Two people are alleged to commit strikingly similar offenses. One, Wen Ho Lee, spends a year in prison only to have an embarrassed government agree to a plea bargain. Another, John Deutsch, hasn't been charged with anything but watches as the investigation against him expands, and expands. Both cases perfectly illustrate one of the American intelligence community's greatest problems: the cult of classification, in which information both rare and commonplace is safeguarded with equal zeal. Both cases also illustrate the intense political pressures on intelligence and counterintelligence agencies, polluting the value of the nation's intelligence. These are the parts of our system that are broken, the ones that no one in Washington wants to talk about. The Lee case is stunning in many respects. But no more so than for this: Many people regularly mishandle classified information, befuddled at the complexity of regulations over what is classified and what is not. Most of these cases never become legal matters at all; they are handled administratively or simply ignored. Lee was probably singled out because U.S. intelligence discovered in China nuclear data similar to what Lee worked with at the laboratory. His plea bargain may mean that in learning the whereabouts of his missing tapes, the government may get what it wanted all along and uncover an espionage ring. But a second scenario now seems more plausible: Under intense political pressure from Washington, U.S. counterintelligence grabbed the wrong guy. Either someone else passed sensitive information, similar to Lee's, to the Chinese government, or there was no spy network. The information itself was already so widely disseminated that the Chinese government could readily find it. Throughout the Lee case there was a recurrent theme: At least some of the material he was accused of stealing had already been published. This raises a series of fascinating questions. If something has already been published, does mishandling it or stealing it really constitute a crime? Did the Chinese government really have to penetrate Los Alamos National Laboratory to steal nuclear secrets — or could an efficient open-source intelligence operation have yielded what was needed? For the U.S. government, the question is more profound. Does anyone really know anymore which of the millions of bits of classified information are already in publicly accessible databases, books, articles and, of course, the Internet? The intelligence community underestimates the massive amounts of information available in the open source. In 1995, for instance, the Central Intelligence Agency held a competition to see who could gather the most information, most quickly, on Burundi. The winner was a Washington company, Open Source Solutions, which left the CIA team in the dust. In 24 hours, OSS compiled huge amounts of information, ranging from statistics to scholars; the CIA team finished dead last, compiling little more than their own World Factbook. There is certainly vital information that must be protected from foreign espionage. These secrets worth saving should be held closely indeed. Far too much effort is being wasted protecting non-secrets, which allows vital secrets to slip through. In Washington, classification has led to a sort of game, creating those in the know and those not in the know. This game heightens the power of bureaucrats. But so much is classified, that it is impossible for people with security clearances to know what is derived from a spy satellite and what is plucked out of a newspaper. But social status derived from clearances is even more insidious. A delightful Washington game is sitting around at lunch chatting with various officials about Paraguay or Cambodia. The conversation turns to details when suddenly, the person you are talking to gets a faraway look in his eye and says, "Sorry, I can't talk about that." Conversation over, and he's the winner. You see, he knows things that he can't tell you. He's wired. You're not. The delightful pause, indicating that he is sifting through his vast store of classified information, trying to determine the source of the particular nugget he can't impart, and the reluctance with which he refuses to go on, is part of the pure joy of holding a clearance. It is not what you know, but what you can't talk about that makes you cool. Compounding the problem is political pressure. If nearly everything is a state secret, then what to do about and with those secrets is of paramount political importance. The Lee case is increasingly turning out to have been deeply impacted by Washington's political mood, the search for a China conspiracy in Congress and the hyper-defensive political calculations of the Clinton administration. Lee may not just have been the wrong guy; he was the wrong guy at the wrong place, at the wrong time. On a day-to-day basis, members of the intelligence and counterintelligence communities are bombarded with highly political demands. Indeed, the politics and policy demands of the administration and Congress are never far from the minds of investigators and analysts. In this environment, the power and value of intelligence itself becomes twisted and biased. Ultimately the utility of the country's base of national security knowledge is compromised. Consider the case of Deutsch, a former top official of both the CIA and the Defense Department. As a high official he has had every right to read classified information in his office, but not at home. There is no indication that he ever intended to sell or pass the information, just that he mishandled it, much like Wen Ho Lee. But here's where the politics comes into play: If they're going to hammer Lee, then they're going to hammer Deutsch. It would look too bad not to do so. So, what is a secret? Nuclear secrets should be kept secret. The names of U.S. agents in other countries must be kept secret. Operational capabilities of U.S. weapons should be kept secret. Unlike today's situation, a secret requires that there be not the slightest hint that the secret even exists. To do that, the government would need to follow just a few simple rules, instead of the myriad complexities it has erected. First, there must be few secrets; unless you are willing to stash people at Area 51, it's easiest to keep a small number of secrets. Second, give secrets to fewer people. The idea of hundreds of thousands of people wandering around with secrets is absurd. Do not use access to classified materials as the justification for doing background checks on military officers. Just do background checks. Don't classify as secret that which is in the The New York Times and on the Internet. Don't use secrecy as a shield, to protect idiotic political and policy decisions. The Lee and Deutsch cases turn out to be identical in the sense that, in the end, the most either man can be charged with is a procedural violation. Both were careless at most. And both incidents call Napoleon to mind: He who defends everything defends nothing. Anyone trying to protect all of the government's classified information protects nothing. Washington has succeeded in making the vital secrets of the republic indistinguishable from banal drivel. The reason to de-classify is not to make civil libertarians happy. It's to stop cases like Lee's and Deutsch's from proliferating, leaving everyone puzzled as to whether a real secret has been compromised. And while we obsess over these, the true secrets will fly out the door to the four corners of the earth.