Failure to prevent the Sept. 11 attacks or find weapons of mass destruction in Iraq have put intelligence issues in an unusual position - at the center of a closely contested U.S. presidential campaign.
All the attention creates both opportunity and danger. If there were ever a moment when public demand might overcome the entrenched institutional interests that block major reforms, this should be it. But there is a huge gap between the urge to "do something" and figuring out just what that something should be.
Public dismay has encouraged the search for a technical fix - radical reorganization of the huge intelligence bureaucracy, or removal of intelligence leadership from high-level politics. Such drastic changes could create as many problems as they solve.
The most commonly heard proposal for major structural change is to replace the director of central intelligence, who has coordinated the intelligence community since 1947, with a new and stronger director of national intelligence.
This new czar would give up the old post's second hat as CIA director to focus more intently on managing the entire intelligence community - the 15 agencies that lie mostly in departments such as State, Treasury, Energy, and especially Defense.
The director of national intelligence would also get more direct authority over these various intelligence units, a move that would radically reduce the role of the Pentagon, where about 80 percent of intelligence activities and budgets now lie.
This sounds good in theory, but it would be costly in practice. Trying to wrest units like the National Security Agency from the Pentagon would leave Capitol Hill and Pennsylvania Avenue awash in blood.
The military would never accept dependence on other departments for performance of its core functions, which include tactical intelligence collection, and politicians will hesitate to override military protests that their combat effectiveness is being put at risk. The likely compromise would split up the agencies in question, giving parts of them to the director of national intelligence, leaving others to the Defense Department.
Then the Pentagon would almost certainly regenerate the lost units, producing expensive redundancy. Or, it could prove politically easier for Congress to take the CIA away from the director of central intelligence - stripping him of his main institutional support - than to give him effective control of other agencies. Such truncated reform would be a step backward: a director of national intelligence even weaker than the current director of central intelligence.
Should intelligence be insulated from politics by giving the director a fixed term of office, like the head of the FBI? Again this sounds good in theory, but the cost could be less effective integration of intelligence in policy. In reality, high-level policy-makers misuse intelligence less often than they ignore it. All the objective information in the world will not matter if the president and his inner circle do not pay attention to it, and they will not do so if they do not interact frequently and have rapport with the leaders of the intelligence community. And if resistant bureaucracies are to be shaken up and made more responsive to new priorities, they must be strong-armed by politically savvy managers with the trust of the president backing them up.
Granted, politicization is a risk. An effective director of central intelligence has to walk a tightrope between professional responsibility and political effectiveness. Since the revelation of mistakes over Iraq's illicit weapons, a common criticism of George J. Tenet, the director of central intelligence, has been that he became too close to the administration's inner circle, "one of the boys."
Yet even if the criticism is valid, it is not clear how much worse that problem is than its opposite. Did the intelligence-policy connection work better, for example, when Bill Clinton's first director of central intelligence, James R. Woolsey, could not even get an appointment with the commander in chief?
There is a solution to this dilemma in theory, although not in legislation: select a president and intelligence director with particular personality traits. The best chief of intelligence is one who has the personal confidence of the president, but who delights in telling the inner circle what it does not want to hear.
This relationship can be sustained, of course, only if the president likes to have his thinking challenged and his job complicated - something more common among intellectuals than politicians.
Specific improvements in collection, coordination and analysis of intelligence are certainly practical, and some are already under way. But at the end of the day, the strongest defense against intelligence mistakes will come less from any structural or procedural tweak than from the good sense, good character and good mental habits of senior officials.
How to assure a steady supply of those, unfortunately, has never been clear.
Richard K. Betts is director of the Saltzman Institute of War and Peace Studies at Columbia University. He previously served on the staff of the Senate's Church Committee investigation of U.S. intelligence agencies and as a consultant in the intelligence community.