Information gained by torture in my opinion has to be evaluated in the same way that any information gained by any means has to be evaluated. Now the question is, why is it considered unreliable? Because the person can lie to get out of torture? Lying takes place outside of torture also, that is no reason not to use torture. If a person knows something, torture is the most effective method of getting him to reveal it in the same way that the most effective way to prevent crime is by punishing those who commit the crime. I do not see how it is anymore unreliable that information gained by other means. Machiavelli says "...a ruler should be slow to believe what he hears...", this is not an injuction against torture, simply against believing everything you hear.
It is considered unreliable because, perhaps counterintuitively, those being tortured tend to lie MORE than those who are not. Even in closely monitored police interrogations, there is a risk of false confession simply because the human being interrogated wants the process to end. When torture is added to the mix, the drive to end the process is stronger. And if the interrogator/torturer is a poor evaluator of truth, the interrogated is merely going to be looking to provide an answer that the interrogator wants to hear.
And that's just with the innocent.
If you have a genuine malfeasor, his drive to protect his mission will be even stronger. He will provide misinformation up until his success is assured.
In addition, if your regime is known to torture, you're less likely to be able to take prisoners alive TO torture.
There have been numerous interviews & articles on NPR about this, such as this one:
Officer 'Unpopular' For Opposing Interrogations
In another one from around 2006, an interrogator stated that in his 20 years of working in intelligence, he had
never found information gathered by torture (from non-USA sources) to prove true, nor had he ever encountered the "ticking time bomb" crisis so popular in fiction.
In contrast, methods in which interrogative forged bonds of understanding with their subjects generally reaped positive results: IOW, they got good info.
Now, is torture 100% ineffective? I doubt it. But judging from what he pros say, the odds of success seem alarmingly tiny, and building a case for it as a useful tool seems as difficult as asserting that plutonium-laced sarin gas is perfect for home defense...