concurring and dissenting:
I agree with the majority that the government properly invoked the state secrets privilege, the privilege applies in a Bivens action, and the district court properly dismissed Horn’s complaint as to Defendant II.1 But we disagree about the proper legal standard for determining when application of the privilege requires dismissal. The majority’s reversal of the district court’s decision pushes this circuit’s state secrets jurisprudence in a new and troubling direction — one at odds with all other circuits that have considered the issue. Because, in my view, the district court reached the correct conclusion, I respectfully dissent.
I
After the government successfully invoked the state secrets privilege, the district court dismissed Horn’s complaint on three independent grounds: (1) because Horn cannot make out a prima facie case absent the privileged material, (2) because the privilege deprives the defendants of information required in their defense, and (3) because “the very subject matter of plaintiffs action is a state secret.” Mem. Op. 8. The majority approves the dismissal of Defendant II, but otherwise rejects the district court’s conclusions on all three grounds.
As to the district court’s first ground— Horn’s ability to establish a prima facie case against Huddle absent the privileged material — I generally agree with the legal standard applied by the majority. But I am less sanguine than the majority that the unprivileged facts actually suffice to make a prima facie showing. Once the privileged material is removed, Horn is essentially left with three pieces of circumstantial evidence — a cable, a table, and Huddle’s apparent lie. I question whether a reasonable person would seriously entertain the possibility, based on that evidence alone, that Huddle learned of Horn’s statement via a wiretap. One wonders if the atmosphere of government intrigue in this case — an atmosphere carefully cultivated by Horn and unfortunately only exacerbated by the government’s invocation of the *155state secrets privilege — is in fact doing much of the work in the majority’s determination that Horn has established a pri-ma facie case on such skimpy evidence. Would a reasonable person really think Horn had established a prima facie case with the same circumstantial evidence if he was an OSHA inspector in Hoboken?
A
But while I remain skeptical that Horn has a prima facie case once the privileged material has been removed, my fundamental disagreement with the majority relates to the controlling legal standard applicable to the district court’s second and third grounds for dismissal. The majority does not expressly disagree with the district court’s conclusion that, as a result of the invocation of the privilege, Huddle will be deprived of information necessary to mount an effective defense. Instead, relying on Molerio v. FBI, 749 F.2d 815 (D.C.Cir.1984), and dicta from Ellsberg v. Mitchell, 709 F.2d 51 (D.C.Cir.1983), the majority concludes dismissal is inappropriate unless the privileged material contains a defense so “dispositive” as to “require judgment for the defendant.” Maj. Op. 149 & n. 4.
To be sure, the privileged defense in Moleño was dispositive, and the court dismissed on that basis. But the court in Moleño merely determined that a disposi-tive defense is a sufficient basis for dismissal, not that a privileged defense must be dispositive for dismissal to be appropriate. See Molerio, 749 F.2d at 825. Because Moleño was the paradigmatic “easy case,” it is unhelpful in establishing the proper standard for harder cases presenting less than dispositive, but nonetheless meritorious, defenses.
This is one of those harder cases, and the majority has responded by borrowing Moleño’s description of an extreme (and therefore easy) case to establish the new baseline for dismissal. In so doing, the majority relies extensively on speculative language from Ellsberg about the effect of the state secrets privilege on privileged defenses. See Maj. Op. 148-51. Thus, the majority purports to apply “long-settled precedent” in precluding all but dispositive privileged defenses. Maj. Op. 150. But as both the majority and dissent in that case recognized, Ellsberg’s sweeping comments about privileged defenses were dicta. See Ellsberg, 709 F.2d at 64 (noting that in light of its disposition, further discussion was not necessary to resolve the case, but that the court “considered] it prudent to address briefly some of the problems the trial court will confront on remand”); id. at 73 (MacKinnon, J., concurring in part and dissenting in part) (characterizing the part of the Ellsberg opinion relied on by the majority here as “muddled dicta” presenting “novel procedures” that “I would be very surprised if the court on remand even attempted to apply” in light of a likely alternative resolution of the case); see also Cohens v. Virginia, 19 U.S. (6 Wheat.) 264, 399, 5 L.Ed. 257 (1821) (“[G]eneral expressions, in every opinion, are to be taken in connection with the case in which those expressions are used. If they go beyond the case, they may be respected, but ought not to control the judgment in a subsequent suit when the very point is presented for decision.”).
Ellsberg’s abstract discussion of privileged defenses is only controlling, therefore, to the extent it is persuasive. As the majority in Ellsberg itself recognized, the potential for “serious injustice” arises when defenses are compromised by the government’s invocation of the state secrets privilege:
Deprived of the ability in practice to adduce the evidence necessary to mount *156a defense to the plaintiffs’ prima facie case, the defendants could be held liable in damages for what in fact was wholly blameless conduct. Such a result not only would be patently inequitable, but might have an unfortunate long-run impact on the recruitment and behavior of government officials.
Ellsberg, 709 F.2d at 69 (footnote omitted). The Ellsberg majority speculated this “serious injustice” might be ameliorated by “recent developments” in the doctrine of qualified immunity. Id. But, as the majority here recognizes, the doctrine of qualified immunity does nothing for Huddle. Maj. Op. 148. We are thus left with the “serious injustice” identified in Ellsberg, without the supposed “way out of th[e] dilemma” envisioned by the Ellsberg majority. Ellsberg, 709 F.2d at 69. Dictum lacks binding precedential value precisely because abstract musings often fail to produce fully-considered legal rules. When “we accept dictum uttered in a previous opinion as if it were binding law, which governs our subsequent adjudication ..., we fail to discharge our responsibility to deliberate on and decide the question which needs to be decided.” Pierre N. Leval, Judging Under the Constitution: Dicta About Dicta, 81 N.Y.U. L. Rev. 1249, 1250 (2006).
Indeed, perhaps recognizing the potential for “serious injustice” identified in Ellsberg, no other circuit has adopted the severe defense standard applied by the majority here. New state secrets cases have been resolved on the defense prong. How to treat privileged defenses is an exceedingly difficult question, and most cases that have presented defense issues have been dismissed on other grounds. See, e.g., Bareford v. Gen. Dynamics Corp., 973 F.2d 1138, 1143 (5th Cir.1992). But the few circuits to address squarely the issue have not applied the defendant-adverse standard favored by the majority. See, e.g., El-Masri v. United States, 479 F.3d 296, 309-10 (4th Cir.2007) (dismissing in the alternative on the ground that “the defendants could not properly defend themselves without using privileged evidence”); Tenenbaum v. Simonini, 372 F.3d 776, 777 (6th Cir.2004) (dismissing because “Defendants cannot defend their conduct with respect to [the plaintiff] without revealing the privileged information”).
By equating a “valid” defense with a “dispositive” defense, and noting that “[o]ther circuits have ... reified] upon Molerlo to adopt the ‘valid defense’ standard,” Maj. Op. 149, the majority papers over the novelty of the defense standard it is applying. True, other circuits have referenced the “valid defense” standard, and cited Molerlo in support, but it is not at all clear that in doing so they interpreted “valid” as meaning “dispositive,” as opposed to “valid” as meaning simply “meritorious.” See BlaoK’s Law Diotionaey 1586 (8th ed.2004) (defining “valid” as both “[l]egally sufficient” and “fin]eritorious”). Indeed, the Sixth Circuit in Tenenbaum— the only circuit actually to apply a “valid defense” standard — apparently meant the latter, since in dismissing the case, it stated only that “Defendants cannot defend their conduct ... without revealing the privileged information,” and made no suggestion that any of the defenses were dis-positive. 372 F.3d at 111.2
The majority’s privileged-defense standard is troubling both in its sharp depar*157ture from the other circuits and in its potential effect on public officers. If a government officer accused of malfeasance has several meritorious — but not surefire — privileged defenses, those defenses are now simply irrelevant.3
The majority also completely ignores the potential for distortion when valid defenses are excised by invocation of the privilege. As Judge Phillips noted over a quarter-century ago, it is “important to keep in mind that by its very nature” the state secrets privilege “compromises the intrinsic fairness of the adversary litigation process which has been provided for formal dispute resolution” — for both plaintiffs and defendants alike. Farnsworth Cannon, Inc. v. Grimes, 685 F.2d 268, 277 n. 2 (4th Cir.) (Phillips, J., specially concurring and dissenting), maj. op. rev’d per curiam, id. at 281 (1980) (en banc). When application of the privilege so “compromises the intrinsic fairness” of a judicial proceeding— whether because it has removed too much information from the plaintiffs case or from the defendant’s defense, or, as in this case, both — the right solution is not simply to muddle on, but rather “to withdraw from ... litigants their normal right of access to the formal dispute resolution forum provided by the sovereign.” Id. at 279. To permit a grossly distorted case— where the court knows the “facts” being litigated are only a parody of the real facts — to continue in our courts is not justice, and only invites injustice. See id. at 279 n. 5 (“[Dismissal is appropriate] where the judge can sense that the actual dispute as defined by the issues so far differs from the dispute that could be litigated while honoring the privilege as to draw in question the fairness of attempting to apply to the restricted dispute the legal principles appropriate to resolution of the actual dispute.”).4
B
By stripping meritorious defenses from Huddle and leaving gaping holes in Horn’s *158prima facie case, the invocation of the privilege so distorts this case that dismissal is necessary. “[T]he undisclosable scope of privilege lies so completely athwart the scope of proof relevant to resolution of the issues presented that litigation constrained by administration of the privilege simply could not afford the essential fairness of opportunity to both parties that is a fundamental assumption of the adversary system.” Id. at 279. Even assuming, however, that the majority remains indifferent to the distortion caused by the privilege, the district court’s third ground for dismissal — that the “very subject matter” of Horn’s action is a state secret — should be affirmed. The district court expressed concern that state secrets are “so central to the subject matter of [Horn’s case] that any attempt to proceed will threaten disclosure of privileged matters.” Mem. Op. 11-12 (quoting Fitzgerald v. Penthouse Int’l, Ltd., 776 F.2d 1236, 1241-42 (4th Cir.1985)). The majority apparently disagrees, finding “national security concerns ... peripheral to what remains of Horn’s prima facie case.” Maj. Op. 152.
In applying the “very subject matter” ground, other circuits have focused on the threat of inadvertent disclosure of privileged material posed by further litigation. See Kasza v. Browner, 133 F.3d 1159, 1170 (9th Cir.1998); Black v. United States, 62 F.3d 1115, 1118 (8th Cir.1995); Bareford, 973 F.2d at 1143; Farnsworth Cannon, 635 F.2d at 281 (en banc) (per curiam). Specifically, courts have considered the extent to which the non-privileged facts remaining in the case are intertwined with or surrounded by privileged material. See, e.g., Bareford, 973 F.2d at 1143 (noting “the practical reality that in the course of litigation, classified and unclassified information cannot always be separated”); Fitzgerald, 776 F.2d at 1243 n. 11 (explaining that “the merits of this controversy are inextricably intertwined with privileged matters”). Courts have expressed special concern where the plaintiff is in possession of some of the privileged material. See Fitzgerald, 776 F.2d at 1242 n. 8; Farnsworth Cannon, 635 F.2d at 281 (en banc) (per curiam). Similarly, they have recognized the risk of accidental disclosure where plaintiffs’ cases depend on testimony from witnesses with personal knowledge of classified secrets “relevant to the subject matter of the litigation.” Fitzgerald, 776 F.2d at 1242; accord Bareford, 973 F.2d at 1143-44. Courts have taken a practical approach, looking realistically at the “facts necessary to litigate” a plaintiffs case, “not merely [those necessary] to discuss it in general terms.” El-Masri, 479 F.3d at 310-11 (collecting cases). As the Fourth Circuit has explained, “[t]he controlling inquiry is not whether the general subject matter of an action can be described without resort to state secrets. Rather, we must ascertain whether an action can be litigated without threatening the disclosure of such state secrets.” Id. at 308.
This court has had no occasion to apply the “very subject matter” ground. But applying its logic to Horn’s complaint leads inexorably to the conclusion reached by the district court. The few remaining unprivileged facts comprising Horn’s prima facie case are islands surrounded by a sea of privileged material. This case is no different in that regard than Farnsworth Cannon, Fitzgerald, Bareford, or El-Mas-ri, except that here the islands are fewer and smaller.5
*159Moreover, the majority’s assertion that Horn “is not in possession of the privileged material” is mystifying. Maj. Op. 145. Horn clearly knows some of the privileged material, which the majority elsewhere implicitly concedes when it affirms the district court’s dismissal as to Defendant II, noting “there is no unprivileged evidence connecting him to Horn’s allegations.” See Maj. Op. 147. Admittedly, because Horn has never seen the classified portions of the two agency investigative reports, he cannot precisely map the division between what portion of the information he knows is covered by the privilege, and what isn’t. But that only exacerbates the potential for inadvertent disclosure, presenting the same problem identified by the Fourth Circuit in Farnsworth Cannon:
[The ex parte] affidavit [delineating the privileged information] has not been seen by [plaintiffs] counsel, and without some disclosure of the affidavit to counsel, the trial lawyers would remain unaware of the scope of exclusion of information determined to be state secrets. Information within the possession of the parties on the periphery of the suppression order would not readily be recognized by counsel, unaware of the specific contents of the affidavit, as being secret or as clearly having been suppressed by the general order of the district court. In an attempt to make out a prima facie case during an actual trial, the plaintiff and its lawyers would have every incentive to probe as close to the core secrets as the trial judge would permit. Such probing in open court would inevitably be revealing.
Farnsworth Cannon, 635 F.2d at 281 (en banc) (per curiam).
Likewise, the majority suggests the unavailability of the privileged IG reports is of little consequence to Horn’s case, because “there would be no barrier to his calling the [reports’] affiants as witnesses in order to testify to ... unclassified matters.” Maj. Op. 148. The majority is unconcerned that witnesses with relevant knowledge who might be called to testify are also sure to possess privileged information relevant to Horn’s case. Because the demarcation between the privileged and unprivileged information is by no means intuitive and, like Horn, the witnesses themselves would not be privy to the exact scope of the privilege, “the danger that witnesses might divulge some privileged material during [direct and] cross-examination is great.” Bareford, 973 F.2d at 1144.
All of these considerations support the district court’s third ground for dismissal. The majority comments that, at this stage of the proceedings, “Horn need not plead the facts sufficient to prove his allegations and evidence that will ultimately be used at trial.” Maj. Op. 147-48. That is true; however, the government’s invocation of the state secrets privilege in this case requires us to frankly consider whether Horn’s case “can be litigated without threatening the disclosure of ... state secrets.” El-Masri, 479 F.3d at 308. Relying on the same case-and fact-specific approach favored by every other circuit that has considered the issue, and based on my review of the entire record in this case, I would find the risk of disclosure too great. Here, the few unprivileged facts remaining are so entwined with privileged matters, and the risk of disclosure of privileged *160material so unacceptably high, that the very subject matter of this action is a state secret.6
II
This circuit’s state secrets cases have predominantly turned on the first of the three grounds relied on by the district court in dismissing Horn’s case.7 The majority’s disposition requires it to address all three grounds, but in doing so it gives short shrift to the important issues of distortion and disclosure. In my view, both of these issues are most effectively considered when analyzing whether the “very subject matter” of a case is a state secret. Thus, I would analyze the effect of invocation of the state secrets privilege as follows: First, can the plaintiff establish a prima facie case absent the privileged material? Second, if so, is a dispositive (i.e., Molerlo-type) defense barred by the invocation of the privilege? Third, is the very subject matter of the case a state secret? In evaluating this final prong, I would consider the issues of distortion and disclosure: Has removal of facts relevant to the plaintiffs prima facie case or the defendant’s defense, or both, so distorted the case that the litigation no longer even approximates reality? And does further litigation threaten inadvertent disclosure?
While I find this framework helpful, another might work equally well. The problem with the majority’s approach is its elevation of the rhetoric of perfect justice over the realities of distortion and disclosure. The question is not whether we like or approve of the state secrets privilege. It exists. The question is how the existence of the privilege, properly invoked, reshapes the case. In reversing the district court’s conclusion that the very subject matter of Horn’s case is a state secret, the majority rejects the standard consistently used by other federal courts and fails to offer any alternative.8 To make matters worse, the majority announces a new and troublingly high threshold for dismissal when invocation of the privilege *161compromises the defenses of government officials.
I respectfully dissent.
. The majority also properly assumes without deciding the important question of whether "the Fourth Amendment protects American citizens abroad,” Maj. Op. 143, because the question is not squarely presented here. While the district court in an earlier phase of this case did find the Fourth Amendment applicable, the government voluntarily dismissed its appeal of that ruling, and the issue was not litigated or briefed on this appeal.
. The majority also cites In re United States, 872 F.2d 472, 476 (D.C.Cir.1989), for its discussion of the "valid defense” standard. See Maj. Op. 149, 150. The court in In re United States, however, had no occasion to apply any privileged-defense standard — certainly not the standard advanced by the majority. See In re United States, 872 F.2d at 482 (D.H. Ginsburg, J., concurring and dissenting) ("Here it is not at all clear that the Government's secret defense is dispositive (or even meritorious under New York law).” (alterations in original *157omitted)). In any event, any privileged-defense discussion in In re United States is irrelevant; unlike here, the government itself was the defendant in In re United States, and, "[i]n Ellsberg, this court made clear that a government party does not forfeit a meritorious defense merely because it would need to rely on privileged materials in order to assert it.” Id. at 481 (emphasis added).
. The majority further states that "Huddle has already revealed his defense — that he learned of Horn’s conversation through Stubbs — and it is unprivileged.” Maj. Op. 149. Surely the majority cannot mean to imply that Huddle is limited to only one defense. As the majority is aware, the district court in a classified portion of its opinion recounted specific aspects of the privileged material Huddle would require to mount an effective defense at trial. Mem. Op. 11 (redacted). None of that material relates to Horn’s disputed conversation with Stubbs.
. The majority argues that considering the " 'distortion' effects of ... omitted defenses .... abridge[s] the rights of plaintiffs,” Maj. Op. 150, and "thwart[s] a citizen’s efforts to vindicate his or her constitutional rights,” id. at 151. Of course, that argument begs the very question that divides us: Do plaintiffs have a "right” to use the courts to press a case against a defendant when the available "facts” of that case no longer approximate reality? Ironically, it is the majority's unprecedented privileged-defense standard that creates "a system of conjecture,” Maj. Op. 150, where fact-finders are forced to invent the missing parts of the story. The majority's standard, not mine, "impose[s] a presumption.” Id. I am advocating a case-by-case assessment of how the privilege has affected the shape of the case being presented to the fact-finder, not "dismissal of a complaint for any plausible or colorable defense.” Id. The majority, in contrast, is effectively establishing a presumption that plaintiffs able to make a prima facie showing deserve to prevail against defendants relying on meritorious privileged defenses, unless those defenses are dispositive.
. The majority characterizes those cases as "inapposite,” noting the centrality of the privileged information to each plaintiff’s case. Maj. Op. 152-53. But this case is no different — here, the clearly "sensitive details,” Maj. Op. 152, of Huddle’s ability to conduct or *159order a wiretap are plainly central to Horn’s case. And the allegations in Horn’s complaint necessarily rely on the involvement of Defendant II; thus, dismissing Defendant II does not alter his centrality to Horn’s case.
. The majority characterizes the declaration of the Director of Central Intelligence as "rejecting] the notion that all of Horn's lines of inquiry are inextricably interwoven.” Maj. Op. 152. But the declaration, which in the portion cited explains only that some potentially relevant evidence presents “no risk to U.S. national security” once segregated, says absolutely nothing about whether attempting to litigate a specific case involving that evidence might present an unacceptably high risk of disclosure. To state that specific pieces of evidence are unprivileged is obviously not tantamount to stating that any litigation involving that evidence could never run an unacceptable risk of disclosure of state secrets. The majority’s attempt to wrest the latter implication from the declaration far exceeds that document’s purpose and scope.
. Molerlo turned on the second ground, but, as noted, Molerio was an easy case presenting a clearly determinative privileged defense. See 749 F.2d at 825.
.The majority comments “there is no need to usurp ... from the district court” the judgment of whether the very subject matter of Horn's case is a state secret. Maj. Op. 153. Yet that is exactly what the majority has done, explaining that "the district court has not yet evaluated the case as it now stands.” Maj. Op. 153 (emphasis added). But the case "as it now stands” is no different than the case the district court dismissed except the majority has put Huddle back in. In dismissing Horn’s entire case, the district court certainly considered whether an action against Huddle alone risked disclosure of state secrets, stating that ”[a]t the heart of plaintiff’s claim is ... information that is at the center of the state secrets privilege” and that "any attempt to proceed will threaten disclosure of privileged matters.” Mem. Op. 12 (emphases added) (citation omitted).