Jack Goldsmith's excellent New Republic article on executive branch secrecy during the War on Terror raises a difficult conundrum: neither the press nor the executive branch can be trusted to make unilateral decisions about secrecy. As Goldsmith points out, the press cannot always be trusted to decide for itself which classified information it might get its hands on to report and which not to. The New York Times and other media outlets claim to "balance" national security interests against the public's right to know when they decide what to publish. But, as Goldsmith notes, they have strong incentives to err on the side of revealing too much:

[New York Times reporter Eric] Lichtblau assures us that it does, noting that the Times editors serve as a "built-in backstop, a check and balance" on reporters, and adding that for "the editors to even consider running a piece, we knew that there had to be a legitimate public interest that outweighed any potential harm to national security."

But . . . there are many reasons to be skeptical. Saying the editors are a check on what the Times publishes is like saying we can trust the president to curb the excesses of his subordinates. It affords little comfort, especially since the public has no access to the process of the editors' decision-making. One wants to know precisely how the editors weigh the public interest in knowing against national security harm. Even if the editors possess the expertise to identify and to assess these trade-offs (something that is doubtful), is their judgment distorted by the pursuit of fame and profit? The separation of powers, the institution of elections, and the free press help to ensure that government's self-directed motives do not get out of hand. But there are few checks on the press itself. The most powerful constraint on the press is the marketplace of ideas--but this marketplace is designed to sort out truth, and it is no corrective when journalists irresponsibly disclose sensitive national security information.

As Goldsmith notes, reporters and editors stand to gain "fame and profit" if they reveal a major "scoop." By contrast, they get little if any benefit from refusing to publish classified information that may benefit the enemy. Goldsmith suggests that the Times crossed this line in some of their reporting on the Administration's surveillance program, reporting which he claims ended up benefiting Al Qaeda by giving them valuable information.

Obviously, as Goldsmith recognizes, there is also a flipside to this. If the executive branch can decide unilaterally what information to reveal, they can use the pretext of national security to cover up human rights abuses, spying on their political opponents, and other misconduct. As Goldsmith explains, "press scrutiny of secret government activity is important to keeping government accountable. Fear of leaks causes national security officials to think twice about what they do, and deters them from doing things that they should not do."

Thus, the executive branch cannot be trusted to make unilateral decisions in this area; but neither can the press. The former is likely to keep too many secrets, the latter too few. There is no easy solution to this dilemma.

One possible approach is to recognize that we need an arbiter for these issues that is as neutral as possible in its incentives, free of both the press' incentive to overreveal and the executive's incentive to engage in excessive secrecy. Though it has flaws of its own, the judicial branch does have the advantage of lacking either of these perverse incentives. So we may want to allow judicial review of classification decisions, perhaps similar to the FISA system of judicial preclearance for warrantless surveillance. If the courts rule that a classification decision was unjustified, reporters could be shielded from prosecution; if they rule in favor of the government, reporters who go ahead and publish nonetheless should perhaps face more severe sanctions than under traditional classification laws.

Alternatively, we can try affect reporters' and officials' incentives through after the fact sanctions rather than by trying to fine-tune the classification system. As Goldsmith points out, it is often difficult or impossible to punish reporters who reveal classified information, even in cases where there has been real damage to national security. As standard law and economics of crime suggests, a difficulty in ensuring certainty of punishment might be partially obviated by increasing its severity.

By the same token, it is often also difficult to punish executive branch officials who use secrecy as a tool for covering up crimes and violations of civil liberties. Here too, we might want to consider increasing the severity of punishment for offenders, so as to at least partially offset the lack of certainty.

None of these proposals can fully "solve" the problem and it may well be that there are other, superior alternatives. The beginning of wisdom, however, is to at least recognize that we have a double-edged dilemma here.