Deep Secrecy

David Pozen

Posted in , , ,

Donald Rumsfeld has been unfairly ridiculed. During a 2002 press conference, a reporter asked about the link between Iraq and al Qaeda, and he gave the following response:

[A]s we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns—the ones we don’t know we don’t know. And if one looks throughout the history of our country and other free countries, it is the latter category that tend to be the difficult ones.1

These lines quickly became infamous, an emblem for the administration’s general lack of candor and for Rumsfeld’s particular brand of obscurantism. Watchdog groups called it “the most nonsensical remark made by a public figure” in memory.2

Yet while Rumsfeld’s remark may have been unusual for the setting, it was not nonsensical, and to the contrary it contained an important insight. There are things we know we know, things we know we do not know, and things we do not know we do not know. And, in fact, the secrets in the latter category tend to be the most difficult ones for a free society. They are deep secrets.

The distinction between deep and shallow state secrets has received hardly any scholarly attention. I believe this reflects a serious hole in the literature on government secrecy, because once elaborated, the deep/shallow distinction can shed light on a wide range of theoretical and practical debates in the field.

Deep Versus Shallow Secrets: The Basic Distinction

In the interpersonal context, the deep/shallow distinction has been recognized at least since sociologist Kim Lane Scheppele explored it two decades ago.3 As Scheppele observed, sometimes the target of a secret knows or suspects that information is being concealed from her, even though she does not know the content of the information. She has learned that a house she is interested in purchasing failed an inspection last year, but she has not learned why. The findings of the inspection report are a shallow secret.

Other times, the target of a secret does not realize that something is being kept from her. The purchaser has been given no clue, and has no reason to suspect, that the house is haunted. Its hauntedness is a deep secret. Scheppele associates these terms with metaphors of light. Whereas the deep secret’s target is “completely in the dark, never imagining that relevant information might be had,” the shallow secret’s target “has at least some shadowy sense” that she is lacking relevant information.4

The deep/shallow distinction is not quite as neat as it might first appear. If a secret is deep at the outset, how much information must be disclosed or made discoverable for the secret to cease being deep? Suppose that the seller of the haunted house tells the prospective buyer that there are occasional strange noises, or mentions that “one never feels alone” on the premises, or alludes to a newspaper article that describes the property as haunted. Suppose that the buyer starts to develop a sense that something is fishy, but she can’t quite place it. At what point, exactly, does the secret pass from deep to shallow? Examples such as this help us to see that there is no precise line separating deep from shallow secrets, but rather a continuum running from one extreme (maximally opaque) to the other (maximally intelligible).

In the government context, things get significantly messier. Many different internal groups may be protecting the same information and, in a democracy, many more outside groups may claim an entitlement to it. The “keepers” and “targets” of state secrets are not monistic categories; we need to disaggregate them. It also becomes more complicated to fix an appropriate conceptual frame for the analysis. To measure secretiveness against any set of external standards, one must first determine what constitutes the secret and the relevant informational background against which to understand it. Modern governments are sufficiently open and complex that virtually every nonpublic action they take will be embedded within a vast web of public rules, norms, policies, and pronouncements. At a high enough level of generality, nothing they do is a complete surprise.

To make the deep/shallow distinction useful at the state level, it is necessary to tweak it. We need to ask not just whether a secret is deep or shallow in the Scheppelian sense to any particular individual or group, but also how deep or shallow it is to society as a whole. Thus, if no one outside of a particular executive branch unit knows of a program the unit is running, the program should be deemed an extremely deep secret. If the general public knows all but a few details, the program is an extremely shallow secret. If the program is disclosed to the other branches of government, but not to the public, it lies somewhere in the middle of the depth continuum. Illegal programs will tend to be deeper secrets than legal ones, all else equal, given the assumption that laws are followed.

Although many borderline cases may arise, we can coherently apply the “deep” and “shallow” labels to secrets that lie sufficiently far to one side of the depth continuum. We can say that a government secret is deep if a small group of similarly situated officials has concealed the fact of its concealment from the public and from other officials. A state secret is shallow if ordinary citizens understand they are being denied relevant information and have some ability to estimate its content.

This framework permits us to discriminate among state secrets on the basis of their overall accessibility, without passing judgment on their content. And it carries the important implication that some secrets may be known only to government actors and yet nevertheless be less deep than other secrets that are known to a different set of government actors. While one fairly bright line continues to separate deep from shallow state secrets—the public always is aware of the existence of the latter, ignorant of the former—the depth discrepancies between many secrets will be more quantitative than qualitative.

Why It Matters

A vast literature explores the pros and cons of government secrecy, and there are plainly good arguments on both sides. Deep secrets upset this balance. Compared to their shallow counterparts, they threaten to do significantly more damage to a polity, without offering concomitant advantages. They do this in part because they are more secretive, and the normative concerns raised by state secrets tend to increase in rough proportion to their obscurity. But they can also create a host of distinctive harms.

I am, of course, speaking in generalities. Some shallow secrets will be highly destructive on account of their content. Some deep secrets will be benign. Appreciating the deep/shallow distinction cannot answer many of the normative questions raised by state secrecy, though it may help to clarify them.

My Article explores these questions at length and connects them to recent controversies. Here, let us consider just a few basic concerns:

Utilitarianism. Sometimes deep secrecy may help officials achieve a worthwhile goal more fully or efficiently than they could have in the presence of greater input and oversight. More reliably, it will exacerbate the standard consequentialist problems of state secrecy. In virtue of their insularity, deep secrets will tend to foster corruption and abuse, groupthink and error, and violations of privacy.

They may also be more likely to breed leaks. Unauthorized leaks typically occur when a secret-keeper is unhappy with the activities of colleagues and cannot have her position vindicated internally. When a secret is shallow, disaffected keepers may be able to argue their position to a wide range of people without compromising the secret. When a secret is deep, however, disaffected keepers will have fewer such channels for dissent and may have no recourse to deter the policy short of full publicity. Something like this appears to explain the case of Thomas Tamm, the conscientious Justice Department lawyer who exposed the Bush administration’s remarkably deep Terrorist Surveillance Program by placing a phone call to the New York Times.

This possibility complicates the link between shallowness and publicity. As a general matter, there are good reasons to assume that alerting a greater number of people in greater detail to the existence of a secret will increase the odds that the secret’s contents will be compromised. But this causal relationship need not hold true in all cases, and toward the upper end of the depth spectrum it may reverse itself. Shallow secrets tend to unravel gradually, accretively, as outsiders piece together clues and pursue further disclosures. The keepers have some time to react. Those deep secrets that become public tend to do so in dramatic fashion.

Democracy. Commentators frequently overstate the tension between state secrecy and democratic values. Secret decisions may have inherent deliberative and participatory deficits that preclude robust democratic legitimacy. But if the public expressly grants someone the authority to keep certain secrets, and if the grantee faithfully applies that authority subject to external review, it is fair to see this exchange as a victory for democracy, not simply as a concession to practical necessity.

This logic breaks down with deep secrets. Deep secrets entail severe deficits in ex ante authorization by citizens and their elected representatives; partly as a result, they also entail severe deficits in ex post accountability. There can be no reliable expectation that they will foster equal treatment or even rationality. They facilitate an excessive concentration of power, which can constitute domination and which upsets the balance of mixed authority that gives the government a claim to being politically egalitarian. Deep secrets undermine not only criticism and deliberation but also the possibility of any meaningful form of consent or precommitment. Given the impenetrability of deep secrets, a decision to authorize their keeping involves not so much a delegation of administrative discretion as an abandonment of self-rule on the matters at stake.

From this perspective, a popular decision authorizing the President to keep deep secrets would not suffice to legitimate the practice, as it would license him to violate norms that are themselves constitutive of democratic rule. It would be like a vote to give away the right to vote, or to protest, or to demand fair and reasonable treatment from government. Like an attempt to contract away an inalienable right. It is not even clear that a statute purporting to authorize deep secrecy could have sufficiently meaningful or intelligible limiting principles to warrant the label “law.”

The Constitution. If you asked most American legal scholars about secrecy’s role in the Constitution, they would probably say that it has none. The text mentions “secrecy” only once, in a relatively obscure clause,5 the Framers did not have all that much to say on the subject, and throughout the vast part of our history “few have seen constitutional issues in governmental concealment.”6

I do see constitutional issues in governmental concealment, many issues, but I believe that secrecy’s constitutional status is fundamentally ambiguous. For every textual, historical, structural, prudential, or doctrinal argument one can posit that would support a public “right to know” what the government is up to, there is a plausible counterargument to show how the existence of that right might be disputed or its scope severely limited. Indeed, it seems fairly clear that the Constitution does not demand total transparency in government. Textual clues such as the discrepancy in the Article I and II vesting clauses,7 historical clues such as the Framers’ concern to strengthen the central government and to enable unilateral action in times of emergency, and structural clues such as the allocation of treaty- and war-making powers to the body in government best able to keep secrets, converge to suggest a constitutional principle authorizing occasional executive branch secrecy.

Not so with deep secrets. Perhaps the President holds a reserve power to use deep secrecy when doing so might save the nation. Then again, perhaps the President holds a reserve power to bypass every last constitutional restraint when doing otherwise would render the constraint self-defeating. Absent a state of exception, the Constitution’s logic strongly suggests that the President was not meant to keep outsiders totally in the dark about his activities.

One important reason is that deep secrecy threatens the relationships among the coordinate branches. If the executive maintains deep secrecy about a program it is running, Congress will have no ability to fulfill its legislative, oversight, or appropriations duties related to the program. Judges will not be able to adjudicate cases or controversies that might have arisen. All of the democratic and legal checks built into our tripartite structure, it turns out, depend on each branch being aware of what the other is doing. Deep secrecy threatens this sufficiency condition by allowing executive branch aggrandizement to occur unnoticed.

To put the point a little more colorfully, we might say that there is a principle to be derived from the Constitution that forbids the existence of informational “black holes” in government: activities taken by one branch that the others have no means of learning about. Shallow secrets may function as singular points of privilege that preserve the separation of powers. Deep secrets can open up black holes into which the powers of a coordinate branch and the rule of law disappear. There is no specific textual hook on which to hang this claim, but the structural harm caused by black holes can be inferred from basic principles of accountability, diffused authority, and checks and balances that underlie our constitutional system.

America’s postratification practices bolster this view. Although they have not been conceptualized as such, landmark open-government statutes such as the Foreign Intelligence Surveillance Act (FISA) proscribe the use of deep secrecy. The central feature of FISA, and its central advance over prior practice, is that it subjects the executive’s secret plans to approval by a separate and nonaligned group of officials in the judiciary—the classic means of moderating depth within government while preserving the advantages of opacity beyond. FISA represents a decisive repudiation of the deep-secret surveillance practices of earlier administrations. The Bush administration’s warrantless surveillance program revived that discredited tradition.

What Can Be Done?

If it is correct that deep state secrecy ought to be avoided whenever feasible, the question arises how we might achieve this aim.

One possibility is for Congress to pass generalized disclosure requirements. Congress can demand public disclosure of the second-order policy choices made by the executive branch (the types of activities it has taken, for what reasons, pursuant to what processes and legal authorities), while forgoing disclosure of the first-order choices embedded within (the specific content of those activities).

Recent events have increased interest in such measures. Until the leaks started to appear, often several years after the fact, the Bush administration refused to disclose the basic contours of the new detention, rendition, interrogation, and surveillance programs it had established. These secrets involved not merely tactical decisions about particular targets or methods, but high-level legal and policy decisions about the nature of the “War on Terror.” Perhaps the closest the administration came to providing notice of these enormously controversial moves was Vice President Cheney’s cryptic statement shortly after the 9/11 attacks that, in fighting terrorism, the administration would “have to work sort of the dark side,” to “spend time in the shadows.”8

Second-order disclosure laws are limited tools, however. As the example of FISA suggests, many of the most promising solutions are internal to government. By increasing and diversifying the pool of officials who are aware of a secret, we can capture some of the benefits of publicity without incurring all of the costs. Even when a secret must remain an unknown unknown to the world at large, internal checking measures can serve a whistleblowing function (rooting out unlawful or nefarious conduct), a deliberative function (improving the quality of decision making), and a representational function (enhancing the democratic credibility of the policies chosen).

The general strategy has a nested structure. When public disclosure and deliberation cannot be permitted in advance, interbranch disclosure and deliberation ought to take its place; when interbranch disclosure is not feasible, interagency deliberation ought to substitute; and when interagency disclosure is not feasible, more robust intra-agency deliberation should be pursued. By disaggregating the government, by seeing official secret-keepers as a plural “they” rather than a singular “it,” we open up the possibility of partial solutions that do not entail the public’s involvement. Although widening and diversifying the circle of keepers may increase the risk of spilling secrets, this risk can be managed. Institutional depth is remediable in a way that interpersonal depth is not.

In the end, though, there is no escaping the influence of bureaucratic culture. To minimize deep executive secrecy, formal mechanisms can only do so much. Congress and the judiciary can only do so much. If the keepers are determined to keep their secrets deep, no matter the cost, there is not much the outsider can do. Consequently, it is vital that the government works to foster a culture that rejects deep secrecy. A bureaucratic culture that looks down upon deep secrets need not be a culture that looks down upon shallow ones. To the contrary, an institution that draws this distinction can breed greater respect for shallow secrets both within and beyond its walls.


State secrecy is such a vast and heterogeneous—and secretive—phenomenon that it is difficult even to talk about it systematically. In my experience, people rarely try.

The concept of depth provides a vocabulary and an analytic toolkit with which to describe, assess, and compare state secrets, a consilient framework for evaluating all that the government does without the public’s full knowledge. The analysis trades on a simple distinction, but it permits us to see that these secrets differ along a variety of stable dimensions; that we can draw meaningful conclusions about them without judging their content; and that there is a vast space between total public disclosure and maximal internal stealth, between sunlight and darkness. Secrecy is a topic that by its nature tends to confound measurement and criticism. It need not confound them quite so much.


Copyright © 2010 Stanford Law Review.

David Pozen is a 2007 graduate of the Yale Law School.

This Legal Workshop Editorial is based on the following Law Review Article: David Pozen, Deep Secrecy, 62 STAN. L. REV. 257 (2010).

  1. U.S. Dep’t of Defense, News Briefing—Secretary Rumsfeld and Gen. Myers (Feb. 12, 2002), (remarks of Secretary of Defense Donald H. Rumsfeld).
  2. Rum Remark Wins Rumsfeld an Award, BBC NEWS, Dec. 2, 2003,
  4. Id. at 21, 76.
  5. U.S. Const. art. I, § 5, cl. 3 (Journal Clause).
  6. Louis Henkin, The Right to Know and the Duty to Withhold: The Case of the Pentagon Papers, 120 U. PA. L. REV. 271, 276 (1971).
  7. Compare U.S. CONST. art. I, § 1 (vesting in Congress legislative powers “herein granted”), with id. at art. II, § 1, cl. 1 (vesting in the President “{t}he executive Power”).
  8. Interview by Tim Russert with Richard Cheney, Meet the Press (NBC television broadcast Sept. 16, 2001), quoted in Jane Mayer, The Dark Side 9-10 (2008). These lines had a double meaning. The counterterrorism effort, they insinuated, would be not only observationally dark but morally dark, not only opaque but unpleasant. A (vice-)head of state faced with an elemental threat to his country, Cheney was expressing the atavistic desire to operate without scrutiny or constraint. His imagery foreshadowed an intent to maintain through deep secrecy a dimension in which the normal rules do not apply.

Post a Comment (all fields are required)

You must be logged in to post a comment.