Ethics and Landsburg's Runaway Train

I recently had the pleasure of reading a book called The Big Questions, by Steven Landsburg. Landsburg is an economist of some note, who teaches at the University of Rochester. (You can follow his blog - largely along the same lines as his most recent book - here.) Landsburg has the delightful distinction of being well-versed in economics, philosophy, mathematics, and physics. If that sounds like a diverse range of expertise to you, you may want to brush up on whichever of those subjects is your particular weak point. :) They are remarkably complimentary.

Landsburg's most recent book is fully of gems to get you thinking, but I want to be clear that I object outright to many of his affirmations. His book is so darn good, though, that I'm recommending it even though I disagree with much of it. 

In particular, Landsburg summarizes a common ethics problem that goes something like this:
Five total strangers are kidnapped by a mad philosopher and tied to a railroad track. A speeding train is en route to run them over. You can divert the train by throwing a switch, but if you do, the train will take an alternate route that will run over one other person tied to the other set of tracks. What should you do?
Landsburg considers this an easy problem. Before I tell you how he answered, please take a moment to consider how you would answer. 

Landsburg's answer is as follows: Saving five people is better than saving one, therefore the moral thing to do is to divert the train by throwing the switch. Landsburg's view seems to be based on a "greatest good" principle. Because throwing the switch means maximizing total human happiness in the system, the "right answer" is to throw it.

Now here's my answer: Before the mad philosopher came along, no one was tied to any railroad tracks. Obviously, the mad philosopher is the bad guy here. Now the choice you face is to murder one person to save five, murder five people to save one, or murder no one by saving no one.

Landsburg is quite critical of deontological ethics, but let's take a deeper look at his greatest good principle. If any one of the five people tied to the first tracks is a serial killer, the greatest good principle is shot to hell. If the one person tied to the alternate route is, for example, a heart surgeon, he saves lives virtually every time he goes to work (or she), so running him (or her) over also thwarts the greatest good. Landsburg would likely respond with an exercise in expected value; in other words, what are the odds that one or more of the five people are extremely bad people, and what are the odds that the one person tied to the alternate route is a an extremely good person? After weighing the "expected payoff" of this, one could arrive at the most likely scenario supporting the greatest good, and that would be that. (Of mathematical note, I believe the five people would still win out, since there are "more chances for good." On the other hand, there are also "more chances for evil," so maybe it really comes down to whether there are more good people or evil people in the world.)

I have a couple of objections to all of this. First, the train is speeding, and there is likely no time to calculate expected values to determine the correct answer. Second, and more importantly, there is no such thing as probability for goodness. In other words, a person is either good or not, and there is no "probability" attached to it. It is a personal choice, it is a matter of integrity. It is a question of free will, if you believe in it, or years of environmental stimulus, if you don't believe in free will. 

In short, human action cannot be calculated. It is impossible to know in advance if the person you condemn to death would have saved dozens of others (or even five others). On the other hand, killing someone always ends in the termination of a life. And that - assuming you respect human life - is an unequivocal tragedy. (Side note: I do not believe in capital punishment, but I'll save that for another day.)

The bottom line is that it is extremely presumptuous, and perhaps even megalomaniacal - to assume that you know better than six other potential murder victims which of them is more "worth saving" than the other(s). The absolute most moral course of action in this case, then, would be to avoid the switch entirely and try some other means of either saving the people or stopping the train. Anything else would be murder.

Landsburg's book criticizes deontological ethics, and I admit that I don't know enough about ethics to defend them. But if rejecting them means making the choice to murder even one person, as opposed to zero, then I can't say much for their value when it comes to ethical outcomes. 

What's your opinion?