dissenting:
As the majority aptly explains, Maj. Op. at 1028-81, the current state of Congress’s back-and-forth with the courts over federal jurisdiction to consider claims by detainees at Guantanamo is this: claims that sound in habeas may be heard; all others may not. Today we decide which category embraces a challenge to a detainee’s conditions of confinement. The majority concludes that such a claim sounds in habeas. I disagree. Although we once toyed with that idea (in dictum), we have never held habeas to reach a prisoner’s conditions of confinement. And the majority provides no persuasive reason why we should reach that decision for the first time today. Congress has repeatedly and forcefully sought to withdraw the federal courts’ jurisdiction over Guantanamo detainees. I would not enlarge the writ to encompass a novel theory in the face of such clear congressional intent.
The Supreme Court’s most recent position on whether habeas encompasses prisoner challenges to their conditions of confinement has been one of agnosticism. Maj. Op. at 1031-32 (citing Preiser v. Rodriguez, 411 U.S. 475, 93 S.Ct. 1827, 36 L.Ed.2d 439 (1973), and Bell v. Wolfish, 441 U.S. 520, 99 S.Ct. 1861, 60 L.Ed.2d 447 (1979)). The majority thus turns to decisions of this court, finding Hudson v. Hardy, 424 F.2d 854 (D.C.Cir.1970), a precedent for the view that habeas covers such claims. Maj. Op. at 1032-34. I find no such holding in Hudson.
Hudson’s background is simple. In an action styled a petition for a declaratory judgment, Hudson sought an order granting him certain privileges, release from a control cell, or outright release from custody. In our initial pass at the case, we held that the district court had been too hasty in granting summary judgment against Hudson, applying the standards for summary judgment with a “strict literalness” that was inappropriate for a pro se prisoner. Hudson v. Hardy, 412 F.2d 1091, 1094-95 (D.C.Cir.1968). We then granted the defendants’ request for rehearing, in order to consider not only the merits but also defendants’ claim that Hudson’s transfer to Leavenworth had mooted the case. Hudson, 424 F.2d at 855-56. We adhered to our initial decision that the district court had been too hasty, but added an instruction to that court to canvas the facts relevant to mootness. In articulating the district court’s mission on remand, we discussed a number of circumstances that might avert dismissal for mootness.
*1045There are many reasons to reject the view that our theorizing in Hudson established a precedent extending habeas to conditions of confinement — so many that the reader deserves a short road map. First, we left completely unresolved the question whether the federal courts had jurisdiction at all; that being so, we were in no position to issue a final merits ruling. Second, we noted that 42 U.S.C. § 1983 was available to the plaintiff; we thus had no need to examine whether § 1983 or habeas best fitted plaintiffs claims, to the extent that they might have related to conditions of confinement. Third, of the various circumstances that we suggested might save the case from mootness, it is doubtful whether any can properly be characterized as involving “conditions of confinement” (a phrase we never used in Hudson).
First we noted that if plaintiff sought money damages, the case was not moot. 424 F.2d at 855. But because we were uncertain whether he sought damages, we went on to discuss the situation if he did not, saying that even in that case “it is by no means certain that the case has become moot.” Id. In remanding to the district court, we identified a handful of reasons Hudson might face ongoing injury; not one of those reasons depended on our observation about habeas, id. at 855 n. 3, which was therefore dictum rather than holding.
Indeed, nowhere in the opinion did we purport to actually find jurisdiction. Although we didn’t explicitly invoke the principle and practice that in determining jurisdiction a court assumes the validity of plaintiffs merits claims, see, e.g., Coleman v. Miller, 307 U.S. 433, 446, 59 S.Ct. 972, 83 L.Ed. 1385 (1939); Doe v. Harris, 696 F.2d 109, 113-14 n. 7 (D.C.Cir.1982); Smith v. Bd. of Comm’rs of D.C., 380 F.2d 632, 634 (D.C.Cir.1967); see also Linda R.S. v. Richard D., 410 U.S. 614, 618, 93 S.Ct. 1146, 35 L.Ed.2d 536 (1973) (applying the principle without stating it), we certainly never abjured the principle, and our discussion was fully consistent with the practice.
And with good reason. For us to have applied substantive law before finding jurisdiction would have flouted the rule that on any appeal “the first and fundamental question is that of jurisdiction.” Steel Co. v. Citizens for a Better Environment, 523 U.S. 83, 94, 118 S.Ct. 1003, 140 L.Ed.2d 210 (1998) (quoting Ex parte McCardle, 7 Wall. 506, 514, 19 L.Ed. 264 (1869)). “Every federal appellate court has a special obligation to satisfy itself not only of its own jurisdiction, but also that of the lower courts in a cause under review.” Id. at 95, 118 S.Ct. 1003 (internal quotation marks omitted). The rule long antedated Hudson. “This Court’s insistence that proper jurisdiction appear begins at least as early as 1804.” Id. And the principle applies as much to mootness as to any other issue of subject-matter jurisdiction. Already, LLC v. Nike, Inc., - U.S. -, 133 S.Ct. 721, 726-27, 184 L.Ed.2d 553 (2013). Of course jurisdiction may depend on the merits claims; the longstanding solution is to assume the merits of plaintiffs position. We have no basis for now declaring that our decision in Hudson deviated from these principles: jurisdiction first, and for that analysis, merits merely assumed.
Recognizing that an absence of jurisdiction would preclude Hudson from having precedential effect, the majority seeks to characterize the merits of Hudson’s claim as itself a jurisdictional question, so that the Hudson court (permissibly) resolved it before resolving mootness. Maj. Op. at 1033-34. Some elements of 28 U.S.C. § 2241 doubtless are jurisdictional. For example, issuance of the writ requires (absent waiver) personal jurisdiction over the *1046custodian, Rumsfeld v. Padilla, 542 U.S. 426, 434 & n. 7, 442, 124 S.Ct. 2711, 159 L.Ed.2d 513 (2004); id. at 451-52, 124 S.Ct. 2711 (Kennedy, J., concurring), and subject-matter jurisdiction depends on the petitioner’s being in custody, Maleng v. Cook, 490 U.S. 488, 490, 493-94, 109 S.Ct. 1923, 104 L.Ed.2d 540 (1989). But that does not mean that a claim’s cognizability under habeas is also jurisdictional. While the Court in Rasul v. Bush, 542 U.S. 466, 124 S.Ct. 2686, 159 L.Ed.2d 548 (2004) (decided the same day as Padilla), rejected the government’s defense that federal courts lacked habeas jurisdiction beyond the United States’ sovereign territory, it nowhere suggested that the merits of whether a claim is cognizable in habeas is itself jurisdictional. The majority’s attempt to analogize habeas to federal question jurisdiction similarly confuses merits and jurisdiction. Where a claim “will be sustained if the Constitution and laws of the United States are given one construction and will be defeated if they are given another,” the issue is one of merits, not jurisdiction. Steel Co., 523 U.S. at 89, 118 S.Ct. 1003 (quoting Bell v. Hood, 327 U.S. 678, 685, 66 S.Ct. 773, 90 L.Ed. 939 (1946)).
The majority reads the discussion of potential alternative grounds for jurisdiction in Bell v. Wolfish to stand for the proposition that the scope of habeas is normally a jurisdictional issue. Maj. Op. at 1033-34 (citing 441 U.S. at 527 n. 6, 99 S.Ct. 1861). All I can extract with confidence from that footnote is that in meeting its obligation to be sure of jurisdiction, the Court found 28 U.S.C. § 1331 sufficient and thus saw no need even to consider whether habeas’s scope was a merits or a jurisdictional question. Especially given the courts’ loose usage of “jurisdiction” before the last decade, see Kontrick v. Ryan, 540 U.S. 443, 124 S.Ct. 906, 157 L.Ed.2d 867 (2004), I am skeptical that offhand references to jurisdiction in a footnote prove that habeas’s exact scope is a jurisdictional question. Nor does the majority’s reference to Kiyemba v. Obama, 561 F.3d 509 (2009), shed light on the correct reading of Hudson; our holding there merely reflects Congress’s enactment of § 2241(e)(2) and the. Supreme Court’s holding in Boumediene v. Bush, 553 U.S. 723, 128 S.Ct. 2229, 171 L.Ed.2d 41 (2008), which together made the scope of habeas a jurisdictional issue for Guantanamo detainees by divesting the courts of jurisdiction to grant any non-habeas relief. It tells us nothing about whether a court’s reading of the habeas statute in effect in the Hudson era had any jurisdictional character.
Given that the Hudson court never suggested that its ruminations on Hudson’s possible causes of action touched on jurisdiction, and that the habeas statute in effect at the time used no jurisdictional language, see 28 U.S.C. § 2241 (1970), there seems no reason to suppose that Hudson’s decision to remand for a mootness determination constituted a resolution of any jurisdictional questions that habeas may entail.
But even if we put the jurisdictional question aside, Hudson’s claim that officials of the District of Columbia had subjected him to “ ‘unjust and cruel’ disciplinary action,” 412 F.2d at 1092, was, as to state officials and those of the District of Columbia, the sort of claim that could be brought under 42 U.S.C. § 1983. E.g., Edwards v. Sard, 250 F.Supp. 977, 978 (D.D.C.1966). Twice in the opinion we explicitly recognized the availability of § 1983. 424 F.2d at 855 & n. 3. Once again, I see precisely nothing that turned on classifying Hudson’s claims — insofar as they may have addressed conditions of confinement — as sounding in habeas.
Finally, regardless of the unresolved status of our jurisdiction and the availabili*1047ty of § 1983, Hudson is quite unclear whether it addresses conditions of confinement at all. In order to “sketch” the principles we thought should guide the mootness inquiry, we reviewed a number of possible claims that might have survived cessation of the allegedly unlawful conduct and the plaintiffs removal from the defendants’ reach. Id. at 856. Those claims stemmed from the fact that a prisoner’s “disciplinary record may follow him throughout the prison system.” Id. It therefore might “affect his eligibility for parole.” Id. Such eligibility of course presents a classic subject of habeas, a claim that would be “squarely within th[e] traditional scope of habeas corpus,” Preiser, 411 U.S. at 487, 93 S.Ct. 1827. See Peyton v. Rowe, 391 U.S. 54, 88 S.Ct. 1549, 20 L.Ed.2d 426 (1968) (cited by Hudson, 424 F.2d at 856). We also mused that Hudson’s prior discipline might compound his future punishment, 424 F.2d at 856. But as an example of future punishment we pointed to Hudson’s transfer to the prison at Leavenworth. Id. at 856 & n. 8. As we had decided a few years earlier that habeas is available to challenge “not only the fact of confinement but also the place of confinement,” Lake v. Cameron, 364 F.2d 657, 659 (D.C.Cir.1966) (emphasis added), this may well have been the root of our speculation that relief was still possible. We concluded that “[i]f [Hudson] desires that the case be treated as a petition for habeas corpus, the court should inform itself of the extent to which appellant is, or is likely to be, still subject to disabilities because of the unlawful acts alleged.” 424 F.2d at 856. Assuming we reached a holding on habeas, it was that it encompassed the “disabilities” we specified — examples fully in line with historical habeas practice — not the loose talk about “form of confinement” that we consigned to a- footnote, id. at 855 n. 3.
In short, in framing the district court’s future jurisdictional inquiry, we tossed up a salad of possible merits claims. This was a perfectly proper way to guide the district court’s exploration of mootness. But that does not mean that any of these speculations constituted a holding. Even assuming the court meant habeas to encompass conditions of confinement, we had neither jurisdiction nor occasion to settle any substantive legal issue, and in the two short pages of F.2d that Hudson occupies (other than caption, headnotes, etc.), I do not see that we did so.
And I also agree with the majority’s acknowledgement that its other cases fail to do so. Maj. Op. at 1034. Two of the cases, Miller v. Overholser, 206 F.2d 415 (D.C.Cir.1953), and Creek v. Stone, 379 F.2d 106 (D.C.Cir.1967), see Maj. Op. at 1034-35, address conditions of confinement that “vitiate the justification for confinement.” 379 F.2d at 109, 206 F.2d at 419 (claim renders confinement “not authorized by the statute”). If the basis for confinement is eliminated altogether, outright release would be the remedy, and the petition would fall within the mine run of habeas challenges. Miller, for example, involved a civil commitment statute intended to rehabilitate sex offenders. The court held that Miller’s confinement with the criminally insane and without treatment was therefore a confinement “not authorized by the statute,” rendering his confinement illegal. 206 F.2d at 419. But the decision was clear that it would apply only in cases that challenge the “legal validity of confinement,” id., which petitioners do not do. Cf. Maj. Op. at 1040-41.
Creek too depends on the proposition that the challenged conditions vitiate the justification for confinement. 379 F.2d at 110. And perhaps more importantly, the court in Creek lacked jurisdiction because the petitioners’ transfer out of the condi*1048tions complained of rendered the case moot. Id. By now it should be clear that the absence of jurisdiction is a common theme among the majority’s cases. United States v. Wilson, 471 F.2d 1072 (D.C.Cir.1972), is no different. The majority relies on Wilson for the proposition that petitioner “unquestionably” may challenge his conditions of confinement in ha-beas. Maj. Op. at 1034-35 (quoting Wilson, 471 F.2d at 1081). Yet right before that observation, the court held that it lacked jurisdiction over Wilson’s petition, concluding that “no remedy is available in this Court” because the place of confinement was not within its territorial jurisdiction. 471 F.2d at 1081 (citing Ahrens v. Clark, 335 U.S. 188, 68 S.Ct. 1443, 92 L.Ed. 1898 (1948)). Here again Steel Co.’s teachings are critical. “For a court to pronounce upon the meaning or the constitutionality of a state or federal law when it has no jurisdiction to do so is, by very definition, for a court to act ultra vires.” Steel Co., 523 U.S. at 101-02, 118 S.Ct. 1003. I see no basis for relying on ultra vires statements to determine the appropriate bounds of habeas.
More recent cases from this circuit suggest that the availability of habeas to challenge conditions of confinement is a murkier question than the majority’s cases suggest. In Blair-Bey v. Quick, for example, we entertained the possibility that habeas itself “might be available” for challenges to prison conditions. 151 F.3d 1036, 1039-42 (D.C.Cir.1998) (Wald, Williams & Tatel, JJ.) (emphasis added). We solved the problem by saying, “Such claims, if they are permissibly brought in habeas corpus, would have to be subject to the PLRA’s filing fee rules.” Id. at 1042 (emphasis added). It would be odd to treat 1998’s mere possibility as a rule clearly established by the time of the MCA, less than ten years later. Yet that is exactly the conclusion that the majority reaches.
Because not one of these cases holds that habeas encompasses claims based on the conditions of a detainee’s confinement, I conclude that no precedent of ours controls the outcome of this case.
The majority soft-pedals the distinction between challenges to the fact or place of confinement and ones to conditions of confinement by observing that one remedy unquestionably available under habeas (in this case, the prisoner’s release) “may” redress both claims, so that the distinction between the claims is “largely illusory.” Maj. Op. at 1035-36. After all, the majority explains, a court can always order release if the petitioner’s custodian does not remedy the defect in the place of confinement. But to suggest that courts should feel complacent in expanding an ancient writ — confined for centuries to attacks on the fact or place of confinement — to reach any unlawful aspect of the confinement merely because the illegality could, in ex-tremis, be cured by an order of release, seems in effect to discard history as a guide.
In any event, a focus on remote, unsatisfactory and implausible remedies of release is a far cry from how an inquiry into the availability of habeas normally proceeds. As the majority observes, Maj. Op. at 1029-30, in determining the scope of our jurisdiction in Kiyemba we first needed to assess the effect of the Supreme Court’s decision in Boumediene. Notwithstanding the fact that an order of release would have sufficed to grant the Kiyemba petitioners’ requested relief, we reviewed the congressional authorization to consider their claims as well as the traditional boundaries of habeas. Kiyemba, 561 F.3d at 512-13. In doing so, we followed the path laid out by the Supreme Court, which *1049determines its habeas authority not by reference to the potential remedy, but instead by reviewing historical practice under statutory and common law. INS v. St. Cyr, 533 U.S. 289, 305-08, 121 S.Ct. 2271, 150 L.Ed.2d 347 (2001); see Boumediene, 553 U.S. at 739-52, 128 S.Ct. 2229; Rasul, 542 U.S. at 473-84, 124 S.Ct. 2686. Thus the majority’s focus on remedies seems completely orthogonal to the method by which the Supreme Court determines the limits of habeas; it should not influence our decision here.
This case itself illustrates the skewed fit between a substantive attack on conditions of confinement and a remedy of release. The petitioners understandably never seek “the writ’s more traditional remedy of outright release,” Maj. Op. at 1035. See J.A. 1, 3 (requesting injunction prohibiting force-feeding); J.A. 158 (requesting injunction prohibiting alleged deprivation of right to communal prayer); Aamer Br. 5 (requesting both forms of injunctive relief). And the majority, rightly acknowledging the legality of the petitioners’ detention, Maj. Op. at 1040-41, focuses only on whether to enjoin the practice of force-feeding. The theoretical effectiveness of an implausible remedy seems a thin basis for shoehorning litigation over conditions of confinement into habeas.
Under the majority’s view, it need not consider petitioners’ alternative theories supporting jurisdiction, but I must. They are no more convincing. Relying on a pair of cases from the Seventh Circuit, petitioners first contend that because the force-feeding protocol requires transfer from “communal living quarters” to “single cell operations,” it constitutes a quantum change in his level of custody, rendering his petition cognizable in habeas. Aamer Br. 24-25. They quote the Seventh Circuit as follows: “If the prisoner is seeking what can fairly be described as a quantum change in the level of custody — whether outright freedom, or freedom subject to the limited reporting and financial constraints of bond or parole or probation, or the run of the prison in contrast to the approximation to solitary confinement that is disciplinary segregation — then habeas corpus is his remedy.” Graham v. Broglin, 922 F.2d 379, 381 (7th Cir.1991). But they omit the critical next sentence: “if [the petitioner] is seeking a different program or location or environment, then he is challenging the conditions rather than the fact of his confinement and his remedy is under civil rights law, even if, as will usually be the case, the program or location or environment that he is challenging is more restrictive than the alternative that he seeks.” Id. at 381. Putting aside the fact that this court has not recognized the “quantum change” theory, the petition here falls squarely within the claims described by the second sentence; he seeks only an alteration to his program and thus his claims sound in civil rights law, even under Graham. Accordingly, § 2241(e)(2) bars Aamer’s claims. Al-Zahrani v. Rodriguez, 669 F.3d 315, 319 (D.C.Cir.2012).
Aamer also asserts that we have jurisdiction because force-feeding constitutes a “severe restrain^] on individual liberty.” Aamer Br. at 26-27 (citing Hensley v. Mun. Court, San Jose Milpitas Judicial Dist., Santa Clara Cnty., 411 U.S. 345, 351, 93 S.Ct. 1571, 36 L.Ed.2d 294 (1973)). While it is true that habeas’s custody requirement can be met by restraints falling short of incarceration (in Hensley, the Court allowed a petition by a defendant released on his own recognizance pending the start of his challenged sentence), it doesn’t follow that any liberty interest enjoying protection under the Fifth Amendment, e.g., the interest in resisting involuntary subjection to pharmaceuticals, Sell v. *1050United States, 539 U.S. 166, 177-83, 123 S.Ct. 2174, 156 L.Ed.2d 197 (2003), is a liberty protected by habeas. Petitioners’ characterization of the affected values as liberty interests is therefore not enough to create habeas jurisdiction. See Janko v. Gates, No. 12-5017, 741 F.3d 136, 145-47, 2014 WL 184729, *7-8 (D.C.Cir. Jan. 17, 2014).
I close with a brief consideration of where we are and how we got here. In § 7 of the MCA Congress sought to all but extinguish federal courts’ jurisdiction to hear claims by detainees at Guantanamo. Subsection 2241(e)(1) purported to remove all habeas corpus jurisdiction over such aliens; subsection (e)(2) eliminated all other jurisdiction, except for the judicial process that Congress had previously established for review of executive branch decisions on the lawfulness of detention. See Maj. Op. at 1028-29. By the two subsections taken together, then, Congress sought (with the exception noted) to exclude Guantanamo detainees from United States courts.
In Boumediene the Supreme Court held that subsection (e)(1) violated the Suspension Clause, 553 U.S. at 733, 771, 792, 128 S.Ct. 2229, but it did not address whether the MCA was valid insofar as it ousted courts from habeas jurisdiction not protected by that clause. We answered this question in Kiyemba, 561 F.3d at 512 n. 2, interpreting Boumediene to have struck down the MCA’s attempt to withdraw ha-beas jurisdiction over detainees at Guantanamo in its entirety, regardless of whether the suspension clause required that invalidation. That decision of course binds this panel, though it self-evidently opens the courts to actions by detainees under circumstances where Congress intended that they be shut (this case being one example), and was concededly not in any way compelled by Boumediene. See id. at 512; id. at 523 (Griffith, J., concurring in the judgment in part and dissenting in part). Yet although Kiyemba restored habeas to its “status quo ante” the MCA, id. at 512 n. 2, nothing in that decision requires that we further expand the writ to encompass ha-beas claims that did not predate the MCA.
The majority does precisely that. To determine just how much the courts will open themselves to habeas independently of the Suspension Clause’s constitutional pressure, it relies on our reflections in Hudson — a case in which we had not found any federal court jurisdiction, where any discussion of the fine points of habeas versus § 1983 was unnecessary, and where it is unclear whether our speculation about plaintiffs claims even addressed conditions of confinement. Because neither Hudson nor any other case of ours establishes the availability of a conditions of confinement claim under habeas, Kiyemba’s restoration of the status quo ante does not compel us to recognize such a claim. And Congress has made quite clear that we shouldn’t. Subsection (e)(1) may be a dead letter, but that does not compel us to ignore Congress’s intent behind subsection (e) as a whole, which unmistakably sought to prevent the federal courts from entertaining claims based on detainees’ conditions of confinement. Cf. Janko v. Gates, No. 12-5017, 741 F.3d at 141-42 & n. 4, 2014 WL 184729, *4 (looking to (e)(1) to interpret (e)(2)). Such evident congressional intent would seem to counsel a cautious rather than a bravura reading of Hudson.
Respectfully dissenting, I would affirm the district courts’ dismissal of the petitions for want of jurisdiction.