I'm just about finished with Tim Weiner's phenomenal "Legacy of Ashes: The History of the CIA." Two themes are at the heart of the book.
First, the Agency has been incompetent from its inception. The roster of incompetence includes subversion operations that cost the lives of hundreds of agents and accomplished nothing; CIA-managed coups that backfired, the Bay of Pigs and many others. Even operations that "succeeded" were pyrrhic. Installing the Shah via a CIA-sponsored coup in Iran in 1953, for example, created enmity that resulted in the Khomeini revolution and hostage crisis of 1979 and continues to this day.
Second, the Agency and its political masters have consistently lied to the American public about CIA domestic lawbreaking. Anyone horrified at the notion that the modern CIA kidnaps and tortures terror suspects at secret prisons should understand that these activities aren't aberrant, but are in fact the legacy of programs like Project Artichoke and Project MKULTRA, in which the Agency built secret prisons in Germany, Japan and the Panama Canal Zone - prisons where suspected double agents were tortured and dosed with heroin, amphetamines, sleeping pills and LSD. And, like the interrogation videotapes the CIA now claims it destroyed in 2005, the CIA also destroyed its records of these earlier illegal activities.
It's tempting to conclude from all this that the CIA should never have been in the operations business - after all, incompetence measured against subversion of the Constitution seems a bad bargain. But it's hard to see what CIA analysis has accomplished, either. Mostly the analysts have been disastrously wrong (on the Soviet invasion of Afghanistan, for example, the Agency continued to insist even after Russian tanks crossed the border that it couldn't be a full-scale invasion), but even when the Agency has been right, it hasn't made a difference. When policy makers agree with CIA conclusions, they use those conclusions to justify what they were going to do anyway. When policy makers disagree with those conclusions, they simply ignore them. Either way, the conclusions become irrelevant. You can have the best information and analysis in the world, but if it has no impact on policy, it's still a waste of resources.
Counterproductive operations, activities that subverted the rule of law, irrelevant analysis ... it's hard to read "Legacy of Ashes" and conclude other than that America would be better off today if the CIA had never existed.
Of course, no politician will ever abolish the Agency. The CIA is too useful a tool for demonstrating to the public that a politician is doing something about a problem, and an iron law of American politics (perhaps all politics) is that a politician can never say, "We're doing as much as can reasonably be done about this problem, and attacking it further would only make things worse." Also, the CIA is too easy to ignore when ignoring it is convenient, too easy to manipulate when CIA support is useful, and too easy to blame when something goes wrong (say, a wrong-headed and unjustified war).
So what can be done? The solution, I think, lies in a critique of Weiner's book by Nicholas Dujmovic, available on the CIA's web site. Dujmovic writes:
"The intelligence services that are often judged to be superior to CIA - the Israeli Mossad, the Cuban DGI, the East German Stasi, and even the British SIS - are far more limited in focus and scope. CIA from the beginning was charged with worldwide coverage in all intelligence areas, something no other service, except perhaps the Soviet KGB, was required to do. If making no mistakes is Weiner's only standard, he has adopted an unrealistic one - a Platonic ideal for intelligence - that CIA, dealing with the world as it is, could only have failed to meet."
That last line is just a straw man: Weiner doesn't require that the CIA make no mistakes. No reasonable person would. But if Dujmovic's point is that the CIA is too diffuse to be effective, why not focus its mission? Eliminate its operations arm, which has consistently done more harm than good. As for analysis, do politicians really need secret information to formulate sensible policy toward, say, China? And even if they did, history suggests they wouldn't use it except to justify what they were going to do anyway. So, eliminate operations and ruthlessly focus on questions that only good intelligence can answer: the whereabouts of Pakistani nukes, for example, or the nature of terrorist financial networks, or how close Iran is to acquiring nuclear weapons. Resources are always finite, and an organization that's focused in part on China will inevitably be less focused on Pakistan - and will probably perform poorly on both.