15 May 2005

Avoiding torture

This goes out of this blog's usual mode, but what the heck.

Over at Marginal Revolutions, Tyler Cowen poses this question:
Let us say that you have been captured and threatened with torture.  You are, for whatever reason, entirely willing to betray the information you hold.  Your primary goal is to avoid pain, and perhaps you positively want to squeal.  How should you present what you know?  I see a few options: 1. Break down immediately, beg for mercy, humiliate yourself, and spill the beans.  (If you talk right away, will they torture you anyway?  And since no further good information can be offered why should they stop?) 2. Go in acting tough, really tough.  At the first sign of serious pain, start crying and switch to strategy #1. 3. Wait until they apply their "best shot" torture, and then talk.  They will feel they have done their job and stop. 4. First offer (or make up) compromising information to show your disloyalty to the cause your torturers are fighting.  Your confession will then be more credible. 5. Say you don't know anything, try to fight the torture, but break down when you can't stand it any more.  You can't fool them, so the best you can do is to actually "go through the wringer."  You are stuck in the pooling equilibrium, and trying to deviate only makes you worse off. Which of these is the most credible signal that you have told all you know?  Can you do any better than number five?  And how does your best answer depend upon the hypothesized motives of the torturers?  Is there anything you can say to the U.S. to avoid being sent out for rendition?  I do't see any simple answer here, the question is which behavior your torturers will interpret as an unlikely tactic from a truly determined trickster.

Which he answers:
Spill all the beans as quickly as possible. Here is the perspective of one interrogator. Two key reasons: you don't signal that torture works on you, and false information can be checked against other sources. Here is my earlier post posing the question.


I should say that I can't get myself too worked up over the question. My interest as a moral philosopher pretty much ends with the assumption that you are willing to give up the information.

But, over on another blog, someone with a bit of training in interrogation gives this answer:
Tyler Cowen over at Marginal Revolution asks what the best strategy for avoiding torture is. Assuming that that is your only goal, and that you're completely willing to sell out anything and everything needed to keep from being tortured.

His question assumes without stating directly that the goal of your captors is to extract information. Statistically and historically speaking, that seems like a dubious assumption. Groups that use torture, in general, have a more political objective to the torture than an information gathering one. Torture, in the usual case, is not a device for getting your prisoner to give out information.

However, I believe that Mr Cowen is really asking how to convince a captor that you have indeed given up all the information that you have. That's an interesting question. He offers five suggestions. I will put on my interrogators hat and assess each in turn, from the standpoint of "what would I find convincing". The questions also have the necessary implication that I will, if useful, use torture. The entire exercise is moot for an interrogator that won't use torture in the first place.

By way of disclaimer, my training as an interrogator was minimal. I was a signals intelligence guy. I received cursory cross-training as an interrogator, and would act as a subject for interrogation impact training exercises, and that's about it. Professional interrogators will probably laugh themselves silly at the ideas presented below.

1. Break down immediately, beg for mercy, humiliate yourself, and spill the beans. (If you talk right away, will they torture you anyway? And since no further good information can be offered why should they stop?)

This isn't a bad strategy. At the point my subject is talking, I'm listening and recording. After we're done here, I will need to cross check and confirm the story. But if it all checks out, that's a good sign. I would certainly go in and do a second or third session, asking the same questions, to see if the story changes at all, or to see if anything else comes up.

If you talk right away, will I torture you anyway? Maybe. In this case the point of torturing you wouldn't be to get you to start talking. It would be to break down your psychological defenses and to make it difficult for you to think clearly. The entire story you are telling me may well be a pre-conceived cover story designed to keep me busy. By using some form or torture, I want to make it harder for you to keep your story straight. Physical exhaustion and sleep deprivation would be the way to go here.

2. Go in acting tough, really tough. At the first sign of serious pain, start crying and switch to strategy #1.

This also might work. Frequently the tough demeanor types crack the fastest. However, the sudden conversion would be fishy. If I had a lot of time to work on you, I would consider, after you broke, giving you time to rebuild your defenses and see if the tough demeanor returned. If it did, that would be more convincing. That is, if you are really the tough guy type, that won't go away from one session where you broke, and would return after you had time to put yourself back together. It might go away after sustained torture over a long period, but not from one session where you broke at the first sign of serious pain.

The other problem with this strategy is that it might actually work. That is, if you are convincing, what I have learned is that I can get you to alter your behavior by inflicting serious pain. Even if all I am looking to do is extract information, I will probably torture you to the serious pain level at each session, as a sort of warm up.

3. Wait until they apply their "best shot" torture, and then talk. They will feel they have done their job and stop.

How are you going to know what that is?

This might work as well. If my only goal is to get whatever useful information you have, then once I have broken you, I've got what I want, and you're of no further use. So the torture would stop at that point. The trouble that I see here, is that how are you going to convince me that this isn't strategy two?

4. First offer (or make up) compromising information to show your disloyalty to the cause your torturers are fighting. Your confession will then be more credible.

Making up the information is probably a mistake. If your information is at all important, or I think it is, I am going to be checking out your stories against other sources. When your made up information doesn't check out, I am going to assume that you are hiding something. I am going to use whatever methods I think best to get that something out of you. Also, once you've told me one fish tale, anything else you tell me is all that more suspect. On that basis, I may use torture just to wear you down.

This seems too much like a ploy to me to be really effective. I would think that you're trying to run some sort of game on your end. I should probably just torture you in some fashion until you're too exhausted to keep it going.

5. Say you don't know anything, try to fight the torture, but break down when you can't stand it any more. You can't fool them, so the best you can do is to actually "go through the wringer." You are stuck in the pooling equilibrium, and trying to deviate only makes you worse off.

Ah. The null strategy. Well, it's guaranteed to work, at least in the sense that I'll stop when I have convinced myself that you are broken. The only problem is the initial denial. I am likely to think (potentially, anyway), that by applying just a little more stress, I can find out whatever you have managed to conceal so far. You would be better off just answering the questions to the best of your ability from the beginning. I might torture you for the various reasons above, but when I keep getting the same answers when I ask you when you're fresh, and when I've keep you awake for four days, and when I've applied some sort of pain, and when I've worked you to exhaustion, I'm going to quit soon.

The problem with all of these strategies is that the depend on, firstly the idea that the interrogator is only interested in your information, an unlikely assumption, secondly the idea that convincing your interrogator is just between you and him. However, an interrogator is going to take your stories and compare them against other sources, looking for inconsistencies.

If the first idea is wrong, then being willing to give up your information is no help. If the second idea is wrong, and it will be for any competent interrogator, I think your best bet is to just answer as completely and truthfully as you can whenever you are asked something. That way your stories will always check out, and since it doesn't affect your answers if they torture you or not, it will be more efficient to just ask. They will figure that out quickly enough, though you might have to go through some torture initially as they check you out.