No-win

The Fields:

By now we've all seen the Air Marshal story out of Miami International. There are, of course, recriminations. For example, "Federal Airport Nazis Execute Unarmed Citizen." Or these comments recorded at Mark in Mexico's blog.

Yet the Air Marshals Service says, "This was a textbook scenario and they acted instinctively* based on the training." It's worth taking a look at why the training runs this way.

Essentially, this is an equation with two variables. Variable X is, "Is this guy really an armed terrorist, or not?" Variable Y is, "Do the Marshals shoot him, or not?"

If X=1 (he's a real terrorist) and Y=1 (you shoot him), nothing bad happens.

If X=0 (not a terrorist) and Y=1 (you shoot him), your career is in question and you might go to jail, based on how the inquiry into your actions pans out.

If X=1 and Y=0, the terrorist is free to act and will carry out whatever plan he has come to execute.

Finally, if X=0 and Y=0, you will answer questions about your reasoning before you're clear to return to duty -- but since you were right, nothing bad happens.

Obviously, the training has to focus on trying to get the X and Y values to match whenever possible. However, perfect knowledge is not possible, and mistakes will occur. Therefore, when choosing how to train Air Marshals, you have to decide if you will preference X=0 Y=1 situations (i.e., non-terrorists getting shot) or X=1 Y=0 situations (i.e., terrorists being free to act).

As a point of pure logic, harm is minimized by training to settle hard cases in a way that preferences X=0 Y=1 over X=1 Y=0. The harm to be done in (X=0 Y=1) is limited to two people: the non-terrorist getting shot, and the Air Marshal whose career and liberty are called into question. The harm to be done in (X=1 Y=0) is unknowable, but potentially quite high. Air Marshal training expects an entire jet full of people to be at the mercy of a terrorist, after all, as that is the normal situation in which they are likely to encounter terrorists. You could lose dozens of lives, or more yet if the terrorists should gain control of a plane 9/11 style.

For that reason, the training is appropriate. Furthermore, by exactly the same logic, the inquiry should accept cases such as this one as "textbook" and "justified." This is because the other Air Marshals will watch the outcome of the inquiry, and modify their actions accordingly. Since it is logical and necessary to prefer X=0 Y=1 cases over X=1 Y=0 cases, you want to conduct the after action in a way that will continue to maintain that preference. Unless there is clear evidence that the Air Marshal acted recklessly, the Service is acting in a logical fashion if it tends to back his actions.

What I find interesting is that, in addition to a field of practical results, there is also a field of political results. The Air Marshals Service can't afford to consider these in their training, but it happens that they encourage the same preference in training:

X=1 Y=1: No politician will condemn the Service; only fringe speakers in the media/blogspace will do so.

X=1 Y=0: Almost every politician will condemn the Service; almost every speaker will do so.

X=0 Y=1: If the Air Marshal was clearly acting recklessly, there will be stiff condemnations; however, the Service can derail most of these with an inquiry and punishment for the individual Marshal. Thus, harm is localized. In situations where the Air Marshal was not plainly reckless, most politicians will avoid comment, and speakers will tend to favor the shooting out of an understanding of the practical considerations (i.e., the first set of possible results, above).

X=0 Y=0: Since we are talking about cases where someone appeared to possibly be a terrorist, but turned out not to be, the reasoning of the Air Marshal will be called into question widely. People will ask "What if?" questions that challenge the training of the Service. Certain ultra-partisan politicians (e.g., Ted Kennedy) will state that this proves that the Service, and indeed the Administration, does not take terrorism seriously.

A rational reading of this field is this: Where Y=1, the result will either be positive entirely, positive on balance, or bad only in a way that is easily compartmentalized by punishing the individual. Where Y=0, the results will either be entirely and bitterly negative, or negative in a more balanced way that nevertheless still calls into question the usefulness and dedication of the entire Service.

Thus, as both a practical and a political matter, the training ought to be what it is. It is only logical.

* "acted instinctively" -- Not precisely, but the confusion of terms is telling. Instinct properly refers to biological responses, not trained responses. These are what Aristotle called "First Nature." However, as Aristotle himself noted, with proper and intensive training you can create a response that, while learned, feels exactly like instinct. This is your "Second Nature." This is why we say that such-and-such was "second nature to him," when we mean that a thing was so ingrained in a man's thinking and habits that it had simply become a part of his character.

Second Nature will develop, by the way, whether you train with it in mind or not. If you aren't actively and rationally thinking about what your Second Nature ought to be, and training yourself mindfully, it will become whatever your habits are. This is why Aristotle spent so much time on the subject of considering what the proper Second Nature was. He felt that it should be carefully considered using the rational part of the soul, and then carefully put into practice until it was fully adopted. Obviously, the Air Marshals Service has done its logical reasoning here, and likewise done its training. On those grounds, the Service deserves praise.