Chapter 2

Should We Rely on Judges?

Transparency and the Problem of Judicial Deference

The difficulty in regulating state secrecy, we have seen, is that there is a mismatch between who should and who does serve as regulator. One reason for the mismatch is judicial deference toward the executive’s claims about the harm likely to be caused by the disclosure of classified information—a posture, critics argue, that encourages and justifies unauthorized disclosures. Hence, if we want secrecy to be regulated by more orderly and lawful means, they argue, then judicial review must be expanded.

Why Do We Need Judges?

There are a variety of means that citizens can use to pressure the president to share the information they need to conduct oversight and enforce accountability: they can threaten to vote him out of office, they can signal suspicion through protests and demonstrations, and they can ask him to justify continuing secrecy. But citizens are hard-pressed to make use of these checks—elections, public opinion, and public deliberation—unless they have some prior indication of wrongdoing, since they will not want to call for the unwarranted disclosure of state secrets. In other words, before citizens can demand that the curtain be raised, they need to know at least something about what has been going on behind the curtain. This is what makes the idea of qualified transparency appealing. This conception of the norm of transparency calls on officials to disclose information unless secrecy is genuinely required to protect national security. But who can be trusted to enforce this norm? The challenge we confront when we ask officials to abide by this norm is that we have no way to tell whether their claims about the costs of making a particular piece of information public are reasonable. Indeed, there is something paradoxical about relying on officials to enforce the norm of qualified transparency: it places citizens in the position of a detective who expects the suspect to surrender the evidence of wrongdoing.

A number of commentators have argued that attending to the framework within which declassification decisions are made can lessen this regulatory challenge. The idea here is not to divest the executive of the final say over classification. Rather, it is to subject the classification decisions made by officials at the CIA or NSA—who may be tempted to use their classification power to shield their organizations from critical scrutiny—to internal review by peers from other intelligence agencies or legal officers such as the attorney general. If a classification decision survives such scrutiny, these commentators argue, then we have reason to be more confident about its merit, since the reviewers are not likely to share the parochial interests of the original classifier.1 As David Pozen has put it, “admitting a mere handful” of lawyers into the classification process “may dramatically change” how secrecy is employed, because their “distinctive skill-sets and socialization” and “institutional responsibilities” will bring “a different perspective to the contemplated policy.”2 Or as Christina Wells has claimed, requiring officials to justify secrecy claims to lawyers from the Justice Department “could alleviate a number of cognitive biases,” including “the tendency to rely only on confirmatory evidence while ignoring disconfirmatory evidence” about the harm likely to be caused by disclosure.3 We could build on these insights by requiring government attorneys to publish sworn statements affirming that the classification decisions they have reviewed are justified and would survive public scrutiny. This requirement would raise the stakes for the internal reviewer: should subsequent disclosures reveal activity that the public deems unacceptable, her concurrence in the concealment of those activities would expose her to professional and legal sanction.

The idea of internal review is appealing because it promises to allow executive officers who have expertise in intelligence matters to balance the costs and benefits of disclosure. But can widening the circle of reviewers really improve the credibility of internal review in hard cases? It may well be the case that the officials responsible for reviewing classification decisions will not have exactly the same incentives as the officials responsible for the original classification decisions—but this does not mean that the former will have an incentive to promote transparency. Broader administration-wide objectives could lead internal reviewers to play along with the original classifiers, especially when the stakes are high. An embarrassing disclosure may not be an embarrassment exclusively to the parent organization of a particular classifier—it may embarrass or even weaken the administration as a whole and thus prompt a loyal silence from internal reviewers. Consider, for example, Kathleen Clark’s complaint that the Bush administration’s refusal to provide security clearances to investigators from the Office of Professional Responsibility—on the advice of Attorney General Alberto Gonzales—blocked an investigation into whether Justice Department lawyers acted inappropriately in approving the NSA’s warrantless surveillance program.4

Furthermore, even if broader interests and loyalties do not influence internal reviewers, how can we be confident that their understanding of the appropriate balance between secrecy and accountability is in fact reasonable? As Dennis Thompson has pointed out, the justifiability of covert actions depends “on judgments about the value of the ends they promote.” But these judgments are usually “too contestable to be resolved through assumptions about human nature, shared beliefs and interests under hypothetical conditions.”5 In practice this means that a government attorney may not know what might be a reasonable standard against which to compare the classification decision under review. A likely outcome under the circumstances is that, convinced of the righteousness of her employer’s cause, she will consider the classification of a controversial policy well justified (if not, she may well find herself replaced with a more like-minded lawyer, who does).6

Are there other ways to bind the hands of those charged with reviewing classification decisions? One answer that many observers fall back upon is to impose limits on the duration of secrecy, principally in the form of “automatic” disclosure rules.7 A rule of this kind effectively establishes a “shelf life” for a classification stamp. For example, a rule might require records stamped “confidential” to be treated as such for one year from the date of classification, following which the need to maintain confidentiality is presumed to have expired and the record is deemed declassified. The utility of such disclosure rules is that they seem to lessen the discretionary power that officials enjoy over declassification decisions, thus increasing the likelihood that citizens and lawmakers will obtain access to records that might otherwise have been withheld by officials wishing to cover up wrongdoing. However, such rules have two obvious shortcomings. The first is that they tend to have long fuses. That is, they tend to be premised on the idea that the harm likely to be caused by the disclosure of secret information diminishes only with the passage of a substantial period of time, typically between twenty to thirty years. Needless to say, requiring citizens and lawmakers to wait this long before they can examine how decision-makers have acted greatly constrains their ability to enforce accountability. Indeed, a long delay could mean that decision-makers will not face moral censure or legal penalties during their lifetimes. Conversely, the knowledge that declassification is likely to take place after a decision-maker’s death, or at least well after the completion of his tenure in office, may well serve to undermine one of the key reasons to pursue accountability: to deter wrongdoing in the future. Then there is the problem of irretrievability: the greater the delay before declassification, the harder it becomes to remedy in any meaningful sense the harm caused by poor decision making.8

A further problem with automatic disclosure rules is that we cannot demand that they be followed absolutely, as we cannot be certain of the length of time after which the disclosure of a secret will prove harmless. Consider Aftergood v. CIA (2005), a case brought under the Freedom of Information Act (FOIA) by Steven Aftergood of the Federation of American Scientists, who had sought from the CIA information about the size of its Cold War–era intelligence budgets. In its declaration before Judge Ricardo Urbina of the District Court for the District of Columbia, the CIA argued that it was entitled to withhold the information in question in order to “protect the classified intelligence methods used to transfer funds to and between intelligence agencies.” Aftergood challenged this claim, arguing that he had previously published the intelligence budgets for 1953, 1954, 1955, and 1972 that the historian David Barrett had discovered in the personal archives of former congressmen. The publication of these documents, Aftergood claimed, had “had no identifiable” negative consequences for the United States.9 Hence the court, he argued, ought to order the CIA to disclose the requested information. However, Judge Urbina concluded otherwise. “The plaintiff invites the court to conclude that the plaintiff is more knowledgeable than the ADCI [assistant director of central intelligence] about what disclosure of information would harm intelligence sources and methods,” he observed. “The court declines the plaintiff’s invitation.”10

If the passage of time cannot eliminate the discretion officials enjoy to withhold information, then officials will have the opportunity to stall the disclosure of classified information if they so wish. Can we prevent an abuse of this discretionary power by rotating decision-makers or by utilizing term limits? This proposal certainly has one benefit—a change of administration may well terminate the ongoing misuse of secrecy. Bear in mind, though, that this step will not prevent new misuses of state secrecy. Nor should we assume that an incoming administration will have the opportunity or the incentive to disclose the wrongful conduct of its predecessors. It may lack the opportunity since prior incumbents will have a strong incentive to avoid leaving a paper trail or else to destroy incriminating documents before departing—as they did during Iran-Contra.11 Moreover, any incentive that incoming officials may have to disclose wrongdoing by their predecessors will surely be counterbalanced by the desire not to establish a precedent that could be used against them when they leave office. Indeed, a new administration will have an incentive to actively conceal past wrongdoing because such revelations are likely to lead to calls for oversight (thereby making it harder for the new administration to employ state secrecy as it sees fit) and to embarrass the United States on the world stage (thereby making it harder for the new administration to pursue its own policies). It is also possible that party affiliations will align the interests of administrations. An example of this is provided by Executive Order 13233, which authorized President Bush to withhold the papers of his predecessors. Though the order was justified on grounds of national security, archivists and historians have pointed out that the order was passed shortly before President Reagan’s papers were due to be made public.12 The papers are believed to contain information relating to the Iran-Contra affair and its protagonists, which include some prominent personalities in the Republican establishment.

The Problem of Judicial Deference

For the reasons outlined above, internal review mechanisms and automatic disclosure rules do not provide much assurance that embarrassing or incriminating information will be made public, especially when the political stakes are high. Scholars therefore argue that we ought to transfer the final say on classification to an institution that will have a less immediate incentive to misuse this authority. Given the partisan character of Congress’s membership, many scholars favor vesting this authority in the courts. But is this a wise choice? Can judges, far removed from the cut and thrust of diplomacy and international intrigue, really challenge the president’s contentions as to what information should not be made public? The evidence, as we shall now see, is not promising.

As we saw in chapter 1, in 1966 Congress passed the Freedom of Information Act (FOIA) in order to provide the public with “access to official information” and to “create a judicially enforceable public right to secure such information from possibly unwilling official hands.”13 Congress acknowledged, however, that unconditional transparency or publicity would not be in the public interest. Hence FOIA exempts nine specific categories of information from disclosure. From our perspective, the relevant provisions are §552(b)(1) and §552(b)(3), commonly referred to as Exemptions 1 and 3. The former allows officials to withhold records that are “specifically authorized under criteria established by an Executive order to be kept secret in the interest of national defense or foreign policy”; the latter allows officials to withhold records “specifically exempted from disclosure by statute.”14 Though the latter provision may seem innocuous, it actually creates a substantial exception to FOIA because the statutes that can be invoked—such as the Central Intelligence Agency Act and the National Security Agency Act—allow intelligence agencies to withhold vast swathes of information about sources and methods, operations and personnel, and organizational features and functions.15

It is not difficult to see that officials who have something to hide may try to take advantage of these provisions. How far have the courts been willing to go to ensure that invocations of these exemptions are justified? The Supreme Court first tackled this issue in EPA v. Mink (1973), a case that arose after Congresswoman Patsy Mink sought access to a National Security Council (NSC) report addressing a dispute within the Nixon administration over whether to conduct an underground nuclear test in Alaska. Having been turned down by the administration, Mink appealed to the Supreme Court, asking it to examine in camera whether the report really needed to be kept secret in the interest of national defense. A number of the justices expressed sympathy for Mink’s cause. As Justice Potter Stewart observed, “One would suppose that a nuclear test that engendered fierce controversy within the Executive Branch of our Government would be precisely the kind of event that should be opened to the fullest possible disclosure consistent with legitimate interests of national defense” because, in the absence of such information, “the people and their representatives [are] reduced to a state of ignorance, [and] the democratic process is paralyzed.” Nonetheless, the court came down on the administration’s side on the grounds that FOIA limited the courts to ascertaining only whether an exempted document was in fact classified. As Justice Stewart underscored, FOIA provided “no means to question an Executive decision to stamp a document ‘secret,’ however cynical, myopic, or even corrupt that decision might have been.”16

In 1974 Congress responded to Mink, and to growing public outrage over real and apparent abuses of secrecy, by amending FOIA to authorize the courts to conduct in camera reviews of documents that officials claimed were exempt from disclosure. But Congress also constrained in camera review in two key respects. The conference report accompanying the legislation advised that “before the court orders in camera inspection, the Government should be given the opportunity to establish by means of testimony or detailed affidavits that the documents are clearly exempt from disclosure.” Furthermore, the report emphasized Congress’s expectation that, when evaluating the validity of an exemption claim, the courts “will accord substantial weight to an agency’s affidavit concerning the details of the classification status of the disputed record,” since “the Executive departments responsible for national defense and foreign policy matters have unique insights into what adverse effects might occur as a result of public disclosure of a particular classified record.”17

Congress’s instructions did not go unheeded. Shortly after the 1974 amendments had passed into law, the courts were confronted in Weissman v. CIA (1977) with a plaintiff who, having learned through news reports that the CIA had been spying on political activists within the United States, requested the CIA to disclose any files it might have on him. The CIA admitted to having files on the plaintiff, but argued that it was entitled to withhold them as they revealed intelligence sources and methods. After the district court found in favor of the CIA, the plaintiff approached the D.C. Circuit, asking it to conduct an in camera review to “check the truthfulness of Agency claims.” But Judge Gerhard Gesell rejected this appeal, arguing that since “few judges have the skill or experience to weigh the repercussions of disclosure of intelligence information,” once the court was satisfied “that proper procedures have been followed, and that by its sufficient description the contested document logically falls into the category of the exemption indicated,” then “it need not go further to test the expertise of the agency, or to question its veracity when nothing appears to raise the issue of good faith.”18

But what if there were doubts about the sufficiency of the government’s explanation? In Halperin v. CIA (1980) an appellant sought the names of private attorneys retained by the CIA as well as the fees paid to them. The CIA responded to this request by arguing that disclosing the identity of an attorney who is an agent of the CIA in intelligence activities “might expose him to adverse action from hostile powers” and thereby create “a strong disincentive to those who are considering future employment or continued affiliation with the CIA.” The appellant, Morton Halperin, a former assistant secretary of state, retorted that this “projection of potential harm” was “pure speculation.” But the D.C. Circuit disagreed, observing that “any affidavit or other agency statement of threatened harm to national security will always be speculative to some extent, in the sense that it describes a potential future harm rather than an actual past harm.” Under the circumstances, the only question before the court, Judge Malcolm Wilkey wrote, was whether “the predicted danger is a reasonable expectation,” and on this front, he noted, “a court, lacking expertise in the substantive matters at hand, must give substantial weight to agency statements.”19

But how far ought such judicial deference to extend? Could a judge challenge officials’ claims that they needed to withhold even seemingly innocuous information on national security grounds? Halperin was instructive on this count too. The CIA argued that it was entitled to conceal even the total fees paid to the unnamed private attorneys because disclosing this figure “could give leads to information about covert activities that constitute intelligence methods.” “If a large legal bill is incurred in a covert operation,” it explained, then “a trained intelligence analyst could reason from the size of the legal bill to the size and nature of the operation.” Though Judge Wilkey averred that the showing of the harm in this case was “not so great,” he still upheld the CIA’s claim, citing what has now come to be termed “mosaic theory.” “We must take into account,” he wrote, “that each individual piece of intelligence information, much like a piece of jigsaw puzzle, may aid in piecing together other bits of information even when the individual piece is not of obvious importance in itself.”20

The cautious stance adopted in Halperin was subsequently affirmed in the landmark case CIA v. Sims (1985). As in other important FOIA cases, Sims involved a plaintiff who had been alerted to potential wrongdoing via a leak of classified information—in this case about the CIA’s MKULTRA project, which involved research into brainwashing. Here the plaintiff sought from the CIA “the names of the institutions and individuals that had performed research” as part of MKULTRA. The CIA declined to comply with this request, citing the 1947 National Security Act, which states that “the Director of Central Intelligence shall be responsible for protecting intelligence sources and methods from unauthorized disclosure.” After the district court and the D.C. Circuit disagreed over whether the institutions and individuals that participated in MKULTRA constituted the kind of “intelligence sources” that the CIA was authorized to protect under the act, the matter reached the Supreme Court, which unanimously concluded that the CIA ought to have wide latitude in defining what constitutes an intelligence source. Chief Justice Warren Burger invoked “the realities of intelligence work, which often involves seemingly innocuous sources as well as unsuspecting individuals who provide valuable intelligence information.” Hence the CIA was entitled to withhold even “superficially innocuous information” because “a foreign government can learn a great deal about the Agency’s activities by knowing the public sources of information that interest the Agency.” In this domain, the court instructed, “the decisions of the Director, who must of course be familiar with ‘the whole picture,’ as judges are not, are worthy of great deference given the magnitude of the national security interests and potential risks at stake.”21

The effect that Sims has had on subsequent FOIA litigation can be seen most clearly in Fitzgibbon v. CIA (1990). In this case, historian Alan Fitzgibbon had sought from the CIA information relating to the disappearance in the United States of a critic of the Dominican Rafael Trujillo regime. After the CIA released a limited number of documents featuring heavy deletions, Fitzgibbon sought relief from the District Court for the District of Columbia in 1979. As in other FOIA cases, the CIA filed a classified affidavit justifying its decision to withhold information. But unlike in prior FOIA cases, District Judge Harold Greene subjected the CIA’s claim to careful scrutiny. Having examined the original documents in camera, Judge Greene determined that some of the deletions that the CIA had made were “unexplained” or “unrelated” to the justification offered, and hence ordered the disclosure of the deleted matter.22

Taken aback, the CIA appealed, but with Sims having been decided in the interim, the case was remanded to the district court. On remand, Judge Greene now reversed his previous order in light of Sims. He persisted, however, in ordering the disclosure of a document identifying the location of a CIA station that had already been disclosed in a congressional committee report. Now both Fitzgibbon and the CIA appealed. Before the D.C. Circuit, Fitzgibbon argued that the CIA ought not to be allowed to withhold information that Judge Greene had previously described as “so basic and innocent that its release could not harm the national security or betray a CIA method.”23 The CIA, on the other hand, sought reversal of the order to disclose the document identifying the location of its station. The D.C. Circuit rejected Fitzgibbon’s appeal outright, noting that Sims’s endorsement of mosaic theory had “vaporized” the merits of his case. In addition, it rapped the district court on the knuckles for having “performed its own calculus as to whether or not harm to the national security or to intelligence sources and methods would result from disclosure” of the location of the CIA station. “The assessment of harm to intelligence sources, methods and operations,” Judge David Sentelle emphasized, “is entrusted to the Director of Central Intelligence, not to the courts.”24

It is not necessary to trace FOIA case history further than this, as the standard of review developed over the course of Weissman, Halperin, Sims, and Fitzgibbon has been consistently reaffirmed in recent decades.25 However, lest this deferential standard be viewed as a consequence of the guidance that Congress issued alongside FOIA in 1974, let us briefly examine how the courts have reacted to invocations of the state secrets privilege. As we shall see, in this domain the courts themselves have fashioned a highly deferential standard of review.

The state secrets privilege is an evidentiary privilege that allows the United States to “block discovery in a lawsuit of any information that, if disclosed, would adversely affect national security.”26 The foundational case here is United States v. Reynolds (1953), which involved a lawsuit brought by the widows of individuals who were on board a military aircraft that crashed during a flight to test secret electronic equipment. In order to assess whether the Air Force had been negligent, the plaintiffs requested access to the accident investigation report. After the Air Force failed to comply, citing a right to withhold documents in the public interest, the district court treated the claim of negligence as justified and awarded damages. The United States then appealed this decision all the way to the Supreme Court, which concluded in a 6–3 decision that the United States did enjoy an evidentiary privilege, and that “even the most compelling necessity cannot overcome the claim of privilege if the court is ultimately satisfied that military secrets are at stake.”27

So what would it take to satisfy a court that state secrets are in fact at stake? Here the Reynolds court discerned a “real difficulty”—namely, that “too much judicial inquiry into the claim of privilege would force disclosure of the thing the privilege was meant to protect, while a complete abandonment of judicial control would lead to intolerable abuses.”28 Not surprisingly, the court then went on to formulate a compromise. Crucially, it rejected the contention that it was for the executive to decide whether the requested evidence could be produced: “judicial control over the evidence in a case,” Chief Justice Fred Vinson wrote, “cannot be abdicated to the caprice of executive officers.” “Yet we will not go so far as to say,” he added, “that the court may automatically require a complete disclosure to the judge before the claim of privilege will be accepted in any case.”29

So when would it be appropriate for a judge to subject a claim of privilege to close analysis? “It may be possible to satisfy the court, from all the circumstances of the case,” the chief justice observed, “that there is a reasonable danger that compulsion of the evidence will expose military matters which, in the interest of national security, should not be divulged.” “When this is the case, the occasion for the privilege is appropriate,” he instructed, “and the court should not jeopardize the security which the privilege is meant to protect by insisting upon an examination of the evidence, even by the judge alone, in chambers.”30 The italicized portion of this passage is important because the outcome in Reynolds ended up hinging upon it. The Reynolds court ruled—without the benefit of an in camera examination—that the accident investigation report was privileged because “there was a reasonable danger” that it “would contain references to the secret electronic equipment” undergoing testing on the Air Force plane.31

This passage in Reynolds has been criticized by scholars who have argued that the Supreme Court made an inexplicable error here—it examined whether there was a reasonable danger that the crash report contained secret information, instead of examining whether there was a reasonable danger that the disclosure of this information to the plaintiffs would harm national security.32 According to these critics, the Reynolds court failed to do what it said a court ought to do—which is to determine whether the claim of privilege was actually justified. But this criticism misapprehends the court’s reasoning. Arguably, the Reynolds court believed that there are two risks that a court needs to account for when handling a state secrets privilege claim. First, there is the risk that national security could be harmed by the unauthorized or inadvertent disclosure of evidence during in camera proceedings; and second, there is the risk that national security could be harmed by the authorized disclosure of evidence to litigants. Though Robert Chesney and Louis Fisher, among others, assert that the court was unduly deferential because it failed to investigate the latter risk by examining the report in camera, another way to understand Reynolds is to see the court as having ceased its investigation after completing the first of the two steps outlined above. That is, the court reasoned that since even an in camera, ex parte examination of privileged evidence comes with a risk of inadvertent or unauthorized disclosure, a court is not obliged to undertake such an examination unless the plaintiff can prove that she has no way of proceeding without the evidence in question. The court believed that this necessity was lacking in Reynolds, because the plaintiffs could have established negligence on the Air Force’s part via depositions. Hence, it concluded, it was not obliged to take on the risk of examining the report in camera.

Since the Court in Reynolds had not seen the need to closely examine whether the Air Force’s invocation of the state secrets privilege was justified, its opinion shed little light on how far judges ought to defer to the executive’s claims in the event of uncertainty about whether the privilege had been invoked correctly. The first real opportunity to develop a standard of review came in Jabara v. Kelley (1977), a case brought by an individual seeking damages for having been subjected to warrantless wiretapping. Frustrated by the refusal of officials to respond to queries about the presumptive wiretaps, the plaintiff sought the assistance of the District Court of the Eastern District of Michigan, in turn prompting the Defense Department to invoke the state secrets privilege. Since there was a clear showing of need in this case—the plaintiff’s action could not proceed without the requested information—Judge Ralph Freeman agreed to examine the Defense Department’s affidavits in camera in order to ascertain whether the claim of privilege was justified. Following this examination Judge Freeman upheld the claim of privilege with respect to all information “that would reasonably tend to reveal foreign intelligence sources and capabilities.”33 But he also ruled that the privilege could not be invoked to shield information that did not relate to the conduct of foreign intelligence, citing in particular the Pentagon’s refusal to disclose the name of the “federal agency” that had helped the FBI wiretap the plaintiff.34 Noting that the name of this federal agency (the NSA) had already been revealed in a recent congressional report, Judge Freeman opined that “it would be a farce to conclude that the name of this other federal agency remains a military or state secret.”35

The Jabara court’s willingness to challenge the executive’s claim about what constitutes a state secret was striking. But barely had Jabara been decided than the D.C. Circuit fashioned a rather different, and ultimately more influential, standard of review in Halkin v. Helms (1978). This was a case brought by antiwar protesters who believed that they had been subjected to warrantless surveillance by the NSA. In order to prove that their rights had in fact been violated, the plaintiffs requested the NSA to disclose the presumed intercepts of their communications. The NSA, however, declined to comply on the grounds that “identification of the individuals or organizations whose communications have or have not been acquired presents a reasonable danger that state secrets would be revealed” because it “would enable foreign governments or organizations to extrapolate the focus and concerns of our nation’s intelligence agencies.”36 Following an in camera review of classified affidavits, the district court concluded that the plaintiffs’ requests for information about intercepts from one NSA operation (MINARET) could not be compelled “because the ultimate issue, the fact of acquisition, could neither be admitted nor denied.”37 However, the district court also ruled that the NSA was obliged to respond to the plaintiffs’ requests for information about intercepts from another operation (SHAMROCK), because congressional investigations had already revealed so much about this operation that national security would not be endangered if the NSA were to admit its existence.

Both the protesters and the NSA appealed from this decision. The former challenged the district court’s deference to the executive’s claims about the harm caused by the disclosure of intercepts from MINARET, whereas the latter challenged the order to disclose intercepts from SHAMROCK. Upon review, the D.C. Circuit rejected the protesters’ appeal on the grounds that the standard of review for a state secrets privilege claim must be “a narrow one.” Judges, the court instructed, must be cautious in challenging claims about the need to withhold information that might seem trivial or unimportant to ordinary observers. As Judge Roger Robb explained, “the business of foreign intelligence gathering in this age of computer technology is more akin to the construction of a mosaic” because “bits and pieces of seemingly innocuous information can be analyzed and fitted into place to reveal with startling clarity how the unseen whole must operate.”38 Hence the courts must “accord the utmost deference to executive assertions of privilege upon grounds of military or diplomatic secrets.”39 Accordingly, the D.C. Circuit concluded by affirming the order to protect intercepts from MINARET and overturned the order to disclose intercepts from SHAMROCK.

Clearly, if the courts must show utmost deference to the executive’s claim about the harm likely to be caused by discovery—even when “seemingly innocuous information” is involved—then there is little to prevent officials from shielding potentially incriminating information by claiming that practically any discovery threatens national security. This is evidenced in Kasza v. Browner (1998), a case in which the Ninth Circuit was confronted with an appellant who sought to compel the Air Force to disclose details about allegedly unlawful handling of hazardous waste at a secret facility in Nevada. The Air Force responded to this action by invoking the state secrets privilege and refusing to share “any evidence tending to confirm or disprove that any hazardous waste had been generated, stored, or disposed of at the operating location.”40 The appellant then challenged this use of the privilege as “overbroad,” arguing that the Air Force’s claim that “the existence or nonexistence of hazardous waste is a state secret is absurd.”41 However, following an in camera review of affidavits that explained how the disclosure of unclassified information about the facility could serve to expose “security sensitive environmental data,” the Ninth Circuit ruled that “if seemingly innocuous information is part of a classified mosaic, the state secrets privilege may be invoked to bar its disclosure and the court cannot order the government to disentangle this information from other classified information.”42 Since the appellant could not move forward without requiring the disclosure of classified information, the Ninth Circuit dismissed the entire action.43

There have now been a number of suits akin to Kasza, where entire cases have been dismissed because officials have claimed that discovery cannot proceed at all without revealing state secrets. The moral and political implications of deference to such broad privilege claims have been made starkly clear by the recent series of cases brought by individuals seeking redress for the practice of extraordinary rendition. The most revealing of these cases are Mohamed v. Jeppesen Dataplan Inc. (2009) (Jeppesen I) and Mohamed v. Jeppesen Dataplan Inc. (2010) (Jeppesen II). Both of these rulings emerged out of a complaint against Jeppesen Dataplan, a subsidiary of Boeing Corporation, which had provided the CIA with the logistical support required to transport suspected terrorists to secret detention facilities overseas where these detainees were subjected to torture. Before the original complaint reached the trial stage, the CIA intervened and sought dismissal on the grounds that the “core” of the plaintiffs’ complaint involved covert CIA operations—“clearly a subject matter which is a state secret.”44 After the District Court of the North District of California agreed with the CIA, the plaintiffs appealed to the Ninth Circuit.

In Jeppesen I, the Ninth Circuit drew a line. The notion that a lawsuit should be dismissed “any time a complaint contains allegations, the truth or falsity of which has been classified as secret by a government official,” Judge Michael Hawkins observed, implies that “the Judiciary should effectively cordon off all secret government actions from judicial scrutiny, immunizing the CIA and its partners from the demands and limits of the law.”45 This theory, he argued, threatened “the principles of the separation of powers and judicial review,” which called for a careful “item-by-item” analysis to determine whether the privileged evidence was genuinely “indispensable” to the maintenance of the suit or whether the plaintiffs could make their case by reference to nonprivileged evidence.46 Of course such close analysis would serve little purpose if judges simply deferred to the executive’s claims about what counts as a state secret since, as Halkin and Kasza show, even information that has already been made public by Congress (the SHAMROCK operation) or is unclassified (the presence of hazardous waste) can be part of a classified “mosaic.” Hence, not surprisingly, Judge Hawkins also pushed back on the question of who was to judge what counted as a state secret. “A rule that categorically equated ‘classified’ matters with ‘secret’ matters,” he wrote, “would perversely encourage the President to classify politically embarrassing information simply to place it beyond the reach of judicial process.” Hence, “while classification may be a strong indication of secrecy as a practical matter,” he instructed, “courts must undertake an independent evaluation of any evidence sought to be excluded to determine whether its contents are secret within the meaning of the privilege.”47 Having said this, Judge Hawkins went on to offer an evaluation of his own—classified information about the extraordinary rendition program that had made its way into the public domain could not, he contended, be considered a state secret. “The government could not seriously argue,” he wrote, “that the Pentagon Papers remained ‘secret’ and therefore subject to the state secrets privilege even after having been published in the New York Times, simply because the government itself refused to declassify or otherwise ‘officially disclose’ the content of the papers.”48

The decision in Jeppesen I was soon reversed, though. Given the “exceptional importance” of the subject, the Ninth Circuit voted to rehear the case en banc and subsequently affirmed the district court’s decision by a margin of 6–5. Writing for the majority, Judge Raymond Fisher conceded that the existence of the extraordinary rendition program could no longer be considered a state secret—not because leaks had revealed “alleged” details of the program, but rather because its existence had been “publicly acknowledged” by “numerous government officials.”49 But the “partial disclosure of the existence and even some aspects of the extraordinary rendition program,” he emphasized, did not “preclude other details from remaining state secrets if their disclosure would risk grave harm to national security.”50 The existence of these valid state secrets, Judge Fisher argued, posed a quandary for the courts. “Because the facts underlying plaintiffs’ claims are so infused with these secrets,” he claimed, “any plausible effort by Jeppesen to defend against them would create an unjustifiable risk of revealing state secrets.”51 But what about Reynolds’s instruction that the greater the showing of necessity on the part of the plaintiff, the greater the obligation of the court to take on the risk of examining whether the claim of privilege was appropriate? Here Judge Fisher begged off, noting that even though “district courts are well equipped to wall off isolated secrets from disclosure, the challenge is exponentially greater in exceptional cases like this one, where the relevant secrets are difficult or impossible to isolate and even efforts to define a boundary between privileged and unprivileged evidence would risk disclosure by implication.”52

Is Judicial Deference Justified?

So far we have seen that the courts have displayed “great deference” and “utmost deference” to the executive’s claims about the harm likely to be caused by disclosure, even where “seemingly innocuous information” is concerned. This record has attracted criticism from commentators who have argued that such sweeping deference is unjustified, especially when civil liberties are at stake. Fisher, for instance, asserts that for judges “to defer to agency claims about privileged documents and state secrets is to abandon the independence that the Constitution vests in the courts and place in jeopardy the individual liberties that depend on institutional and public checks.”53 It has also been argued that judicial deference tends to be premised on faulty grounds. In a pathbreaking essay, Chesney warns that officials and judges alike have a tendency to “oversimplify” the reasons why deference is owed to the executive in the national security context.54 A closer inspection, he argues, would reveal that judges have reason to scrutinize the executive’s claims more carefully than they do at present.

These are weighty charges. However, before we advise judges to subject the executive’s claims to close scrutiny, we ought to be clear about the reasons that the courts themselves have offered in defense of deference. The first reason in favor of judicial deference is the prudential concern we touched upon during the discussion of Reynolds and Jeppesen II—namely, the worry that the examination of classified materials even in camera could lead to unauthorized or inadvertent disclosures. Importantly, the concern here is not that judges and their clerks cannot be personally trusted with state secrets. Claims along these lines have been made in the past, and they have rightly been rejected by the Supreme Court in United States v. United States Dist. Court (1972) (also known as the Keith case) where Justice Lewis Powell underscored that “judges may be counted upon to be especially conscious of security requirements in national security cases.”55 The concern we have, by contrast, is about the limited institutional capacity of the courts, which, judges themselves have argued, are ill-equipped to handle large volumes of classified materials on an ongoing basis. As the D.C. Circuit explained in Ellsberg v. Mitchell (1983), to assert that an examination in camera “is not entirely safe” is not meant to “slight judges, lawyers or anyone else.” Rather, the problem is that “in our own chambers, we are ill equipped to provide the kind of security highly sensitive information should have.”56

Notably, the passage of time and prolonged experience with cases involving classified information have not altered this view. For instance, in Sterling v. Tenet (2005), a high-profile racial discrimination case brought by an employee of the CIA, the Fourth Circuit recited Reynolds’s injunction against conducting in camera review of highly secret materials as the key reason to refrain from calling for the CIA’s personnel files, a necessary step to evaluating the charge of discrimination. The courts, the Fourth Circuit warned, “are not required to play with fire and chance further disclosure—inadvertent, mistaken, or even intentional—that would defeat the very purpose for which the privilege exists.”57 The Fourth Circuit has been far from alone in taking this view. As noted above, the Ninth Circuit reversed Jeppesen I precisely because it was convinced that “the risk of disclosure that further proceedings would create cannot be averted through the use of devices such as protective orders or restrictions on testimony.”58

We have also already touched on the second reason in favor of judicial deference—namely, the concern for expertise as highlighted in Halkin and Sims. As the Fourth Circuit has recently summarized in El-Masri v. United States (2007), “deference is appropriate not only for constitutional reasons, but also practical ones: the Executive and the intelligence agencies under his control occupy a position superior to that of the courts in evaluating the consequences of a release of sensitive information.”59 Once again, it is not the intellectual competence of judges that is in doubt. To question the expertise of judges is not to “see the judiciary as a hapless incompetent.”60 Rather, the concern is about specialization or comparative advantage in making judgments about the harm likely to be caused by the disclosure of information. As the Fourth Circuit explained in El-Masri, deference in the areas of military and foreign affairs and secret intelligence is founded on the assumption that executive officers who are continuously immersed in these domains are likely to make more refined or accurate predictive judgments.61 The federal judiciary, by contrast, “is a generalist institution composed of generalist judges,” as John Yoo has emphasized. “Very few judges,” he points out, “have significant foreign affairs experience before their appointment to the federal bench,” and the organization of courts along geographic lines “not only prevents specialization, but also retards the accumulation of experience.”62

To be sure, a few courts have tried to circumvent the question of expertise by allowing cases to proceed on the basis of evidence that is already in the public domain. In two such widely cited cases, Hepting v. AT&T (2006) and ACLU v. NSA (2006), courts were confronted with invocations of the state secrets privilege that prevented the plaintiffs from proving that they had in fact been subject to warrantless wiretaps. Rather than compelling the disclosure of the relevant information, which would have required dismissing the NSA’s concerns about harm to national security, the courts decided that official acknowledgment of the existence of a warrantless surveillance program was sufficient to grant the plaintiffs standing.63 This is an interesting precedent, but its utility over the long term seems questionable, because it creates a strong incentive for officials to say as little as possible in public. Moreover, legal action based on public statements remains vulnerable to the problem that even secrets that have been publicly acknowledged will usually be too closely tied up with undisclosed secrets to permit the sort of evidentiary disclosures required to allow plaintiffs to actually proceed with their case. This is precisely what prevented Jeppesen II from proceeding in spite of extensive public discussion about the extraordinary rendition program.

So why have these arguments in favor of deference been rejected? Briefly put, critics have argued that the prudential and epistemic concerns cited above can be addressed via “procedural innovations”—either by modifying trial procedures by, for example, calling upon outside experts, or by establishing special courts staffed by judges and clerks deeply versed in intelligence matters and able to conduct trials in camera and ex parte where necessary.64 Meredith Fuchs, for instance, has argued that the appointment of “a neutral and experienced [special] master with the appropriate security clearances could … relieve the court of its expertise and burden concerns,” while Pozen has submitted that “extrajudicial assistants” could help judges “evaluate the plausibility” of claims based on mosaic theory, which tend to be “highly speculative.”65 More radically, Chesney has suggested that Congress could create “a classified judicial forum” where Article III judges could hear cases involving state secrets “in camera on a permanently sealed, bench-trial basis,” where plaintiffs could be represented by a guardian ad litem “selected from among a cadre of, for example, federal public defenders with the requisite clearances.”66

There are, however, reasons to be skeptical about these proposals. As Fuchs admits, the use of outside experts raises concerns about the “improper delegation” of judicial authority and consequently about the integrity of the judicial process. Her response to the problem is twofold. First, she instructs that an outside expert must be “truly autonomous, lacking any current or known future relationship with either the government or plaintiffs.” Second, she advises that “experts should be used primarily as case administrators” tasked with “winnowing down voluminous records.”67 Both these instructions are problematic, though. An expert who is familiar with the challenges associated with national security will presumably have worked extensively for the government and will therefore likely have ties to serving officials.68 Furthermore, since a security clearance can be taken away, one must also question the disinterestedness of an outside expert who wants to retain her clearance. And if the outside expert has long since retired or has no intention of being employed in the near future, then one must wonder how knowledgeable she can be about contemporary security concerns. Finally, if the purpose of an outside expert is to be limited to merely winnowing records, and “judges must still make the hard decisions over what information is rightfully released,” as Fuchs puts it, then how does the appointment of such an expert actually help the judge overcome his lack of expertise?69

Chesney’s proposal too is problematic. Most immediately, it relies upon public defenders armed with the requisite security clearances to uphold the interests of plaintiffs. Presumably, these lawyers will want to retain their clearances, especially if they have spent their careers developing expertise in the area of national security law. No doubt the termination of the military career of Charles Swift, the navy lawyer assigned to represent Salim Hamdan, will give them some pause.70 More fundamentally, once we move to a system where entire trials are conducted in camera, the distinction between an independent court and an internal executive branch review process begins to fade. It may be objected that the accoutrements of national security will not interfere with the independence of an Article III judge. But consider the likely outcome once a judge goes from occasionally dipping his toes into the world of secret intelligence to embracing it on a permanent basis. It is not unreasonable to expect that a court that routinely handles classified information will become a target of espionage. Nor is it unreasonable to expect the occasional leak from a headstrong clerk or public defender. Additionally, there is the near certainty that these judicial officers will come under increased scrutiny when the protagonists in a national security emergency belong to the same ethnic or racial or religious group as they do. As a result, we should be prepared for the eventuality that judges, clerks, and public defenders will be subject to counterintelligence surveillance and may be liable to prosecution for disclosing classified information. How will the resulting intrusions into the private lives of these actors affect their decisions? What sorts of actions will count as troubling enough to justify the loss of a security clearance? What security will there be against “dirty tricks,” especially those aimed at subduing zealous public defenders? Scholars tend not to discuss these scenarios, but they should, because these are precisely the sorts of threats that employees in the intelligence world must contend with when they expose or challenge perceived wrongdoing.

Suppose the prudential and epistemic concerns outlined above are rejected as overblown. There is still one more reason to support judicial deference—namely, a concern for accountability. The starting point for this argument lies in recognizing the fragility of the idea of expertise. There is a tendency on the part of the critics of judicial deference to assume that there is something akin to a science of intelligence. That is, they seem to believe that there are objective answers to questions about the harm likely to be caused by the disclosure of a given piece of information. However, if intelligence professionals are to be believed, it is not uncommon for intelligence assessments to diverge sharply. This is because, as the Supreme Court observed in Department of Navy v. Egan (1988), the business of intelligence “is an inexact science at best”—one that frequently involves making judgment calls about the consequences of particular choices under conditions of highly incomplete information.71 There is no better example of how such calls can turn out badly than Weatherhead v. United States (1998), one of the very few FOIA cases in which an appellate court has ruled against the executive branch.

Weatherhead emerged out of a FOIA request for a copy of the correspondence between the British Foreign Office and the U.S. government about the extradition and impending trial of a British national. After the British government deemed the correspondence to be confidential in nature, the State Department withheld it, prompting the plaintiff to seek review by the District Court for the Eastern District of Washington. Upon review, the district court ordered that the letter be disclosed because the State Department’s affidavits justifying the withholding had “failed to provide a particularized explanation of how disclosure of the letter would damage the relations between the United States and the United Kingdom.”72 However, following an appeal for reconsideration, Judge Frederick Van Sickle conducted an in camera review and then proceeded to reverse his earlier decision. Now, having read the letter in question, he declared that he “knew without hesitation or reservation that the letter could not be released,” adding that “there is no portion of it which could be disclosed without simultaneously disclosing injurious materials.” The plaintiff, Judge Van Sickle concluded, “would have to be satisfied with the solace of knowing that not only do two high ranking [Department of State] officers believe disclosure of the subject material injurious to the national interest, but so does an independent federal judge.”73

Unimpressed, the plaintiff appealed to the Ninth Circuit, which also examined the document in camera—and reached precisely the opposite conclusion. Writing for the majority, Chief Judge Procter Hug observed that having reviewed the letter, “we fail to comprehend how disclosing the letter at this time could cause harm to the national defense or foreign relations of the United States.” The letter, he added, is “innocuous.”74 The embarrassing contrast between the language used by the district court and the Ninth Circuit did not pass unnoticed. In a sharply worded dissent, Judge Barry Silverman of the Ninth Circuit asserted that the contrast showed that “we judges are outside of our area of expertise here.” “It’s one thing to examine a document in camera for the existence of facts,” he argued, but “it’s a whole different kettle of fish to do what the majority has presumed to do here, to make its own evaluation of both the sensitivity of a classified document and the damage to national security that might be caused by disclosure.”75 For his part, Judge Silverman concluded by siding with Judge Van Sickle on the grounds that the Ninth Circuit should not have treated the “sobering assessment” provided in the government’s affidavits “with so little regard.”76

If only this were the end of the story. Following the Ninth Circuit’s decision, the United States appealed to the Supreme Court. However, as the court prepared to hear oral arguments, the litigants discovered that the British consul in Seattle had long ago transmitted a “significant portion of the contents of the supposedly confidential letter” to the plaintiff.77 Following this embarrassing discovery, the case was deemed moot and the decisions of the lower courts were vacated.

Arguably, what Weatherhead makes clear is that there often is a strongly subjective aspect to the business of predicting the harm likely to be caused by the disclosure of classified information. Before we can strike a balance between the costs and benefits of disclosure, we must take the vital first step of deciding what the costs and benefits are, and here much depends on one’s moral intuitions, political awareness, and common sense—in other words, on one’s political judgment. The classic explanation as to why the courts ought to steer clear of making such judgment calls comes in Chicago & Southern Air Lines, Inc. v. Waterman SS Corp (1948), where Justice Robert Jackson observed that such decisions should be made “only by those directly responsible to the people whose welfare they advance or imperil.”78 This warning may seem overblown in the context of Weatherhead, which was more a farce than a national security emergency, but it does loom large when we consider cases involving extraordinary rendition and secret imprisonment, practices that appear necessary to some but frightful to others.

Here it may be objected that even if decisions about whether and when to release classified information are indeed political in nature, it is still preferable to have this decision made by judges rather than executive officers, as the latter have an obvious conflict of interest. In other words, it can be claimed that the subjective but disinterested decision of a judge is preferable to the subjective and likely self-interested decision of an executive officer, especially since the presence of secrecy means that citizens may not actually be able to hold the latter any more accountable than the former. But is it prudent to assume that judges will remain disinterested for long once they are drawn into the business of substantively evaluating whether secret information ought to be disclosed to the public? On the contrary, it is likely that vesting such politically significant authority in the courts will make appointments to the relevant benches a point of strong contention. And since the appointed judges will need to keep the grounds for their decisions secret, it will be difficult for the public to ascertain that they have not been influenced by partisan or ideological considerations.

An illustration of this difficulty is provided by questions that have been asked about the record of the Foreign Intelligence Surveillance Court (FISC). For obvious reasons, this court examines warrants in camera and ex parte. Should we be disturbed by the fact that this court has reportedly ruled against the United States only a handful of times in the more than twenty thousand cases brought before it?79 Or, on the contrary, should we take the court’s track record as evidence of the government’s unwillingness to bring weak cases before it? And if the latter interpretation seems the more reasonable one, then how do we make sense of reports that the FISC has retrospectively accused the government of misleading it in at least seventy-five instances since 2001?80 We have no informed basis on which to answer any of these questions. Equally, should we be troubled by the decision of the chief judge of the FISC to cooperate with the NSA’s warrantless surveillance program? A number of scholars seem convinced that the NSA program was unconstitutional. Should we be worried then about what other potentially unconstitutional activities the FISC may have permitted? How would we even know if such decisions have been made?

The point, in short, is that compelling judges to take on the responsibility of assessing the harm likely to be caused by the disclosure of classified information is not likely to further our confidence that the norm of transparency is being observed. On the contrary, this step is likely to merely shift the locus of fear—away from officials and toward judges. This is not to suggest that judicial intervention will replicate the conflict of interest that arises when the final say over classification is left in the hands of the executive. Rather, the point is that under conditions of secrecy citizens will lack a reason to be confident that judges are behaving disinterestedly, since the public will not have access to the information necessary for rational trust. This problem would persist, moreover, even if we were to appoint an “independent” Declassification Commission, as scholars and lawmakers have proposed.81 Indeed, the extensive regulatory capture at bodies such as the Securities and Exchange Commission and the Federal Communications Commission, which operate in public sight, suggests that establishing an opaque and unelected “secrecy regulator” ought to actually deepen fears about the misuse of secrecy, seeing as the financial, political, and ideological interests at stake in important national security decisions would give the affected parties a powerful motive to try to “capture” the regulator.

The concern about unaccountable judicial intervention may be challenged in two ways. First, it may be argued that judges are in fact accountable—even when they judge in secret—because unlike those of executive officers, their judgments may be appealed to higher courts. But, as the prognostication of the harm likely to be caused by the disclosure of state secrets depends heavily on political judgment, what exactly can a higher court hold a lower court accountable for? Certainly, a lower court might be sanctioned for outright bias or gross negligence. But how can it be penalized for a difference of opinion over the harm likely to be caused by the disclosure of a piece of classified information? And is the Supreme Court even likely to be able to hold lower courts accountable for “mistaken interferences with regard to foreign and national security policies set by the political branches”?82 There are, as Yoo has reminded proponents of judicial review, hundreds of federal judges, dozens of district courts, and thirteen appellate courts. If the Supreme Court, which currently hears fewer than a hundred cases each year, cannot devote more time to resolving differences between these decentralized actors, then it is reasonable to conclude that “judicial involvement in foreign and national security policy will create disharmony where uniformity is crucial.”83

Second, it may be argued that concerns about unaccountable decision making can be tempered by having members of the intelligence committees in Congress serve in an “advisory capacity,” principally by offering judges their views about whether a request to withhold information is well founded.84 But this proposal is not attractive either. It is widely recognized that Congress tends to defer to the president in the domain of national security. Indeed, this is precisely why proponents of transparency criticize judicial deference—they see the courts as the only safeguard against undue secrecy, especially during emergencies. In their view, judges are insulated from everyday political pressures precisely so that they will ask the hard questions that Congress will not.85 Moreover, presidents have long reserved the right to keep national security information from Congress on the grounds that a lawmaker from a rival party might disclose such information with an eye to gaining political advantage.86 This claim, as we shall see in chapter 3, is not without foundation. Consequently, even if involving Congress lessens the fears that judges are making arbitrary decisions about what information should be disclosed to the public, this may well turn out to be a pyrrhic victory, since it will actually make it harder for the courts to maintain the secrecy they are supposed to protect.

Can Judges Moderate Secrecy?

So far I have been making the case that judges are not well positioned to promote transparency because they are not qualified to challenge the executive’s claims about the harm likely to be caused by the disclosure of secret information. I have also argued that judges should not be asked to make what are effectively subjective judgments about the costs and benefits of disclosure, as this will merely encourage the politicization of the relevant benches and thereby defeat the whole point of turning to the courts, which is to obtain an impartial adjudicator. This is not the end of the road, though. There is another way in which judges may be able to help promote transparency: by policing norms about how officials ought to employ secrecy rather than about what precisely ought to be made public. In particular, by requiring officials to offer reasons in defense of secrecy, judges may be able to moderate the scope and scale of state secrecy—that is, they may be able to help make secrecy shallower.

The concept of shallow state secrecy emerges from Amy Gutmann and Dennis Thompson’s observation that when secrecy prevents the content of policies from being exposed to public scrutiny, it becomes especially important to make sure that citizens “have a chance to decide in advance whether the policy is justified and to review the details of the policy after it is implemented.”87 Of course citizens can have such a chance only if they know that something has in fact been kept secret. Hence Gutmann and Thompson argue that the exercise of secrecy must abide by two conditions: only “the details” of the relevant policy, and not the existence of the policy, must be kept secret; and “the fact that the details of the policy are secret should itself not be a secret.”88 Thus, for example, if the government wishes to establish a spy satellite program, then the norm of shallow secrecy would require it to inform the public or Congress and obtain approval for the policy of having spy satellites even if it does not disclose the details of the program. Only if the government acts in this way, Gutmann and Thompson argue, will citizens have the “opportunity to challenge the keepers of secrets and ultimately to decide whether the secret should be kept.”89

But must officials always keep secrets shallow? That is, are they obliged to declare the existence of every state secret? It is not difficult to imagine cases where announcing the existence of a state secret can itself have undesirable consequences—for example, merely acknowledging the existence of a spy satellite program could prompt terrorist organizations to take countermeasures like moving their training facilities underground. Gutmann and Thompson acknowledge this possibility when they discuss “deep secrets”—secrets whose very existence is kept secret from citizens and lawmakers out of a concern for national security. They argue that such deep secrets can be “justified only if they can be shown to be necessary to safeguard the democratic values of basic liberty, opportunity, and deliberation, and only if this showing can meet the test of accountability.”90 But to whom should this showing be made? It cannot be made publicly, as this would lead to the disclosure of the very information the government wants to protect. Yet if officials are not required to make a public showing, then how can we know whether they are exercising deep secrecy for the right reasons?

It may be objected that it is unlikely that officials will need to maintain many deep secrets. But even if this were true, it remains the case that even an “ordinary” state secret can be maintained at varying depths. For instance, the United States could either declare that it has a spy satellite program or it could declare that it has a dozen spy satellites that cost X and feature Y and Z groundbreaking capabilities. Before we can determine whether officials are obliged to make the latter, more informative declaration, we need to know something about the harm this declaration may engender (such as whether it will in fact prompt terrorist organizations or rival states to move their training facilities or weapons factories underground). It seems reasonable to assume that the executive alone can supply this estimate. But in that case what is to prevent officials from overstating the cost of the latter declaration, leading us to conclude that they are obliged to make only the former, less informative declaration? Needless to say, if we cannot prevent such “threat inflation,” then demands for shallow secrecy will not greatly increase the extent of information available to the public.

So how can we enforce the norm of shallow secrecy—that is, how can we make sure that officials will say all that can safely be said about the secrets in their possession? Are the courts any more likely to be able to help on this front? At first glance the evidence seems promising, especially in the FOIA context where the courts have demanded that officials submit “a detailed public justification for any claimed right to withhold a document.”91 The purpose of this requirement, as the D.C. Circuit explained in the landmark case Vaughn v. Rosen (1973), is that it allows judges to determine whether the records that officials wish to withhold are in fact related to the exemption being claimed, and it also provides plaintiffs an “opportunity to contest the withholding of the documents,” and to identify and seek the disclosure of those records that are less likely to be exempt from disclosure.92 To this end, Vaughn obliges officials to submit a detailed and itemized affidavit (now referred to as a “Vaughn index”) that “must describe the material being withheld, state the justification for nondisclosure, and cite each exemption asserted.”93 This requirement nicely parallels the idea of shallow secrecy. However, the picture is complicated by the fact that Vaughn also allows affidavits to exclude “factual descriptions that if made public would compromise the secret nature of the information.”94 This concession—that an explanation as to why a secret ought to remain a secret could itself harm national security—has given rise to an array of cases in which officials have been excused from having to even acknowledge the existence of secrets, much less to justify keeping the secret itself.

The foundational case here is Phillippi v. CIA (1975), one of a series of cases in which plaintiffs sought records pertaining to the Glomar Explorer, a ship that had reportedly been employed in a covert operation aimed at recovering a sunken Soviet submarine. In the case at hand, a journalist, Harriet Phillippi, requested from the CIA information about its contacts with reporters covering the Glomar Explorer story. The CIA responded that it could not fulfill Phillippi’s request as the information sought involved an activity that “in the interest of national security” could “neither be confirmed nor denied” (a response that has since been named a “Glomar response”).95 Phillippi then approached the district court, asking it to compel the CIA “to provide a detailed justification for each document claimed to be exempt from disclosure.” The CIA responded by submitting classified affidavits, which the district court then examined in camera. After the district court subsequently issued a summary judgment in favor of the CIA, Phillippi approached the D.C. Circuit, arguing that courts were entitled to examine only the withheld documents in camera, not the affidavits themselves. However, the D.C. Circuit rejected this claim. It concluded that courts were entitled “to examine classified affidavits in camera and without participation by plaintiff’s counsel” because “when the Agency’s position is that it can neither confirm nor deny the existence of the requested records, there are no relevant documents for the court to examine other than the affidavits which explain the Agency’s refusal.”96

I have been arguing that efforts to enforce the norm of shallowness—that is, to compel government officials to declare the existence, if not the content, of state secrets—can be frustrated by the fact that officials can claim (1) that acknowledging the existence of a state secret could harm national security, and (2) that explaining why acknowledging the existence of a state secret could harm national security could itself harm national security. The courts, to their credit, have pushed back against the latter of these claims. Although the Phillippi court allowed officials to submit classified affidavits explaining why they could not confirm or deny the Glomar Explorer operation, it instructed the CIA to “provide a public affidavit explaining in as much detail as is possible the basis for its claim that it can be required neither to confirm nor to deny the existence of the requested records.” The purpose of doing so, the Phillippi court said, was to “create as complete a public record as is possible.”97

But how detailed can such a public record be? Can it reliably procure the shallow secrecy that Gutmann and Thompson recommend? The prospects appear dim. As the D.C. Circuit acknowledged in Hayden v. NSA (1979), when “itemization and justification are themselves sensitive … to place them on public record could damage security in precisely the way that FOIA Exemption 1 is intended to prevent.”98 In other words, Hayden confirms that in some cases officials may legitimately refuse to offer any public explanation whatsoever for their decision to withhold information from a FOIA requester. Of course not every FOIA case involves the supersecret NSA as Hayden did. In two well-known cases, Gardels v. CIA (1982) and Miller v. Casey (1984), officials refused to confirm or deny the existence of the requested records, but they did offer public affidavits explaining why confirming or denying the existence of the state secret in question could harm national security. In Gardels, which involved a request for information about covert CIA contacts with members of the University of California, the CIA explained that it could not confirm or deny such contacts because if it “were required to indicate those schools with which it had had no covert contact,” then foreign intelligence organizations “could and would concentrate their efforts on the remaining American colleges and universities” in order to “discover what the CIA is up to” on those campuses.99 Arguably, this explanation satisfies the norm of shallowness—it offers the public a plausible reason as to why the CIA cannot confirm or deny the existence of a secret activity. However, at this degree of abstraction we have no way of knowing whether campus contacts are being used in ways we might consider improper (for instance, to entrap foreign students). At the same time, there is little we can do to obtain more detail. So long as we defer to the CIA’s factual claim that “foreign intelligence agencies are zealous ferrets,” we have little basis on which to demand greater specificity in the public affidavits filed by officials, lest this alert foreign observers and thereby endanger national security.100

The justifications for secrecy tend to be even more vague in cases where the state secrets privilege has been invoked. The first significant effort to extend the Vaughn standard to this domain came with Halkin v. Helms (1982) where the plaintiffs, who were seeking to discover whether they had been subjected to unlawful surveillance, argued that the public affidavit submitted by the CIA justifying its invocation of the state secrets privilege “was too vague to establish the privilege.”101 The D.C. Circuit disagreed. The CIA’s public affidavit, it argued, “while necessarily unspecific, set forth the grounds requiring secrecy in this context.”102 The grounds were twofold: first, that “revelation of particular instances in which foreign governments assisted the CIA in conducting surveillance of dissidents could strain diplomatic relations,” and second, that information that allowed the plaintiffs to know they have been the subjects of surveillance, might “when combined with knowledge of the individual’s other activities,” allow foreign intelligence agencies “to identify CIA operatives.”103 As in Gardels, the FOIA case examined above, these justifications for invoking the state secrets privilege are pitched at such an abstract level that they shed little light on the CIA’s conduct. Indeed, these justifications could theoretically support any invocation of the state secrets privilege by the CIA, since the revelation of almost any of its covert activities could “strain diplomatic relations” in one way or another.

So ought the courts to demand the submission of a more detailed public justification for the invocation of the state secrets privilege? This question was addressed frontally in Ellsberg v. Mitchell (1983), a case in which the protagonists in the Pentagon Papers episode sought damages after learning that they had been subjected to warrantless surveillance. After the District Court for the District of Columbia sided with the attorney general on the basis of in camera submissions justifying the invocation of the state secrets privilege, the plaintiffs approached the D.C. Circuit, challenging the “meager public justifications” offered by the government. The public affidavits, the plaintiffs argued, alleged “that the national interest would be prejudiced by release of the requested materials,” but did not specify “what that ‘interest’ consists of and how it might be damaged.” Hence the government should be compelled to “provide a fuller public account of why disclosure of the information would harm national security” or, alternatively, to explain “why a more specific description of the anticipated adverse consequences would itself damage national security.”104

Judge Harry Edwards expressed sympathy with the plaintiff’s complaint, observing that “the more specific the public explanation, the greater the ability of the opposing party to contest it.” Yet there could not be, he argued, “a strict rule that the trial judge must compel the government to defend its claim publicly before submitting materials in camera,” because it is “imperative that the procedure used to evaluate the legitimacy of a state secrets privilege claim not force disclosure of the very thing the privilege is designed to protect.” The most that could reasonably be demanded, he wrote, was that “the government’s public statement need be no more (and no less) specific than is practicable under the circumstances.”105 And who is to decide what is “practicable”? Though there could be “no abdication of a judicial role in connection with proposed applications of the state secrets doctrine,” Judge Edwards declared, the courts “should accord considerable deference to recommendations from the executive department.”106

What Is the Alternative?

There are, as we have seen, good reasons why judges will not—and indeed should not—take the lead in compelling the revelation of secret information. Consequently, we should expect judges to set only a low procedural hurdle rather than a high substantive barrier before the executive’s pleas for secrecy on national security grounds. This does not mean that judges can make no valuable contribution whatsoever in combating the abuse of secrecy. As Chesney has underscored, procedural requirements ought to be an important “precondition” for judicial deference, because they allow judges to verify that harm-claims have not been made carelessly.107 And as Pozen has rightly emphasized, deference is not the same thing as abdication; a deferential standard of review does not preclude “demanding plausible arguments tailored to specific documents withheld.”108 There is no reason to believe that the courts are unqualified to conduct close scrutiny in this more limited sense. As Justice Powell dryly stated in Keith, if a purported threat “is too subtle or complex for our senior law enforcement officers to convey its significance to a court,” then one must wonder whether it really exists.109

Happily, there is evidence that procedural requirements have had a sobering effect on officials from the executive branch. In the FOIA context, for instance, the Justice Department has warned government lawyers that judges will not grant summary judgments when affidavits contain “boilerplate explanations” that are not “tailored” to justify the “particular information” that officials want to withhold.110 Meanwhile, a remarkable recent case, Horn v. Huddle (2009), has reminded officials that procedural requirements associated with the invocation of the state secrets privilege are not to be taken lightly either. In this case District Judge Royce Lamberth of the District of Columbia moved to penalize government lawyers and senior CIA officials after discovering that the state secrets privilege had been invoked fraudulently: one of the litigants was not in fact a covert operative at the time of the trial (as the CIA had claimed).111 Shaken, the CIA reached a three-million-dollar settlement with the plaintiff in short order so as to preempt further legal proceedings.

Though Horn certainly sends out a clear signal that federal judges are not to be trifled with, it is important to recognize the limits of procedural requirements. Arguably, Horn indicates that such requirements will allow judges to guard against only ham-handed uses of secrecy. Indeed, had a government lawyer not voluntarily informed Judge Lamberth of the CIA’s fraudulent claim, Horn might have turned out very differently.112 Much the same lesson holds as far as FOIA is concerned. That is, there is no denying that once embarrassing or incriminating information has been made public, FOIA has had a powerful countervailing effect, as Seth Kreimer has detailed in two excellent essays on the subject.113 For example, though the American Civil Liberties Union (ACLU) has not prevailed in FOIA cases where it has requested the disclosure of highly classified documents pertaining to the United States’ controversial detention policies, it has nevertheless been able to use FOIA to shake loose reams of associated documents—cumulatively upwards of a hundred thousand pages—that have helped shed light on a controversial practice. Still, we must not lose sight of the fact that the ACLU was able to make good use of FOIA only after leaks brought reports of potential wrongdoing to its notice. The “ecology of transparency,” as Kreimer has nicely put it, is that “leaks provide the basis of FOIA requests, and FOIA requests in turn provide an occasion for courts to address the justifiability of continued secrecy.”114 But the worry here, as noted in chapter 1, is that the practice of leaking violates the law and is therefore prone to suppression. There is, to stay with Kreimer’s metaphor, a standing threat to the sustainability of the “ecology of transparency.” We will address this vulnerability in chapters 46. But let us first examine whether it is possible to bolster the effectiveness and credibility of congressional oversight.