If the purpose of detainment and questioning is to humiliate, demoralize, and terrorize a group of people, then torture is a very effective tool.
Torture may provide useful information, but there's no way of distinguishing between authentic confessions and false "make it stop" confessions. The only way to assess the merits of information gleaned via torture is by testing it against information acquired via old-fashioned intelligence gathering, which defeats the point of torture since it adds little or no value to the stock of intelligence.
In the run-up to the war in Iraq, for example, I was able to establish by some time in 2002 that Iraq was effectively disarmed of its WMD. I managed this despite having no secret sources of information, no global network of agents, and a research budget of approximately $0.
I used the radical method of reading published reports by internationally recognized experts on WMD - the International Atomic Energy Agency, the UN Special Commission (UNSCOM) weapons inspectors, and the two UN coordinators for Iraq (Hans von Sponeck and Dennis Halliday), both of whom resigned in disgust when the Security Council refused to lift the devastating sanctions that killed half a million children over the 1990s.
The experts agreed that aside from a few unaccounted dregs that would not in themselves constitute a real threat, Iraq was widely considered to be disarmed by 1998.
The US government, determined to convince the US public that it was justified invading Iraq, disregarded these exhaustively researched, extensively documented, publicly available reports and listened instead to bogus intelligence provided by interested parties like Ahmed Chalabi, who later turned out to be a double agent for Iran, and to the testimonies of "enemy combatants" who had been subjected to interrogation methods that used to be called "torture" when the Soviet Union used them.
Torture provides the answers you want to hear. It consists of inflicting extremes of discomfort, pain, and distress on detainees and then drilling them with leading questions until they agree - anything, anything, just let me sleep/sit down/eat/warm up/stop degrading myself in front of this female guard!
When lawyers ask leading questions during court cases - for example, "Did you see the red car?" as opposed to, "Did you see a red car?", their opponents can justly object, and decent judges sustain the objection. This is because they recognize that people are remarkably likely to tell you what you want to hear, even if they aren't being tortured.
So-called "push-polls", administered by partisans for the purpose of influencing public opinion, accomplish the same thing. If it's unethical and misleading to do this, how much worse is it to torture people first and then ask leading questions?
If the purpose of detainment and questioning is to obtain information, then torture is a poor tool because it does not provide reliable, verifiable information.
However, if the purpose of detainment and questioning is to humiliate, demoralize, and terrorize a group of people, then torture is a very effective tool. The military wunderkinder behind the "War on Terror" understand this principle well, and apply it with remarkable sophistication.
So yes, torture does work, as long as you understand its purpose.
January 9, 2006