WHY SNOWDEN AND NOT GREENWALD? ON THE ACCOUNTABILITY OF THE PRESS FOR UNAUTHORIZED DISCLOSURES OF CLASSIFIED INFORMATION

. In 2013, following the leaks by Edward Snowden, The Guardian published a number of classiﬁed NSA documents. Both leaking and publishing leaks violate the law prohibiting unauthorized disclosures. Accordingly, there are two potential targets for prosecution: the leakers and the press. In practice, however, only the leakers are prosecuted: Snowden is facing a threat of 30 years’ imprisonment; no charges have been made against The Guardian . If both leaking and publishing leaks violate the law, why prosecute only the leakers and not the press? I consider and reject two arguments. The ﬁrst claims that the press has special moral claims by virtue of its rights (press freedom) or its role (the Fourth Estate; conduit for information). The second argument states that the leakers commit a greater wrong than the press. I conclude that the current prosecution practice is inconsistent: prosecutors should either prosecute both or neither.

Apparently, as Greenwald approvingly notes: There are both formal and unwritten legal protections offered to journalists that are unavailable to anyone else. While it is considered generally legitimate for a journalist to publish government secrets, for example, that's not the case for someone acting in any other capacity. 3 The difference in treatment that leakers and press receive regarding unauthorized disclosures raises the following question: If both leaking and publishing classified information violate the law prohibiting unauthorized disclosures, why prosecute only the leakers and not the press? Why Snowden, not Greenwald? I am interested in the moral dimension of the puzzle: Are both parties not equally morally culpable? Is leaking a secret not the same thing as publishing a secret? In addressing this puzzle, this essay searches for a moral disanalogy between the sources of the leaks and their distributors, which would justify their different treatment. Here, I enter unexplored territory. Philosophical discussions have addressed the moral and legal accountability of civil servants for leaking classified government information, but have not attended to press accountability. Likewise, the difference in treatment accorded civil servants and press with regard to unauthorized disclosures has not received attention. This essay fills this gap. I consider two juridical arguments, which set out to establish a moral disanalogy between leakers and press, thus justifying the practice of selective prosecution. I refute both, concluding that we have no moral reason for treating the press differently from leakers.
In developing my argument, I assume that state secrecy is, under certain conditions, legitimate. When it is not legitimate (concealing, for example, power abuse or wrongdoing), its disclosure is justified and this essay's question does not arise: We have a moral reason for dropping prosecution of leakers and press rather than differentiating them. 4 Assuming that state secrecy may be a legitimate exercise of democratic authority, I do not defend this claim: Given that the considerations justifying secret uses of power remain the same irrespective of who discloses classified information, spelling out these considerations will not indicate a possible difference in the status of the disclosers. For a discussion of the legitimacy of state secrecy, I refer the reader to my earlier work. 5

I. THE FIRST ARGUMENT FROM DISANALOGY: THE PRESS HAS A MORAL JUSTIFICATION UNAVAILABLE TO LEAKERS
One way to justify the different treatment civil servants and press receive for disseminating classified information relies on the claim that unauthorized disclosures by the press have a status differing from those by civil servants. By virtue of freedom of the press, the press may publish whatever content editors choose, including leaks. Freedom of the press gives it the prerogative to balance compliance with the law against its conviction regarding the importance of publishing leaked information. If the press has a justification unavailable to leakers, we should not prosecute the press even if we prosecute leakers.

A. Freedom of the press
Contemporary literature on press freedom appropriates John Stuart Mill's reasons for protecting individual rights of self-expression and free speech. Mill argued that 'absolute freedom of opinion (…) on all subjects, practical and speculative, scientific, moral, or theological' implies 'liberty of expressing and publishing opinions' because this is 'practically inseparable' from the freedom to express oneself. 6 Following Mill, the literature on the freedom of the press extends individual rights of self-expression to the press. As Judith Lichtenberg observes: [W]hen we examine the most famous arguments for freedom of the press, we find nothing to distinguish them from those for freedom of speech or expression more generally. The press is treated as a voice, albeit a more powerful one, on a par with individual voices, and defending press freedom is then tantamount to a general defense of free speech. 7 Assuming that press freedom is a species of freedom of speech and self-expression, any extraordinary privileges it would confer upon the press with regard to publishing classified information must be privileges that the press enjoys in virtue of freedom of speech. Does freedom of speech confer such privileges upon the press? This question is usually answered 'yes'. While leaking classified information constitutes a criminal offense, publishing leaks is seen as protected by 'First Amendment values'. 8 From the perspective of freedom of speech legislation, that an act is ordinarily unlawful is not conclusive. For example, as Geoffrey Stone explains, it may be unlawful to obstruct the draft. Correspondingly, an individual who physically blocks access to a selective service office can be punished, 'but an individual who distributes leaflets criticizing the draft as immoral cannot be punished. The criminal law is the same but the pamphleteer is protected by the First Amendment'. 9 Similarly, a journalist who publishes classified information is committing a crime, unless it is shown that the First Amendment affords her protection. The press has not yet been prosecuted; thus, the impression arises that its prosecution is indeed incompatible with such legislation. The fact is, however, that the courts have had no opportunity to resolve this issue. In the US, the court addressed the question in the Pentagon Papers case but left it unanswered. 10 Nor has the question received systematic theoretical attention. The conclusion must be that free speech legislation has not yet been demonstrated to impede prosecution of the press. The next two sections address this unresolved problem from a moral perspective.

B. Freedom of speech: Interests of the speaker
One major theory of free speech sees it as a right protecting the autonomy and self-expression of the speaker. 11 Restrictions on what individuals say and write are an affront to their autonomy and, thus, the autonomy-related interest implicit in the right to free speech requires that individuals have a right to speak without interference. Assume for the sake of argument that the press too can exercise a right to free speech so understood: the autonomy-related interest implicit in the press's right to free speech requires that it has a right to publish without interference. Does this account of free speech support the press's privilege to publish classified information?
Two reasons show this cannot be so. First, this account does not support a press privilege to publish classified information because it does not justify disclosure of classified information at all. Second, even if it did so, it would not establish the moral disanalogy between press and leakers required to justify selective prosecution.
So, the claim that appealing to freedom of speech justifies disclosures of classified information presumes that such disclosure constitutes speech and, therefore, should be protected. A prohibition on disclosure would be a prohibition on free speech and, hence, an affront. Yet a defense of unauthorized disclosures in terms of this account would be problematic. On this account, as Wojciech Sadurski put it, '[s]peech as a vehicle of self-fulfillment has no other immediate aims than to be uttered and heard, while we believe that often the 'point of the speech' lies in its capacity to bring about the results desired by the speaker'. 12 Saying that Greenwald has a right to publish classified information by virtue of his right to free speech commits us to saying that Greenwald is justified in publishing classified information because doing so is his way of expressing himself. Alternatively, The Guardian is justified in publishing leaks because this is how its editorial board exercises its autonomy. The problem with this is that it trivializes the importance of leaks. We believe that the point of publishing leaks is the impact they will have. Yet on the free speech defense presented above, such disclosures have no relevance other than advancing the self-expression of the journalist or the autonomy of the newspaper's editorial board. Onora O'Neill has articulated this concern more generally: this approach overlooks the significance of the content of press reporting. As the media are in the communication business, its driving force should be concern about content, not self-expression. 13 Defending disclosures of classified information by appeals to selfexpression and autonomy creates yet another problem. The argument implies that Greenwald is justified in publishing classified information, that is, in violating the law prohibiting unauthorized disclosures, if this law limits his self-expression. More generally, it implies that the law is binding only as it accords with our selfexpression. But this is to be very confused about law's normative force. By analogy, the rule setting the speed limit at 100 km/h is binding only when it coincides with my self-expression: I have a Maserati, express myself by speeding at 200 km/h, therefore the speed limit isn't binding. To understand law's normative force like this makes a legal system inconsistent, bidding us to think there is a right to violate the law. A like problem has surfaced in the debate on the justification for whistleblowing. Despite there being no legal right to whistleblowing, lawyers and legal scholars often argue that unauthorized disclosures may enjoy protection under the right to freedom of expression. Thus, sanctions against whistleblowers are described as 'violations' of their rights. As Eric Boot persuasively argues, however, it is impossible to reconcile the general prohibition on whistleblowing with the claim that there is a freedom-of-expression-based legal right to whistleblowing: Such a legally recog- nized right would mean the law recognizing a right to break the law. 14 Even if the appeal to free speech did justify disclosures of classified information, this would not ground a disanalogy in the normative situation of leakers and press. Freedom of speech is directed at speech in general and, thus, applies to both leakers and press. To claim it applies particularly to journalists' speech suggests that the press enjoys the right to free speech more than others, committing us to the unreasonable claim that journalists are more important than others. 15 In conclusion, this account of free speech based on the interests of the speaker fails as a defense of the press's privilege to publish unauthorized disclosures: Quite aside from its implausibility, this argument excludes preemptively any difference between press and civil servants with regard to disclosures of classified information. 16 Before exploring another theory of free speech, let me briefly consider the suggestion that the speaker's interests may have nothing to do with freedom of speech. Judith Lichtenberg notes we may be dealing with the speaker's claims to property: The publisher may say: 'It's my newspaper and I can print what I want.' (…) Such a claim to editorial autonomy is really (…) the assertion of a property right in the guise of a free speech right. 17 Is the press's appeal to property rights a better defense of its privilege to publish classified information? If the claim is that a newspaper is free to publish classified information because it is a private business, and can publish whatever it wants, then this argument fails because governments regulate private property.  , p. 387. 16 Note that the argument overlooks the fact that free speech is usually hedged by restrictions. Article 10 of the ECHR indicates the social and private interests for which speech may be restricted: national security, prevention of disorder/crime, protection of health, morals, or the reputation of others. Given that classifying government information is a matter of protecting social interests (law enforcement/national security), classified information-related speech is unlikely to be protected. 17

C. Freedom of speech: Interests of the listeners
According to a second influential understanding, free speech protects the interests of the speaker only insofar as her exercise of free speech serves the interests of listeners. 19 As Stephen Holmes put it, 'At issue is the access of listeners to information not the access of speakers to the microphone'. 20 Greenwald's self-expression requires protection only insofar as it serves the people's access to information.
The importance of people's access to information is explained in terms of a more fundamental argument: People's access to information is relevant in democracies so they can make informed, autonomous decisions. In its early version, put forward by Alexander Meiklejohn and Thomas Emerson, the democratic account of free speech presented people's access to information as a condition of people's self-government. 21 Drawing on the idea of popular sovereignty, Emerson argues: The public, as sovereign, must have all information available in order to instruct its servants, the government. As a general proposition, if democracy is to work, there can be no holding back of information; otherwise, ultimate decision-making by the people (…) becomes impossible. 22 The idea of popular sovereignty in its purest form has been dropped in modern democratic theory, replaced by concepts of participation, deliberation, accountability, oversight, and public opinion. Despite this shift in the conceptual framework, democratic justification of freedom of speech retains its appeal. Free speech is entitled to protection because it fosters communication, contributes to forming citizens' political preferences and reasoned political deliberation, and exposes people to the information they need to hold decision-makers to account. 23 On this account, the press has a right to publish classified information as this right serves the interests of democratic citizens by, for example, triggering debate on the desirability of classified government programs, without which government decisions and policies would suffer a democratic deficit. 24 19 Mill, On Liberty; Raz, J. 1991 Let me briefly compare this defense of special privileges of the press to the one discussed in the previous section. Theories of free speech concerned to protect the audience argue that a reporter must enjoy special protections not because she is entitled to them but in order to secure some general benefit to the public as a whole. Ronald Dworkin refers to arguments of this kind as arguments of policy. 25 This is different from freedom of speech arguments granting the reporter special protection in terms of her interests as a speaker: there the reporter is granted protection not in order that others benefit, but because she as a speaker would suffer some unacceptable injury or insult if censored (she would be granted protection even if the community as such suffered from allowing her to speak). Dworkin refers to arguments of this kind as arguments of principle.
As Dworkin observes, the defense of free speech based on arguments of policy is weaker than the one based on arguments of principle. Arguments defending freedom of speech in terms of certain benefits which the exercise of free speech procures (for example, political participation) must yield when there is reason to believe that other private and social interests of the community (for example, public security) demand constraints on it. They thus invite the reply that in some cases the public's interest would be better served, on balance, by censorship rather than by publication. 26 Arguments of principle in support of freedom of speech cannot be defeated in this way; if there is a conflict between the rights of the individual speaker as an individual and the competing interest of the community, the former trumps the latter. However, Dworkin argues, even though the defense of freedom of speech as a matter of principle is stronger, it is less promising with regard to establishing special privileges of the press. Offering the reporter special protections in terms of the principle-based account of free speech leads, as we saw in the previous section, to the unreasonable conclusion that the reporter's interests, as an individual speaker, are 25 Dworkin, A Matter of Principle, ch. 18 and 19. 26 ibid, p. 389. more important than the similar interests of others. In this respect, it is more plausible to support special privileges of the press in terms of the policy-based account of free speech arguing that the reporter's exercise of freedom of speech has democracy-enhancing consequences and, thus, secures an important benefit to the community.
Does this democratic, policy-based account of free speech provide us with tools to justify the different legal treatment of leakers and press with regard to unauthorized disclosures? It is not necessary to engage with this argument to say that, even if correct, it would not justify the practice of selective prosecution. Even if an appeal to free speech justifies the press's publication of classified information by virtue of its democracy-enhancing consequences, the same consequences follow disclosures by civil servants. Indeed, this argument has been invoked by lawyers in cases of unauthorized disclosures by civil servants. The European Court of Human Rights, in its first whistleblowing case, argued that disclosure is justified if the public interest is served by the information disclosed. 27 In the US, defendants have appealed to their First Amendment rights with the defense construing the leaker's right to free speech in terms of the democracy-reinforcing consequences of her disclosures (not her individual interests in autonomy or self-expression). 28 Given that an appeal to freedom of speech can be invoked to protect leakers just as it is invoked to protect the press, it will not privilege the press over civil servants with regard to disseminating classified information. Of course, when ascertaining whether an unauthorized disclosure will enjoy protection under the right to freedom of expression, the courts typically balance the interest of the public in having the information revealed against the considerations underpinning the law prohibiting unauthorized disclosures. In effect, the right to free speech the courts establish in these cases contains provisions limiting its scope in order to account for, say, national security or confidentiality of diplomatic negotiations. 29 However, if the press exercises the same right, its 27  exercise of this right should be subject to the same limiting provisions. Thus, the different treatment accorded press and leakers for violating the law prohibiting unauthorized disclosures remains a puzzle. 30

D. Chilling effect
If the appeal to freedom of speech does not justify a privileged position for the press with regard to unauthorized disclosures, a different component of freedom of speech adjudication could perhaps do the job. In particular, one could argue that the law prohibiting unauthorized disclosures has the incidental effect of deterring, or chilling, speech that society has reason to encourage such as, for example, disclosures of information that reveal abuses of power. Strict execution of the law prohibiting unauthorized disclosures could deter such speech, especially when the speakers are hesitant about the value of the information they disclose, a situation likely to arise when the (il)legitimacy of a government's actions may 30 Defending press freedom in terms of its democracy-enhancing consequences resembles another famous argument for it viz. the marketplace of ideas. While democracy-enhancing consequences are the thrust of the former, discovery of the truth is at stake in the latter. Originating with John Milton (Milton, J. (1918(Milton, J. ( [1644) Areopagitica. A speech for the Liberty of Unlicensed Printing to the Parliament of England, ed. R. Jebb. Cambridge University Press), the argument found systematic articulation in John Stuart Mill (1989). Free, unfiltered and uninhibited circulation of information and opinions allows discovery and testing truth and detecting and rejecting falsehood. In American First Amendment jurisprudence, the argument was subsequently phrased in terms of the 'marketplace of ideas' metaphor. Thus, free trade of ideas, like an economic market, can mobilize resources otherwise dormant, stimulating them into use, bringing the truth into political controversies enabling the best policies to emerge. From this perspective, publication of classified information has epistemic advantages because it triggers exchanges of ideas in the public forum which have beneficial effects on the quality of political decisions. As a justification of the practice of selective prosecution, this argument fails for the same reason the appeal to the democracy-enhancing consequences of free speech does viz. it establishes no difference between press and leakers. Moreover, the assumptions upon which the argument rests make it problematic on its own count (see Schauer, F. 1982. Free Speech: A Philosophical Inquiry. Cambridge University Press). First, it assumes that truth always drives out falsehood and prejudice. Yet confronted with many historical examples, this assumption appears empirically unfounded (Holmes, 'Liberal Constraints on Private Power?', p. 34). The phenomena of fake news, echo chambers, and filter bubbles demonstrate that there is no empirical basis for believing that when ideas are tested publicly, truth will necessarily prevail. Second, in claiming that broad press freedoms, including freedom to publish classified information, will result in more true beliefs, the argument assumes that the search for truth should enjoy precedence over other social goods. This assumption is problematic because other interests can at times predominate (Schauer, Free Speech, p. 33). For example, as Boot (Boot, E. 2019. 'Leaks and the Limits of Press Freedom'. Ethical Theory and Moral Practice 22(2), p. 14) claims, 'at times, (…) national security concerns may very well outweigh the search for truth and thus justify state secrecy as well as limits upon the scope of press freedom'. Finally, the position presupposes that more information is better. However, '[u]ntrammeled exchanges in that famous marketplace are as likely to lead to a Babel of voices as to comprehension, let alone discovery of truth' ( be a matter of reasonable disagreement. 31 As we have reason to avoid the chilling effect, we are justified giving leaks 'strategic protection' viz. abstaining from the imposition of sanctions in cases in which such sanctions would otherwise be permitted. Now one could argue that the chilling effect of the law prohibiting unauthorized disclosures is greater on the press than on leakers because leakers have the advantage of disclosing information anonymously, while the press does not. Anonymity weakens the probability of prosecution and thereby weakens its chilling effect. The press, without the option of anonymity, is more vulnerable to the chilling effect of the law and more likely to censor itself and refrain from publishing instances of government wrongdoing. As the risk of the chilling effect is greater with regard to the press, there is reason for a more lenient prosecution of the press. This is controversial on two counts. According to the argument, it is necessary to protect speech disseminating even harmful leaks because any prospect of punishment would create an impermissible chill on speech that discloses information that society has reason to value. The reasoning is problematic: To arrive at this claim, one must weigh the benefits of the law prohibiting unauthorized disclosures, on the one hand, against the probability that valuable leaks will be put at risk, on the other hand. This balancing exercise is practically impossible because establishing the quantity and value of leaks that would be chilled is entirely counterfactual -the leaks in question have not yet been leaked. The counterfactual calculations involved in the argument cast doubt on its validity. This problem resonates in more general concerns raised about the chilling effect argument. The behavioral assumptions for the chilling effect argument rest upon predictions about the behavior of subjects of the law under counterfactual conditions. Yet, as many scholars concede, no empirical evidence has been produced to justify those predictions. It has not been clearly established that individuals are deterred or become overly cautious as a result of the existence of particular regulations. For example, as Leslie Kendrick admits, 'It is difficult to establish either the presence or the absence of a chilling effect, let alone to measure the extent of such an effect'. 32 If such empirical assumptions have little demonstrable basis, then the legitimacy of a policy to prevent the effect is questionable. Second, the empirical claims the argument makes about the vulnerability of leakers and press to the chilling effect are problematic: In the light of new surveillance/investigation techniques the claim that leakers are likely to stay anonymous is doubtful. Heidi Kitrosser observes, for example, that new technological and legal tools (phone/e-mail records, electronic indicia of travel and in-person meetings; third party subpoenas to banks, communications and credit card companies) enable governments to track leakers more easily than ever. 33 Journalists report that these new realities have chilled their communications with leakers. The claim regarding the relative vulnerability of the press to the chilling effect is equally problematic. As Frederick Schauer explains, the behavior of reporters, editors and publishers is more impervious to a chilling effect than commonly thought. 34 For example, libel law is often believed to have publication-deterrent consequences. However, Schauer reports significant journalistic disregard of the potential legal liability for libel. It is not evident the press is less willing to publish; rather, 'publishers treat libel as a cost of doing business'. 35 Moreover, he claims, journalistic awards, opportunities for professional recognition, routes to professional esteem, and avenues for promotion reinforce or require risky behavior on the part of journalists, editors and publishers, and, in effect do more than counteract the chilling effect of libel law. There is no reason to assume that the situation is any different with regard to other areas of legislation such as the law prohibiting unauthorized disclosures.
Thus, the chilling effect argument is unsuccessful and cannot be used to justify the practice of selective prosecution.

E. The press as fourth estate
For the press to enjoy a special dispensation to publish classified information, the press must enjoy a special status compared to leakers. 33  It is often argued that the press is entitled to special protections because it is the 'fourth branch of government'. As Lee Bollinger observes: The dominant image of the press views the press as the Fourth Estate, as an entity serving a critical quasi-official function in the political system. It elevates the press to the highest rung on society's organizational chart and anoints it as the public's representative. 36 This view has been endorsed by a number of American First Amendment scholars and has occasionally been defended by the US Supreme Court: 'The primary purpose [of the Press clause was] to create a fourth institution outside the Government as an additional check on the three officials branches. (…). The relevant metaphor (…) [is that] of the Fourth Estate'. 37 This idea is also a potent part of journalists' self-identity. Among others, Greenwald sees the role of the press thus: 'One of the principal institutions ostensibly devoted to monitoring and checking abuse of state power is the political media. The theory of a 'fourth estate' is to ensure government transparency and provide a check on overreach'. 38 The press's role as a 'fourth estate' is commonly presented as correlative to the 'people's right to know'. The 'right to know' is understood as a right held by any individual, as against a government, to know about the workings and dealings of that government. 39 According to the fourth estate argument, the press, in its capacity as the fourth branch of government, serves as a trustee of the 'people's right to know'. As American constitutional scholar Alexander Bickel put it, 'The reporter's access is the public's access. (…). The issue is the public's right to know. That right is the reporter's by virtue of the proxy which the freedom of the press (…) gives to the press in behalf of the public'. 40 Sometimes this role is spelled out in terms of the adversarial role of the press viz. the press is presented as a competing power to that of government. Greenwald subscribes to this militant model of journalism when he approvingly quotes The New York Times' war correspondent David Halberstam for whom, as Greenwald phrases it, 'infuriating the government was a source of pride, the true purpose and calling of journalism'. 41  the struggle between press and government produces an equilibrium: The goals of the democratic system are served, as if by an invisible hand, when the government is doing all it can to stop the press from getting information, and the press is publishing just about anything it can get its hands on. '[T]he press is (…) one party to a contest', Bickel claimed. 'It is the contest that serves the public interest'. 42 What does the fourth estate model imply for the publication policy of the press with regard to classified information? On the adversarial version of this argument, the press, in virtue of this role, has no duty to comply with government calls for non-publication. Given that the press has no duty to keep state secrets, it commits no wrong in disclosing them. In effect, the press is justified in publishing state secrets whenever it sees fit: 'the press', Bickel argues, 'may publish materials that the government wishes to, and is entitled to, keep private'. 43 An excellent summary of this view was recently issued by the deans of the leading US schools of journalism. After The New York Times revealed the Terrorist Financing Tracking Program, they declared: 'It is the businessand the responsibility -of the press to reveal secrets'. 44 When fourth estate proponents drop their adversarial rhetoric and present the press simply as an element of a system of checks and balances, it implies that any government decision on classification must be subject to a check and approval by the press. In this spirit, Leonard Downie, the Executive Editor of The Washington Post, asserted: 'it's important (…) in our constitutional system that (…) final decisions [about publishing classified material and whether publication would indeed threaten the nation's safety] be made by newspaper editors and not the government'. 45 Or, as editors of the New York Times and the Los Angeles Times claimed, 'Decision on whether or not to publish classified information (…) is not one we can surrender to the government'. 46 The fourth estate argument presents the press as occupying a unique role in the political system, one that leakers do not occupy. If correct, this would justify the different treatment press and leakers receive for disseminating classified information. Below, I claim that 42  this argument cannot succeed because we cannot make sense of the press as a 'fourth estate'. A model that has been widely used to illuminate the relationship between citizens and the three branches of government is the principal-agent relation, in which one party (principal) authorizes another (agent) to represent her interests and to act on her behalf and the latter provides an account of her actions with respect to those interests. 47 In this approach, citizens are viewed as principals and government actors as agents. Presenting the press as a 'fourth branch' of government would imply that it too stands in an agency relation to the people. Indeed, when the US Supreme Court entertained the idea that the press is analogous to the fourth branch of government, they presented journalists as 'agents' of the public at large. For example, in 1974, Justice Lewis Powell declared that 'in seeking out the news, the press (…) acts as an agent of the public at large. (…) The underlying right is the right of the public generally. The press is the necessary representative of the public's interest'. 48 Such a description of the relationship of the press to the people, however, is misleading on several counts.
For one, in the context of the principal-agent relation, agents are prohibited from acting in their own interests and required to advance the interests of the principals. Thus, government actors/agents are prohibited from acting to further their own interests; they must act in the interests of their citizens/principals. The press, by contrast, is a private actor free to act in its own interests as determined by market dynamics. 49 47 Christiano, T. 1996. The Rule of the Many: Fundamental Issues in Democratic Theory. Westview Press. 48 Cited after BeVier, 'The Journalist's Privilege', p. 469, emphasis added. 49 BeVier, 'The Journalist's Privilege', p. 475. One could object to this argument by claiming that the fact that the press is acting on its own interests does not preclude its standing in a principal-agent relation to the people. In particular, the argument could go, we do not hesitate to describe political parties as agents of the people even when they act on special interests, e.g., interests of ethnic minorities, regional interests or separatist movements. If we don't hesitate to describe political parties whose main aim is to protect special interests as agents of the people, what would prevent us from so describing the press? To answer this objection, it is sufficient to demonstrate that the analogy between political parties and the press does not hold. While both press and political parties may act on special interests, these are of different kinds. Special interests can be private or partial; while all private interests are partial, not all partial interests are private. Political parties do not act on private interests even though they act on partial interests. Moreover, the existence of political bodies representing certain segments of society is part of the political division of labor in a representative democracy. It ensures that all social groups can ultimately be represented in decision-making bodies which, in turn, guarantees the principal-agent relation between political representative institutions and the people. To say that the press acts on special interests is to say that they act on the private interests of their owners, e.g., the personal commercial interests of media tycoons. Organizations representing the private commercial interests of specific individuals are not part of the political division of labor in a representative democracy. The media, acting on private interests, rather than on the interests of (certain sections of) the people cannot be seen as an agent of the people.
Secondly, while the people can be said to authorize the three branches of government as their agents, they cannot be said to so authorize the press. Unlike the powers of government, the powers of the press are neither conferred upon it by the people -they are not subject, for example, to regular elections -nor by their representatives. 50 From this perspective, if the press claims the authority to reverse classification decisions taken by government officials through its publication decisions, then its view of its competences is flawed because representative democracies grant such authority to elected government officials and not to editors. When the press discloses classified information, then, it violates the democratic allocation of power. 51 50 The thrust of this argument is that the press is not an agent of the people because it is not authorized by them. In response, one could claim that the people don't authorize at least one branch of government viz. its judicial branch. To the extent that this claim holds, it puts pressure on the argument that the press cannot be an agent of the people: If the fact that the judiciary is not authorized by the people doesn't prevent us from construing it as an agent of the people, the fact that the press is not authorized by the people cannot prevent us from construing it as an agent of the people either. In response to this objection, first, I question the claim that the judiciary is not authorized by the people: Just like the other branches, it is. If correct, this argument breaks the analogy between press and judiciary and allows us to present the latter but not the former as an agent of the people. In representative democracies, people can authorize office holders directly or indirectly. People authorize office holders directly when they, e.g., vote for them. They authorize them indirectly when they do so through their representatives. Even if the judiciary is not directly authorized by the people, it is authorized indirectly by the people's representatives. Moreover, the appointment process of judges is determined by a constitutional procedure that is authorized by the people. The process differs per legislative system. In the US all federal judges are nominated by the President and must be confirmed by the US Senate by majority vote. Following confirmation, the President finally makes the decision to commission the judges to their position. In most European countries the executive or legislative branches play a decisive role in the appointment of judges. Federal judges in Germany are appointed by the Federal President, after being elected. The judges are elected by the Judges Election Committee consisting of sixteen ministers of the substates and sixteen members elected by the German Parliament. In the UK, judges are elected by a judicial appointments committee consisting of twelve members of the Ministry of Justice and three members of a judicial council. Thus, all branches all government, including the judiciary, are authorized either directly by the people or by the people's representatives. No such authorization processes obtain with regard to the press: Neither the people nor their representatives confer any decision-making powers on the media, and the analogy between the media and the judiciary does not hold. Nor does the argument it was meant to support viz. that the press, just like the three branches of government, is an agent of the people. Second, I argue that the press cannot be a fourth branch of government because it lacks the kind of action-guiding powers commonly ascribed to government. The authoritative powers exercised by governments have traditionally been described as content-independent: It is the fact that the action-guiding directives have been issued, not their content, that is intended to provide the subject with a reason for action (Hart, H.L.A. 1997. The Concept of Law. Facts concerning the democratic allocation of power aside, the hypothetical proposition to recognize the press as professionally entitled to reveal state secrets would build an inconsistency into the system of democratic authority. On the traditional conception of political authority, the exercise of authority creates correlative obligations to obey its directives (within limits). From this perspective, authorizing the government to classify information creates a correlative obligation to obey its classification directives. According the press the power to disclose information classified by the government would deny such obligations, defying the state's authority to classify. As Sagar argues: [we] cannot simply authorize (…) reporters and publishers to disclose classified information as they see fit: to do so would create a contradiction in the structure of public authority, with one bearer of public authority ((…) reporters and publishers) allowed to disclose what another bearer of public authority (the executive) has been allowed to conceal. 52 Thirdly, if the press were in some sense an agent of the public, providing people with information they have 'a right to know', the people should also be able to hold the press to account for its performances. Yet the idea of the accountability of the press to the public is problematic. Whereas the accountability of executive, legislative, and judicial institutions to society is ensured, the media in almost all democracies are in private hands, free to operate according to their own will, subject only to general legal restrictions. There are no formal, enforceable mechanisms of holding the press to account for how well it performs as a trustee of the 'people's right to know'. Imagine a journalist discovering information the public might consider relevant. Is she therefore obligated to provide this information to the public, and can the public hold her to account if she does not? Ronald Dworkin drove this point home when reflecting on claims that the public could legitimately raise against journalists. Imagine, he says, that The New York Times chose not to publish the Pentagon Papers. Would this entitle someone to sue the newspaper for not publishing them? 53 This suggestion, he concedes, would be met with horror by most journalists. 52 Sagar, Secrets and Leaks, p. 117. Not much would be gained by adopting an alternative view of authority, in which it only has moral permission to issue directives for action and attach sanctions to disobedience. In that case, the press still has a duty to accept the risk of bearing sanctions. 53  Similarly, if special newsgathering privileges were conferred upon the press by the public, the public might require that the press meet certain standards of training, knowledge, and objectivity in reporting. Could the public choose to institutionally enforce these standards? 54 Could special newsgathering privileges be withdrawn if the press does not measure up to certain standards such that if, say, The New York Times fails to meet these standards, then its newsgathering privileges will be withdrawn? 55 Finally, even if one could surmount these problems and make sense of the idea that the people have entrusted the press with protecting and enforcing their 'right to know', this argument would not confer upon the press the right to disseminate unauthorized disclosures. This is because the 'people's right to know' does not extend to information legitimately classified by government. In effect, the press as agent of the people cannot appeal to that right by way of justifying its publication decisions: The agent cannot have a right to provide the principal with information that the principal has no right to receive. As I detailed this argument elsewhere, below I indicate only its main lines. 56 F. The press as a trustee of the 'people's right to know' There are two arguments commonly invoked to support the 'people's right to know' government-held information. According to the first, this 'right to know' is a human right. Thus, for example, Patrick Birkinshaw claims that access to government-held information is 54 Schauer, 'Rights and the Right to Know',p. 73. 55 In response to this, one might suggest that the principle-agent relation between media and people involves moral, not legal accountability; to do so, one could refer to the professional codes of journalist ethics. The moral guidelines that the codes articulate are presented as requirements that befall the media in their capacity as a trustee of the people implying that the people can hold the media morally accountable for living up to those moral standards (e.g., the British National Union of Journalists' Code of Conduct states: 'All members of the press have a duty to maintain the highest professional and ethical standards. […] they should have regard to the provisions of this Code of Practice and to safeguarding the public's right to know'). This reading of the codes is unwarranted. Codes of journalist ethics are the outcome of self-regulation by the media (Koene, D. 2009. Press Councils in Western Europe. Netherlands Press Council Foundation/The Netherlands Press Fund, The Hague). Moreover, submission to them is voluntary. This means that whatever moral responsibilities the media recognize, they are responsibilities that are self-imposed rather than imposed by the people. That is, whatever moral accountability mechanisms the professional codes imply, they do not indicate mechanisms affecting the profession from without, by the public at large. 56  'fundamental to my membership as a full member of the human race'. 57 A human right to know government-held information is also mentioned in various political documents. 58 Scholarship underpinning this position conceives of human rights in two different ways. The first understands human rights on the natural rights model. Proponents of this approach claim that all human rights, like natural rights, are generated out of the urgent, universal interest in the exercise of human agency and autonomy. 59 It is not immediately clear how a claim to access government information satisfies this description.
Presumably, an argument to this effect could begin with a general moral claim emphasizing that access to information is necessary for the capacity for autonomous and intentional action. As Matthew Liao and Adam Etinson say: 'The relevant basic human right may be that of acquiring the knowledge necessary to be an adequately functioning individual in one's circumstances, or, perhaps even more basic than that, the right to effective agency'. 60 Secondly, one could argue that information under government control belongs with the information necessary to the capacity for autonomous, intentional action. Whether this argument holds, however, depends on how one specifies the general moral claim to the information necessary to autonomous action in modern societies. Under the conditions of pluralism and disagreement characterizing modern societies, general moral claims can be specified in many different, competing ways. Starting with disagreements about the value of autonomy and conditions for autonomous action, people will disagree about what information they need in order to act in autonomously. Some may consider intelligence information necessary to making autonomous political choices; others may not. Those who agree that intelligence information is necessary will disagree about its scope. If information about electronic surveillance is necessary to make autonomous political choices, is all other intelligence information similarly nec- The few human rights theorists who have addressed the problem of translating general moral claims into concrete, action-guiding rules do so by reference to political decision-making. As Samantha Besson says, it is 'the law [that] turns universal moral rights into human rights'. 61 According to Seyla Benhabib, under conditions of disagreement regarding the content of human rights, people should be given an equal say, that is, universal claims should be specified in the process of democratic decision-making. 62 This answer to the specification problem has important consequences for the scope of the human right to access government-held information. Even if the right to know has a normative status prior to political decision-making, it is the decisionmaker, viz. the state, that determines the exact contents of the right to know by resolving the disagreement about the boundaries of the right to know. Before I proceed let me qualify this argument. To the extent that the state exercises (arbitrating) authority over people, it is in virtue of the reasons that individuals have to be subject to it. 63 Such reasons, therefore, articulate the side constraints on the political decision-making process; decisions that do not serve those reasons are not binding. The state's authority to determine the content of the human right to know is similarly restricted. On one traditional view, political authority is instituted for the preservation of the lives and property of the citizens. Correspondingly, political decisions that undermine these aims are illegitimate. From this perspective, denying people access to information without which they would run serious health risks is not legitimate. The state's power to determine the content and scope of the human right to know is, thus, not absolute. Yet, as long as the state, in making the right to know effective in its political decision-making, exercises its powers within the boundaries of legitimate authority it also determines the limits of this right. A classification regime is one way of doing this. In effect, people have no human right to access information that the government has classified legitimately. Correspondingly, the press cannot appeal to this right to justify its publication decisions on classified information: as the people have no right to receive it, the press is not entitled to give it to them.
Similar conclusions follow when one understands the 'people's right to know' along the lines of a second dominant conception of human rights, the 'political conception'. 64 On this view, human rights introduce a system of standards for the domestic conduct of governments that reasonable states would agree to adopt even though disagreeing about their justification. From this perspective, to claim that access to government information is a human right is to claim that access to government information is a pre-institutional moral right to which every reasonable state would give effect in its legislation lest it lose its claim to legitimacy. Just as in the conception of human rights modelled on natural rights, on this conception a human right to access to government information acquires its specific content and scope in national legislation. To the extent that the human right to know is made effective in the particular legislative setting, the state, in institutionalizing a human right to access some government information, also sets limits to it (within the boundaries of legitimate authority). From this perspective, the 'people's right to know' does not extend to information access to which the state limits. In effect, the press cannot appeal to that right to justify its publication of that information.
Above I have considered the press's claim that by publishing classified government information it provides people with information they have a 'right to know'. This claim, I argued, does not hold if we understand the 'people's right to know' as a human right. The 'people's right to know', however, can also be understood as a right of democratic citizenship. 65 Let me briefly consider whether this makes the press's appeal to this right more plausible.
An appeal to citizenship as the ground of the people's right to access government information is as common as an appeal to human rights. Birkinshaw mentions it in one breath with an appeal to human rights: 'The right to information (…) is fundamental to my position as a citizen and a human being'. 66 This, too, is included in various political documents. 67 Here, the people's right to know arises from the relations between citizens and the state, the principal-agent relation, as explained above, where citizen-principals authorize state-official-agents to rule on their behalf, the agents providing an account of their actions. Now, as Bernard Manin, Adam Przeworski, and Susan Stokes point out, to the extent that state officials have the authority to rule, they also have the authority to establish rules of information access: The peculiarity of the principal-agent relation entailed in the relation of political representation is that our agents are our rulers: we designate them as agents so that they would tell us what to do, and we even give them the authority to coerce us to do it. And the rules that our agents impose on us include access to information. (…). The principal-agent model entailed in the relation of representation is a peculiar one, insofar as it is the agents who decide what principals will know about their actions. 68 This implies that if state agents have the authority to establish rules of information access, they also have the authority to limit its scope. Thus, the people have no right to know the information the state lawfully withholds and, consequently, the press cannot appeal to that right to justify its publication of classified information. In objecting to this argument, one may argue that the people must have a right to full access to all government information as a condition of their right to call state officials to account. Note, however, that full disclosure of government information to citizens at large is not necessary to ensure accountability; there are mechanisms of accountability that do not require full transparency (oversight committees, retrospective disclosures). 69 As long as such mechanisms are in place, the scope of the 'people's right to know' does not extend to information that the people, in their capacity as principals, have authorized their agents viz. state officials to classify.
Above I have considered the claim that the press, in virtue of its role as the 'fourth branch of government' and a trustee of the 'people's right to know', enjoys a privileged position with regard to disclosing classified government information. I reject this claim because we cannot make sense of the role of the press thus described. This does not mean that the press plays no role in democratic society, only that the fourth estate model does not properly describe it. Below I consider another argument regarding the role of the press and ask whether it entails a special privilege to disclose classified government information.

G. The press as expert communicator
There is a broad consensus that the role of the press includes, minimally, reporting events of public importance: Providing 'news' to the public. This informational role of the press includes both transmitting specialist knowledge from government and giving voice to public opinion by enabling the principal organizations and groups in society to be heard. 70 In that, the press sets in motion processes of political communication between government, interest groups and citizens. It has been widely recognized that in contemporary pluralistic, mass societies, activating this public communicative sphere cannot be done by means other than the press or, more generally, the media. For example, James Bohman observes that '[r]ather than (…) addressing each other in face-to-face interaction, most [political] communication passes through various expert communicators who package exchanges and discussions for audiences'. 71 Given the 'sheer size of the body of citizens and the complexity of social problems', the press is uniquely positioned to do this mediating job: 'because of their breadth, the media necessary for these tasks are thus 'mass media'. Informal mechanisms alone would be empirically and normatively insufficient for this task'. 72 This view of the press avoids the pitfalls of the fourth estate argument. Unlike that model, whatever privileges and responsibilities attach to the role the press performs in the public domain, they do not arise from a principal-agent relation. Rather, they reflect the necessity of mediated communication in modern democracies, and the unique position the press occupies in the public space to perform this job. In other words, it is only insofar as the press plays a unique role in the public communicative space, which is, as Habermas would say, 73 a shared good in a democracy, that it holds special privileges and responsibilities. Does the unique role of the press described thus confer upon it a privilege to disseminate classified information?
The argument presupposes that democracy is the ultimate foundation of the informational role of the press. This implies that the press, in performing its role, must respect democratic principles including the principle of the democratic allocation of power: If democratic government has authority to classify, then the role of the press cannot extend to disclosing classified information because, as I argued with Sagar above, this would introduce a contradiction into the structure of democratic authority with one political actor (reporters/publishers) allowed to disclose what another political actor (government) has been allowed to conceal. The role of expert communicator does not endow the press with a privilege to disclose information the government is authorized to classify.
So far, I have reviewed one class of arguments defending the different treatment that leakers and press receive for violating the law prohibiting unauthorized disclosures. According to them, leaking and publishing leaks are equally wrong, but the wrong committed by the press is justified in a way that the wrong committed by leakers is not. None of the arguments considered -an appeal to press freedom; the chilling effect of regulation; press property rights; the special role of the press as the 'fourth estate' and trustee of the 'people's right to know'; the press's role as expert communicator -succeeds in establishing a privileged position for the press with regard to disclosing classified information. If the practice of selective prosecution is justified, it would have to be in terms other than those considered 73 Habermas, 'Political Communication'. so far. Below I examine a different strategy to make sense of the difference in treatment that leakers and press receive.

II. THE SECOND ARGUMENT FROM DISANALOGY: LEAKING IS WORSE THAN PUBLISHING LEAKS
Perhaps this difference in treatment reflects a moral difference between leaking and publishing leaks viz. that leaking constitutes a more serious wrong than publishing leaks and, therefore, deserves harsher punishment. Lawyers and legal scholars have suggested two variants of this argument. According to one, leaking should be considered worse than publishing leaks in the same sense in which, for example, stealing is considered worse than distributing stolen goods (fencing). According to the second version of the argument, leaking is worse than publishing leaks because leaking, unlike publishing leaks, involves an additional wrong on top of violating the law prohibiting unauthorized disclosures.

A. Stealing information is worse than distributing stolen information
Legal scholars who entertain the idea that leaking is the worse offense hint at the difference between stealing information and distributing stolen information. Stealing and distributing objects which one knows (or believes) to be stolen are different acts with a different moral import. Whereas both are wrong, they differ in degree of wrongness, stealing being more serious than fencing. The suggestion is that leaking constitutes the more serious wrong because leaking involves theft and publishing leaks does not. The point was clearly articulated by a US federal judge who ruled in 1990 in favor of a newspaper that published a leaked document from a local police department: There is ''a distinction in the criminal law between thieves and receivers of stolen goods (…). This court does not mean to say that the media are liable to the same degree as those who act improperly for it, or those who supply it with improperly obtained material. But the distinctions are of degree, not of kind. '' 74 This way of explaining the difference in treatment between leakers and press is problematic on two counts. First, the legal foundation of the claim that fencing is a lesser offense than theft is not obvious. 74 Scheetz v. Morning Call, Inc., 747 F. Supp. 1515(E.D. Pa. 1990).
The status of fencing in Anglo-American law has fluctuated. Early Anglo-American law did not register fencing as a crime at all. In the 19th century, fencing was considered an offense subject to lesser penalties than theft. In the 20th century fencing has come to be treated as interchangeable with theft and subject to the same punishment. Thus, the US Model Penal Code treats fencing as an alternative means of committing theft and subjects it to the same penalty. 75 The 1968 English Penal Code even includes the possibility of a harsher penalty for fencing. 76 Given its mutable status on the statute books, and especially its status in current law, fencing is not necessarily a lighter offense deserving lighter treatment than stealing goods. By the same token, publishing leaks is not necessarily a lighter offense deserving lighter treatment than leaking. Second, while the difference between theft and fencing rests on uncertain legal foundations, it can perhaps be made on moral grounds. Even so, it is not clear we can extend it to the distinction between theft and distribution of stolen government information. To explain why, let me first reconstruct the moral content of the wrong involved between theft and fencing.
The thief's wrong consists in the harm he inflicts on the owner by depriving her of her property rights. With regard to the fence's wrong, we can adopt two interpretations. First, the fence's wrong consists in perpetuating the wrongful deprivation of the victim owner's property rights first effected by the thief. Second, the fence's wrong consists in encouraging the commission of future thefts by helping to create a market for stolen goods: 'without receivers [and distributors]', as Stuart Green puts it, 'there would be no thieves '. 77 Placed in this conceptual frame, the leakers-thieves would deprive government of its property rights in certain classes of information. The press, as distributors of stolen information, would either perpetuate the prior harm conferred by the leakers-thieves and/or encourage more information theft by creating a market for leaks. The latter reading is problematic: The argument that the distributor encourages further thefts commits us to claiming that there would be no leaks without the press. However, we can imagine cases of 75  stealing government information without the intention of publishing it. I suggest ignoring this perspective and paying closer attention to the former. On this reading, as Green puts it, the thief harms the property owner by causing a radical, involuntary change in her relationship to her property: From possession to non-possession. 'Unlike the thief, the receiver [and distributor] does not cause any change in the status of the owner's rights', Green argues. He merely perpetuates the harm inflicted by the thief. Given that 'perpetuating an owner's loss of property is a lesser wrong than causing him to lose his property to begin with', 78 Green says, the receiver/distributor of stolen property deserves less blame than the thief. Applied to the divide between the leakers and the press, the leakers-thieves harm the government by causing the radical, involuntary change in the government's relationship to its property in information: From possession to nonpossession. The press, as distributor of stolen information, merely perpetuates the prior harm effectuated by the leakers-thieves and, thereby, deserves less blame and punishment. But this description is inaccurate in the context of goods such as information. Whereas stealing, say, a bicycle, involves a transfer of property, in the case of stealing information it would be incorrect to say that any transfer of property has taken place. This is because even if the leaker steals government information, government is not deprived of information in the sense in which a bike owner is deprived of her bike. The government still has the information; what it has lost is exclusive access to it: now, not only the government but also the leaker has access to information. If there is a change in the government's relationship to certain classes of information effectuated by the leaker-thief, then it is a change from exclusivity of possession to nonexclusivity of possession.
This way of spelling out the wrong involved in unauthorized disclosures challenges the claim that leaking is worse than publishing leaks and, thus, that leakers should be prosecuted more harshly than the press. The challenge comes in either of two ways. In the first scenario, it reverses the moral book-keeping and presents the press's distributing information as worse than the leaker's stealing it. In the 78 ibid, pp. 45, 37. second scenario, it shows that press and leaker are implicated to the same degree.
Let me start with the first scenario: The wrong involved in leaking/stealing government information consists of depriving government of exclusive access to certain classes of information. Note that exclusivity of access to information has degrees. Whereas a bike cannot be more or less stolen, information access can be more or less exclusive: We speak of a higher degree of exclusivity when fewer individuals have access to information, and of a lower degree of exclusivity when more individuals do. Given that exclusivity of access is a matter of degree, its loss is a matter of degree too. Now the degree to which government is deprived of exclusivity of access is lower when only the leaker can access it and higher when it can be accessed by millions of citizens. Correspondingly, any harm the government suffers is more serious when its information can be accessed by millions of citizens. If this is the case, then it is incorrect to claim that leaking-stealing government information is worse than publishing-distributing stolen government information. Given that it is the press that discloses the stolen information to millions of its readers, the press does not merely perpetuate the prior harm inflicted by the leakers-thieves; it aggravates it by increasing the degree of the deprivation of exclusivity of access. This is recognized by most commentators when they say that '[l]eaks of classified government information pose little threat without the press as a medium for public disclosure'. 79 This line of argument takes us to the following conclusion: Even if distributing stolen bikes is less bad than stealing them, it is incorrect to argue that publishing stolen government information is less bad than stealing government information. If the comparison holds at all, the press should be penalized to a higher degree because the wrong it commits is more serious than the wrong involved in leaking.
The line of argument presented above assumes that the leaker's stealing classified government information and the press's publishing it are two unrelated events. Whether this is indeed the case depends, however, on the intentions that the leaker has in disclosing classified information to the press. If the leaker steals government information with the intention that the press discloses it to the public at large 79 Xanders, 'A Handyman's Guide ', p. 767. (rather than the press merely further investigating the matter), leaking and publishing are inherently related. As the leaker is not separate from the publication of the classified information by the press but rather implicated in it, the leaker is also involved in the supposed wrong committed by the press, namely the publication of classified information. In other words, the leaker foresees and intends the result of millions having access to the classified information. In this scenario, the wrong committed by press and leaker is the same. As there is no difference in the wrong thus committed, there is no reason to differentiate prosecution. Importantly, independent of whichever scenario we take, the actual practice of selective prosecution focusing on leakers and leaving the press unpunished is unjustified.

B. Two wrongs versus one
Another way of explaining why leaking is worse than publishing leaks is to say that while both leaking and publishing leaks violate the law, leaking involves an additional wrong on top of it. Namely, the leakers not only violate the law prohibiting unauthorized disclosures, they also violate their contractual obligations. Snowden consented to nondisclosure as a condition of his employment at the NSA. When he leaked classified information, he not only violated the law but also his contractual obligations. Sometimes, instead of pointing to their contractual obligations, one points to the obligations that civil servants and government employees acquire in virtue of the professional role they occupy. On this view, civil servants are roleobligated to comply with their superiors' directives, including directives to respect the classified character of certain documents; when they leak these, they not only violate the law but also their professional obligations. 80 The fact that leakers violate contractual or professional obligations in addition to violating the law seems to implicate leakers more than the press. When publishing classified information, Greenwald, unlike Snowden, was not bound by a government contract nor was he a government employee. Thus, he was not subject to corresponding obligations. Not bound by such obligations, he did not 80 Boot, 'Classified Public Whistleblowing', p. 549. violate them. In effect, whereas Snowden committed two wrongs, Greenwald committed only one. Given the difference in the degree of wrongdoing, the difference in prosecution seems logical as a matter of proportionality: More severe wrongs should be punished more and less severe wrongs less.
Reasoning along these lines, Heidi Kitrosser argues that the protection offered to civil servants leaking classified information should be weaker than the protection offered to the press for publishing it. 81 In Kitrosser's view, unauthorized disclosures by both civil servants and the press should be seen as political speech protected by freedom of speech legislation. 82 This is because unauthorized disclosures, unless they risk grave damage to national security, constitute a check on government supplementing the structural oversight mechanisms. This means that while both civil servants and the press violate the law prohibiting unauthorized disclosures, this wrong can be mitigated by presenting it as an exercise of political speech. However, while both civil servants and the press can appeal to their right to political speech when they disclose classified government information, the exercise of this right by civil servants conflicts with their duties in their capacities as part of the executive branch to contribute to its efficacy (including its secret-keeping efforts), a duty the violation of which the press is not guilty of. The wrong civil servants commit when disclosing classified information is therefore greater than the wrong committed by the press and their claim to freedom of speech protection is weaker.
Note that the 'two-wrongs-versus-one' argument, if correct, applies only to cases in which the additional wrong the leaker commits, viz. violation of her contractual and/or professional obligations, cannot be justified. If civil servants are justified in violating their contractual and/or professional obligations, there is no sense in which this violation implicates them more than the press. Under these circumstances, the only moral wrongs that would enter the moral book-keeping refer to the violation of the law prohibiting unauthorized disclosures, and the argument registers no difference between leakers and press in this respect. Moreover, if the violation of the contractual obligation by leakers is justified (because, for 81 Kitrosser, 'Leak Prosecutions', p. 926. 82 ibid, p. 908. example, the information the civil servant leaks reveals an abuse of power by the government), so too is her violation of the obligation to obey the law, in which case no moral wrong can be registered at all. If disclosures of classified information by civil servants satisfy these criteria, the same disclosures subsequently published by the press must satisfy them too. Therefore, the violation of the law on the part of the press must be justified too. Again, we have no reason to treat these two classes of agents differently. If anything, we have reason to refrain from prosecution in both cases.
The 'two-wrongs-versus-one' argument holds, then, only on the assumption that disclosures lack justification of the kind indicated above. If this is the case, does the disclosure of classified information implicate the leakers more than the press and does this difference justify the difference in prosecution? To answer these questions, we must measure and compare the wrongs committed by leakers and press. C. Calculating wrongs I argued earlier that the wrong involved in unauthorized disclosures consists in the government's loss of exclusive access to certain classes of information. Accordingly, its severity is measured by the number of people who acquire access to information formerly in the exclusive possession of government: The wrong is greater when more people acquire access to it; it is lesser when fewer people acquire access. Having specified the wrong of disclosure as depriving government from exclusive access to certain classes of information, I argued that the wrong involved in leaking is not greater than the wrong involved in publishing leaks. Rather, even if leakers commit two wrongs compared to one wrong committed by the press, the total wrong they commit is either less than the wrong committed by the press (the leaker reveals classified information only to the reporter; the press multiplies the unauthorized access by millions of readers) or equal to it (both leaker and press reveal classified government information to millions of readers). I concluded that the 'two-wrongs-versus-one' argument fails to explain why we should prosecute leakers more than the press. The argument above specifies the wrong involved in the leaker's breach of contract in terms of its consequences viz. the number of people acquiring unauthorized access to classified information. Perhaps, however, this consequentialist reading is insufficient as a description of the wrong involved in unauthorized disclosures by leakers. Instead, we should see it in non-consequentialist terms as a violation of the civil servants' duty to act as specified by their contract and/or professional role. This non-consequentialist reading of the wrong committed by leakers identifies a difference in their normative situation compared to the press: As the press is not a party to the wrongdoing understood like this, there is no sense in which it can aggravate the wrong committed by civil servants; as it does not stand in the employment relationship to the government, it does not commit a similar wrong. This non-consequentialist reading of the 'two-wrongs-versus-one' argument identifies a disanalogy in the normative situation of leakers and press. Does this disanalogy justify the difference in prosecution?
It does, but not in the form it takes in current practice. The 'twowrongs-versus-one' argument sets out to explain the difference in prosecution by pointing to the difference in wrongdoing. In postulating that the difference in prosecution should reflect the difference in wrongdoing, it relies on the proportionality principle, which requires that more severe wrongs should be punished more. Recall that currently leakers are prosecuted, and the press goes unpunished. On the logic of the proportionality principle, the absence of prosecution should reflect the absence of a wrong to be prosecuted. Hence, in order to justify refraining from prosecuting the press, the argument would have to claim either that the press commits no wrong or that the wrong it commits viz. violation of the law should be discounted. By claiming that the press commits no wrong, the 'two-wrongs-versus-one' argument would contradict itself: It explicitly concedes that the press is guilty (even if only of one wrong). The latter option viz. discounting the violation of the law as an offense deserving prosecution is problematic because any reasons for discounting the violation of the law by the press as an offense deserving prosecution (e.g., democracy-enhancing consequences of disclosures or disclosing information revealing government wrongdoing) apply to the leakers too. Now if such reasons justify the violation of the law by the leakers, then they must also justify the additional wrong they commit in disclosing classified information viz. violation of contract/role obligations. In effect, there is no moral wrong that either the press or the leakers commit. Ergo, the argument cannot provide grounds for differentiating prosecution because it would commit us to dropping prosecution altogether.
In conclusion, even if the leaker's breach of contractual and/or professional obligations establishes a difference in the moral situation of leakers and press, this difference is insufficient to justify the current practice of prosecuting leakers and not the press.

D. Unauthorized disclosures and professional duties of the press
The 'two-wrongs-versus-one' argument stipulates that the normative situation of leakers and press is disanalogous. Whereas both are guilty of unauthorized disclosures, leakers, unlike the press, commit an additional wrong: violating their professional duties. This section inquires into whether a similar argument can be made with regard to the press. Journalists, like doctors, lawyers, engineers or accountants, consider themselves professionals in their own field, claiming to adhere to a set of ethical standards governing their role. 83 As argued above, the professional obligations of the press cannot be understood on the fourth estate model; however, they can be understood in terms of the informational role of the press, a role the press acquires in virtue of its unique position in the public space of mediated communication. Following Sagar, I argued that, on pain of creating a contradiction in the structure of democratic authority, the informational role of the press does not extend to disseminating information the democratically elected government has classified legitimately. Here, I draw a further implication from that argument: When the press publishes information legitimately classified by government, it violates its professional role. This puts the press on the same footing as leakers. Thus, there is no reason for treating the press differently when it comes to accountability for unauthorized disclosures: prosecutors should either prosecute both (even if the violation of pro-83 Sigma Delta Chi (US); the UK Code of Professional Ethics of the Institute of Journalists, etc. fessional duties by civil servants and by journalists calls for different sanctions) or neither. III. CONCLUSION In this essay, I have reflected on the moral puzzle that the difference in treatment that leakers and press receive with regard to unauthorized disclosures of classified information poses. Whereas both leaking and publishing classified information violate the law prohibiting unauthorized disclosures, the prosecution focuses on the leakers, leaving the press unpunished. Assuming that state secrecy is, under certain conditions, legitimate, I have argued that as far as normative assessments are concerned, there is no moral reason for making a difference between the civil servants who leak classified information and the press who publishes it: If the prosecution targets leakers, it should also target the press. If the press goes unpunished, so should leakers.
If selective prosecution is not a matter of moral principle, its resilience can be explained in terms of the pragmatic interests guiding prosecutorial discretion. Social scientists have suggested that if the press is not prosecuted it isn't because it does no wrong or is less culpable than leakers, but because this might be government information management. 84 Or that the executive has a strong incentive to avoid criminalizing reporters and publishers rather than compromise the government's instrumental use of the press, 85 the press retaliating by publicizing the foibles of officials. 86 Full discussion of the pragmatic considerations guiding prosecutorial discretion goes beyond the scope of this paper. The point here was to search for a moral principle justifying the practice; one has not been found. (DEMSEC GA 639021). I thank Govert den Hartogh and Tim Meijers for discussions on the earlier versions of this essay. I would like to acknowledge insightful comments and suggestions of two anonymous reviewers for this journal.

OPEN ACCESS
This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.