Column: A Starting Point for Curbing Government Secrecy
A little-known presidential advisory committee recently issued recommendations for transforming the way government officials classify national security information. The goal is to rein in the growing problem of overclassification — the classification of information that poses no threat to national security. Several of the committee’s proposals hold great promise for shifting the culture of secrecy that pervades many agencies. But President Obama can, and should, go further.
Last year alone, there were 92 million decisions to classify information, and top military and intelligence officials agree that between 50 percent and 90 percent of such material could safely be released. Such massive overclassification threatens national security rather than enhancing it. It impedes information-sharing that can be vital to preventing terrorist attacks. It also promotes leaks by undermining officials’ respect for the system. And when too much information is classified, too many people need security clearances to do their jobs.
In December 2009, Obama took a first crack at the problem, directing officials not to classify documents if they had significant doubts. He also specified that no document could be classified indefinitely, and he strengthened classification training requirements. These steps were important but modest. Recognizing this, the president directed his advisory committee on classification policy — the Public Interest Declassification Board — to develop recommendations for a “more fundamental transformation” of the classification system.
The board approached its task with commendable diligence and transparency. It solicited public input, met regularly with outside experts, and shared its thinking at every stage. The results include proposals that could dislodge entrenched biases toward secrecy.
For instance, officials currently face stiff penalties if they fail to protect information that later proves (or is deemed) sensitive. The board recommended a “safe harbor” for officials who make a good faith decision not to classify. It also addressed the poor quality of many agencies’ training programs, recommending that an inter-agency committee develop best practices for training and that agencies conduct units on avoiding overclassification. The board further suggested that agencies create incentives for employees to challenge improper classification decisions. All of these changes would make it easier for officials to classify responsibly.
But until it also becomes harder for officials to classify irresponsibly, overclassification will continue. Too many perverse incentives remain. Dealing in secret information confers power and status, and it is easier for busy officials to classify indiscriminately than to pause for thought. Moreover, classification is a convenient way to hide embarrassing facts or evidence of misconduct. There are no disincentives to balance the scales, as classification decisions normally go unreviewed and agencies do not penalize officials for overclassifying.
It would be simple enough to provide a measure of accountability. Classifiers could be required to provide more information about their decisions, and agencies could conduct “spot audits” of those decisions, with meaningful administrative penalties for officials who flunk the audits. The board’s report, however, includes no proposal to hold officials responsible for overclassifying.
Worse, the report promotes one idea that could dramatically exacerbate overclassification: automated decision-making. Under the board’s vision, computer programs would use previous classification decisions to identify information in a new document that should be classified. If an official disagreed with the recommendation, she would have to refer the issue to specialists for resolution.
The problem is that computer programs would often make faulty recommendations. The task of matching information in one document to differently worded information in another document carries obvious potential for error. Although classifiers in theory could appeal the computer’s recommendation, no official would volunteer for such an inconvenience.
And even if the program always matched information correctly, its recommendations would be based on past decisions, perpetuating the current intolerable level of overclassification.
Translating the board’s varied recommendations into a genuine, positive transformation of the classification system will take three actions on the part of Obama. First, he should embrace those recommendations that tackle the culture of secrecy. Second, he should adopt the board’s idea of an inter-agency committee, and he should charge the committee with developing measures to hold classifiers accountable for overclassification. Third, he should eschew any proposal that would replace, rather than encourage, thoughtful human deliberation.
In all of these matters, the president will no doubt face resistance from military and intelligence agencies that jealously guard their prerogative to operate in secret. The president is understandably reluctant to pick a fight with these agencies. But by stepping back from his pledge to transform the classification system, the president will not avoid a battle; he will merely concede it. The stakes for our national security are too high.
Elizabeth Goitein is co-director of the Liberty and National Security Program at the Brennan Center for Justice at NYU School of Law.