The New Yorker's Jane Mayer has a fascinating article on the CIA's "black" prisons, focusing on the case of Khalid Sheikh Mohammed, the man who confessed to beheading Wall Street Journal reporter Daniel Pearl in Pakistan in 2002 and involvement in at least 30 other terrorist plots.

Or did he? How reliable is the information he provided under what appears to be the conditions of torture? An excerpt:

A complete picture of Mohammed’s time in secret detention remains elusive. But a partial narrative has emerged through interviews with European and American sources in intelligence, government, and legal circles, as well as with former detainees who have been released from C.I.A. custody. People familiar with Mohammed’s allegations about his interrogation, and interrogations of other high-value detainees, describe the accounts as remarkably consistent.

Soon after Mohammed’s arrest, sources say, his American captors told him, “We’re not going to kill you. But we’re going to take you to the very brink of your death and back.”

Here's what Mayer had to say to Democracy Now!

One of the findings that, in this story, that really stunned me was that a top CIA official, who I can’t name, but somebody who really knows a lot about this program, said to me that 90% of what they got from every kind of technique they used was bogus. So 10% of what they got was accurate. And they are arguing that that 10% certainly made it worthwhile, and they think it saved people's lives.

But I think the question, finally, that I have and that I think that Philip Zelikow asks in this story, who was the legal counselor to Condi Rice, I think is the question that the country should be asking maybe, is not “Do these techniques work?” but “Are these the only techniques that work?” And the answer, if you talk to the military and you talk to the FBI, is that there are many other ways to get more reliable information. So we may not need to go to these lengths. And I think it's certainly something that I’d like to see some public debate on.