dissenting:
My colleagues have determined that no “jurists of reason would find ... debatable” whether Carlton Blount’s habeas petition is timely. See Maj. Op. at 736 (quoting Slack v. McDaniel, 529 U.S. 473, 478, 120 S.Ct. 1595, 146 L.Ed.2d 542 (2000)). As a jurist formerly known as reasonable, I disagree. In the form of a Certificate of Ap-pealability (“COA”) analysis, the majority performs a full-blown merits review on the timeliness of Blount’s petition—exhausting legal and factual arguments, crafting a new rule for interpreting state law, and invoking the court’s rarely-used discretion to forgive the government its concessions. Such an analysis is inappropriate at the COA stage, as the Supreme Court clearly held only a few months ago. Buck v. Davis, — U.S.-, 137 S.Ct. 759,197 L.Ed.2d 1 (2017). In fact, Blount’s contentions easily pass the Court’s threshold COA test. More importantly, once over that modest hurdle, Blount’s claims on appeal warrant a remand to the district court. I first explain why Blount’s claims entitle him to a COA, and, having done so, propose a merits disposition.
[[Image here]]
In 2001 a District of Columbia jury was asked to determine if Blount committed murder, a specific intent crime. Over defense counsel’s objection, the trial court instructed the jury that Blount could be convicted even if he did not intend to commit murder. Transcript, United States v. Blount, No. F-983-00, at 73-75 (D.C. Super. Ct. Feb. 20, 2001). All the jury needed to find was that (1) Blount “intentionally participate[d]” in some crime and (2) that murder was a “natural and probable consequence” of that crime. Id. Blount was, quite unsurprisingly, convicted. On appeal, his new counsel neglected to raise the rejection of trial counsel’s objection to the jury instruction, and the conviction was affirmed. Blount’s conviction is thus marred by two prima facie constitutional errors. The jury instructions omitted necessary elements of the crime. See United States v. Gaudin, 515 U.S. 506, 509-10, 115 *744S.Ct. 2310, 132 L.Ed.2d 444 (1995); see also Wilson-Bey v. United States, 903 A.2d 818, 826, 843-44 (D.C. 2006) (en banc) (finding the same jury instructions unconstitutional). And Blount’s appellate counsel did not challenge the substance of these instructions, though the error was compelling and properly preserved. See Smith v. Robbins, 528 U.S. 259, 288, 120 S.Ct. 746, 145 L.Ed.2d 756 (2000) (articulating Sixth Amendment test for appellate counsel’s “failure to raise a particular claim”); see also Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). His appellate lawyer’s error, bound up with the trial court’s error, is the subject of Blount’s current petition.
Blount filed this petition (his second) over a dozen years after his conviction. Absent some reason to stop the clock, it would appear untimely against the one-year statute of limitations in 28 U.S.C. § 2244(d)(1). But Blount has offered two reasons for us to find him entitled to have the merits of his petition addressed.
First, Blount claims that he is entitled to equitable tolling of the statute of limitations as of the date he filed his first federal habeas petition. That petition contained the same claim as the current one, but it was dismissed for lack of exhaustion (Blount had not yet employed the District’s device for seeking relief on his claim of ineffective appellate counsel, i.e., asking the D.C. Court of Appeals to recall its mandate). Assuming that the first petition was timely (a fact the government conceded), and assuming that it was improperly dismissed rather than stayed (a highly probable conclusion), Blount would be entitled to equitable tolling for a reasonable period, thereby overcoming the time barrier.
Second, and independent of the first argument, Blount asserts that the D.C. Court of Appeals re-opened his case on October 11, 2012, when it adjudicated his motion to recall the mandate. As per Jimenez v. Quarterman, 555 U.S. 113, 129 S.Ct. 681, 172 L.Ed.2d 475 (2009), a state court’s decision to re-open an appeal resets the limitations period. If the D.C. Court of Appeals in fact reopened Blount’s case, then it would not have become “final” again until January 9, 2013, when the 90-day window closed on petitioning for cer-tiorari of the October decision; § 2244(d)(l)’s one-year limit would thus give him until January 2014 to file a new petition. Blount filed his current petition in 2013, so it would also be timely if his re-opener theory is correct.
My colleagues examine both issues, but they create a chimerical beast, neither COA nor merits analysis and in disregard of the sequence mandated by statute and the Supreme Court. Before Miller-El v. Cockrell, “[m]any Courts of Appeals decisions ha[d] denied applications for a COA only after concluding .that the applicant was not entitled to habeas relief on the merits.” 537 U.S. 322, 348, 123 S.Ct. 1029, 154 L.Ed.2d 931 (2003) (Scalia, J., concurring). The Supreme Court put an end to that practice, declaring that a COA is a “threshold question” that “should be decided without ‘full consideration of the factual or legal bases adduced in support of the claims.’ ” Buck, 137 S.Ct. at 773 (quoting Miller-El, 537 U.S. at 336, 123 S.Ct. 1029). A court'cannot defacto “decid[e] the case on the merits'” and then use that determination to deny the COA. Id. In Buck the Fifth Circuit had issued a brief unpublished opinion that “phrased its determination in proper terms” of the COA standard, id., insisting that it was performing an “[ijnitial examination” and merely looking for any “plausible argument” on the merits, see Buck v. Stephens, 623 Fed.Appx. 668, 673-74 (5th Cir. 2015). But the Court looked past those recitations to find *745that the Fifth Circuit’s analysis was “essentially” the same as would be performed on an actual merits review. Buck, 137 S.Ct. at 773.
The majority today also focuses squarely on the merits—but does so with even less pretense of a COA analysis than in the Fifth Circuit’s Buck opinion. Although a “thorough” consideration of the issues certainly suggests that we’ve moved beyond the COA stage, see Buck, 137 S.Ct. at 774, the primary fault of the majority opinion isn’t that it’s lengthy and detailed, compare Maj. Op. at 742-43 n.6; the problem, as we’ll see, is that it resolves a number of subtle issues, clearly debatable among reasonable jurists (as here).
That is a “fundamental” error, Miller-El, 537 U.S. at 342, 123 S.Ct. 1029, though one whose correction would likely do Blount no good—the majority opinion leaves little doubt that my colleagues would deny the claim if they recognized the merits as being properly before the panel. In any event, the merits analysis they perform is not simply inappropriate at this stage; it is, as I explain after discussing the COA, incorrect.
To warrant a COA Blount need only show that reasonable jurists could disagree about one of his two procedural arguments, along with, of course, his constitutional claim of right. See Slack, 529 U.S. at 478, 120 S.Ct. 1595 (COA should be granted when the prisoner shows “at least” that reasonable jurists could disagree over both the procedural propriety and the substance of the petition). On the equitable tolling point, the government has insisted to us that Blount’s first habeas petition was timely. Gov’t Opposition to Mot. for COA, No. 15-5056, at 4 n.5 (July 27, 2015) (“Gov’t Opp.”); see Maj. Op. at 741. Given this concession, tolling for a reasonable period from that date could render the second petition timely as well.
But relying on our discretionary authority to forgive “the government’s [ ] failure,” my colleagues resolve that the first petition was time barred and that any tolling from that date would be pointless. Maj. Op. at 741-42. How such a forgiveness becomes indisputable among reasonable jurists I cannot fathom. In Day v. McDonough, 547 U.S. 198, 126 S.Ct. 1675, 164 L.Ed.2d 376 (2006), the Court found discretion in the courts to overlook a state’s mistaken assertion that a habeas petition was timely; it agreed with the state’s contention that the exercise of that discretion should depend on “whether the administration of justice is better served by dismissing the case on statute of limitations grounds or by reaching the merits of the petition.” Id. at 208-09, 126 S.Ct. 1675. The Court later cautioned in Wood v. Milyard, 566 U.S. 463, 132 S.Ct. 1826, 182 L.Ed.2d 733 (2012), that “appellate courts ordinarily abstain from entertaining issues that have not been raised and preserved” and that they “should reserve” their Day “authority for use in exceptional cases.” Id. at 1834.
The majority disregards the principle of Day v. McDonough, paying no attention to its instruction to consider whether the administration of justice is better served by holding the government to its concession. Instead the majority insists that Blount’s argument would require deviating from certain “correct dates” and expresses doubt that a “waiver or forfeiture could ever lead a reasonable jurist” to deviate so, at least not without justification beyond the concession itself. Maj. Op. at 741-42. But under Day and Wood those doubts are misplaced. And treating Blount’s first petition as timely would not require changing any of the majority’s reconstructed dates. As the majority agrees that equitable tolling would begin “when Blount filed his first federal habeas petition,” id. at 741, *746and the government conceded the timely filing of that petition, Gov’t Opp. at 4 n.5, the majority’s calculations merely show that Blount needs what is being requested: equitable tolling for every day between the first and second petitions.
It makes no difference that the government never waived its statute of limitations defense to the current, second petition. Maj. Op. at 742-43. The government’s explicit concession that Blount’s first petition was “filed ... within the one-year statute of limitations,” Gov’t Opp. at 4 n.5, if accepted by us, opens the door to equitable tolling and defeats the government’s general timeliness objection to the second. Under the Court’s cases, the government’s concession (even if erroneous) should be excused only in “exceptional” circumstances. See Wood, 132 S.Ct. at 1834; Day, 547 U.S. at 209-10, 126 S.Ct. 1675. The majority offers nothing remotely exceptional here.
In exercising the discretion allowed by Wood and Day, my colleagues have chosen to respond to the government’s errors with easy magnanimity, and to Blount’s diligence with merciless exactitude. This approach is particularly troubling given the strength of Blount’s underlying Strickland argument. See Rhines v. Weber, 544 U.S. 269, 278, 125 S.Ct. 1528, 161 L.Ed.2d 440 (2005) (courts should be wary of dismissing “potentially meritorious” habeas claims). And regardless of how my colleagues treat the government’s concession, they are incorrect to insist on fully resolving that issue at the COA stage. “[A] court of appeals should not decline the application for a COA merely because it believes the applicant will not demonstrate an entitlement to relief.” Miller-El, 537 U.S. at 337, 123 S.Ct. 1029.
In some cases, a choice might be discretionary but nonetheless obvious to any reasonable jurist. Not so here, where the default rule lies in Blount’s favor (a party must live with its concessions) and Blount’s constitutional claim is likely to succeed. Blount’s equitable tolling argument is hardly one we can say all reasonable jurists would reject. He has thus made the necessary showing on the procedural part of the COA. (Blount’s Jimenez theory would also survive the threshold analysis suitable for a COA, but as Blount’s equitable tolling argument is enough to earn him a COA for review of the district court’s procedural holding, I discuss Jimenez only below, as an alternative route for reversal in the appeal proper.)
The substance of Blount’s Strickland claim also deserves a COA. When it adjudicated Blount’s motion to recall the mandate, the D.C. Court of Appeals “did not decide” the first Strickland prong, deficiency, but may have ruled on the second prong, prejudice—reasonable jurists would therefore decide deficiency de novo and review prejudice with deference to that court. See Porter v. McCollum, 558 U.S. 30, 38-39, 130 S.Ct. 447, 175 L.Ed.2d 398 (2009); see also 28 U.S.C. § 2254(d). In Robbins, the Supreme Court adopted the Seventh Circuit’s formula for ascertaining deficiency when counsel omitted an issue on appeal: appellate advocacy is constitutionally deficient when “ignored issues are clearly stronger than those presented.” 528 U.S. at 288, 120 S.Ct. 746. Blount has offered more than enough evidence on this front. His appellate counsel neglected to challenge the unconstitutional jury instructions even though that challenge had been preserved by trial counsel and those instructions “w[ere] under active criticism and discussion at the time,” Blount v. United States, No. 08-CO-959, 983 A.2d 1064, slip op. at 1 (D.C. Nov. 10, 2009) (“Blount IF) (affirming denial of post-conviction relief for the jury instruction error, because Blount’s appellate counsel had no “cause” to omit that claim). The minor *747weaknesses of this argument—e.g., the possibility that the error could be found harmless, see Neder v. United States, 527 U.S. 1, 4, 119 S.Ct. 1827, 144 L.Ed.2d 35 (1999)—offer scant reason for omitting it. And it remains far more compelling than the points Blount’s appellate counsel did raise, contentions almost bound to lose, such as that the court admonished trial counsel too harshly during cross-examination and that it was error to give an accomplice instruction because the indictment failed to allege accomplice liability. See Blount v. United States, No. 01-CF974, 859 A.2d 651, slip op. at 3-7 & n.12 (D.C. June 29, 2004) (“Blount I”) (affirming conviction on direct appeal).
As for the prejudice prong, the D.C. Court of Appeals stated that “based upon the entire record, any error was harmless due to the overwhelming evidence presented at trial.” Blount v. United States, No. 01-CF-974, 859 A.2d 651, slip op. at 1 (D.C. Oct. 11, 2012) (“Blount III”) (order adjudicating motion to recall the mandate). That analysis misses the nuance of the test mandated by Robbins: whether there is “a reasonable probability that, but for his counsel’s unreasonable failure ... he would have prevailed on his appeal.” 528 U.S. at 285, 120 S.Ct. 746. Thus it would suffice for Blount to show, as he has, a reasonable probability that the original appellate panel would have rejected the government’s claim of harmlessness beyond a reasonable doubt. When you multiply a fraction of a burden (reasonable probability) by a fraction of a burden (rebutting government’s contention that the jury instruction made no difference beyond a reasonable doubt), you get a smaller fractional burden: Blount need establish only a reasonable probability that the government might not have established harmless error. The D.C. Court of Appeals was incorrect to require Blount to rebut the government’s harmlessness contention entirely. Whether that mistake amounts to an “unreasonable application” of Strickland, reasonable jurists might debate. See 28 U.S.C. § 2254(d). And that is all that Blount need show “at this stage,” where “we only ask whether ... deference” to the state court decision is “debatable amongst jurists of reason.” Miller-El, 123 S.Ct. at 1042.
For these reasons, the majority should have granted the COA rather than issue, as they did, an advisory opinion on the merits followed by a summary denial of the COA. As I find grounds for granting the COA, I now proceed to the appeal proper, on which I find the majority’s advisory opinion mistaken. Readers may wonder why I did not simply grant the certificate myself and thus provide my colleagues with jurisdiction to reach the merits decision that they do. A circuit judge acting alone is authorized to grant a COA. 28 U.S.C. § 2253(c)(1); Fed. R. App. Pro. 22(b)(l)-(2). But as far as I can tell, no judge has ever taken that step in this court, certainly not after panel oral argument. I have chosen not to do so partly out of comity but more out of futility; a move to the more difficult appeal stage would—so far as appears—not yield a difference in my colleagues’ conclusion. Nonetheless, I now move on to the merits, the logical sequel to my view on the COA.
[[Image here]]
Turning to “full consideration” of the procedural merits, Buck, 137 S.Ct. at 773, I find Blount entitled to equitable tolling— covering the time spent adjudicating Blount’s first federal petition as well as the short periods it took him to go from our court back to the D.C. Court of Appeals and from there back to federal court. (Blount’s time litigating the motion to recall the mandate can be tolled as a matter of course. See Maj. Op. at 740 n.4.) Equitable tolling is appropriate under the Anti-terrorism and Effective Death Penalty Act *748of 1996 (“AEDPA”), Pub. L. No. 104-132, 110 Stat. 1214, when the prisoner “has been pursuing his rights diligently” and “some extraordinary circumstance stood in his way.” Holland v. Florida, 560 U.S. 631, 649, 130 S.Ct. 2549, 177 L.Ed.2d 130 (2010). Blount has certainly been diligent. He filed his motion to recall just 31 days after the COA on his first petition was denied. Compare Blount v. Wilson, No. 11-7060 (D.C. Cir. Sept. 27, 2011) (denying COA) with Mot. to Recall Direct Appeal Mandate, Blount v. United States, No. 01-CF-974 (mailed October 28, 2011). As for an extraordinary blocking circumstance, that can be found in the ill-advised handling of Blount’s first petition, which the district court (along with this court) construed too narrowly and dismissed too readily, rather than holding it in abeyance for Blount to exhaust his District remedy.
Blount had not exhausted his Strickland claim before he brought his first petition. But “pleading errors may be corrected through the liberal construction or amendment we are accustomed to providing a pro se prisoner.” Davis v. U.S. Sentencing Comm’n, 716 F.3d 660, 667 (D.C. Cir. 2013). Blount’s Strickland claim was intertwined with one that had been exhausted in District post-conviction proceedings: that the jury charge was unconstitutional. See Blount v. Wilson, No. 11-7060, at 1-2 (noting that the first petition appeared to raise both claims). Blount had already challenged those instructions via the District’s procedure for post-conviction relief, D.C. Code § 23-110. See Blount II at 1. Although the D.C. Court of Appeals acknowledged that the instructions were “improper,” it affirmed the denial of that challenge because he had not raised the issue on direct appeal. Id. When his federal habeas petition is read to focus on that claim, the Strickland argument still plays a role; it serves to excuse his procedurally defaulting in his direct appeal. See Edwards v. Carpenter, 529 U.S. 446, 451, 120 S.Ct. 1587, 146 L.Ed.2d 518 (2000). Blount’s first petition then, was a mix of both exhausted and unexhausted claims. And “it likely would be an abuse of discretion for a district court to deny a stay and dismiss a mixed petition if the petitioner had good cause for his failure to exhaust, his unexhausted claims are potentially meritorious, and there is no indication that the petitioner engaged in intentionally dilatory litigation tactics.” Rhines, 544 U.S. at 278, 125 S.Ct. 1528. Blount’s mixed petition satisfied each Rhines element.
First, Blount’s unexhausted claim had considerable merit; second, his failure to exhaust was due, at least in part, to the District’s byzantine process for pursuing Strickland claims of appellate representation; and third, he did not—and had no incentive to—engage in “dilatory litigation tactics,” as those tactics would prolong his time in prison. Dismissing Blount’s first petition also ran counter to our precedent on unexhausted Strickland claims flowing to us from the D.C. Court of Appeals. In Streater v. Jackson, 691 F.2d 1026 (D.C. Cir. 1982), a petitioner for federal habeas for ineffective assistance of appellate counsel had been stymied by the then prevailing uncertainty in the District over where such a claim could get review. Recognizing the interaction between that confusion and the requirement of exhausting state remedies, we held the federal petition in abeyance so that the petitioner could pursue whatever the District might offer. Id. at 1028. Both Rhines and Streater show that Blount’s first petition should have been stayéd, not dismissed.
Equitable tolling would correct this unfortunate mistake and retroactively stop the clock during the pendency of Blount’s first petition and during the reasonable amount of time it took Blount to file his motion to recall and his second federal habeas petition. The district court was in*749correct to deny Blount equitable tolling, and its decision should be reversed.
Blount’s Jimenez claim also warrants reversal. The D.C. Court of Appeals views its procedure for Strickland-based motions to recall the mandate (its sole remedy for constitutionally defective appellate counsel) as a two-step process. There is a decision on the motion itself, and then, if the motion is granted, full consideration of the Strickland claim and the merits of the recalled appeal. See Long v. United States, 83 A.3d 369, 378 & n.15 (D.C. 2013).
The parties here appear to agree that the first stage does not qualify as a reopening under Jimenez but that the second does. That shared assumption seems a reasonable reading of Jimenez, where the Court specifically found that, as a matter of plain meaning, ' 28 U.S.C. § 2244(d)(l)(A)’s trigger for the AEDPA’s one-year habeas limitations period—“the date on which the judgment became final by the conclusion of direct review or the expiration of the time for seeking such review”—encompasses not only standard conclusion of direct review but also instances where a state court re-opens and re-concludes direct review. 555 U.S. at 119-20,129 S.Ct. 681.
But as Long itself illustrates, the D.C. Court of Appeals may collapse into a single proceeding its grant of the motion to recall and its adjudication of the re-opened appeal. 83 A.3d at 378 n.15.1 A composite process would also count as re-opening direct review, and Blount claims this occurred in his case. In Long the movant ultimately prevailed in the reopened appeal, but there is nothing preventing the use of Long's streamlined process in cases
where the movant ultimately loses (as did Blount). The two steps in the D.C. Court of Appeals’s process impose successively increasing burdens on the defendant: he need only show “sufficient merit” to pass the first stage but can only prevail in the second stage after the court “fully explores” his Strickland claim. Griffin v. United States, 598 A.2d 1174, 1175-76 (D.C. 1991) (emphasis and internal quotation marks omitted). As a matter of logic, a court can, in a single proceeding, find that a movant clears the lower hurdle but not the higher one. Indeed, this ladder of increasing burdens should sound familiar: it is the process AEDPA’s COA provision calls for us to follow here—analyze the COA and then (if we grant the COA) address the merits.
My colleagues recognize the possibility of a combined grant-and-reject process. See Maj. Op. at 738-39. But they insist that when the D.C. Court of Appeals claims it has dismissed a motion at the first stage, we must end our inquiry, even if that court’s “reasoning ... mirrors the reasoning it would use on a,reopened appeal.” Id. at 739. They reject Blount’s “notion” of analyzing what a state court actually did, preferring to treat as final what it said it did. See id.
Blount’s “notion,” though, flows from the Supreme Court’s command: we are to analyze the state court decision to see if “in fact” it re-opened the appeal. Jimenez, 555 U.S. at 120 n.4, 129 S.Ct. 681. “[F]or purposes of applying” AEDPA provisions “that interact! 1 with state procedural rules, we look to how a state procedure functions, rather than the particular name that it bears.” Carey v. Saffold, 536 U.S. *750214, 223, 122 S.Ct. 2134, 153 L.Ed.2d 260 (2002) (emphasis added). Federal courts applying AEDPA are admonished not to take what the state court said it was doing “as an absolute bellwether” for what it actually did. Id. at 226, 122 S.Ct. 2134. Doubtless there may be close cases, such as this one, but accepting the burden of adjudicating these cases is, in the Supreme Court’s judgment, radically preferable to willful ignorance. Closing ourselves off to reality does no honor to state courts and no justice to habeas petitioners.
Still, my colleagues contend that horrendous consequences would flow from treating a denial of a motion to recall the mandate as a reopening of the appeal— even in cases where the state court’s reasoning “mirrors” what it “would use on a reopened appeal.” Maj. Op. at 739-40. The majority envisions prisoners evading time limits altogether as they endlessly filed frivolous motions to recall the mandate. See id. But state and federal courts have formidable tools to prevent such gamesmanship. First, the D.C. Court of Appeals invites little chance of re-opening when it follows its standard practice of dismissing motions to recall when they are facially insufficient. Here though, the court issued multiple orders, noted the apparent sufficiency of Blount’s claims, and applied the legal standard appropriate for a re-opened appeal. See Dissent, infra, at 740-41. Second, there is the District’s 180-day deadline for motions to recall the mandate, after which the D.C. Court of Appeals may dismiss a motion procedurally with no hint of re-opening. D.C. Ct. App. R. 41(f). The court waived that bar in Blount’s case, further indicating its desire to plumb the merits. Finally, any fear of meretricious reopening must assume extraordinary sloppiness on the part of federal courts in interpreting state court decisions. We are quite capable of comprehending that though all cats have four legs (with a few exceptions), not all animals with four legs are cats. Characterizing a motion to recall as frivolous may “mirror[ ]” a merits dismissal for frivolity, but the reasoning would be inconsistent with any idea that the court went on to the second stage of its procedure. Compare Maj. Op. at 739-40.
I would follow the Supreme Court’s example in Carey, looking to the substance of the D.C. Court of Appeals’s decision, not its form. Judged from this perspective, the decision almost certainly re-opened Blount’s appeal. After Blount filed his motion, the D.C. Court of Appeals released an order noting that “the jury may have relied upon the aiding and abetting instruction which included the ‘reasonable and probable consequences’ language this court held to be unconstitutional”; requesting a response from the government; and concluding that Blount “appear[ed]” to have raised a viable Strickland claim. Order, Blount v. United States, No. 01-CF-974 (D.C. Aug. 29, 2012). Thus the court seems to have concluded that “on their face” Blount’s claims had “sufficient merit,” see Watson v. United States, 536 A.2d 1056, 1060 (D.C. 1987) (plurality); see Griffin, 598 A.2d at 1176, and that it was necessary “to consider more fully and then decide” whether a Strickland violation had occurred, see Stratmon v. United States, 631 A.2d 1177, 1178 (D.C. 1993) (internal quotation marks omitted). That is precisely the reasoning the D.C. Court of Appeals uses when it “recall[s] the mandate.” See Stratmon, 631 A.2d at 1178.
Consistent with this interpretation of the first order, the court’s next order on Blount’s motion appears to involve the type of “full[ ] exploration]” one expects in a recalled appeal. See Griffin, 598 A.2d at 1176. As mentioned in my discussion of Blount’s Strickland claim, the second order invoked the precise legal test, harmless error, that should govern the reopened appeal. See Blount III at 1. Thus it *751went well beyond what would have been necessary simply to deny recall of the mandate. Compare Maj. Op. at 738-39. To be sure, the D.C. Court of Appeals’s precedents on this issue are somewhat confusing. Despite case law asserting that the motions to recall are decided “on their face,” Watson, 536 A.2d at 1060, the D.C. Court of Appeals once ordered “an eviden-tiary hearing by the trial court” on such a motion, Stratmon, 631 A.2d at 1180. But the best reading of the decisional array appears to be that the D.C. Court of Appeals’s disposition of Blount’s motion comprised both a decision to recall and a decision on the merits. Thus this panel would have done well to find Blount’s petition timely by virtue of Jimenez as well as the equities (and certainly within the realm of reasonable disagreement).
Finding grounds for granting the COA and for ultimately finding Blount’s petition not time-barred, I would reverse on the timing issue and remand for the district court to address the merits of the constitutional claim. See Mickles-El v. D.C. Gov’t, No. 98-7019, 1998 WL 846649, at *1 (D.C. Cir. Nov. 24,1998).
I respectfully dissent.
. There are, practically speaking, three steps combined in Long: an initial Strickland inquiry when reviewing the motion to recall the mandate; the post-recall complete Strickland analysis; and the adjudication of the claim that should have been brought in the original appeal. Id. at 378-84 (performing initial Strickland analysis, full Strickland analysis, and plain error analysis simultaneously, then remanding to district court because there was plain error). Performance of either of the latter two steps would indicate re-opening.