Telling Secrets

Washington is bracing for another Wikileaks document dump later this week and the Pentagon is urging reporters not to publish the secret files from the Iraq war — once again, the conversation has turned to whether or not there’s a danger in releasing this information. But in a city full of fractious disagreements, there is one issue that nearly everyone in Washington agrees on: The overclassification of information in the name of national security has run amok. We need "effective measures to address the problem of overclassification," President Barack Obama stated last year. "We do overclassify," affirmed Director of National Intelligence James R. Clapper Jr. at his confirmation hearing this summer. "We can be a lot more liberal, I think, about declassifying, and we should be," he added.

Excessive government secrecy is an evergreen concern — as far back as 1956, a Defense Department study complained that overclassification had "reached serious proportions." This problem has serious ramifications throughout the vast national security bureaucracy. It impedes the flow of information across agency boundaries, obstructs the feedback mechanisms that keep policies and programs on track, conceals error and incompetence, undermines oversight and accountability, and fosters public ignorance on vital matters of national security and foreign policy.

Given the severity of this problem and the seemingly bipartisan will to devise a solution, it seems fitting that Congress passed — and the president signed into law on Oct. 7 — a set of remedial measures called the Reducing Over-Classification Act.

The new law mandates, among other things, that classifiers receive formal training in the proper use of classification, enlists agency inspectors general in overseeing the classification system, encourages the release of unclassified versions of certain intelligence, and creates a new position at the Department of Homeland Security to assist state and local officials in accessing information.

Facing a problem so deeply entrenched in the U.S. government, will the new act make any tangible difference? Forget eliminating the problem completely — will it even "reduce" overclassification, as its title modestly proclaims?

It almost certainly will. There is a real need for training in the proper exercise of classification authority, as the law requires, because tens of thousands of people currently implement classification procedures with minimal supervision. And by tasking inspectors general to participate in oversight of the classification system, it will multiply the number of sharp eyes on what material is classified several times over. In fact, each provision in the act is useful, and none seems likely to do any harm. Its passage is a notable achievement in a policy field that is littered with failed proposals.

Nevertheless, the new law also has profound shortcomings that illustrate the depth and complexity of this problem.

The law’s fundamental flaw lies in the fact that it does not define "overclassification." Because it does not provide any criteria for identifying the problem, it cannot meaningfully inform the training or oversight that is supposed to reduce it. Providing such a definition is intrinsically difficult because it involves practical judgments about the requirements of national security and also raises politically delicate questions about legislating the boundaries of executive authority. But without clear definitions, the ensuing policy is unavoidably vague, if not altogether toothless.

In practice, the term "overclassification" is used in two distinct ways. Of course, it refers to information that should be unclassified and publicly available, but is kept secret. But the term is also used more narrowly to refer to information that is classified at too high a level, thereby impeding sharing within the U.S. government. At a lower classification, it would still be classified — and withheld from public disclosure — but it would no longer be "overclassified."

The act blurs this distinction, but its provisions will do more to promote information sharing among authorized persons than force the public disclosure of wrongly classified information. This law isn’t going to trigger a new wave of declassification.

To the contrary, the act actually tends to reinforce the status quo in significant ways. In particular, it embraces the president’s executive order on classified national security information as the reigning standard of proper classification. In effect, the classification system is whatever the president says it is. So under the act, classifiers will be trained in the correct application of the president’s policy, and oversight will consist of verifying that the executive order was implemented correctly.

What this means is that the law does not contemplate the possibility that the president’s classification policy could be wrong! If the executive order is being faithfully implemented, then as far as the act is concerned, all must be well by definition.

But this leaves the roots of overclassification untouched. Agency officials often make absurd, erroneous, or self-serving classification judgments — and not just because they lack training or supervision. For example, ever since 2007, the Office of the Director of National Intelligence (ODNI) has publicly disclosed the total budget for the National Intelligence Program. But last year, when I asked for the size of the 2006 budget, ODNI’s staff asserted that the 2006 figure is still classified and would damage national security if disclosed. This is self-evidently ridiculous and reflects the widespread inability of classifiers to re-evaluate past practices in the light of new circumstances.

The new act provides no mechanism for correcting overclassification of this kind. For that, a different approach will be needed.

As it happens, the Obama administration has recently initiated one such approach to correcting classification errors. The president’s executive order, which took effect last June, established a procedure called the "fundamental classification guidance review," and it requires every agency to start reviewing all its current classification guidelines in order "to identify classified information that no longer requires protection and can be declassified." Not only that, but a directive states that the review is supposed to involve "the broadest possible range of perspectives" to validate classification policies and eliminate obsolete classification requirements.

The success of this review process, to be completed by mid-2012, is not guaranteed. It might be implemented perfunctorily, halfheartedly, or perhaps not at all — despite the presidential order. And if, for example, items such as the 2006 intelligence budget total remain classified at that point, we will know it was a failure. But the review represents a novel attempt to systematically challenge and adjust the inherited classification judgments that are the legacy of a now distant past.

Beyond mere adjustments and corrections to the status quo, Obama has also raised the possibility of designing "a more fundamental transformation of the security classification system." But this notion has not yet been given any substance.

To point out the limitations of the new act is not to devalue it. Others, such as the late Sen. Daniel P. Moynihan, who proposed to legislate a statutory foundation for classification, have aimed higher and ended up achieving less. But the incremental progress represented by this law is a reminder that, when it comes to addressing the problem of overclassification, much remains to be done.

Secrecy in matters of national security is easily mistaken for security itself. But they are not the same thing, and sometimes they are opposites. This is particularly true when secrecy corrupts the policy process, diverts resources into bottomless black holes, and leads the country where it would never have voluntarily chosen to go. Reducing secrecy to a minimum is simply good democratic hygiene.