I’m skeptical of reports that “enhanced interrogation techniques” — referred to by some as torture, minimized by others as frat-boy hijinks — yielded uniquely useful intelligence. By that, I mean that I’m skeptical that the techniques yielded information that had not been, or could not be, obtained by other methods or through other sources.

That skepticism involves a healthy dose of mistrust for the intelligence apparatus of the United States as led and organized for the last decade. But it is also based on my experience with the presentation and manipulation of intelligence and the institutional and cultural factors that guide it.

No, I’ve never been a CIA agent. I’ve never been an “intelligence analyst” in the national security sense — though in the late 1990s I did obtain a security clearance necessary to view classified data in a terrorism prosecution. (The most highly classified data I viewed established that a target I will call Billy Joe Bob was actually named Joe Billy Bob. Look out, Jack Ryan!) But in the course of about six years as a federal prosecutor, I did obtain, review, and evaluate much information provided by cooperating defendants and witnesses. Flippers and snitches, in other words. I evaluated the intelligence such people provided for two purposes. The first purpose was to determine whether to call them as witnesses in cases I was prosecuting, whether to conduct follow-up investigations and searches and interrogations based on the intelligence they provided, and generally how to shape operational strategy based on what they said. The second purpose was an entirely different one — it was how to portray their cooperation to my superiors within the U.S. Attorney’s office, to the snitches’ attorneys, and to the court for the purpose of advocating how much sentencing credit they should get for their cooperation. The lesson I learned? The first sort of assessment is vastly more reliable than the second.

In federal prosecutions, at least for the last two decades, defendants have gamed the arcane and mechanistic U.S. Sentencing Guidelines through cooperation to get the best sentence possible. A prosecutor can reward a cooperating defendant in two substantial ways: by empowering the court to drop below the harsh mandatory-minimum sentences created by the federal drug trafficking statutes (which otherwise would bind the court, unless the defendant had no substantial record and met the exacting standards of the so-called “safety valve”), and by asking the court to depart downward from the established guideline range under Section 5K1.1 of the sentencing guidelines. Even now — after the Supreme Court ruled that the Sentencing Guidelines must be treated as hortatory, not mandatory — judges are substantially influenced by a prosecutor’s 5K1.1 departure recommendation. As a prosecutor, I made recommendations that took eight years off of a sentence; as a defense lawyer, I’ve represented clients whose cooperation made the difference between a 20-year mandatory minimum and a six-year term.

Assistant U.S. Attorneys generally need approval from a supervisor to make a departure recommendation for a cooperator. Before I could ask a court to give a cooperator sentencing credit, I had to write a memo to my direct supervisor and to the Chief of the Criminal Division of my office. That’s the process in which I learned the difference between evaluating intelligence for operational purposes and for purposes of reporting to, or convincing, someone else. within a bureaucracy. As a line prosecutor asking for approval to make the argument I wanted to make to the court, I was acting ostensibly as an analyst, but practically as an advocate. I knew the facts of the entire investigation better than my direct supervisor and vastly better than the Chief of Criminal, who might not know about the case at all. There is effectively no chance that either my direct supervisor of the Chief of Criminal could find the time to immerse themselves in the facts to the extent I did. My conclusions went largely unchecked — that the intelligence I obtained from the defendant was true, that it was substantial, that it was useful, that it helped me induce other defendants to plead guilty. Similarly, my factual predicates and the level of confidence with which I made them could not easily be verified by my supervisors — whether the intelligence came from the defendant first, whether the defendant simply confirmed previously known intelligence, whether the defendant gave up information easily or had to be coaxed for a long time before ratting out friends, whether all of the information provided could be verified or proved true, etc. Did I lie? No. But I’m sure that different experienced prosecutors could have viewed the same cooperation different ways, and could have written the memoranda different ways. The case agents — the people with the most intimate knowledge of the nitty-gritty of the cases, and of interrogation techniques — may have viewed it another way still.

Institutional and cultural factors influenced me to write memoranda in a way that supported the conclusion I wanted my supervisors to reach: I wanted defense lawyers to continue to bring their clients in for cooperation and therefore wanted to reward defendants in a manner I thought to be adequate, I wanted supervisors to agree with my conclusions and therefore conclude that I was a reliable analyst and prosecutor, and I wanted to close cases. All of these factors had an impact on my description of the value of intelligence.

The same factors controlled my description of the value of cooperation to the court. I wanted the court to give a defendant enough credit — so that other defendants would see the value of cooperation — but not too much credit — to avoid being seen by supervisors as weak or unreliable. Therefore I described the value of defendants’ cooperation as an advocate, sometimes downplaying it when defense attorneys sought to puff it beyond its true value. Since then, as a defense attorney, I’ve done the opposite — I’ve tried to spin my clients’ cooperation as case-changing, ground-breaking, and crucial. Sometimes I’ve convinced courts to give clients more credit than the government wanted them to get. Like my supervisors, the judges only got what information about the case that we gave them, and their conclusions were necessarily shaped by our advocacy.

All of this led me to place reasonable faith in operational assessments of the value of intelligence (conducted by people actually making decisions about how to use the intelligence, and facing the consequences if it proves unreliable), but to view bureaucratic and advocacy-style intelligence evaluations with substantial skepticism. The value of intelligence is stunningly easy to spin, massage, parse, and manipulate in a way that is nearly impossible to falsify by anyone who is not immersed in the day-to-day operations of the investigation in question. Moreover, institutional and cultural factors strongly promote such spinning, massaging, and parsing. Nobody — and especially nobody in government — wants to be seen as doing things that are useless, counterproductive, a waste of resources, or needlessly cruel. Intelligence analysts and their supervisors and advocates face enormous institutional and cultural pressures to present their efforts as bearing fruit. Moreover, they experience the same human inclinations as anyone else to try to convince their superiors that their conclusions are reliable and their analysis accurate.

Hence, when I see reports that the CIA believes that “enhanced interrogation techniques” yielded actionable intelligence, I am skeptical in a way that is not merely rooted in mistrust of the American intelligence apparatus. Conclusory statements that torture yielded uniquely useful information — and even factual claims that the information was new and could not have been obtained from another source — do not convince me. This is not to say that the CIA is necessarily lying or even necessarily advocating an incorrect conclusion. But the conclusions should be viewed with a healthy skepticism.

However, as Patterico suggests in the course of accepting claims that enhanced interrogation yielded useful results , the moral and policy question of whether torture is justified if it yields actionable intelligence is separate from the factual question of whether it works.

Last 5 posts by Ken

• Towards A Comprehensive List of Things Obama Is Pushing Us To - July 29th, 2010
• And Don't Get Him Started On Charlie the Tuna - July 27th, 2010
• Jeffrey Lord Is Not A Writer - July 27th, 2010
• If The Government Says That It's Not About Freedom, Then It's Just NOT - July 26th, 2010
• Will Refrain From Snark For Food - July 26th, 2010