why not assassinate Hitler?

Reading a comment from a reader of Krugman's blog about the recent death of Paul Samuelson made me rethink my attachment to utilitarianism. You know the train scenarios for utilitarianism?

Scenario 1: You're standing next to a fork in the traintracks, with access to the lever that changes the direction the train will take. The train can not be stopped, and if left as it is, it will run over and kill 5 people standing at the end of the tracks. If, however , you pull the lever the train will change directions, killing the 1 person standing at the end of that fork. What do you do?

Scenario 2: The situation with the train is the same, however this time you're standing on a bridge over the tracks. There's only one direction the train can go, and at the end there are 5 people that will be killed. You know that the only way to stop the train is by throwing some sort of heavy object in it's path. The only heavy object at your disposal at the moment happens to be the very fat man standing next to you. What do you do?

Most people pull the lever in scenario 1, but refuse to kill the fat man in scenario 2. But the consequences of pulling the lever and killing the fat man are the same! So are we all just cowards? Somehow we have an instinct against violence that is hard to overcome. The discomfort of the situation leads many people to change the rules. For example, here is one response that totally refuses to accept the premises:

In both scenarios, I'd run down and tell morons from moronville to not stand on train tracks even when a train is coming.

But, if forced into the only given options, I would pull the lever in scenario 1 so less people would be killed (unless it was like, 5 old people and 1 child or something). In the second I would not throw the extraordinarily fat man into the tracks. Fat and already-in-danger-of-dying he may be, anything heavy enough to stop a train's momentum like that would cause the train to violently stop and derail. Many more than 5 people would die and be seriously injured that way.

Here is a slightly more realistic example. Suppose you lived in Germany in the late 1930's and found yourself with a chance to kill Hitler. Suppose further that you either wouldn't get caught, or else wouldn't mind sacrificing yourself for the greater good. And of course, assume that you agree with today's consensus about the Nazis being evil.

Of course, it'd be great to stop the deaths of fifty million people. So shouldn't you go for it?

I think you can make an argument against killing Hitler even on utilitarian grounds. The problem is that the future is never clear. This is why we don't like to kill the fat man, and why doing so wouldn't be defensible in court: in practice, we're never really sure that the train wouldn't have stopped anyway, that other methods of stopping the train aren't possible, or that killing the fat man would help instead of even leading to some worse outcomes. The only time we really know for sure is in a thought experiment, which is why these experiments feel so artificial.

But c'mon! It's Hitler! Nobody likes Hitler. What could possibly be bad about killing him?

Well, what if popular opinion otherwise would have turned against the Nazi party, and killing Hitler drew more popularity to the party? Certainly Hitler wasn't the only Nazi evildoer. And even in hindsight it's hard to know what would have happened. More importantly, Hitler made many mistakes in fighting the war, often related to not listening to the advice of his generals. A different leader might have been much more effective, and not much less evil. Indeed the British studied the question and decided to cancel their assassination plans because, as one British officer put it, Hitler's incompetence made him "of the greatest possible effort to the British war effort." And the German generals who tried to assassinate him were arguably more motivated by Hitler's incompetence than by his warmongering.

Of course, if you're just some random German barber in 1940, you might not know all this. But that's the point! We never know exactly what the consequences of our actions will be. In what seems to be even the most clear-cut case for murder, the situation gets murkier and murkier the more you look into it. And often we have to make decisions without looking into them. Personally, I hadn't even heard the arguments about Hitler's incompetence until a few years ago, and before I had heard them I thought killing Hitler would have been a purely good thing. So for most of my life I was (probably) totally wrong about this question, despite knowing more about WW2 than most other parts of history. Extrapolating to the future, this means that I shouldn't overstate my certainty about things which today I confidently believe.

Finally, let's come back to the comment on Paul Samuelson:

Samuelson was just another Eichmann. He is responsible for propagating a destructive economic dogma.

Whoever wrote this is convinced that Samuelson's economic theories (or maybe his contributions to linear programming) made him an evil person, maybe even evil enough that we would have been better off if he had been assassinated earlier. Obviously whoever wrote this is a nutjob, but they clearly don't think so. They probably think that they alone have the moral clarity to see that Scenario 1 equals Scenario 2.

But the problem is that any of us could be that nutjob. And that's why we shouldn't kill people, even if we're utilitarians and are really pretty sure that it's all for the best.