Information theory, human psychology and torture

Lets start with the now famous thought experiment that some claim "justifies" torture in at least some cases:

A captured terrorist has placed a nuclear time bomb in a populated area. If you torture him he will give up the location allowing you to save the lives of millions of people. Isn't it ethical in this case to torture him?  

The problem with this argument is information theory which undermines it's basic premise:  how did you know the terrorist has placed a nuclear bomb in a populated area in the first place and how do you know who he is and where to find him. Knowing all that means you are well on your way to finding the bomb yourself without any additional direction from him. Is there any evidence that torturing him will result in the extraction of a correct answer vs. a deliberately wrong or deceptive answer that wastes your time when you are well along the way to finding the bomb in the first place? The answer from the experts is no: a tortured person will say say something, but whether or not it's correct is highly speculative.

The psychology of someone who would put a nuclear bomb in a populated area would have to be pretty hard core. I do not think it would be prudent to base any short term actions on *anything* he says - under duress or not. Such a person would probably say nothing or give you deliberately false information.  Maybe if you had the luxury of time so that you could give him successively worse rounds of torture in response to him giving you incorrect information you might have a slightly better chance of (eventually) finding the bomb via torturing him. Except that as time moves on his information becomes stale (cohorts would probably move the bomb knowing one of them has been compromised) and in this thought experiment we do NOT have the luxury of time, if the bomb is not found right away it explodes. Not only is torture ineffective in this circumstance it's imprudent. It's imprudent to base anything on what the terrorist says. It's better to use the same information that you used to find him to find the bomb.

Here is another thought experiment -one to be presented to (typically conservative) people who throw out the thought experiment above: lets assume that tripling the spending on food stamps in year X will cause an increase in work and tax revenue from recipients in subsequent years that more than makes up for the spending on the food stamps in year X. Wouldn't it make sense in this case to triple food stamp spending? The answer would inevitably be "I don't agree with the premise that giving people more food stamps will increase their work ethic."

Likewise the premise that a terrorist would give the correct locaction of the bomb under torture is a false premise.