Monday, January 16, 2006

Moral Luck: Part I

Two would-be assassins take shots at their intended victims. One hits and is guilty of murder. The other misses and is guilty of attempted murder—a crime, but a less serious one. And, legal distinctions aside, most of us will see the successful murderer as morally tainted by his act—a taint that his failed colleague, by pure chance, escaped.

Why the difference? Being a bad shot is not a moral virtue.

For a second example of the same puzzle, consider two drunk drivers, one of whom hit and killed a child, one of whom (barely) missed doing so. Again, to both the law and individual moral feelings, the former is worse than the latter. Actual blood stains, potential blood does not--even if the difference is a matter of pure chance. Why?

For a third example, consider our feelings towards someone who was a Nazi concentration camp guard. Suppose you are convinced that most people in his society, offered the job, would have taken it. Does that make him less guilty? Does it mean that most of his contemporaries are, morally speaking, just as guilty—having escaped only through the good luck of not having the opportunity to commit his crime? Suppose you are convinced that most human beings, probably including yourself, if born and brought up in his society and offered the job would have taken it. Does that mean we are all equally guilty? Nobody I know feels that way, and yet the argument looks convincing. Why should someone’s moral status, praise or blame, depend on factors over which he had no control?

That is the puzzle that philosophers refer to as the problem of moral luck. I do not think I can fairly deal with the puzzle, and my answers, in one post of reasonable length, so will stop here, await comments, and return to the topic in a few days.

29 comments:

Anonymous said...

Since a few would have the integrity to refuse the guard job, even if it equated to suicide, and since we don't know who would exhibit such integrity, we must give benefit of the doubt to those who did not encounter the choice. Presumed innocent until proven guilty, even if most are guilty.

Anonymous said...

Isn't the point simply to provide an incentive to fail at distructive acts (even if it only operates at a subconscious level)?

Anonymous said...

randall randall: Some guy decides to blow up a building containing hundreds of people. As he reaches for the button to set of the explosion the police come charging in and manages to prevent him from blowing the building up.
Personally I would claim that his actions were wrong, even though he didnt actully manage to kill people, simply from the fact that he was allmost certainly going to.
And if I understand your previous post correctly it seems to follow that you would not agree?

There needs to be somekind of a distinction though. Imagine that we somehow could forsee the future (a determinists wet dreem) would we then be in the right to simply lock the guy up years before he attempted to blow the named building up?

Anonymous said...

One difference between the assassin and the would-be assassin is that the former has done more damage than the latter, and for that reason deserves more punishment. It may be circumstance that Hinckley missed his target and Sirhan hit his, and I might think they are both equally bad people, but Sirhan demonstatably did more damage.

We certainly do things that way when we account the goodness of people. I intend to donate 10% of my World Series Of Poker winnings this year to the Red Cross. I will get much more kudos for that if I actually have WSOP winnings, a highly unlikely event.

Anonymous said...

On the assassin: there are two relevant viewpoints here, the moral and the legal. To me, both assassins are equally at fault, morally; as you put it, bad aim is not a moral virtue. It's our legal system that assigns lesser punishment to the one who missed.

Now, in theory, a legal system should reflect the morality of its practitioners. However, when you get down to questions of intent, you have to start making compromises, because intent is so hard to prove. If attempted murder carried the same penalty as murder, it should carry the same standard of proof; in many cases, prosecutors simply couldn't meet that standard. Either many attempted murderers would go free, or legislators would lower the standard of proof to the point where innocent people would be punished. Our current setup seems like a reasonable intermediate case.

Anonymous said...

Individuals will only retaliate against those who have done them harm (or close shaves). Imagining a pure abstract justice system impartially viewing each person's behavior is a vestige of statism. The CC guard knows when he takes the job certain people will wish his death for the murder of friends/family. This is part of the risk he accepts when taking the job.

john(classic) said...

Aren't you confusing legal punishment with morality?

Legal punishment serves more purposes than just upholding morality. For instance, one of its goals is to supplant private revenge.

My desire for revenge is substantially different if one successfully or unsuccesfully attempts to kill someone in my family.

Anonymous said...

You seem to be suggesting that (perhaps) punishment should reflect the "expected value" of the damage done by an action. That seems quite reasonable, except for the problems of assessing expected value. Perhaps our criminal knows he's a bad shot and couldn't possibly hit -- why should he be punished more because there are other criminals who are good shots?

In fact, I think the current system achieves the objective of making the punishment fit the expected damage if you think of punishments as statistical. At the moment someone pulls the trigger, they know they have some chance of being caught, and if so some chance of being convicted of murder and some chance of being convicted of attempted murder. At that moment, while the bullet is in the air, his expected punishment fits his crime.

By punishing people according to outcomes the justice system can completely skirt the need to assess probability distributions.

Of course, as a policy matter, certain small punishments for acts with probabilities of bad outcomes may be better (from a variety of perspectives, including effectiveness of deterrence) than lottery punishments. I wouldn't want to do away with DWI punishments where no actual harm resulted, for example.

David Friedman said...

randall randall writes:

"Personally I would claim that his actions were wrong, even though he didnt actully manage to kill people, simply from the fact that he was allmost certainly going to.
And if I understand your previous post correctly it seems to follow that you would not agree?"

You did not correctly understand me. What I'm arguing as that most people would feel that what he did was wrong--but not as wrong as if he had succeeded.

Do you disagree? Would you think the unsuccessful mass murderer deserved the same punishment, and the same degree of moral opprobrium, as the successful one?

David Friedman said...

John (classic) asks:

"Aren't you confusing legal punishment with morality?"

No.

I'm pointing out that both the legal system and most people's moral intuitions follow the same pattern. Both degree of punishment and degree of moral opprobrium depend in part on considerations that seem to be morally irrelevant--the accident of hitting or missing, of whether the police get to you just before or just after you set off the bomb.

John T. Kennedy said...

"Being a bad shot is not a moral virtue."

What is?

Your positive account would seem more to explain away vice and virtue than to explain them.

Anonymous said...

Suppose I own a dog that viciously attacks people on a regular basis. What should be done? Should it matter whether my dog is a rottweiler or a chihuahua?

In practice it does. The rottweiler will be euthanized. The chihuahua will be kicked regularly.

This basically reduces to the voodoo killer problem in your book. I feel that the obvious solution that you give is actually the right one. If voodoo does not kill people then attempted murder by voodoo should not be punished. It costs the victim nothing and presumably has some value to the hoodoo.

If we punish people for attempted voodoo we should also punish murderous people for reading about poisons. In each case the only possible negative effect is that the murderer has gained dangerous knowledge.

Plots foiled by random chance are harder to argue. If I fail to blow up a car through no fault of my own (Say detonators fail with frequency 1/10000) why shouldn't I be punished as though I had succeeded?

Given that we can't impose a high enough punishment to properly discourage killings it makes sense to punish people who made skilled attempts, but I don't think punishing people for impossible attempts will discourage possible ones.

I suppose you can argue that you don't want attempted murder and murder to carry the same punishment since there would then be no disincentive to go through with it. For the detonator case that arguement is weak however.

Richard P. said...

Let me offer a case where moral consequences are not so dire. I currently live in Slovenia which by most measures is an extremely Catholic country. So much so that in a referendum last year voter chose to close shops of a certain size on Sundays. Now the question remains as to why. If you wanted to make sure that lowest paid workers had the power to refuse to work on Sunday then passing a religious freedom law would be more appropriate. But if you wanted to make respect for the Christian Sabbath mandatory that would be another thing all together. It seems to me that the second reason is the one voters had in mind when they voted yes. But don't you need the freedom to choose not to respect the Sabbath to make respecting the Sabbath a moral act? I think this is a question of freedom.

Anonymous said...

Your examples disregard incompetence. Intentions may not change outcomes but they are the defining factor between malice and negligence. For instance, if a police sharp-shooter aimed at a hostage taker but hit the hostage instead, is he immoral?

An intended but incompetent killer is immoral.
A drunk driver, never intending to kill but killing anyway, is negligent, but not necessarily immoral.
In the example of the Nazi concentration camp guard, if this is a preferred post and not a forced on, then the individual is immoral. If it was a situation of "take the post or your family gets locked up" or some such Sophie's Choice, then I cannot judge.

Anonymous said...

Bayesian inference.

In your examples, you claim that the different outcomes were due to pure luck, but observers in real life situations wouldn't know this with confidence. Maybe, instead of by chance, the concentration camp guard was slacking off on purpose, and that's why he didn't kill anyone.

Plus, we seem to need to find someone to blame for actual damage done, but we can be philosophical about mere statistical chances of danger. Perhaps our notions of blame need a concrete action to attach to to really be motivating.

Lippard said...

I recall reading an excellent David Lewis paper on this subject, "The Punishment That Leaves Something To Chance," in the Winter 1989 _Philosophy and Public Affairs_. I'll have to see if I can dig it up, only the first page appears to be available online.

Anonymous said...

David asks about two people with the same intent - to murder - one of whom is successful and the other is not. What about two people with different intents - a murderer and a drunk driver - who both kill someone. Should they face the same criminal penalties? The same civil penalties? Both murder and drunk driving are immoral and illegal. In this case, both victims are equally dead. But we treat one crime much more harshly than the other because of the intentent of the criminal. Should we?

Anonymous said...

The David Lewis Paper that Mr. Lippard refers to was reprinted in Papers in Ethics and Social Philosophy (Cambridge, 1999).

Loquitur Veritatem said...

In the example of the assassins your unstated premise is that most of us would not see the unsuccessful assassin as morally tainted. On the contrary, I think most of us would see both assassins as morally tainted because of their intentions.

And that -- intentions -- is the key to the example of drunk drivers. Neither driver intended to kill a child. The one who killed a child had bad luck, and the one who missed had good luck. There's no "moral luck" involved.

The third example, of the concentration camp guard, hinges on the assumption that the concentration camp guard had "no choice." But he did have a choice; it might have been a choice between accepting the job or being shot -- but it was a choice. The guard's moral status depends very much on a choice over which he had control. None of us should feel superior or inferior to the guard because none of us (or very few of us) is likely to face the same kind of choice.

In sum, the concept of moral luck is an empty concept because it doesn't hold up to strict scrutiny. I have written more about its shortcomings here.

Anonymous said...

Thomas: “There's no "moral luck" involved.”

Are you denying that the person who hits the child is considered more immoral than the one who does not? Perhaps it is true where you live but it is certainly not true everywhere. Drunk driving is relatively common and many people do not regard it as a moral evil, most people do regard drunk driving killing children as a moral evil. The difference is luck and people have different attitudes to the morality of each. Moral luck is an entirely appropriate term.

Anonymous said...

Your question leads on to this....

Should compensation be due to the object of a failed attempt or (more relevantly in the real world) to somebody who could have been (but was not) injured as a result of somebody else's very risky behaviour?

If yes, how do you calculate it?

If no, how else would free market courts deal with failed attempts and very risky (but in the event, non damaging) conduct?

Tricky.

Todd Mitchell said...

David Friedman writes: "What I'm arguing as that most people would feel that what he did was wrong--but not as wrong as if he had succeeded."

I think that's true, and the law seems to draw the distinction as well. In some states (Georgia, for one) there is not even an attempted murder statute. If I shoot someone in the head, with express intent to kill them, and they live, the worst I can be charged with is Aggravated Assault. That would seem to put a "moral" judgment on the rightness or wrongness of act as well.

Anonymous said...

"some states (Georgia, for one) there is not even an attempted murder statute. If I shoot someone in the head, with express intent to kill them, and they live, the worst I can be charged with is Aggravated Assault."

As with so many things in life, this would be fascinating if true, but sadly it is not. See the Georgia Criminal Code Title 16 section 16-4-1 and following:

http://www.legis.state.ga.us/
legis/GaCode/Title16.pdf

Julius

Todd Mitchell said...

Julius Blumfeld writes: "As with so many things in life, this would be fascinating if true, but sadly it is not. See the Georgia Criminal Code Title 16 section 16-4-1 and following: http://www.legis.state.ga.us/
legis/GaCode/Title16.pdf"

The statute cited: "A person commits the offense of criminal attempt when, with intent to commit a specific crime, he performs any act which constitutes a substantial step toward the commission of that crime."

You are correct in that the statute certainly implies attempted murder, but the penalties for "criminal attempt" and "aggravated assault" appear to be incongruent. Agg. Assault carries a penalty of 1-20 years (16-5-2) while attempt carries a penalty of 1-10 years (16-4-6).

The prosecutor would probably still go with aggravated assault charge, seeing as the penalty is stiffer.

Anonymous said...

Interesting. Here in England, I believe the position to be that the maximum permitted punishment for an attempt is the same as for the substantive crime.

Tanner said...

I'm a law student in Georgia. (Actually, Todd, I live in the same town you do.) We definitely try people for attempted murder all the time. If an attempted murder is also an aggravated assault, of course, a defendant is typically charged with both, not merely the one with the greater sentence. See Jones v. State, 288 S.E.2d 795 (Ga. App., 1982).

Todd Mitchell said...

Tanner writes: "I'm a law student in Georgia. (Actually, Todd, I live in the same town you do.)"

Excellent...we should commisserate some time at the Globe!

"We definitely try people for attempted murder all the time. If an attempted murder is also an aggravated assault, of course, a defendant is typically charged with both, not merely the one with the greater sentence. "

Absolutely, and I should have written that as well...the charges certainly aren't mutually exclusive, and to make sure the guy was put away a long time, multiple counts of agg assault would probably be piled on, along with the attempt.

Peripatético said...

I’m agree with Kyle Bennet; under a economic perspective of ethics (as mean to create incentives) the moral valuation of an act should be correlated with the outcome in both ways a priori(expected) and a posteriori(historic). The both are important and must be considered together when it is possible. Marxism and its offspring is a straightforward example: Judged by its intentions (peace, wealth, equality, liberty, etc…) could be acceptable, by its means (Economic pacification, state control of society, proletarian dictatorship, etc…) was a controversy but judged by its effects (death, poverty, crime, etc…) it’s simply unacceptable. So, moral as and adaptive machine that solves problems takes account of all those informations. Therefore, two acts might have different valuations not only by its expectations or results but also by its means (death by lethal injection vs. death by 3-day torture session) by each individual.

The weight of each part of information of this hypothetic value equation is impossible to estimate but I think (this is an intuition) that expectations are a big part of the pie.

Anonymous said...

Perhaps the resolution to the paradox is that we never know the whole story, and the outcome itself becomes part of the evidence as to the person's mental state.

Maybe the would-be assassin missed because he didn't really want to murder someone. The evidence of intent to kill is stronger for the one who succeeded than it is for the one who failed.

Perhaps the one drunk wasn't as intoxicated as the other -- and that's why he missed hitting the child.