One more note on the importance of counterfactual reasoning in the debate over the effectiveness of torture, which I've recently highlighted in two posts (here and here). In a Fresh Air interview with Terry Gross that I only heard recently, New York Times reporter Scott Shane offered the most detailed exposition of this point that I've seen thus far (my emphasis):
GROSS: And you write that even the most exacting truth commission may have a hard time determining for certain whether brutal interrogations conducted by the CIA helped keep the country safe. Why do you think it will be so hard to determine for sure if these techniques actually resulted in information that helped deter a terrorist plots?
SHANE: Well, I think, if there's strong evidence that valuable information came from this program and ultimately led to the capture of a lot of key al-Qaida leaders and that probably prevented future attacks - but whether these particular methods were necessary to get that information is a very different question. It's, sort of an uncontrolled experiment. They used these methods and they got the information. Many experienced FBI and military interrogators will tell you that they believe you could have gotten the same information, possibly more information, using traditional rapport building methods. They don't think these harsh methods were necessary and they think they risked producing false information as well.
I would like to believe that other journalists have thought through the issue as carefully as Shane, but it's unlikely -- the counterfactual reasoning that is necessary to evaluate causal claims is not well understood even among quantitative social scientists.