In yesterday's NY Times, Ali Soufan, an F.B.I. supervisory special agent from 1997 to 2005 who worked on counterterrorism, wrote a devastating indictment of the failure of torture to collect useful intelligence. To me, the most critical part of the op-ed is about how torture fails to reveal novel information:
There was no actionable intelligence gained from using enhanced interrogation techniques on Abu Zubaydah that wasn't, or couldn't have been, gained from regular tactics. In addition, I saw that using these alternative methods on other terrorists backfired on more than a few occasions -- all of which are still classified. The short sightedness behind the use of these techniques ignored the unreliability of the methods, the nature of the threat, the mentality and modus operandi of the terrorists, and due process.
Defenders of these techniques have claimed that they got Abu Zubaydah to give up information leading to the capture of Ramzi bin al-Shibh, a top aide to Khalid Shaikh Mohammed, and Mr. Padilla. This is false. The information that led to Mr. Shibh's capture came primarily from a different terrorist operative who was interviewed using traditional methods. As for Mr. Padilla, the dates just don't add up: the harsh techniques were approved in the memo of August 2002, Mr. Padilla had been arrested that May.
This confirms what I have argued all along--torture can't reveal novel information because you need other data to confirm the results of torture (and if this corroborating evidence exists, then torture isn't really necessary, is it?):
Figuring out what is reliable intelligence is difficult even with voluntary walk-ins; it will be much harder with someone who is actively trying to hinder you. So how do you figure out whether or not the torture victim is telling the truth? Corroborating evidence.
In other words, evidence acquired through torture will never reveal unique, unknown information. It is only useful when there is other confirmatory evidence. So torture serves no purpose in the Jack Bauer situation, since it, at best, further supports what you already knew or suspected. For torture to have any possibility of being an effective interrogation technique, you have to torture lots of people in the hope that multiple torture victims will confirm each others' information. This means that torture has to become a routine component of interrogation.
Prosecute the torturers.
Related note: Roger Cohen argues that torture prosecutions would divide the country and prevent the reinstitution of checks and balances. There are a lot of people in positions of authority who acquiesced to a torture regime. They should lose their jobs. This isn't disruptive or "retribution", it's restoring a modicum of decency to this country's leadership.
Guess what? The country is all ready divided. They AG should do the right thing and try to the torturers and the administration officials who justified torture, and do it do the fullest extent of the law.
Matt, I agree definitely 100% with your statement "It is nice that ... I do it because I like it...and I am very lucky to have the opportunity to do something I like!"
And yet, I would also agree 100% with the contrasting view "It is unfortunate that few QIS researchers are asking 'Why am I lucky to have the opportunity to do something I like, when so many others are unlucky?'"
There was a time when many information theorists asked (and answered) that class of question ... Norbert Wiener's 1950 treatise The Human Use of Human Beings is an example that is still readable today, fifty-nine years after it was written.
It seems odd (to me) that prepending the word "quantum" to "information science" could ever be regarded as restricting the scope of information science.
The situation (to me) seems more like prepending the word "algebraic" to "geometry" ... the historical result was an immensely expanded scope for both algebra *and* geometry.
It is nice that "I like quantum theory" is doing well, as is the similar answer about "intellectual merit". Both of these answers boil down to "I do it because I like it...and I am very lucky to have the opportunity to do something I like!"
I will second Ian in asking for: (X) in service of other/external goal(s).
The point is that math and science are (often) regarded as goals in themselves ... certainly they are commonly taught that way. Engineers (in contrast) often explicitly embrace external goals; this tradition is much closer to my personal point of view regarding QIS/QIT.
In my case, these external-to-QIS drivers are: (1) regenerative medical capability, (2) universal access to molecular-scale resources in biology and materials science that is (a) comprehensive and (b) free-as-in-freedom, and (3) the acceleration of global-scale enterprise in service of job creation.