CALABRESI, Circuit Judge.

When an asylum applicant himself has submitted two or more affidavits in support of his application that, he says, have been provided by different persons, but which are strikingly similar in their structure or language, our court has allowed an Immigration Judge (“IJ”) to treat those similarities as evidence supporting an adverse credibility finding. See Surinder Singh v. Bd. of Immigration Appeals, 438 F.3d 145, 148 (2d Cir.2006) (per curiam) (holding that the IJ’s adverse credibility finding was properly based on “the nearly identical language in the written affidavits [petitioner] submitted,” which the petitioner had alleged were “provided by different people in India in support of [petitioner’s] applications”). We have repeatedly allowed IJs to take into account such “intra-proceeding” similarities 1 because, in most cases, it is reasonable and unproblematic for an IJ to infer that an applicant who herself submits the strikingly similar documents is the common source of those suspicious similarities.

In the case before us, we are confronted with a related but far more difficult question: whether an IJ may consider “inter- proceeding” similarities — that is, striking similarities between affidavits that were submitted separately by ostensibly unrelated asylum applicants — as evidence of incredibility. To assume that one asylum *520 applicant is responsible for, or even aware of, the striking similarities that appear in an unrelated applicant’s submissions is much more problematic. This is because, in inter-proceeding cases, it may well be, inter alia, (1) that both applicants have inserted truthful information into a similar standardized template; (2) that the different applicants employed the same scrivener, who wrote up both stories in his own rigid style; (3) that “the other” applicant plagiarized the truthful statements of the petitioner; or (4) that the similarities resulted, not from the original documents themselves, but rather from inaccurate or formulaic translations — which unaffiliated applicants would not be in a position to discover or contest. 2

In light of these possibilities, it is clear that any reliance an IJ places on inter-proceeding similarities must be met , by a reviewing court with an especially cautious eye. Nonetheless, for the reasons here stated, we conclude that an IJ may, in appropriate situations, take such similarities into account. Because the IJ in this case carefully considered the particular similarities in question and rigorously complied with the procedural protections of Ming Shi Xue v. Board of Immigration Appeals, 439 F.3d 111, 125 (2d Cir.2006) (see infra at note 5 for a description of these), we deny the petition for review as to the merits of petitioner’s asylum claim. And because petitioner either waived or failed to exhaust key issues with respect to her withholding of removal and Convention Against Torture (“CAT”) claims, we deny review of these as well.

BACKGROUND

Petitioner Mei Chai Ye (“Ye”), a native and citizen of the People’s Republic of China, entered the United States in April 2002. She was placed in removal proceedings shortly thereafter, and, in January 2003, filed an application for asylum, withholding of removal, and relief under the CAT. In a statement attached as an addendum to her 1-589 form, Ye claimed that she had been subjected to two forced abortions in China, and that she feared that, if returned to that country, she would be involuntarily sterilized.

Ye first appeared before IJ Alan A. Vomacka on January 28, 2003, and then testified at length on June 13, 2003. On both occasions, she recounted the details of her two forced abortions, as well as her eventual escape from China. Ye’s husband, Qiu Peng Hu (“Hu”), also testified at the June 13 hearing. Hu asserted that he and his wife left China together after she had been subjected to the two forced abortions. When IJ Vomacka asked Hu why he had not filed for asylum on his own behalf, Hu replied that he had been receiving assistance from the “Huang Li Li law firm,” and that the firm advised him that he and his wife “cannot apply together.” Hu also acknowledged that Huang Li Li was “helping [him] with [his][i]mmigration case,” and that the firm helped his wife Ye with her asylum application. IJ Vomacka commented, in passing, “I see a lot of lawyers in Court, but I don’t think I’m familiar with Huang Li Li.”

Toward the close of the June 13 hearing, IJ Vomacka mentioned to counsel for both parties that he seemed to recall an asylum application filed by a different petitioner— in a case also pending before IJ Vomac-ka — that strikingly resembled Ye’s own asylum application. Moreover, both Ye and the unidentified petitioner were then *521 being represented by the same lawyer, Baird Cuber (“Cuber”). Because IJ Vo-macka believed privacy concerns were implicated in the sharing of affidavits across unrelated cases, he asked the Department of Homeland Security (“DHS”) to prepare redacted versions of the two applications. DHS agreed to do so, and the hearing was then adjourned for the day.

On June 17, 2003, Cuber submitted to IJ Vomacka a handwritten statement that was in the Chinese language. IJ Vomacka stated that he could not read it, to which Cuber replied:

As your Honor has stated that there are some similarities in this case, I just wanted to present basically the handwritten statements of the respective respondents just to, just to show that they did individually make out their own statements.

IJ Vomacka admitted the documents into the administrative record, and then asked Cuber to respond to the fact that there appeared to be striking similarities between the two petitioners’ supposedly-unrelated affidavits. Cuber attempted a response:

I believe that any similarities in the two cases would really relate more to a pattern of practice of the Chinese Government with regard to their coercive family planning policy. I do not believe that there are unique details included in the statements that, that might lead one to believe that, that the two, the two cases have striking similarities that would, that would strike one as strange. I, I do believe that although both female respondents received abortions, I believe that that is a common occurrence in, in the People’s Republic of China. The, they do have a population problem there. I, I believe that, I think it’s the Country Reports that state that they have about one quarter of the world’s population and only seven percent of, of the farmable land in, in the world. And, and so they — ....

IJ Vomacka interrupted at this point and pushed back: “Well, that might be some explanation of why they might have a birth control policy, but the Immigration law indicates that that is an acceptable congress in terms of forced birth control.” To this, Cuber said, “Yeah.” IJ Vomacka continued:

So the reason isn’t really relevant. It seems to me there are some noticeable similarities in terms of the narrative statements.... Similar phrasing, similar structure, many things that are mentioned which wouldn’t necessarily have to be mentioned but are mentioned. And it seems as though the, the parallel nature of the structure of the statements is pretty noticeable.... But I still think that the, the explanation, whatever explanation there might be for why the statements are so similar in structure and vocabulary, I guess, would be what’s puzzling the Court. You might assume that there is a country where a lot of people are persecuted the same way, but you wouldn’t expect two people in the United States to write down the history of their persecution in such a noticeably similar way.

Cuber chimed in and argued that, “I guess, you know, I guess, Your Honor, you know, with regard to translating, I, I think that when somebody, somebody starts translating documents, I, I think they develop a certain style.” IJ Vomacka acknowledged this possibility, but rejoined, Well, I don’t have any evidence that that’s so. I don’t even know that the documents were translated by the same person.” And Cuber conceded, “Right, that’s true.” The June 17 hearing was then adjourned.

The parties reconvened on June 30, 2003. Cuber did not attend, but Ye was *522 represented at the hearing by David X. Feng (“Feng”). As IJ Vomacka explained,

the Court has provided the attorneys [including both Cuber and Feng] with copies of the narrative statements of the respondents in this case and the other case, in which I’ve made notes in capital letters of what seems to me to be extremely similar. And, as we’ve discussed off the record, I expect this case to be reset in case Mr. Cuber finds anything he needs to explain, present, et cetera, in terms of documents, after studying the Court’s notations and so on about these similarities.

On August 8, 2003, counsel reconvened, although Ye — who had requested a waiver of appearance — was not herself present. As to the striking similarities that the IJ had identified in his careful annotations, Cuber had no explanation to offer. Instead, Cuber requested a withdrawal from his representation of Ye:

Your Honor, I do not speak the Chinese language and I do have to rely on other individuals to translate documentation as well as addendfa]. And after reviewing the notations that Your Honor has made in these two separate addend[a], I would request a withdrawal from my representation of [Ye],

IJ Vomacka asked, “Now, without trying to lead the counsel, is this due to possible conflicts between you and your client about the subject matter of the case, is that one way to sum it up?” Cuber agreed. IJ Vomacka proceeded to grant the motion, and allowed Feng — who was also present at the hearing — to substitute as counsel for Ye.

On August 8, 2003, IJ Vomacka issued an oral decision in In the Matter of Ye, Mei Chai, No. A 78 974 047 (Immig. Ct. New York, N.Y. Aug. 8, 2003). First, IJ Vomacka identified several supposed inconsistencies within Ye’s testimony and between her testimony, affidavit and corroborative documentation. IJ Vomacka made clear, however, that he did not consider these inconsistencies to be major. In fact, the IJ acknowledged that “[t]he basic story of the respondent is set out fairly well in the narrative statement attached to her asylum application^ and while s]he did not tell the story exactly the same way in her testimony ... the differences are not radically different in terms of major elements of the story.”

Then, IJ Vomacka turned to what he considered to be “the main issue in the present case,” namely, “the striking similarity” between Ye’s affidavit and the affidavit submitted by a different petitioner in an unrelated case. IJ Vomacka recalled the annotations he had provided to the parties, which identified twenty-three separate places at which the two affidavits were strikingly similar in language and grammatical structure. Moreover, these identical portions appeared, with only two minor exceptions, in the exact same order in both affidavits. IJ Vomacka concluded, “a reasonable person familiar with the presentation of such documents would find that the similarities here are much too striking and far too many to likely be the result of an accident.” The IJ continued,

[I]t is theoretically possible that two different people from China who had suffered problems under a birth control policy might explain their stories in a remarkably similar way. That could happen. But I think given the number of words and sentences in these statements, and given the number of possible variations that might have been introduced in terms of how people tell a story that happened to them, the chance that these two statements are the result of accidental similarity is a very small chance, and I do not feel bad in saying it might be one in a million.

*523 But IJ Vomaeka did not rest his conclusion on this impression alone. Rather, he noted (1) that he had given Ye various opportunities to respond to the striking similarities, and that she failed to provide any convincing response; (2) that he had received no evidence to suggest that the translations were not accurate — and Ye offered no proof in support of such an argument; and (3) that there was no evidence suggesting that “one respondent might have somehow obtained a copy of [Ye’s] Chinese narrative and then used it to make up a story by, in other words, plagiarizing someone’s true experiences.” In addition, IJ Vomaeka observed that Hu’s testimony about the assistance he received from Huang Li Li, which IJ Vomaeka suspected was not a veritable law firm,

raise[d] ... a very clear explanation of how this similarity might have resulted, that is to say, the agency might simply be preparing stories for would-be asylum applicants and selling the story, so to speak, as a service so that they have a basis to apply for asylum regardless of whether they actually had any problem in China.

Given all this, IJ Vomaeka concluded (1) “that the respondent has not met her burden to prove that the factual basis for any of her three applications is more likely true than not,” and (2) that “the Court further finds that the respondent did submit a fabricated application for asylum.” 3 Accordingly, Ye was ordered removed from the United States.

Ye appealed the IJ’s ruling to the BIA, and on December 22, 2004, the BIA summarily affirmed. See In re Ye, Mei Chai, A 78 974 047 (B.I.A. Dec. 22, 2004), aff'g No. A 78 974 047 (Immig. Ct. New York, N.Y. Aug. 8, 2003). This petition for review followed.

DISCUSSION

“Where the BIA expressly adopts the IJ’s findings and reasoning, as it did here, we review the decision of the IJ as if it were that of the BIA.” Chun Gao v. Gonzales, 424 F.3d 122, 124 (2d Cir. 2005). The IJ’s factual findings, including adverse credibility findings, are reviewed under the substantial evidence standard of 8 U.S.C. § 1252(b)(4)(B). While this standard of review is especially deferential, 4 we *524 have made clear that “the fact that the [agency] has relied primarily on credibility grounds in dismissing an asylum application cannot insulate the decision from review.” Ramsameachire v. Ashcroft, 357 F.3d 169, 178 (2d Cir.2004). An adverse credibility finding that is “based on flawed reasoning ... will not satisfy the substantial evidence standard.” Secaida-Rosales v. Immigration and Naturalization Serv., 331 F.3d 297, 307 (2d Cir.2003).

In her briefing to this court, Ye argues that “[t]he similarity the Immigration Judge mentioned between two applicants is irrational.” We disagree.

Although this court has never spoken to the precise issue of whether inter-proceeding similarities may support an adverse credibility finding, our case law on intra-proceeding similarities has firmly embraced the commonsensical notion that striking similarities between affidavits are an indication that the statements are “canned.” See, e.g., Surinder Singh, 438 F.3d at 148; see also supra note 1. We believe it is likewise reasonable, in appropriate circumstances, to draw an inference of falsity from inter-proceeding similarities.

To be sure, it is far more dangerous to draw such an inference from inter-proceeding similarities. And this is worth emphasizing. As explained earlier, see supra, inter-proceeding similarities may result simply because, inter alia, applicants are inserting wholly truthful information into standardized templates; or, because the different applicants were illiterate and related their stories to the same scrivener who wrote them up in his own — unchanging — locution; or, even if plagiarism did occur, it may be impossible to determine who copied whom. Alternatively, such similarities may have been inserted into the documents by the translators rather than by the applicants themselves.

In light of these dangers, it is clear that inter-proceeding cases call for caution. We therefore encourage the BIA to address more formally and systematically the issue of inter-proceeding similarities, so that, upon its consideration of the frequency, type, and manner of such cases, it might provide us with expert guidance as to the most appropriate way to avoid mistaken findings of falsity, and yet identify instances of fraud. But the BIA has yet to address the issue. Until it does, this court must determine for itself whether inter-proceeding similarities have been properly considered.

In the present case, IJ Vomacka meticulously followed certain procedural safeguards which, taken together, sufficiently addressed the dangers inherent in relying on inter-proceeding similarities. We believe it appropriate, in explaining this holding, to identify and describe the features of IJ Vomacka’s decision that have led us to *525 deny Ye’s petition for review. To that we now turn.

IJ Vomacka carefully annotated the twenty-three strikingly similar portions of the two affidavits, and considered the possibility that the similarities might have been the result of mere coincidence. He concluded, quite reasonably, that the similarities in this case were plainly too pervasive to have resulted from chance, and that the stories were so blatantly similar in both form and substantive details that they could not have been the result of honest applicants inserting truthful information into standardized templates. Equally importantly, IJ Vomacka rigorously complied with the notice requirements of Ming Shi Xue, 5 by (1) notifying Ye of the similarities, and providing her with copies of his annotations; (2) openly and exhaustively expressing to Ye his concerns about the inter-proceeding similarities; (3) granting Ye several opportunities to comment on those similarities; and (4) inviting Ye to offer evidence of plagiarism, inaccurate translations, or any other possible innocent explanation. Once it became evident that Ye would not seek to take advantage of these numerous opportunities to explain, it became reasonable for IJ Vomacka to draw the inference that the remarkable inter-proceeding similarities were evidence that Ye’s asylum application was false. 6

*526 We further conclude that, once IJ Vo-macka had carefully parsed through the remarkable similarities and rigorously complied with the safeguards of Ming Shi Xue, and drew the reasonable inference that Ye’s narrative was falsified, it was appropriate for the IJ to find, in the circumstances of this case, that Ye’s submission of the false document undermined her general credibility — and, by extension, the credibility of her husband Hu, who had offered identical testimony. Indeed, Ye’s willingness to submit a false document is in itself sufficient evidence of incredibility. See Siewe, 480 F.3d at 170-71 (discussing the falsus in uno, falsus in omnibus doctrine, and explaining that an alien’s submission of a false document “redounds upon all evidence the probative force of which relies in any part on the credibility of the petitioner”). 7

We pause to make clear that our holding today does not purport to promulgate and impose a specific set of procedural safeguards which IJs must follow in all respects and in all cases. Rather, and as we indicated above, our holding is necessarily based on the limited information concerning inter-proceeding similarities that this court currently has at its disposal, and does not preclude the BIA from developing more appropriate guidelines of its own. 8

We hold merely that the dangers inherent in relying on inter-proceeding similarities are significantly reduced when, as here, an IJ (1) carefully identifies any similarities; (2) closely considers the nature and number of those particular similarities and determines (a) whether there is a meaningful likelihood that they resulted from mere coincidence, (b) whether it is plausible that different asylum applicants inserted truthful information into a standardized template or, for illiteracy reasons, conveyed it to a scrivener tied to an unchanging style; (c) whether the same translator converted valid accounts into a peculiarly similar story; and (d) whether there is a likelihood that the petitioner was an innocent “plagiaree”; and (3) rigorously complies with the procedural protections of Ming Shi Xue, 439 F.3d at 125, by allowing an alien the opportunity (a) to explain or contest the similarities; (b) to investigate the possibility that her affidavit might somehow have been plagiarized; or (c) to consider whether the seemingly similar affidavits might merely have been translated or recorded inaccurately or formulaically.

*527 Our holding has two implications. On the one hand, when an IJ carefully follows these procedural safeguards 9 — as IJ Vomacka did — a reviewing court can confidently defer to reasonable inferences that an IJ draws from the inter-proceeding similarities — including the generally permissible inference that, when striking inter-proceeding similarities are present, those similarities are evidence of a “canned” story. Cf. Surinder Singh, 438 F.3d at 148; see also Siewe, 480 F.3d at 169 (“So long as an inferential leap is tethered to the evidentiary record, we will accord deference to the finding.”).

On the other hand, our holding indicates that we would view much more skeptically an adverse credibility finding by an IJ who, in relying on inter-proceeding similarities, adopted a less rigorous approach than that employed by IJ Vomacka in this case.

We are mindful that the IJ also identified several parts of Ye’s testimony which he believed were inconsistent. It is unnecessary to analyze whether these supposed inconsistencies were proper bases for the IJ’s adverse credibility finding, however, because the IJ made abundantly clear that he believed them to be minor. See In re Ye, Mei Chai, No. A 78 974 047, at 6 (“The basic story of the respondent is set out fairly well in the narrative statement attached to her asylum application. She did not tell the story exactly the same way in her testimony, and I will discuss the discrepancies. But the differences are not radically different in terms of major elements of the story.”). Because, in effect, the IJ based his adverse credibility finding almost exclusively on his identification of striking inter-proceeding similarities, and because we conclude that those similarities were appropriately treated as substantial evidence of incredibility, “we can state with confidence that the IJ would adhere to his decision were the petition remanded.” Surinder Singh, 438 F.3d at 147-48 (citation and internal quotation marks omitted).

*528 II

It is not entirely clear whether Ye adequately exhausted her withholding and CAT claims before the BIA. If she did not, then we would lack jurisdiction to consider them. See Lin Zhong v. U.S. Dep’t of Justice, 480 F.3d 104, 107 (2d Cir.2007) (“[Ejxhaustion of ... claims for relief, is jurisdictional.”). But we need not resolve whether the jurisdictional claim exhaustion requirement was satisfied because, assuming arguendo that the claims were exhausted, they nevertheless should not be considered by us.

First, Ye does not raise her withholding claim in her briefing to us, and this claim is therefore waived. Second, as to Ye’s CAT relief claim, she now raises, for the first time, the argument that she will be tortured for having left China illegally. And this is the only basis on which she seeks CAT relief before us now. But Ye did not exhaust this issue before the BIA. And, to the extent Ye implicitly relies on our court’s decision in Lin Zhong, her efforts are entirely misguided. As the opinion in Lin Zhong makes perfectly clear, “in holding that, though not jurisdictional, issue exhaustion is mandatory, [we] expect[] that virtually no case will be treated differently from the way it would be were the requirement deemed jurisdictional.” Id. at 107 n. 1; cf. Xiao Ji Chen v. U.S. Dep’t of Justice, 471 F.3d 315, 320 n. 1 (2d Cir.2006) (refusing to consider petitioner’s unexhausted issues on appeal despite government’s failure to raise the issue exhaustion defense, and noting, correctly, that “nothing in Lin Zhong requires” a contrary result). The opinion in Lin Zhong merely clarifies — and thus strengthens — the doctrinal foundations of our court’s issue exhaustion jurisprudence. To suggest that Lin Zhong drains the force of longstanding exhaustion principles would be to read the opinion in reverse. Assuming arguendo, then, that Ye exhausted her claims before the BIA, (1) Ye has waived her withholding claim, and, (2) consistent with Lin Zhong and the longstanding practice it fully embraces, we must decline to consider her unexhausted CAT relief argument.

CONCLUSION

For the foregoing reasons, the petition for review is Denied. Any pending motions are Denied as moot.

Notes

. Indeed, it has become so much a part of our practice that this has been done regularly by summary order.

. In addition, there is the possibility, also present when dealing with intra-proceeding likenesses, that the similarities may be the product, not of copying, but merely of coincidence.

. While the IJ found the document to be "fabricated,” the IJ did not elaborate whether this meant that he was making a finding of frivolousness, with all the draconic consequences of such a finding. And the BIA, in affirming the IJ, nowhere mentioned the IJ’s frivolousness ruling, a result of either Ye's counseled decision not to argue the issue to the BIA or an ambiguity in the IJ's ruling. Under the circumstances, we assume that no issue of frivolousness is before us, and discuss the matter no further. We note, however, that to the extent the IJ’s or the BIA’s decisions might have been read to include a finding of frivolousness, the BIA’s recent opinion in In re Y -L-, 24 I. & N. Dec. 151 (B.I.A. Apr. 25, 2007), would appear to cast doubt on the sustainability of such finding. See id. at 156 (seeming to require a separate and detailed finding of deliberate fabrication).

. Our court has not been consistent in its description of the substantial evidence standard. In Jin Shui Qiu v. Ashcroft, 329 F.3d 140 (2d Cir.2003), a panel suggested, in dicta, that "[substantial evidence review in the immigration context is 'slightly stricter’ than the clear-error standard that the circuit courts typically apply in reviewing a district court's factual findings,” id. at 149. More recently, a panel in Siewe v. Gonzales, 480 F.3d 160 (2d Cir.2007), stated, also in dicta, that the clear error and substantial evidence standards are identical, id. at 168 ("These standards of review bespeak no lesser deference to an IJ than to a district judge when each draws inferences from the evidence as a finder of fact.”).

In the end, this disagreement over formal labels is not the real issue. As our court so presciently stated in Ming Xia Chen v. Board of Immigration Appeals, 435 F.3d 141 (2d *524 Cir.2006), "[w]hile the various statements made in the course of upholding or rejecting the adequacy of a particular finding are often helpful, they cannot become rigid rules of law that dictate the outcome in every case,” because "[w]e know of no way to apply precise calipers to all such findings so that any particular finding would be viewed by any three of the 23 judges of this Court as either sustainable or not sustainable,” id. at 145. In practice, "[pjanels will have to do what judges always do in similar circumstances: apply their best judgment, guided by the statutory standard governing review and the holdings of our precedents, to the administrative decision and the record assembled to support it.” Id. And sound judgment of this sort cannot be channeled into rigid formulae.

. In Ming Shi Xue, our court held that "an IJ may not rest an adverse credibility finding on non-dramatic putative contradictions or incongruities in an alien’s narrative without first giving the applicant a chance to reconcile the testimony.” 439 F.3d at 125. We noted, however, that when "the relevant inconsistency [is] plainly self-evident so that identification of it [is] not needed to make the alien cognizant of the defect,” an IJ was not obligated to bring it to the attention of the alien. Id.; see Majidi v. Gonzales, 430 F.3d 77, 81 (2d Cir.2005) (holding that, where an asylum seeker has given “dramatically different” accounts of his alleged persecution, an IJ may properly find him incredible "without soliciting from the applicant an explanation for the inconsistency”); Hoxhallari v. Gonzales, 468 F.3d 179, 187-88 (2d Cir.2006) (applying Majidi to an IJ’s finding of changed country conditions where that finding was "clearly sufficient” and where "there [was] no doubt that there has been a fundamental change in the political structure and government of Albania” (emphasis added)).

The present case, of course, involves consistencies that appear to impugn the credibility of the applicant. But Ming Shi Xue is also applicable in this context. In fact, when dealing with inter-proceeding similarities, Ming Shi Xue, rather than Majidi, will virtually always apply — and with special force. This is because, absent clear evidence that one alien has accessed the submissions of the other and, given that the similarities arise, not within either aliens' submissions, but in a comparison of them, it is unreasonable to assume that inter-proceeding similarities are "plainly self-evident” to either alien.

At its narrowest, Ming Shi Xue applies to this case because a credibility finding is here involved. But beyond that, the protections of Ming Shi Xue are applicable because the core holding of Ming Shi Xue is that petitioners have a right to be informed of the bases of decisions as to which an explanation is crucial, and as to which the need for an explanation is not obvious. That is true in some credibility situations and not in others, see, e.g., Majidi, 430 F.3d at 81; it is equally true in some changed country conditions cases and not in others, see, e.g., Hoxhallari, 468 F.3d at 187-88.

. IJ Vomacka further supported his conclusion with the evidence suggesting that Ye's submissions were prepared by an illegitimate agency, Huang Li Li. While evidence of an alien's access to "canned” asylum narratives undoubtedly would, as a general matter, tend to bolster an inference of falsity, we believe that, in this case, IJ Vomacka did not sufficiently bring this particular issue to the attention of Ye. Stated differently, under Ming Shi Xue, it was not enough for IJ Vomacka merely to say that “I don’t think I’m familiar with Huang Li Li.” In order to rely on Ye’s contacts with Huang Li Li as evidence offering an “explanation of how this similarity might have resulted,” IJ Vomacka needed to make *526 clear to Ye, not just that he was unfamiliar with Huang Li Li, but also that he was inclined to draw adverse inferences from the fact of his unfamiliarity.

Nonetheless, we conclude that, in this case, the evidence of striking inter-proceeding similarities was strong enough to support the IJ’s adverse credibility finding, even without this additional evidence of "opportunity.”

. Siewe lists five situations in which it would be inappropriate for an IJ to rely on the doctrine of falsus in uno. See Siewe, 480 F.3d at 171. The Siewe panel did not propose that list to be exhaustive, and we do not take it to be as such; in this context, as in any other context in which adverse credibility findings must be assessed, a reviewing court is called on to exercise sound judgment, and hence, any circumstances that might make the doctrine of falsus in uno inappropriate must be taken into account. See supra note 4; see also Ming Xia Chen, 435 F.3d at 145.

. In addition, we do not intend to suggest that, in the context of intra-proceeding similarities, an IJ need not follow any procedural safeguards. The opinion in Surinder Singh, and the numerous summary orders applying Surinder Singh’s holding, did not discuss the issue of an IJ's obligations in that context. Because this case involves inter-proceeding similarities only, we have no occasion to consider what kinds of procedural safeguards are appropriate when an IJ wishes to rely on intra-proceeding similarities.

. There is nothing novel about our insisting on the application of heightened procedural protections to a context in which they are necessary to safeguard the integrity of the agency's fact-finding function. When an IJ finds that corroborative evidence is required to support an asylum application, for example, we have held that "the IJ and BIA [have] no leeway to deny [an alien's] application without first (a) pointing to specific pieces of missing, relevant documentation, and (b) showing that this documentation was reasonably available to [the alien],” Jin Shui Qiu, 329 F.3d at 153, and, in addition, we have required the agency explicitly to "assess [the applicant’s] explanations for his failure to produce the requested corroborative evidence,” Diallo v. INS, 232 F.3d 279, 289 (2d Cir.2000). As our court has explained, "[t]hese directives ensure[] that, before denying an asylum petition because of insufficient corroboration, an IJ g[ives] adequate and meaningful notice to the applicant of evidence that the IJ believed was significant and missing,” and, in addition, "and equally importantly, [the directives] guarantee applicants an opportunity to remedy the supposed evi-dentiary gap,” Ming Shi Xue, 439 F.3d at 122.

In similar fashion, we have held that, when an applicant’s testimony is so vague as to lead an IJ to "fairly wonder whether the testimony is fabricated,” Jin Shui Qiu, 329 F.3d at 152, an IJ should not "reflexively surrender[] to 'nagging doubts about an applicant’s credibility due to the spareness of her testimony,’ ” but rather, an IJ is encouraged to " 'probe for incidental details, seeking to draw out inconsistencies that would support a finding of lack of credibility.’ ” Ming Shi Xue, 439 F.3d at 122-23 (quoting Jin Shui Qiu, 329 F.3d at 152). One reason for our recommendation was our recognition of the dangers of allowing IJs to rely on such reasons — in that context, "the dangers of allowing judges to dispose of potentially legitimate asylum claims for reasons that could be conjured up at will.” Ming Shi Xue, 439 F.3d at 123.