Necessary but Illegitimate: On Democracy's Secrets

Abstract Transparency has become the constant refrain of democratic politics. However, executive branch officials consistently seek to insulate their activities from public scrutiny. A recurrent rationale presents secrecy as a necessary measure called for in circumstances in which the basic interests of the state are at stake. The purpose of this paper is to normatively assess the appeals to the necessity of executive secrecy in democratic governance. The paper fleshes out two ways in which the necessity argument has been framed. It argues that an appeal to necessity fails to confer political and/or legal authority on the state's resort to secrecy because necessity escapes normative codification both in the moral and legal domain (“necessity knows no law”). Drawing a distinction between legitimacy and vindication, it argues, however, that even though a state resorting to secrecy acts beyond its democratic authority, this action may be vindicated.

As Pierre Rosanvallon observes, transparency has become "the new democratic ideal" and the "paramount virtue" of modern politics. 1 Executive branch officials, however, consistently seek to insulate their activities from public scrutiny. A recurrent rationale points to the necessity of secrecy in governance. For example, when the UK government refused to disclose the British Cabinet minutes from March 2003, where military action in Iraq was Dorota Mokrosinska is associate professor of political philosophy in the Institute of Philosophy at Leiden University, Nonnensteeg 1-3, Leiden 2300 RA, Netherlands (d.m.mokrosinska@hum.leidenuniv.nl).
I thank Thomas Fossen and participants in the Political Theory Colloquium at Hamburg University for helpful feedback on earlier versions of this paper. I gratefully acknowledge Ruth Abbey, the editor of the journal, for her generous advice and guidance and two anonymous reviewers for most helpful insights and comments. deliberated, it claimed that secrecy was necessary to ensure the effective operation of Cabinet government. The Attorney General argued that "conventions on Cabinet confidentiality are of the greatest pertinence when the issues at hand are of the greatest sensitivity. . . . Ministers must have the confidence to challenge each other in private. . . . Disclosure of Cabinet minutes . . . has the potential to . . . compromise the integrity of this thinking space where it is most needed." 2 Besides arguments presenting secrecy as a necessary tool ensuring the integrity of decision-making processes and the effectiveness of government action, there are arguments that present secrecy as a necessary condition of national security. For example, when the Polish government refused to either confirm or deny the existence of the CIA black sites on its territory between 2002 and 2004, it claimed that secrecy about the CIA extraordinary rendition program was necessary for national security. While Poland's participation was not motivated by terrorist threats to its national security, Polish decision-makers at the time saw alliance with the United States as a strategic national security guarantee against Russia. 3 Since "the interest of the Polish state [was] based on necessity to participate in the anti-terrorist coalition," according to the Polish prime minister, 4 and since the US-Poland intelligence cooperation fell under NATO's COSMIC Top Secret classification, secrecy was deemed necessary to national security.
"Political survival" in the face of national security threats and "government capacity for action" are two dominant narratives in terms of which the necessity argument has been framed. Each of them, as proponents of the necessity argument claim, makes a convincing case that government's resort to secrecy is a legitimate exercise of democratic authority. The purpose of this article is to normatively assess this claim. When secrecy is necessary for democratic states to survive and effectively operate, is it thereby democratically legitimate?
I start by homing in on the two forms of the necessity argument. I argue that in both versions, the argument follows the logic of the raison d'état tradition. That tradition is antithetical to democracy in that it conceives of secrecy, along with other special powers adopted by the state in situations of necessity, as creating a space for an exercise of power devoid of the moral and legal principles that govern liberal-democratic states. I consider whether the necessity argument can escape this problem first from the perspective of political morality and then from the perspective of legal theory. I argue that an appeal to necessity cannot be codified in terms of the moral and legal 2 Dominic Grieve, "Exercise of the Executive Order under Section 53 of the Freedom of Information Act 2000," https://www.gov.uk/government/uploads/system/uploads/ attachment_data/file/60528/Statement_of_Reasons-31July2012_0.pdf, 3-4, accessed July 7, 2022.

THE REVIEW OF POLITICS
principles that the political authority exercised by democratic governments presupposes. This leads me to conclude that the necessity argument fails to present secrecy as an exercise of democratic authority and fails to escape the antidemocratic implications of reason of state thinking. Contrary to what proponents of the necessity argument claim, necessary secrets are not a legitimate form of democratic governance. Neither the UK government's withholding of Cabinet minutes regarding its involvement in the Iraq war nor the Polish government's refusal to confirm or deny its involvement in the secret CIA extraordinary renditions program can enjoy democratic legitimacy in virtue of them being necessary for, respectively, the integrity of government decision-making processes and national security. I do not mean to deny that executive secrecy in situations of necessity can ever be legitimate, but only that an appeal to necessity does not legitimate it.

Secrecy and the Political Survival of the State
In its first narrative, the necessity argument presents secrecy as an emergency measure employed to ensure the state's political survival. 5 The emergency rationale is tightly linked to national security. It draws on the idea, deeply embedded in bureaucratic and popular culture, that the defense of national security demands strict controls on the flow of information. 6 In situations of national security crises, such as terrorist threat, invasion, or war, disclosure of certain classes of information would make the state vulnerable to its enemies: making troop positions public would make them an easy target for the enemy; revealing sources and methods of intelligence would expose them to countermeasures. Diplomatic relations are another policy area in which secrecy is considered integral. Diplomatic documents, when misused, can harm the vital interests of the state. Thus, even though democratic states commonly recognize a right to access to information, they also widely accept a diplomatic exception to it (e. Over the last decades security and defense factors have been presented as interrelated with other political, economic, and social factors allowing the presentation of issues that go beyond questions of traditional security as potential security threats. 9 As a result of such "securitization" of public policies, the necessity argument has expanded to policy domains ranging from migration, energy, climate change, and critical infrastructure to public health and finance. 10 The British Security Service Act of 1989, for example, imposes on the security service the function of "safeguard[ing] the economic well-being" of the nation; the US National Security Education Act of 1991 similarly makes a direct connection between national security and the "economic well-being of the United States." 11 As the scope of national security issues expands, so does that of emergency powers. Calling immigration a "threat to national security" in European programs of border surveillance and migration policy has opened the door to state secrecy in migration policy enforcement. 12

Secrecy and the Executive Capacity for Action
While the first necessity rationale of executive secrecy presents it as a condition of political survival for the state, the second presents it as a necessary condition of the effectiveness of government action. "Some democratic policies require secrecy," Dennis Thompson writes; "if they were made public, [they] could not be carried out as effectively or at all." 13 The two frames overlap. In the realm of security and defense, for example, secrecy is a strategic resource employed to enhance the effectiveness of intelligence sources and methods. Given their role in national security, its effectiveness therein is also a matter of political survival.
In policy areas other than defense and security, the effectiveness rationale of state secrecy often stands on its own. In economic policy, secrecy is presented as a condition of the effectiveness of government financial market interventions such as currency devaluation or price decontrol. It is also invoked as a condition of effectiveness in the realm of law enforcement: the police could not infiltrate criminal networks if it revealed in advance where it was 9 Barry Buzan, Ole Waever, and Jaap de Wilde, Security: A New Framework for Analysis (Boulder, CO: Lynne Rienner, 1989). 10

76
THE REVIEW OF POLITICS going to place undercover agents and how to recognize them. The effectiveness rationale is also invoked to justify secrecy of the deliberations of executive and legislative bodies. A number of policy-oriented studies demonstrate that opening decision-making processes to the wider public and the media increases the likelihood of deadlock rather than compromise among decision makers and lowers the quality of debate. 14 On this view, secret settings serve democratic decision-making processes because they insulate decision makers from external pressures and stimulate serious discussion and the frank exchange of views, facilitate compromise, and produce better arguments. 15 Cornelia Ulbert and Thomas Risse's and David Stasavage's studies of the proceedings of the European Council of Ministers support these findings. 16 Steiner and his colleagues draw similar conclusions from their studies of the transcripts of parliamentary deliberation in Switzerland. 17 The functional necessity of deliberative secrecy is also used to justify special powers of the executive. In the United States, deliberative process privilege protects from disclosure privileged communications within or between government agencies. The need for deliberative secrecy motivates one of the exemptions to the US Freedom of Information Act which entitles the executive to withhold internal deliberation, policymaking, and inter-or intradepartmental records that are predecisional from the public view. A similar principle is written into the German constitution. The Kernbereich exekutiver Eigenverantwortung was introduced by the German Federal Constitutional Court in 1984 (BVerfGE 67, 101). This principle establishes a constitutional right to deliberative secrecy on the part of the executive by recognizing that the government needs a protected sphere for deliberation, free from parliamentary interference, before making a decision. In all these cases, the effectiveness rationale for secrecy in governance focuses on the state's capacity for action (which need not relate to averting an acute danger to national security). Given that openness would undercut it, secrecy, as a necessary condition of its success, is privileged over transparency.
The two versions of the necessity argument have a similar structure. Both present secrecy as a tool for advancing important state interests. In both cases we also recognize that secrecy undermines transparency and other moral and legal principles that otherwise underpin the liberal-democratic order.

Democratic Deficits of Secrecy
One concern about government secrecy is that secret policies and closed-door decision-making processes generate a knowledge deficit that undermines mechanisms of democratic accountability: people cannot control and hold state officials to account if they do not know what they are doing and why. As Richard Aldrich and Daniela Richterova observe, "secrecy is often interpreted . . . as providing a discretionary space of action . . . opaque to the cleansing effect of democratic scrutiny." 18 Mechanisms of accountability are among the arrangements ensuring representative responsiveness to citizens' wishes and in this sense they are a condition of popular sovereignty. Deprived of the ability to demand that representatives explain and justify how they pursue citizens' aims and, if needed, to demand a change, this distinctive feature of democratic governance is lost.
A related concern is that government secrecy leaves people out of the collective decision-making process. Equality of decision-making power between citizens and representatives and the idea that citizens should remain coauthors of laws and policies to which they are subject are among key democratic ideas. 19 When decisions are made behind closed doors, excluding people from the decision-making process, their equal status disappears. Excluding people from the decision-making process bears on the legitimacy of that process. It is commonplace to think that in a democracy, political decisions are legitimate only if they are authorized by citizens. 20

78
THE REVIEW OF POLITICS power seem to lack democratic authorization because people cannot authorize what they are denied knowledge about. 21 Finally, executive secrecy cuts democratic deliberation short in situations in which democratic self-governance requires it most, namely, when people disagree about whether the situation qualifies as a necessity and about the scope of secrecy that may be required in response. This problem emerged with particular force in the context of the National Security Agency (NSA) secret surveillance program leaked by Edward Snowden. That program relied on a number of assumptions about which reasonable citizens disagree, such as the level of risk to public security and the scope of the trade-offs between security and individual privacy a society is willing to accept. The secrecy prevented public debate on these issues. Without it, according to Eric Boot, the NSA's unilateral decision that its intelligence function was so important to public security that it condoned violating individuals' privacy online lacked democratic legitimacy. 22

Democratized Reason of State?
Whereas the necessity argument recognizes that the state's resort to secrecy in situations of necessity violates moral and legal principles that otherwise constrain political action, its proponents argue that the benefits of secrecy, measured in terms of public security and/or government capacity for action, justify this violation. Such consequentialist arguments granting the state, in situations of necessity, special powers to do something it could not otherwise do have a long history. Scholars link them to the reason of state tradition. "The idea of 'reason of state,'" as Nancy Rosenblum put it, "captures this tension between legal concerns and grim necessity." 23 Reason of state doctrine emerged in early modernity and is commonly associated with Niccolò Machiavelli and the thinking about politics that he instigated. 24 Secrecy has been a defining element of the doctrine since its inception. A distinctive feature of reason of state politics is the recognition that serving the state's vital interests may require a violation of moral or legal norms. As Friedrich Meinecke put it in his classic survey of the concept's intellectual history, Raison d'état is the fundamental principle of national conduct, the State's first Law of Motion. It tells the statesman what he must do to preserve the health and strength of the State. . . . The well-being of the State and of its population is held to be the ultimate value and the goal . . . which must-without any qualification-be procured. Without qualification, insofar as it must even be procured if necessary at the expense of a complete disregard for moral and positive law. 25 The contemporary necessity defense of state secrecy thus restates the classic doctrine of reason of state. Yet there is a caveat. Reason of state thinking has been deemed antithetical to democratic governance because the powers it confers on the state to respond to situations of necessity are unlimited, standing beyond the legal order of a liberal-democratic state and devoid of democratic control. Carl Schmitt, for example, openly acknowledged the antidemocratic character of emergency powers. 26 Were the necessity argument a mere restatement of reason of state, its proponents could not be expected to offer an account of the democratic legitimacy of the special powers of the state. If anything, then, they have to deny the antidemocratic character of the special powers that the state exercises in situations of necessity. Their strategy is to present "political survival" and the "effectiveness of government action" as democratic goods along with accountability, transparency, equality, and self-governance. As long as the ends pursued by the state are democratic, the means-the powers conferred on the state necessary to achieve them-will be democratic too. Thus, even if the pursuit of "political survival" and the "effectiveness of government action" involve infringing on democratic accountability, equality, the people's right to self-governance, and political participation, this infringement is democratically justified.
In this "democratized" version of the reason of state argument, the state's resort to special powers overriding moral and legal principles underpinning the liberal-democratic order in situations of necessity is a legitimate exercise of democratic authority. This is, for example, the thrust of Gabriel Schoenfeld's defense of the Manhattan Project, a military research program that produced the first nuclear weapons undertaken during World War II in the United States. Not only was the American public kept in the dark but even those who worked on the project were often unaware of the nature of the product they were constructing. Schoenfeld claims that "selfpreservation" is "the most fundamental business of democratic governance." 27 Given that the project was considered necessary for the survival of American democracy, its secret character, he argues, was legitimate; the 25

80
THE REVIEW OF POLITICS democratic deficit pertaining to executive secrecy was offset by the positive consequences it enabled. 28 A similar argument is proposed in defense of secrecy as a means of ensuring the effectiveness of government action. Thus one claims that the government's capacity for action is a democratic good and that the benefits of securing it outweigh the costs of sacrificing the democratic commitment to open government. Given that the costs of upholding government's transparency, measured by the negative consequences in terms of the democratic interest in the government's capacity for action, are too great for the liberaldemocratic state rationally to bear, one concludes that secrecy is democratically legitimate. As Jenny de Fine Licht and Daniel Naurin claim, "'Getting things done' is also a core democratic value, which-under certain circumstances-might call for privileging secrecy over transparency." 29 This democratized version of the reason of state argument relies on balancing democratic interests, in particular, political survival and/or the effectiveness of government action, on the one hand, and accountability, political equality, and a right to self-governance and political participation, on the other. To assess the success of the necessity argument is to inquire whether the exercise in balancing different democratic interests it proposes is plausible. I argue that this argumentative strategy is problematic both at the level of philosophical and of legal discourse. It fails, as a democratic defense of state secrecy, for the same reason that raison d'état politics is considered antithetical to democracy.

Necessity Escapes Moral Codification
At the level of moral theory, the necessity argument runs up against the problem that the cost-benefit analysis at its core balances traditionally consequentialist considerations (public security and political efficiency) with deontological considerations (a right to hold decision makers to account anchored in democratic values of equality and self-governance). This balancing exercise is problematic because deontological considerations provide us with a reason to observe them irrespective of the beneficial consequences that may follow from violating them; they resist cost-benefit analysis and the utilitarian metric that goes with it. By extending cost-benefit analysis to deontological considerations, the necessity argument effectively forces a conversion of a deontological framework into a consequentialist one. Whether this can be done places the problem of the legitimacy of executive secrecy in the context of old disputes in moral theory. 28 Ibid., chap. 7. For orthodox consequentialists, the right act in any situation is the one that will produce the best consequences, as judged from an impersonal standpoint, which gives equal weight to everyone's interests. 30 As consequences are the only moral currency in terms of which moral judgments are to be made, whenever the importance of political survival or effectiveness of government action outweighs the importance of accountability, the state's resort to secrecy is legitimate. 31 Orthodox deontologists believe that there are moral principles that are unconditional and may never be violated because they express values that are incommensurable and nonexchangeable. Trading them off against other values is inconceivable as there is no common currency to make the weighing exercise possible. 32 For an orthodox deontologist, then, the necessity of political survival and/or effectiveness of state action can never legitimate the state's resort to special measures that override the deontological considerations underpinning the principle of accountability, namely, equality and people's right to self-governance and political participation.
Both positions suffer from a similar weakness: Neither takes seriously the dilemma that may arise between, security and/or effectiveness of government action that require secrecy, on one hand, and citizens' right to call government to account that requires transparency, on the other. Instead, they define the dilemma out of existence. 33 An intermediate position, endorsed by most contemporary deontologists, holds that deontological constraints apply so long as the negative consequences remain under a certain threshold. 34 Once the threshold is reached, consequentialist considerations dominate.
The trouble with this position is that it fails to explain how deontological considerations such as the values of equality and self-governance underpinning the accountability requirements change their unconditional character once the threshold has been reached: as Allon Harel and Assaf Sharon ask, if "consequences do not determine rightness and wrongness of actions, why does this change when their weight increases?" 35 This problem bears on our discussion concerning the state's authority to resort to special measures in the following way. Special powers claimed by the state in situations 30 35 Harel and Sharon, "Necessity," 851.

82
THE REVIEW OF POLITICS of practical necessity may require violating otherwise unconditional moral principles underpinning the political order. To claim the authority to violate such principles is to claim that there are moral principles that make their violation permissible conditional on necessity. If threshold deontology is successful in stipulating such principles, the state's resort to special measures, including secrecy, could be seen as legitimate. If threshold deontology is not successful, the state's resort to such measures has no plausible ground. I join those commentators who argue that threshold deontology is incoherent. My argument draws on Harel and Sharon, who demonstrate the implausibility of threshold deontology by focusing on the threshold-level permissibility of the state's resort to torture in situations of necessity. 36 I extend their argument to secrecy, showing that it applies to a broader scope of special measures the state adopts when confronted with situations of necessity. My contribution to this debate is to connect it to the discourse on the authoritative powers of the state: I demonstrate that in the absence of moral principles supporting threshold deontology, the state's resort to special measures (secrecy, torture) cannot be seen as an exercise of political authority.
Threshold deontology carves out space for departures in the otherwise unconditional principles once the threshold of necessity is reached. Two problems emerge when violation presents itself as a legitimate option. First, once we grant that it is permissible to violate moral principles when the necessity threshold has been reached, the status of the threshold is eroded. If a new threat arises which the threshold model did not foresee, it is not clear whether a departure from the principle is permissible. To justify action in these circumstances, the content of the threshold rule should be specified. Under the conditions of pluralism and disagreement that characterize modern societies, this can be done in different and competing ways, which presents us with the necessity of settling on one. The solution, as forcefully argued by many scholars, especially Kantians, is political: as Jeremy Waldron, Arthur Ripstein, and Anna Stilz argue, the state is an impartial arbiter charged with laying down the content of the right action and addressing any residual indeterminacy in cases of conflict. 37  addressing the new threat requires lowering the threshold of permissibility for violating moral principles? Since any subsequent threat may be argued to require overriding the previous threshold, this sends the argument down the slippery slope of hollowing out moral principles until they are effectively overridden. Making the permissibility of violations of moral principles conditional on necessity makes it hard to constrain them.
In political practice such concerns resonate with the worries voiced by activists and scholars about the unlimited powers that special measures confer on the state. Regarding executive secrecy, they indicate that the ease with which government officials "exaggerat[e] the need for secrecy" 38 sends government classification practices down a slippery slope towards overclassification, or withholding information in an ever-expanding range of situations in which the necessity of secrecy is increasingly problematic. As Mark Fenster observed, the WikiLeaks and Snowden megaleaks offered an opportunity to test the government's appeals to the necessity of secrecy. The outcome, he claims, does not support government claims made about the necessity of keeping the relevant information secret. 39 A similar concern was raised in response to my earlier example of the UK government's refusal to release the official minutes of Cabinet meetings regarding British involvement in the Iraq war. Arguing that disclosure would have a negative impact on the quality of future Cabinet deliberations, the government resorted to a veto power contained in section 53 of the FOIA that allows the executive to block disclosure in "exceptional circumstances" (MOJ 2012). Commenting on the refusal, the Information Commissioner (the UK's independent regulatory authority for information rights) observed that the veto contained a danger of sending government secrecy on a slippery slope toward permanent withholding of minutes of Cabinet discussions: "If the veto continues to be exercised in response to the majority of orders for the disclosure of Cabinet minutes, it is hard to imagine how the most significant proceedings of the Cabinet will ever be made known before the elapse of 30 years . . . [such] disclosures . . . will, by definition, always be the ones to attract the veto as an 'exceptional case.' 40 The slippery slope dynamics of appeals to necessity are not exclusive to executive secrecy. The ease with which the executives use special powers to suspend fundamental civil rights beyond the purpose for which these powers are initially invoked is a case in point.

84
THE REVIEW OF POLITICS calamity, the French constitutional emergency powers afford the police powers to search and to impose house arrest without prior judicial authorization and suspend a number of fundamental rights. These powers were invoked after the attack on the Bataclan theater in Paris in 2015, but the French executive extended them to control unrest about the United Nations climate summit that was held in Paris two weeks after the attack, prohibiting protests against the summit, canceling marches, conducting home searches, and imposing house arrest on climate activists. 41 The second problem that emerges once threshold deontology builds a deviation space into otherwise unconditional principles is that we treat unconditional principles as if they were conditional. We mold them into a consequentialist framework and deny their deontological status or, as Waldron put it, we stretch and deform them. 42 This corrupts the practical reasoning of the agents and changes our moral and political landscape beyond recognition. Harel and Sharon ask us to imagine a situation in which resorting to torture would avert a catastrophe. 43 By incorporating a principled exception to the rule prohibiting torture, an agent who considers what to do in a particular case is invited to consider the possibility that torture is permissible. Even when torture is eventually rejected on the grounds that the circumstances do not call for it, its permissibility has been elevated to the status of a rule-like directive. Harel and Sharon argue that this would corrupt an agent's practical reasoning because the agent considers a rule permitting torture to be on par with a rule prohibiting it whereas, when the circumstances do not justify torture, torture ought to be not merely rejected, but not even considered as an option to be weighed against other alternatives.
If violations of unconditional moral principles were incorporated into the moral system as principled permissions under a moral rule, they would create rights to act accordingly. Torture conditional on the necessity to avoid a catastrophe would create a right to torture. Once we make it legitimate for the state to exercise this right on the grounds of public security, we make it legitimate for it to treat the people's right to bodily integrity and their human right not to be tortured as conditional. Such a state would not be the liberal-democratic state we are familiar with.
No moral principle can make it morally permissible to do what is otherwise impermissible. Necessity requirements escape codification in terms of moral principles or, as Aquinas put it, "necessity knows no law" (necessitas non habet legem). 44 If no moral principle suspending moral principles is possible, no such principle could underpin the state's authority to resort to special measures that violate otherwise unconditional moral principles.
This need not mean that the state is never permitted to resort to special measures in the face of necessity nor that it cannot sacrifice transparency and accountability when secrecy is necessary to national security. However, we have to stop thinking of infringements of fundamental moral norms as acts permissible under moral rules. Harel and Sharon distinguish between principled and unprincipled acts, a distinction familiar from the dirty hands and just war literatures. In these contexts, scholars recognize that circumstances can make violations of moral principles by political actors necessary and therefore permissible, but claim that such necessary violations are not permissible by some competing moral norm or principle. As Harel and Sharon rephrase this position: "Violations of our most fundamental norms may be unavoidable. But . . . this does not entail a rejection or modification of our basic rules. What allows the [violation] is solely the necessity of avoiding catastrophe, not a different law allowing [violation] under some conditions." 45 Their position claims that moral principles know no exceptions; it concedes, however, that there may be exceptional cases beyond moral principles. In terms of practical reasoning, this means that the political agent violating moral principles is not governed by a moral principle permitting the violation under exceptional circumstances. Even if the agent violating a moral principle is governed by that moral principle in the sense that she decides to go against it, the when and the why of its violation are not so governed. As Harel and Sharon explain, necessary violations are permitted as "singular act[s], particular to the case at hand and not a generalizable norm that may be used in guiding future decisions." 46 The argument that necessary violations remain morally unlegislatable has clear implications for the state's authority to resort to special measures, including secrecy. Given that (a) the political authority of the state is founded on moral principles and subject to corresponding constraints, 47 and (b) necessity-driven departures from fundamental moral norms, pace threshold deontology, resist framing in terms of moral principles or, as Harel and Sharon put it, "resist rule-governed normativity," 48 necessity cannot ground the political authority of the state to violate fundamental 44 Thomas Aquinas, Summa Theologica (London: Penguin Classics, 1999), I-II, q. 96, art. 6. 45 Harel and Sharon, "Necessity," 863. 46 48 Harel and Sharon, "Necessity," 857.

86
THE REVIEW OF POLITICS moral norms. Acts of violation of fundamental moral norms can be performed strictly from the necessity of the circumstances and not as a matter of authorized state policy. This argument lays bare some of the problems involved in the seminal defense of state secrecy proposed by Amy Gutmann and Dennis Thompson and developed in Thompson's subsequent work. 49 While they deny that direct appeals to the necessity of secrecy by the executive make secrecy legitimate, they admit that necessary secrets can be legitimate if their necessity is acknowledged by citizens: "The secrecy is justified not only because it is necessary for the policy, but also because the question of whether and in what form it is necessary is itself the subject of public deliberation." 50 Making the necessity of secrecy the subject of public deliberation does not require that the executive disclose secret policies; it only requires the executive to admit that they resort to secrecy and to indicate the reasons for this even though they do not disclose the specific contents of the secret policies and processes. Such "second-order publicity about first-order secrecy" 51 enables citizens to publicly deliberate about the necessity for secrecy and, at least partly, to authorize it. Thompson acknowledges that the limited scope of information revealed by the executive may be insufficient for an act of authorization to be valid. 52 Even if such problems could be mitigated, however, this argument fails to authorize the state's appeals to necessity for reasons indicated earlier: necessity-driven special measures cannot be seen as exercises of authority because appeals to necessity escape the normative codification that the concept of authority presupposes.
If necessity does not authorize states to resort to special measures violating otherwise fundamental moral and legal principles even if it may permit them to do so, how should we conceptualize the permissibility at issue? David Owen's concept of "vindication" helps to mark the difference between the kind of permissibility to act at stake here and the one involved in being authorized to act in that way. 53 The concept of (political) authority refers to justificatory reasons that can be given independently and in advance of the action; in this sense (political) authority is always prospective. Vindication comes into play when justificatory reasons cannot be given independently and in advance of the action in question, but the agents do not have reasons, all things considered, to regret having performed it despite the moral costs involved; in this sense, vindication is retrospective. Owen emphasizes that vindication does not "retrospectively justify" 54 acts that violate moral principles. Such actions remain morally unjustified, but their performance is vindicated by the value they have brought about. Owen associates vindication with the kind of permissibility pertaining to morally unjustified action the performance of which turned out to be necessary for the establishment of good ends as in Machiavelli's doctrine of reason of state. In his view, the concept of vindication also makes sense of a common reaction to political actors whose hands are dirty: we realize that their action was morally wrong, but are glad that they did what they did. 55 Applying this concept to special measures adopted by the state in the face of necessity, we can say that while not authorized, they may nonetheless be vindicated insofar as the political community would, all things considered, not regret the introduction of special measures despite the moral costs.
An objection can be raised to my conclusion that necessity cannot ground political authority to resort to special measures because it escapes any normative codification that the concept of political authority presupposes. If there are circumstances in which agents can be vindicated in infringing otherwise unconditional moral principles, then agents can be called upon to do so when these circumstances obtain. It follows that agents should identify whether exceptional circumstances obtain or not. For this purpose, they should construct a rule identifying the exceptional circumstances. Insofar as the argument above must be committed to admitting that such a rule can be constructed and acted on, its insistence that necessity considerations escape normative codification and thus cannot ground political authority to violate fundamental moral norms is unfounded. 56 This objection posits that there must be a rule specifying what is to count as exceptional circumstances.
Yet, as Harel and Sharon retort, there cannot be a rule specifying what is to count as exceptional circumstances because then these circumstances would not be exceptional. Moreover, in claiming that a rule is required to determine whether standard moral rules apply (normal circumstances) or not (exceptional circumstances), the objection lapses into a vicious regress: if in order to determine whether a standard moral rule holds we must refer to another rule, we must be able to verify the validity of that other rule. Given that we can do that only by reference to yet another rule, the problem reappears at this (and every subsequent) level as well. As an attempt to reinstate the idea of moral rules permitting violation of moral rules, this objection fails.
The question addressed in this article is whether in a situation of necessity, a resort to secrecy or other special measures that override the moral and legal constraints that normally apply to political action is a legitimate exercise of political authority vested in liberal-democratic states. Having discussed this 54 Ibid., 160. 55 Ibid., 161. 56 Harel and Sharon, "Necessity," 860-61.

88
THE REVIEW OF POLITICS from the perspective of moral theory, my answer is that necessity does not confer political authority on the state to deploy special measures, including secrecy, because necessity escapes the normative codification that the idea of political authority presupposes. To return to the examples opening my article, the appeals to the necessity of secrecy which motivated the UK government's refusal to disclose Cabinet minutes in which British involvement in the Iraq war was deliberated and similar appeals which motivated the Polish government's refusal to confirm or deny its involvement in the CIA extraordinary rendition program are insufficient to render executive secrecy in these cases legitimate. In the next section I consider the question whether in a situation of necessity, a resort to secrecy is a legitimate exercise of political authority vested in liberal-democratic states from the perspective of jurisprudence.

Necessity Escapes Legal Codification
In early modernity, lawyers would place the special powers claimed by government in situations of necessity within the sphere of the prerogative, a power to be exercised in the public good that operates outside and against the existing framework of standing laws. 57 Nowadays, as Giorgio Agamben observes, the idea that special powers which empower government beyond their usual mandate could be extralegal is rejected. 58 One believes that necessity-driven departures from the law can be accommodated in the legal system. We can legally suspend the law by creating carve-outs into law that specify circumstances in which it can be suspended. For example, law protects freedom of expression defined as the right to receive information and ideas without interference by public authorities, but at the same time, it allows states to derogate from it during a state of emergency; thus transparency legislation creates exemption clauses that recognize that there are circumstances under which information should not be released. Paradoxically, as Ben Worthy remarks, FOIA legislation not only creates transparency but also maintains secrecy. 59 Similarly, law protects freedom of assembly and freedom of thought, conscience, and religion but allows states to suspend these freedoms during a state of emergency. There are two main strategies to accommodate the special measures the state adopts to respond to a necessity in the legal framework of liberal- democratic states. The first proposes to subject them to ex ante constitutional provisions: special emergency provisions are added to the constitution for regulating and constraining departures from law adopted by the state in situations of necessity. Emergency constitutions permit the delegation of special powers to the executive or to some other constitutional authority to suspend legal processes and rights, to censor information, issue decrees, and so forth with the aim of restoring the system to its previous state. This model is adopted in a majority of the legal frameworks of liberal-democratic states. 60 The second approach subjects special powers to ordinary law regulation and ongoing or ex post juridical or parliamentary oversight. Ordinary legislation grants power to the executive when a crisis is expected or has already presented itself. This is done either through the adoption of new legislation that grants the government special powers or through the reinterpretation of existing law. This model enables the legislature to stay in control, allowing it (instead of the constitution) to decide when a situation calls for special measures and which powers to hand over to the executive to address it. The legislature and the courts are also expected to monitor the use of the special powers, to investigate abuses, to extend these powers if necessary, and to suspend them if the emergency ends.
To the extent that the law claims to regulate and constrain these special measures, secrecy appears to be a legitimate tool for the exercise of political authority in liberal-democratic states. As Dorothee Riese demonstrates, the idea that legal regulation confers legitimacy on the executive's appeals to necessity prevailed in recent debates in the German Bundestag on granting intelligence agencies extraordinary powers in the context of security policies. MPs agreed that intelligence secrecy was justified insofar as it was necessary. As in reason of state thinking, they believed that necessity trumped the moral and legal principles underpinning the liberal-democratic order, such as protection of individual rights or transparency. Despite the democratic deficits of this course of action, they considered the executive's appeals to necessity legitimate as long as they were legally regulated. In pushing for legal enclosure of necessary secrecy, Riese claims, "German parliamentary practice can be seen as an attempt at democratizing . . . the idea of reason of state which otherwise is often considered as . . . inherently anti-democratic." 61 How successful this strategy to accommodate the special powers of the state in the legal framework of the liberal-democratic state is has been the subject of an extensive discussion in jurisprudence.

90
THE REVIEW OF POLITICS requirements of necessity. Mark Tushnet points out that prospectively codifying, let alone constraining, special measures called for in a situation of necessity is impossible. 62 Legislators cannot predict what crisis situations may arise and what special measures might be necessary to manage them. In addition, with regard to any codification in place, necessity could be invoked to override them. Tushnet asks us first to list the circumstances that would justify the suspension of legality in the face of emergency and then to imagine that a new emergency arises that the list failed to foresee. The government can claim that "the emergency is so pressing that it requires suspension of the legality expressed in the list of criteria for determining whether legality should be suspended, and the procedures for doing so. There is no response to this argument available to those who believe that suspension of legality is sometimes defensible." 63 Like ordinary legislation, emergency legislation must also be subject to suspension should the executive deem this necessary. The conceptual difficulty of codifying the necessity of secrecy and other special powers the executive may adopt in order to avert the threats to political survival and government capacity for action bears on the character of many existing constitutional emergency clauses. William Scheuerman observes that these provisions pay less attention to providing a pregiven substantive definition of what specific events deserve to be described as a necessity and focus instead on establishing procedural mechanisms of delegating and overseeing special powers. This leaves it largely to the executive actors to determine whether a particular development constitutes a necessity situation and what measures are needed to address it. 64 Yet if judgments about necessity are for the executive to make, then legal provisions do not function as constraints on its action but rather as factors about which the executive will decide. In effect, they permit government to do as it pleases while claiming to act within its legitimate authority. As Tushnet argues, "The provisions provide executive officials with a fig leaf of legal justification for the expansive use of sheer power. What appears to be emergency power limited by the rule of law is actually unlimited emergency power." 65 Similar problems confront the model which seeks to legalize special executive powers through ordinary law. Here, when adopting new legislation in the face of an emergency, legislators tend to codify the discretionary powers of the executive in a vague and open-ended way in order to ensure that the government enjoys all the powers needed to deal with the crisis at hand. Or, when interpreting existing law, they adopt expansive readings of their statutory and constitutional powers, on the one hand, and narrow readings of existing laws that might otherwise constrain their behavior, on the other. The legal justification of even the most controversial measures the US government adopted in the "war on terror" provides an apt example of inventive interpretation of existing law. As Clement Fatovic and Benjamin Kleinerman point out, lawyers working in the Office of Legal Counsel claimed legal authority for torture and the targeted assassination of American citizens and foreign nationals suspected of waging war against the United States and claimed that the Geneva Conventions of 1949 and other domestic and international law do not apply to these measures. 66 Seeking legal authority for even the most controversial actions the executive takes in the face of necessity commits us to accepting a situation in which, despite the facade of legality, there is illegality or, in David Dyzenhaus's words, "an absence of law prescribed by law under the concept of necessity." 67 While defenders of the second model reserve an important role for judges in overseeing the executive's deployment of special powers and making sure that it observes the basic values of the rule of law, legal practice shows that judges defer to the executive in times of crisis. Eric Posner and Adrian Vermeule point out that laying the judgment of the legitimacy of executive action in their hands would be unworkable: deciding what qualifies as a situation of necessity and what would be the most appropriate response is a matter of political judgment that goes beyond their competence; to have them do this would be to require them to second guess the judgment of the executive. 68 The legal codification of necessity-driven exemptions to law inherently involves the possibility of its own suspension. As Fatovic and Kleinerman show, 69 if the law says nothing more than "the executive shall have the right to do whatever it pleases," it becomes meaningless as a constraint on the special measures the executive adopts. Unlimited power of this kind is in tension with the concept of legal authority. To the extent that authority is constrained by norms that legitimate it, legal authority to resort to special measures which resist any legal codification is difficult to defend. I conclude that the attempts to confer legal authority upon the executive to resort to special measures overriding legal and moral principles that normally constrain political action fail. My conclusion aligns with the work of those

THE REVIEW OF POLITICS
legal scholars who, like Tushnet, Dyzenhaus, and Oren Gross, oppose the incorporation of special measures into the legal system. My broader discussion reveals a deeper layer to the problem with emergency powers they have identified because I have shown that their argument restates, in juridical terms, the problem we encounter when exploring the political authority of the state to resort to special powers. There it turned out to be difficult, on pain of slippery-slope reasoning, to determine the necessity threshold at which the state could exercise political authority to violate moral constraints that otherwise apply to its action. Here it turns out to be difficult to legally codify special powers of the state to violate law. In both contexts, in the absence of such constraints, the prospect of unlimited power looms large. Doubts about the legitimacy of special measures adopted by the state in situations of necessity spill over into doubts about the status of jurisdiction in related areas, such as whistleblowing. The US Whistleblower Protection Act grants the president authority to exempt any positions from its protection when the president finds it "necessary and warranted by conditions of good administration." 70 Given the inherently open-ended and unrestricted character of necessity and the court's deference to the executive, this leaves government whistleblowers unprotected. This outcome has struck activists and legal scholars as problematic, sparking a discussion on the justification of whistleblowing and the need to revise whistleblowing protection legislation. 71 Does this mean that the state is never permitted to resort to special measures, including secrecy, in a situation of necessity? Some contemporary jurists who oppose the incorporation of special measures into legal rules, for example, Gross, Tushnet, and David Feldman, postulate allowing for emergency measures while conceptualizing them as external to the legal authority of the state. 72 In this, they echo the claim that even though the state does not exercise authority when resorting to special measures, it may be vindicated in doing so. Yet if special measures cannot be accommodated in the legal framework and thus authorized, it remains unclear what expression such vindication could take. Some advocates of the extralegality of special measures, such as Harel and Sharon, propose that courts could exempt public officials from responsibility or grant ex post exemption facilitated by prosecutorial discretion, pardoning, or other tools which highlight the exceptional and unprincipled nature of the circumstances giving rise to the act. 73 Gross and Feldman propose subjecting extralegal emergency action to public judgment ex post, which might excuse the illegal action or even effectively endorse it by more conventional political mechanisms. 74 Authorizing Necessary Secrets without Appealing to Necessity I have argued that appeals to necessity fail to ground the authority of executive secrecy. Necessary secrets are not a legitimate form of democratic governance, but at most its vindicated suspension. I now briefly comment on a recent defense of state secrecy and reflect on whether it can do what the necessity argument cannot.
In previous work, I have argued that a degree of secrecy in democratic governance can be legitimate. 75 My argument draws on the formal features of political authority, namely, its content-independent character, which is an element of the traditional concept of political authority and one endorsed by most influential theories of democratic authority. 76 To define political authority as content-independent is to say that the binding force of its directives is detached from their content and from the considerations which underlie the executive's judgment about what directives it ought to issue. This distinguishes authoritative directives from standard cases of reasons for action in which there is a connection between the reason for action and the action itself, such as when the action is independently desirable, has beneficial consequences, or otherwise has moral merit. Authoritative directives are different in that these factors are not what makes them binding. Their validity "is in the . . . fact that someone in authority has said so," as Joseph Raz puts it, 77 not in what she has said or why she has said it. I argued that presenting the content of the policies as irrelevant to their authoritative character opens the door to saying that acquiring knowledge of their content is not a factor that contributes to their authoritative character. If knowing the content of the policies is irrelevant to whether they are authoritative, then the policies can be authoritative irrespective of whether citizens know their content. This argument, if successful, extends the political authority exercised by democratic states to secret uses of power as it presents secret policies as a special case of policies that have a content-independent authority.
Imagine that the executive's resort to secrecy is driven by considerations of necessity. Given that it is the executive or, in Raz's words, "someone in authority" that has resorted to secrecy, the resort to secrecy is authoritative. This argument fixes in part the problems that pertain to the necessity argument because it authorizes secrets that the executive deems necessary. However, their necessity plays no role in their authoritative character. The executive's judgment about the necessity of secret policies is an evaluative judgment about those policies. On the argument under consideration, however, the authority of government policies is content-independent, detached from evaluative judgments about them. Thus, even if the driving force of secret policies is necessity, it is not necessity that renders executive secrets authoritative but the fact that the decision to resort to secrecy was taken by a political actor with legitimate powers to do so.
Not being a straightforward fix to the necessity argument, the argument deriving the authority of secret uses of power from the formal features of political authority forms an interesting alternative to it. I have argued here that the necessity argument inevitably sends the executive power to resort to secrecy down a slippery slope toward unlimited power. The argument outlined in this section does not legitimate the secret uses of power across the board. Content-independent authority is always limited and given that secret policies can be seen as a special case of policies that have a contentindependent authority, these limits apply to them too. Roughly, the limits relate to (1) the reasons for having political authority in the first place and (2) the fact that its directives must have certain institutional features that make it possible for them to be successful. As Dyzenhaus argues, these limits affect, but do not determine, the content of any authoritative decision. 78 The first class of reasons is substantive and corresponds to the reasons for which political authority has been instituted. Formulated for a liberal-democratic order, they refer to principles such as equality, self-governance, justice, and citizens' privacy. 79 These reasons/principles set limits on the exercise of power in the sense that the directives that contradict the substantive basis of the authority to rule in a content-independent way are invalid. The second class of reasons is procedural. If the authoritative directives are to be successful, they must bear formal marks of authority that distinguish them from arbitrary uses of power. This, Dyzenhaus argues, requires that they satisfy the criteria of the rule of law. 80  constraints on content-independent authority limit the legitimacy of necessary secrets: secret policies and decision-making processes violating these constraints fall short of having democratic authority.
Verifying whether laws and policies remain within such constraints is a matter of democratic control and oversight, which are necessary, but not sufficient, conditions of democratic legitimacy. The control/oversight condition limits the scope of legitimate secrecy to "shallow" secrets, namely, those of which citizens know the existence even though they are ignorant of their content. 81 For example, citizens know that state agencies run military research programs, but they do not know the content of the programs. Given that control of government secrecy requires at least that the fact of secrecy is known, in the case of shallow secrets, the demand for scrutiny can be raised and their democratic authority is not precluded. This condition cannot be satisfied with regard to "deep" secrets, whose existence citizens are not aware of, like the Manhattan Project. Being placed beyond democratic oversight precludes deep secrets from having democratic authority.
The defense of state secrecy outlined in this section does what the necessity argument cannot: it extends political authority to secrets that the executive deems necessary. Whereas the necessity argument has difficulties setting limits to the state power to resort to secrecy, the argument anchoring the legitimacy of state secrecy in the formal features of political authority indicates the substantive and procedural limits to secret uses of power by democratic governments in situations of necessity. Because it does all this without appealing to the concept of necessity, it would be a mistake to see it as an improved form of the necessity argument.
The difference between the two arguments bears on the analysis of the examples opening my article. By the necessity argument, the Polish involvement in the CIA extraordinary renditions program is legitimate. According to the argument outlined in this section, it is not: kept in deep secrecy, it fails to satisfy the procedural constraints on content-independent power exercised by democratic governments. As it was used to cover the violation of human rights involved in the "enhanced interrogation techniques" performed on the prisoners kept at black sites, it also fails to satisfy the substantive constraints on content-independent power exercised by democratic governments. 82

Conclusion
From the claim that secrecy is a necessary condition of the political survival of a democratic state and/or the capacity for action of a democratic government, 81

96
THE REVIEW OF POLITICS proponents of the necessity-based defense of state secrecy infer that secrecy is a legitimate exercise of democratic authority. This article concludes, however, that an appeal to its necessity fails to confer political and/or legal authority upon the state's resort to secrecy, even if it can vindicate it. Necessity cannot ground its authority because it escapes normative codification in both the moral and legal domains. As no moral or legal norm can fully capture and constrain special measures adopted by the state in the face of necessity, such measures open the door to unlimited executive powers operating in a discretionary space of action impenetrable to democratic scrutiny. The necessity argument fails in escaping the antidemocratic implications of reason of state thinking. It succeeds, at most, at presenting secrecy as a vindicated suspension of democratic governance, but not as a form of democratic governance. The overall conclusion of the article, however, is not that secrecy, along with other special measures the state adopts in situations of necessity, is illegitimate, but only that an appeal to necessity does not suffice to legitimate it. An account of the democratic legitimacy of state secrecy must proceed in terms other than necessity.
NECESSARY BUT ILLEGITIMATE 97