke to his father for the first time since leaving Montenegro nearly four years before. His father explained, in Berishaj’s words “that once I left, the [Montenegrin] police came three times and checked the house inside out looking for me after I had escaped. At this time, they asked him “Where is he? Where can we find him?’ And his response was he didn’t know.... After I left, my father was telling me that many incidents they came and checked the house inside out three times, and at one point were also guarding the house overnight to see if I would come home.” The police stopped searching for Berishaj when his father told them that Berishaj had left permanently. Berishaj *320also learned from his father that his brother was serving a five-year sentence for helping the Kosovar resistance during the war. Berishaj’s sister, a naturalized American citizen, confirmed in an affidavit made in late 1997 that she had received similar accounts from their father, brother, and sister regarding police activity at their parents’ home.
B. Documentary Evidence in the Administrative Record
1. Contemporary Country Conditions
Much of the record addresses contemporary treatment of ethnic Albanians in Kosovo or elsewhere in Serbia; as this is not especially relevant to the situation in Montenegro, we will concentrate only on documentary evidence addressing Montenegro. We begin with the State Department’s 1999 Country Reports on Human Rights Practices: Serbia-Montenegro, issued in February 2000 (the “1999 Country Report”); this is the latest country report available in the administrative record. While on the one hand the BIA may not “ ‘hide behind the State Department’s letterhead” ’ and place full and uncritical reliance on a country report, Ezeagwuna v. Ashcroft, 325 F.3d 396 (3d Cir.2003) (quoting Li Wu Lin v. INS, 238 F.3d 239, 246 (3d Cir.2001)), neither is it permissible for the IJ and BIA not to address the relevant country report in some detail. The first line of the 1999 Country Report reads: “Serbia-Montenegro is dominated by Slobodan Milosevic,” though it goes on to note that Milosevic’s primary influence is over Serbia proper (and even there, not in Kosovo) and less over Montenegro. Nonetheless, given Milosevic’s control over the Serbian army and federal police, the 1999 Country Report establishes that a prime force in the persecution (or worse) of ethnic Albanians was still in power at the time the administrative record was compiled.3
On the other hand, the 1999 Country Report represents that Montenegro was making progress toward democracy, holding free and fair elections, and that 1999 saw even further escape from the federal control of Milosevic’s regime. The 1999 Country Report further states that while the Montenegrin government generally respected its citizens’ human rights, there were reports of extrajudicial killings by federal troops, forcible conscription, and violence and discrimination against minorities. Academic freedom is said to have been respected. In early 1999, the government began a program of devolving authority on local government officials in ethnic Albanian communities. The Yugoslav Parliament passed an amnesty for draft evaders and deserters in late 1995, and the Montenegrin Parliament passed a similar law in late 1999. According to documents in the record from Amnesty International, however, there is evidence that at least the latter law was not fully observed, as federal Yugoslav authorities and military police controlled the treatment of evaders and deserters.
The 1999 Country Report generally sounds of relative stability and democratic progress, but other parts of the record suggest that events were very fluid in 1999. For example, a series of news articles from the New York Times and reports from human rights organizations suggests that paramilitary groups associated with ethnic cleansing of ethnic Albanians in Ko-*321sovo had moved into Montenegro, perhaps at the behest of the Montenegrin police.
2. Corroborating Materials
We turn now to materials in the record that could corroborate Berishaj’s accounts from the early 1990s-specifically his time at the illegal university in Kosovo and his military service from 1992 to 1994. Two pieces are worthy of note. First, Beri-shaj’s story about the operation of the illegal university in Kosovo, and police hostility to it, is perfectly corroborated by an Amnesty International Report from 1994, which is in the administrative record. This report describes the creation in 1990 and 1991 by ethnic Albanians of “a parallel educational system using [pre-Serbian] curricula,” with “lessons ... held in private homes.” The report also describes several specific episodes of police violence against ethnic Albanians on account of this parallel educational system.
Second, and also in the administrative record before the IJ, a 1992 article from the Bronx-published English-language Albanian-American newspaper Illyria profiles a young ethnic Albanian, Adem Kras-niqi, whose experience as a forced inductee into the Serbian army closely parallels Berishaj’s. Krasniqi was one of many Albanian “tankers” (i.e., tank operators or gunmen). In an attack on Vukovar, Croatia, he was forced to move forward in the first wave; the story quotes Krasniqi as saying, “Behind us were the Serbian irregulars uniformed as soldiers. Anyone trying to desert would be shot. We had two choices. Keep firing or get shot from people behind you.” Krasniqi also describes indiscriminate shelling, and being forced to fire at innocent civilians; this is again consistent with Berishaj’s account.
C. The IJ’s Decision
The IJ’s decision (which, save for irrelevant introductory and concluding remarks, is recounted in full in the paragraphs that follow) begins with his adverse credibility determination:
The case at bar is afflicted by testimony that is incredible in nature. The Applicant’s statements as to how he was recruited and placed in a position of combat by the Serbs while at the same time adducing to an attitude of total disdain and bias toward the Applicant is just incredulous to the Court. This fact is dramatized and magnified by the Respondent’s testimony that although he was despised by the Serbs in the army he was placed in command of a tank. The testimony further developed how the Applicant learned to operate the tank by reading the instructions in it and how, albeit they were written in a language he did not understand, he was able to familiarize with the operation of the tank in just three months. As fantastic and ludicrous as that statement may appear, the Court was dazzled and astounded by the declaration that although he was in control of the tank he had a Serbian officer behind his back pointing a gun at him at all times! A better script could not have been thought about by kings of comedy like Peter Sellers or Mel Brooks.
This ridiculous testimony is not supported by one scintilla of evidence and in addition to be completely absurd it borders in an offensive and arrogant attitude toward the Court. The Applicant’s demeanor, throughout the sessions of testimony, was characterized by an arrogant disposition in thinking that he deserves what he is asking for.
The IJ next turned to a brief discussion of the then-current conditions in Montenegro:
The amount of time this case has been pending has made the Respondent’s *322claim even weaker. Historically, Montenegro was considered a satellite or puppet nation of strongman Slobodan Milosovic. Under Milosovic’s regime the Respondent’s position of opposition to service in the Serbian army may have had some validity. It was not until very recently that the Government of Montenegro has taken an independent position with regard to the treatment of ethnic Albanians in the region. Once the apparent defeat of the Milosovic administration, the Government of Montenegro has shown signs of self-determination. This change of events, contrary to the Applicant’s position benefits the Respondent and makes his return to Montenegro reasonable. Montenegro has granted an amnesty to deserters and draft dodgers. Nothing in Respondent’s arguments convinces this Court that his return to Montenegro would place him in any type of danger at the present time. In concluding as I do I have determined that the Applicant’s possibility of any future persecution is nil. Respondent’s attorney’s arguments that the new administration of elected president Vojislaw Kostunica is a mirror image of his predecessor is not persuasive nor established.
The final substantive portion of the IJ’s decision reiterates his adverse credibility determination:
The Court has stated its opinion earlier as to how skeptical the Court is about the Applicant’s claim of past persecution. Testimony that has been plagued by fantastic anecdotes and uncorroborated information is very difficult to accept even as plausible. The Applicant’s case is precisely affected by these characteristics and therefore makes it impossible for the Court to accord it any credence.
The BIA affirmed this decision without opinion.
II. Standard of Review
Because the BIA affirmed the decision of the IJ without opinion, see 8 C.F.R. § 3.1(e)(4), the decision of the IJ is the final agency determination, which we are called upon to review. See Dia, 353 F.3d 228. We have jurisdiction under 8 U.S.C. § 1252 over this timely petition for review of a final determination of the BIA.
We review the Agency’s findings of fact-such as the IJ’s credibility determinations, his findings on the CAT claim, and his findings regarding changed country conditions-under 8 U.S.C. § 1252(b)(4)(B), which provides that “administrative findings of fact are conclusive unless any reasonable adjudicator would be compelled to conclude to the contrary.” As we explained in Dia, we have “read this standard to require that the agency support its findings with substantial evidence, as articulated by the Supreme Court in INS v. Elias-Zacarias, 502 U.S. 478, 481-84, 112 S.Ct. 812, 117 L.Ed.2d 38 [ (1992) ].” 353 F.3d at 247; see also Sevoian v. Ashcroft, 290 F.3d 166, 171 (3d Cir.2002) (“[The Illegal Immigration Reform and Immigrant Responsibility Act] codifies the language the Supreme Court used in Elias-Zacarias to describe the substantial evidence standard in immigration cases.”). We concluded in Dia that
the question whether an agency determination is supported by substantial evidence is the same as the question whether a reasonable fact finder could make such a determination based upon the administrative record. If a reasonable fact finder could make a particular finding on the administrative record, then the finding is supported by substantial evidence. Conversely, if no reasonable fact finder could make that finding on *323the administrative record, the finding is not supported by substantial evidence.
353 F.3d at 249.
III. Berishaj’s Application for Asylum
A. The Asylum Claim
1. The Adverse Credibility Determination
Berishaj applied for asylum and withholding of removal based on past persecution and a well-founded fear of future persecution if he is removed to Montenegro. In Gao v. Ashcroft, 299 F.3d 266 (3d Cir.2002), we laid out the statutory frame-: work for asylum claims and the relevance of adverse credibility determinations to the asylum inquiry:
A grant of asylum under § 1158(b)(1) of the Immigration and Nationality Act (INA) allows an otherwise removable alien to stay in the United States. The Attorney General “may” grant-asylum to an alien who demonstrates that he/she is a refugee: a person unable or unwilling to return to the country of that person’s nationality or habitual residence because of past persecution or because of a well-founded fear of future persecution on account of his race, religion, nationality, membership in a particular social group, or political opinion. See INA § 208(b)(1), 8 U.S.C. § 1158(b)(1) (requiring asylum applicant conform to definition of refugee); [§ ] 101(a)(42)(A), 8 U.S.C. § 1101(a)(42)(A) (providing definition of refugee). In order to establish eligibility for asylum on the basis of past persecution, an applicant must show: “(1) an incident, or incidents, that rise to the level of persecution; (2) that is ‘on account of one of the statutorily-protected grounds; and (3) is committed by the government or forces the government is either ‘unable or unwilling’ to control.” Navas v. INS, 217 F.3d 646, 655 (9th Cir.2000).
An applicant can demonstrate that she has a well-founded fear of future persecution by showing that she has a genuine fear, and that a reasonable person in her circumstances would fear persecution if returned to her native country. Elnager v. INS, 930 F.2d 784, 786 (9th Cir.1991). Aliens have the burden of supporting their asylum claims through credible testimony. Abdille v. Ashcroft, 242 F.3d 477, 482 (3d Cir.2001). Testimony, by itself, is sufficient to meet this burden, if “credible.” 8 C.F.R. § 208.13(a), Chand v. INS, 222 F.3d 1066, 1077 (9th Cir.2000). In some cases the INS may require documentary evidence to support a claim, even from otherwise credible applicants, to meet their burden of proof. Abdulai [v. Ashcroft], 239 F.3d [542,] 554 [(3d Cir.2001) ].
.... [A]dverse credibility determinations are reviewed for substantial evidence. Balasubramanrim v. INS, 143 F.3d 157, 161 (3d Cir.1998).... Adverse credibility determinations based on speculation or conjecture, rather than on evidence in the record, are reversible. Salaam v. INS, 229 F.3d 1234, 1238 (9th Cir.2000). Generally, minor inconsistencies and minor admissions that “reveal nothing about an asylum applicant’s fear for his safety are not an adequate basis for an adverse credibility finding.” Vilorio-Lopez v. INS, 852 F.2d 1137, 1142 (9th Cir.1988). The discrepancies must involve the “heart of the asylum claim.” Ceballos-Castillo v. INS, 904 F.2d 519, 520 (9th Cir.1990).
Gao, 299 F.3d at 271-72. Furthermore, an alien who offers credible testimony regarding past persecution is presumed to have a well-founded fear of future persecution. See Abdulrahman v. Ashcroft, 330 F.3d *324587, 591-92 (3d Cir.2003) (citing 8 C.F.R. § 208.13 (“An applicant who has been found to have established such past persecution shall also be presumed to have a well-founded fear of persecution on the basis of the original claim.”)).
Here, if the IJ’s adverse credibility determination is supported by substantial evidence, Berishaj’s asylum claim would arguably fail because the IJ refused to credit significant testimony at the core of Beri-shaj’s story of past persecution-his being subjected to persecution by Serbs on account of his status as an ethnic Albanian. We set aside for the time being that the IJ failed altogether to address Berishaj’s testimony about how he was treated outside the military. That testimony-from his civilian life-arguably could independently support his asylum claim. In light of our conclusion that the IJ’s adverse credibility determination is not supported by substantial evidence, we need not address whether the IJ’s adverse credibility determination with respect to Berishaj’s account of his military service could properly be used to reject his accounts of arguable persecution as a civilian.
As noted, we conclude that the IJ’s adverse credibility determination is not supported by substantial evidence. In terms of the language we used in Gao, the IJ’s credibility determination was “based on speculation [and] conjecture, rather than on evidence in the record.” 299 F.3d at 272 (citing Salaam v. INS, 229 F.3d 1234, 1238 (9th Cir.2000)). Without further elaboration, the IJ simply rejected Berishaj’s testimony regarding his military service as “just incredulous [sic] to the Court,” as “fantastic and ludicrous,” and “astound[ing],” worthy of “kings of comedy like Peter Sellers or Mel Brooks.” The IJ’s comments are not only intemperate but singularly unhelpful. At best they amount to a finding that Berishaj’s testimony was implausible or inherently improbable when, as we will demonstrate, the testimony appears eminently reasonable. At all events, the IJ’s comments are not tethered to the record, owing what little support they have to hyperbole and appeals to popular culture-two utterly inappropriate bases for an asylum decision.
The BIA has of course held, and we have agreed, that an adverse credibility determination may properly be based on implausibility or inherent improbability. See In re S-M-J- 21 I. & N. Dec. 722, 729-30, 1997 WL 80984 (BIA 1997) (holding that an adverse credibility determination may be “appropriately based on inconsistent statements, contradictory evidence, and inherently improbable testimony ... in view of the background evidence on country conditions”); Dia, 353 F.3d at 249 (“Where an IJ bases an adverse credibility determination in part on ‘implausibility[,]’ ... such a conclusion will be properly grounded in the record only if it is made against the background of the general country conditions.” (citing Gao, 299 F.3d at 278-79; He v. Ashcroft, 328 F.3d 593, 603 (9th Cir.2003))). As these cases illustrate, however, there must be record support and specific, cogent reasons for such an adverse credibility determination.
The en banc Court in Dia emphasized exactly this point in rejecting the adverse credibility determination made by the IJ in that case:
“[W]hile we defer to the IJ on credibility questions, that deference is expressly conditioned on support in the record,” Nagi El Moraghy [v. Ashcroft], 331 F.3d [195,] 205 [ (1st Cir.2003) ], and “[d]eference is not due where findings and conclusions are based on inferences or presumptions that are not reasonably grounded in the record.” Id. at 202 (citation and internal quotation marks omitted); see also Abdulrahman, 330 *325F.3d at 597 (stating that “substantial deference” to a finding is to be “afforded ... where it is grounded in evidence in the record”). To this end, it is clear that “[ajdverse credibility determinations based on speculation or conjecture, rather than on evidence in the record, are reversible,” Gao, 299 F.3d at 272, and that an IJ must support her adverse credibility findings with “specificf,] cogent reasons.” Id. at 276; Abdulrahman, 330 F.3d at 597; see also Secaida-Rosales [v. Ashcroft], 331 F.3d [297,] 307 [ (2d Cir.2003) ] (“When an IJ rejects an applicant’s testimony, the IJ must provide ‘specific, cogent’ reasons for doing so.”); He, 328 F.3d at 595 (“[T]he IJ and BIA must offer a ‘specific, cogent reason for any stated disbelief.’ ” (quoting Hartooni v. INS, 21 F.3d 336, 342 (9th Cir.1994))).
If the IJ’s conclusion is not based on a specific, cogent reason, but, instead, is based on speculation, conjecture, or an otherwise unsupported personal opinion, we will not uphold it because it will not have been supported by such relevant evidence as a reasonable mind would find adequate. In other words, it will not have been supported by substantial evidence.
Id. at 249-50 (some alterations in original).
None of the IJ’s reasons for finding Berishaj incredible withstand scrutiny in light of the record. We address them seriatim. Most obviously wrong is the IJ’s conclusion that Berishaj’s testimony about his experience in the military was “not supported by one scintilla of evidence.” As we have already described, see supra Part I.B.2, there is at least one strikingly similar published account from another ethnic Albanian in the Serbian army-Adem Krasniqi’s story as published in Illy-ria. Irrespective of Illyria’s reliability vel non as an unbiased source of news from the Balkans, the article was in the administrative record and the IJ was obliged to address it. Moreover, the IJ is in no position to comment from his own experience on the plausibility of the cruel practices employed in one of the most heinous conflicts of the modern era. If anything, the tactic of forcing one ethnic minority to kill another is entirely consistent with multiparty ethnic warfare, and there is no substantial evidence on which to conclude otherwise. It seems eminently plausible that the Serbians would require the Albanians to be the ones to shoot the Bosnians, and that they would enforce that role in the manner described-by having a gunman behind each Albanian. To describe this as a Mel Brooks scenario seems to us bizarre.
We also cannot understand the IJ’s incredulity at Berishaj’s ability to learn enough Serbian to clean and operate a tank gun as a low-ranking soldier. Beri-shaj testified that he did not find Serbian difficult to learn, and that he had to ..stop speaking Albanian publicly-which suggests that he was effectively immersed in Serbian for several months. These are both candid and credible statements that the IJ did not address. What is more, even though the IJ’s hearing in January of 1998 (no more than a year after Berishaj arrived in the United States) was conducted with an Albanian interpreter, Berishaj repeatedly demonstrated that he was listening to the questions in English, and not waiting for the interpreter’s translation. At times, Berishaj even responded in English. English is a difficult language to learn, but Berishaj apparently has some facility for picking up languages, and this only enhances the credibility of his claim to having learned basic Serbian in a few months. In light of all this, we do not see how a reasonable factfinder could conclude that his claim that he learned enough Serbian to clean and operate the tank gun was not worthy of belief.
*326We also note the basic misstatement of the record in the IJ’s description of Beri-shaj’s testimony-Berishaj never testified that he was “placed in command” of the tank, or “in control of the tank”; he testified that he was at the bottom of the chain of command. There is no evidence contra. In sum, nothing of the IJ’s adverse credibility determination survives even basic scrutiny, and we cannot accept the IJ’s determination as supported by substantial evidence
The IJ also rejected Berishaj’s testimony on the ground that it was “plagued by ... uncorroborated information.” To the extent that the IJ meant that Berishaj’s account of his military experience could not be squared with the experiences of other ethnic Albanians in the Serbian army in the mid-1990s, we think our discussion above amply refutes any notion that Berishaj’s experience was implausible. To the extent that the IJ complained of the absence of testimonial or documentary materials in the record to support aspects of the factual account given by Berishaj, we are at a loss to comprehend the IJ-requiring corroborative evidence in this situation would run counter to our precedent, BIA precedent, and common sense.
In Abdulai, 239 F.3d at 554, we held that “the BIA may sometimes require otherwise-credible applicants to supply corroborating evidence in order to meet their burden of proof.” In so doing, we refused to hold invalid the rule of corroboration laid down by the BIA in S-M-J- 21 I. & N. Dec. 722. We explained that S-M-J-“contemplates a three-part inquiry: (1) an identification of the facts for which it is reasonable to expect corroboration; (2) an inquiry as to whether the applicant has provided information corroborating the relevant facts; and, if he or she has not, (3) an analysis of whether the applicant has adequately explained his or her failure to do so.” Abdulai, 239 F.3d at 554 (internal quotation marks omitted); see also Mulanga v. Ashcroft, 349 F.3d 123, 133-37 (3d Cir.2003).
The IJ plainly did not heed even the first step, which is simple common sense: There are matters on which it is plainly unreasonable to expect any kind of corroboration. Wartime persecution is surely among these matters-exigency, strife, and destruction all conspire to destroy what records there might once have been. Evidence documenting military persecution and abuse is rarely made in the first place. Testimony is nearly impossible to come by because of death and dispersal in the ranks. The IJ had no grounds on which to expect corroboration from Berishaj.
2. Changed Country Conditions
The IJ’s alternative reason for rejecting Berishaj’s asylum claim was that country conditions in Montenegro had improved by 2000 to the point that Berishaj’s stories of past persecution-even if they were credible-no longer provided a basis for a well-founded fear of future persecution. As we have noted, an alien who offers credible testimony regarding past persecution is presumed to have a well-founded fear of future persecution. See 8 C.F.R. § 208.13(b)(1); Abdulrahman, 330 F.3d 587. But 8 C.F.R. § 208.13(b)(1)(i) also provides that
an immigration judge ... shall deny the asylum application of an alien found to be a refugee on the basis of past persecution if [it] is found by a preponderance of the evidence [that] [t]here has been a fundamental change in circumstances such that the applicant no longer has a well-founded fear of persecution in the applicant's country of nationality ... on account of race, religion, nationality, membership in a particular social group, or political opinion
*327The burden of proof in a changed-country-conditions rebuttal is on the government. 8 C.F.R. § 208.13(b)(1)(h).
Other Courts of Appeals have recognized a limitation on the inferences that may be drawn from evidence of changed country conditions. The First, Seventh, Ninth, and Tenth Circuits agree that evidence of changed country conditions can successfully rebut an alien’s fear of future persecution based on past persecution only if that evidence addresses the specific basis for the alien’s fear of persecution; generalized improvements in country conditions will not suffice as rebuttals to credible testimony and other evidence establishing past persecution. The other Courts of Appeals, including this Court, appear to have had no occasion to consider the matter.
As the Court of Appeals for the Ninth Circuit has explained, “[the] INS is obligated to introduce evidence that, on an individualized basis, rebuts a particular applicant’s specific grounds for his well-founded fear of future persecution. Information about general changes in the country is not sufficient.” Rios v. Ashcroft, 287 F.3d 895, 901 (9th Cir.2002) (internal quotation marks and citations omitted); accord Krastev v. INS, 292 F.3d 1268, 1276-77 (10th Cir.2002) (granting petition for review, noting that the country report relied on by the BIA did nothing to rebut petitioner’s fear from local, not national, authorities in Bulgaria); Gailius v. INS, 147 F.3d 34, 36 (1st Cir.1998) (“It is well established that general changes in country conditions do not render an applicant ineligible for asylum when, despite those general changes, there is a specific danger to the applicant.” (citing Fergiste v. INS, 138 F.3d 14, 19 (1st Cir.1998))); Kaczmarczyk v. INS, 933 F.2d 588, 593-95 (7th Cir.1991) (explaining that, though it appropriately rebutted petitioners’ fear of future persecution in that case, the BIA could not use the election of Solidarity Party members in Poland to reject all asylum claims by Polish nationals). We agree with these cases and apply then-precepts to the case before us. The rule is a natural corollary of the more general proposition that the IJ is required to consider the record as a whole in ruling on an alien’s claim. See, e.g., Tarrawally v. Ashcroft, 338 F.3d 180, 184 (3d Cir.2003) (citing Balasubramanrim, 143 F.3d at 161).
To the extent that the IJ proposed to proceed on an interpretation of 8 C.F.R. § 208.13(b)(1) that departs from the one stated by our sister Courts of Appeals, the IJ offered no reasoning and cited no authority-not even the pertinent regulation itself. Accordingly, we have no basis on which to conclude that the IJ’s reading and application of the regulation was “reasonable” and therefore entitled to deference under Chevron U.S.A., Inc. v. Natural Resources Defense Council, 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). Cf., e.g., Valansi v. Ashcroft, 278 F.3d 203, 208-09 (3d Cir.2002) (quoting, inter alia, Lewis v. INS, 194 F.3d 539, 544 (4th Cir.1999) (“If we conclude that Congress has not directly addressed the question at issue in a statute or its intent is ambiguous, we must defer to the Board’s interpretation of the statute provided it is not an unreasonable one.” (internal citations and quotation marks omitted))).
Subject to our discussion of the staleness of the country report in the administrative record here, the IJ’s reliance on changed country conditions was only in part supported by substantial evidence. Substantial evidence does support the IJ’s conclusion that Berishaj could no longer have an objectively reasonable fear of future persecution in the military-after all, as even the 1997 State Department Country Report for Serbia/Montenegro notes, *328the war in Bosnia ended with the 1995 Dayton Peace Accord, and the Yugoslav parliament had approved an amnesty for those who had avoided military service between 1991 and 1995. While there were scattered reports of forcible conscription, it was clear by 2000 that this was the exception.
On the other hand, nothing in the country reports, or elsewhere in the record, rebuts Berishaj’s fear of persecution at the hands of Montenegrin police authorities. Berishaj testified that the police had come to his parents’ home looking for him after he had left Montenegro, and that his brother had been put in jail. Berishaj’s sister corroborated Berishaj’s account of his parents’ report of the police searches. The IJ’s extremely general observation that, in the wake of Milosevic’s withdrawal of influence over Montenegro, “the government of Montenegro has shown signs of self-determination” does nothing to refute Berishaj’s claims of police-initiated persecution.4 Similarly insufficient is the government’s observation (not relied on by the IJ, we note) that ethnic Albanians participate in the political process in Montenegro and have won seats in parliamentary elections. There may be specific reasons to think that Berishaj’s fear of persecution is no longer reasonable, but the IJ offers none, and we will not scour a 700-plus page record (well over half of which is devoted to documentary materials) for evidence unnoticed and unanalyzed by the IJ to uphold the IJ’s decision. The burden of proof in a changed-country-conditions rebuttal is squarely on the government, and no reasonable factfinder could conclude that the government has carried its burden of presenting specific evidence to rebut Berishaj’s presumed well-founded fear of future persecution.
In sum, substantial evidence does not support the IJ’s rejection of Berishaj’s claim of persecution by Montenegrin police authorities; we will therefore grant his petition for review of the IJ’s disposition of his asylum claim. We do not hold that Bershaj’s asylum claim must succeed; it may be that his objective fear of future persecution is rebutted by evidence in the record, but we certainly will not mine the record to invent our own reasons to reject Berishaj’s application. It may also be that the events of which Berishaj complains do not amount to persecution, but it would be manifestly inappropriate for us, rather than the Agency, to undertake that inquiry in the first instance. See INS v. Ventura, 537 U.S. 12, 123 S.Ct. 353, 154 L.Ed.2d 272 (2002). Finally-as we discuss in greater detail infra Part III.B-it may be that the passage of considerable time since the original agency disposition will allow the administrative record to be supplemented in a way that sheds more light on Ber-shaj’s claim for asylum.
B. The Trouble with Stale Administrative Records
It is a salutary principle of administrative law review that the reviewing court act upon a closed record. This modus procedendi secures to an administrative agency the necessary measure of authority and discretion within its sphere of *329special competence, by preventing undue interference by generalist courts that are charged only with ensuring procedural regularity in the agency’s actions. This in turn translates to long-term stability and predictability in outcomes in matters within the agency’s expertise. While the principle yields good results in most cases, in the area of asylum law, where claims are heavily dependent on country conditions, it can become an albatross. More specifically, the dispute often centers on the government’s assertion, based upon a State Department Country Report, that conditions have so changed from those represented in the asylum application that there is no longer a basis for the alien’s claim of persecution in the country of proposed removal.
It has become common that those country reports in the administrative record are three or four years old by the time the petition for review comes before us, and they frequently do not fairly reflect what our knowledge of world events suggests is the true state of affairs in the proposed country of removal, or the region embracing it. It almost goes without saying that, in the troubled areas of the planet from which asylum claims tend to come, the pace of change is rapid-oppressive regimes rise and fall, and conditions improve and worsen for vulnerable ethnic, religious, and political minorities. As a consequence, we become like astronomers whose telescopes capture light rays that have taken millions of years to traverse the cosmos, revealing things as they once were, but are no longer. But unlike astronomers, who can only speculate about what is happening at this moment in a far-off galaxy, we often know very well what has happened in the years since an administrative record was compiled.
As we have suggested above, the process-based review of agency actions is, in theory at least, just that-process-based, without regard to the merits. That should make it easier, not harder, to judge long-cold records. However, in contrast to the traditional administrative law case, this type of review can give rise to potentially devastating consequences to an applicant who faces the possibility of persecution (or worse) if he is removed.
This case is a good example of how much can change in the time between the creation of the administrative record before the IJ and the judgment of this Court. On the one hand, Slobodan Milosevic is now gone from the region, the Federal Republic of Yugoslavia no longer exists, and Berishaj’s native Montenegro is now within the recently formed loose federation of Serbia and Montenegro. On the other hand, Berishaj claims-at least as of his testimony in early 2001-that the leaders in power are in practice “mirror image[s]” of Milosevic. Four-year-old country reports are singularly unenlightening when faced with this kind of situation.'
Specific to Berishaj’s fear of persecution-we are looking now to the State Department’s 2003 Country Reports on Human Rights Practices: Serbia and Montenegro, released in February 2004, which we have downloaded from the State Department’s web site (the “2003 Country Report”)-we note that police occasionally beat suspects during arrest and detention, but there has been generally improved respect by the police for human rights. Recently enacted criminal procedure reforms are aimed toward eliminating arbitrary arrest and detention, and the Montenegrin Helsinki Committee (HCM), a recognized human-rights monitor, did not record any incidents of arbitrary arrest or detention during 2003. In the cases where arrest did not lead to prosecution, the HCM did not find (in contrast to previous *330years) any political, ethnic, or religious motivation by the police. Ethnic Albanians participate in the political process, and though they are proportionately underrepresented, they do have seats in the Montenegrin Parliament. Finally, with respect to Berishaj’s troubles at the illegal university in Kosovo, the State Department notes that the government in Kosovo “did not restrict access to the Internet or academic freedom.” But this 2003 Country Report is not part of the administrative record.
There are some applicants to whom our concerns simply do not apply-applicants from countries where conditions have not changed significantly for the better or worse in many years. And in other countries, the flux of world events is too great to hope for perfect, up-to-date decisions in every immigration case. Surely, however, we can do much better than we are doing now, especially in cases from volatile countries and with exceptionally stale records. The precise problem is not just that the administrative records in so many cases are out-of-date (though that is a contributing factor), but concomitantly that we do not have a reasonably recent final agency determination to review. It is one thing to supplement the record before us; it is quite another to decide a case based on this expanded record.
We are aware that the Court of Appeals for the Seventh Circuit apparently takes judicial notice of post-final-agency-determination developments, in the form of new country reports, and at times rests its disposition on those developments. See, e.g., Pelinkovic v. Ashcroft, 366 F.3d 532, 540-41 (7th Cir.2004) (taking judicial notice that country conditions for ethnic Albanians in Serbia and Montenegro in 2004 are much-improved over conditions in the early 1990s). This practice might go a long way toward solving the problem we face, but with all respect we are unable to square this practice with the clear command from SEC v. Chenery Corp., 318 U.S. 80, 63 S.Ct. 454, 87 L.Ed. 626 (1943), that courts reviewing the determination of an administrative agency must approve or reject the agency’s action purely on the basis of the reasons offered by, and the record compiled before, the agency itself. Moreover, we are not especially sanguine about the Seventh Circuit’s relaxed approach to agency review. It not only carries with it the potential for wholesale relitigation of many immigration-law claims, but the Courts of Appeals are ill-equipped to receive supplementary evidence. At all events, the asylum claimant should have the opportunity to challenge the updated country report that the government would rely on.
Congress could, of course, modify the rules normally applicable to petitions for review of a final decision of the BIA without scrapping the strictures of administrative agency review altogether. Congress could require the Courts of Appeals, in their sound discretion, on motion or sua sponte, to grant petitions for review of the BIA, and remand when it appears from judicially noticeable materials that the record compiled before the agency does not generally reflect contemporary country conditions.
Better yet, the parties to these proceedings might take advantage of the procedures in 8 U.S.C. § 1229a(c)(6) (permitting aliens to move to reopen proceedings on the basis of “new facts”) and 8 C.F.R. § 1003.2 (permitting an alien or the government to move the BIA to reopen proceedings, and authorizing the BIA to do so sua sponte). Indeed, both the statute and regulation seem to explicitly contemplate the situation we comment on here; they permit reopening of asylum proceedings “based on changed country *331conditions arising in the country of nationality or the country to which removal has been ordered.” 8 U.S.C. § 1229a(c)(6)(C)(ii); see also 8 C.F.R. § 1003.2(e)(3)(ii) (permitting reopening “based on changed circumstances arising in the country of nationality or in the country to which deportation has been ordered”). Counsel for aliens generally seem to be zealous in pursuing these motions to reopen when appropriate. But if this panel had to characterize the posture of petitions before it for review of the BIA on an outdated record, we would say that in the majority, country conditions had improved, weakening the alien’s case for relief. Accordingly, we encourage the Department of Justice to adopt a policy that encourages its attorneys to file motions to reopen when the adjudication of an applicant’s claim would benefit from an updated administrative record. The device of the motion to reopen is far from perfect, though, as it may additionally delay an already protracted process.
We come at last to the one actor not directly discussed so far: the BIA. The trigger for the recent spate of out-of-date records is, we suspect, the streamlining regulations noted above, which permit the BIA to summarily affirm an IJ’s decision without issuing its own opinion. See 8 C.F.R. § 3.1(e)(4). The natural-though surely unintended-consequence of the streamlining regulations is summary affir-mance by the BIA of stale, backlogged decisions by IJs. When it does so, the BIA may have shirked its role and duty of ensuring that the final agency determination in an immigration case is reasonably sound and reasonably current. The decision here on review is neither, and it is an embarrassment to the Agency on multiple levels. The “reasoning” of the IJ is open to ridicule, as we think our discussion in Part III.A illustrates; and the administrative record is a hoary relic: For example, the most recent country report was thirty-five months out-of-date at the time the BIA rendered its decision, and as of this writing, is fifty-four months out-of-date.5 Though the en banc Court in Dia approved the streamlining regulations over a statutory and Constitutional challenge, it does not follow that the regulations are not subject to misuse and even abuse.
Setting aside our perplexity at how the BIA apparently thought the IJ’s opinion worthy of being the “final agency determination,” we do not understand why the BIA did not intervene to supplement the record in a weak case, arising out of a highly volatile and evolving region of the world. The streamlining regulations exist to save an overburdened BIA from unnecessary and redundant tasks. They are not a license for the BIA to say “not our problem.” Outdated administrative records are the BIA’s problem, at least as things now stand, and the BIA needs to confront them. We therefore call on the BIA to adopt-by opinion, regulation, or otherwise-policies that will avoid the Court of Appeals having to review administrative records so out-of-date as to verge on meaningless.
In view of this discussion, we direct the Clerk of the Court to send a copy of this opinion, calling particular attention to this *332Part III.B, to the Chair, Ranking Member, Chief Majority Counsel, and Minority Counsel of the Senate Committee on the Judiciary, and the Chair and Ranking-Member of the Subcommittee on Immigration, Border Security and Citizenship; to the Chair, Ranking Member, Chief Majority Counsel, and Minority Counsel of the House Committee on the Judiciary, and the Chair and Ranking Member of the Subcommittee on Immigration, Border Security, and Claims; to the Attorney General of the United States, the Assistant Attorney General for the Civil Division, United States Department of Justice, and the Deputy Assistant Attorney General in charge of the Office of Immigration Litigation; to the Secretary of Homeland Security and the General Counsel of the Department of Homeland Security; and to the Chair of the Board of Immigration Appeals.
IV. Berishaj’s Application for Protection Under the CAT
An applicant for relief under the CAT must show that it is “more likely than not” that he would be tortured in the country of removal. See Wang v. Ashcroft, 368 F.3d 347, 348 (3d Cir.2004) (quoting 8 C.F.R. § 1208.16(c)(2) (2004)); see also Dia, 353 F.3d at 233 n. 1. It is the alien’s burden to show this, and objective evidence is required. See Sevoian, 290 F.3d at 175. The government simply argues that Berishaj’s CAT claim fails because the IJ found him not credible, and it was only his own testimony that formed the basis for the objective likelihood of being tortured. Berishaj counters that the IJ’s analysis of the CAT claim is so cursory that it is impossible to tell whether (1) the IJ thought that a CAT claim could not stand if the asylum claim fell, or (2) the IJ analyzed the country conditions evidence and concluded that it did not support a CAT claim. The first alternative would be a legal error, and would be grounds for granting the petition because asylum and CAT claims are “analytically separate.” See Zubeda v. Ashcroft, 333 F.3d 463, 476 (3d Cir.2003) (quoting Kamalthas v. INS, 251 F.3d 1279, 1283 (9th Cir.2001)).
We think the better reading of the IJ’s decision to be the second alternative-that the record as a whole does not show that Berishaj is more likely to be tortured than not if removed to Montenegro. Preliminarily, we note that Berishaj’s own testimony-whether credible or nob-has nothing to do with his CAT claim; CAT claims are entirely concerned with the objective likelihood of torture in the future, and Berishaj’s testimony did not address contemporary treatment of disfavored persons in Montenegro in any particularized way.
The balance of the record describes mistreatment and indignities, but there is scant evidence-let alone compelling evidence-that it is more likely than not that Berishaj would be tortured if removed to Montenegro. To be sure, the record suggests that, at the time of its making, there was political instability in the fledgling Federal Republic of Yugoslavia (many newspaper articles in the record attest to this), and that Serbs continued to perpetrate abuses and massacres in Kosovo (this is well-chronicled in the 1999 Country Report). But the former does not, of course, amount to torture, and the latter was in Kosovo, not Montenegro. As for Montenegro itself, there are reports of extrajudicial killings perpetrated by the Yugoslav army, but such action seemed to be outside of government control or direction, and at all events, was not “more likely than not” to be the fate of a Montenegrin like Beri-shaj. There was also forcible conscription of ethnic Albanians in Montenegro, but again, this is not torture. Most troubling perhaps is that a human rights group, the *333International Crisis Group, claims that as of 1999, “Yugoslav forces ha[ve] undertaken limited ethnic cleansing campaigns directed against ethnic Albanians in northern Montenegro.” Similar reports are scattered throughout the administrative record.
Ultimately, even if this activity amounts to torture in some instances, there is no suggestion that it is nearly frequent enough to compel the conclusion that Beri-shaj himself would more likely than not suffer torture upon removal to Montenegro. Thus we must deny the petition for review of Berishaj’s CAT claim. We also note that our observations regarding stale administrative records, see supra Part III.B, can apply with similar force to claims for protection under the CAT, even though in this particular case the CAT issue is not presented in as stark a relief as the asylum issue.
V. Conclusion
For the foregoing reasons, we will grant the petition for review, and vacate the IJ’s decision with respect to Berishaj’s asylum claim. Because the IJ’s disposition of Berishaj’s claim for withholding of removal rested on the same grounds that we have found insufficient to support his rejection of Berishaj’s asylum claim, we will also grant the petition for review, and vacate the IJ’s decision, with respect to Berishaj’s claim for withholding of removal. See Mulanga, 349 F.3d at 132 (describing relationship between asylum claims and claims for mandatory withholding of removal). We will deny the petition for review with respect to Berishaj’s claim under the CAT. We direct the Clerk of the Court to send copies of this opinion to the officers and legislators identified in Part III.B above.