Punishment and the Public Interest

There's an interesting debate going on in the comments section of econoholic's post on Polanski. I started writing a rather longish comment, before deciding just to post the whole thing here. The crux of the debate seems to boil down to those who think that only a crime victim's wishes count and those who think that society has some interest in preventing certain types of crime. But there is a great deal of confusion among those holding these views.

Curunir and econoholic defend (rightly IMO) the view that society has some interest in preventing certain types of crime. Here's econoholic:

On the other hand, if you accept that you personally lose something if other people are regularly murdered in your neighborhood and not just the direct victim, then there is a sense in which he has committed a crime against others as well, and he should be punished for it at the discretion of these others.

This, it seems to me, is another way of saying that certain types of crimes carry a negative externality. Where murder is rampant in a neighborhood, there is a (really high) cost to be paid by the direct victims of murder. But that cost doesn't full capture the total cost of the action. People fear for their lives, refuse to go out at night, purchase home security equipment, etc. Those are real costs, and while they are admittedly far less bad than the cost of, you know, dying, that doesn't entail that those aren't still costs.

If you take this notion seriously, then it's not absurd to suggest that, to the extent you think punishment is about restitution, criminal sanctions will have two components: the cost of making the victim whole plus the cost of the externality. Or, crudely, sanctions S = Restitution of Victim (RV) + Cost of Externality (CE).

Now several people point out that RV is more important than CE. And I don't deny that at all. In nearly every case RV > CE. Moreover, there are probably a lot of cases in which CE is close enough to zero as to be negligible. But it doesn't follow from either of those general observations that CE is always zero. And I don't think it's hard to make the case that, at least sometimes, CE is reasonably big.

In the Polanski case, we have a strange situation. For in this instance, we have a victim stating that RV = 0. It's hard to see how someone could argue with that. Who better than the victim knows what it takes to make her whole? Certainly not I.

But, and here's the rub, that doesn't automatically entail that CE is also zero. So when Micha asks "Does the people's interest outweigh the interest of the victim?" that's a sort of category mistake. The people's interest isn't something that one weighs against the victim's interest. RV and CE are on the same side of the equation. IOW, even where RV = 0, there may still be a good case for having some sort of sanctions.

I think that this case probably is one in which jail is merited, whatever the wishes of the victim. There is something to be said for living in a society in which being rich or talented or famous or a citizen of another country doesn't allow you to harm others and then prance away as if nothing happened. (And, no, I don't think that being forced to make all your movies in Paris counts as punishment.)

But whatever our arguments about the value of CE in this particular case, it's worth being clear on how the logic of punishment works. The cost of the externalities of crimes and the cost to the victim are additive. It's a mistake to think that because one outweighs the other, the smaller value just doesn't matter.

Share this

1. I don't think anyone here

1. I don't think anyone here claims that CE is always zero. But not every externality warrants a reaction.

2. The argument you make here is strikingly similar to the argument for hate crimes statues. That is: A hate crime does damage to people other than the direct victim through fear, intimidation, etc. Do you follow the same logic through for hate crimes?

3. Re: a category mistake. Just because RV and CE are on the same side in your equation doesn't mean that they can't conflict. Sometimes - as is true in this particular case - RV is not zero, but negative, and so we have to ask if when added together, RV + CE is positive or negative (assuming we are taking CE into account at all).

Fair Enough

These are reasonable objections. I'll take them in order.

1. I totally agree that not every externality warrants a reaction. Instances in which CE is fairly close to zero, for example, would seem to fall into that category. It's a totally separate discussion as to which sorts of examples of CE ought to merit reactions.

I do think that this may run the risk of conflating different theories of punishment, though. Saying that some instances of CE aren't worth compensating is a fairly utilitarian sort of reply. I know your intuitions run that way, as do mine. But I wasn't really aiming at a specifically utilitarian argument. If you're going to go that route, there's really no need to worry about CE or RV. Making the victim or society whole are more deontological approaches to punishment. If you want to be a utilitarian, you can skip this whole mess, move straight to deterrence and have done with it. Of course, in that case, what the victim wants has very little direct bearing on our calculations. To the extent that your main worry is with respecting the victim's wishes, you're not really playing the utilitarian game any longer.

2. This is sort of a subcategory of (1), I think. Another way to ask the question is to ask whether I think that CE in the case of hate crimes is large enough to warrant additional punishment. And the answer is...well, I don't know what the answer is. My natural inclination is to talk in terms of utility, and I think that there are very good utilitarian reasons against doing anything about CE in such cases.

If I were to stick with the deontological logic of punishment as compensation, though, then I would probably want to frame an answer in terms of competing rights. Even criminals have the right think odious thoughts. And thinking odious thoughts is permitted even where those odious thoughts do cause harm. This would require more thought to fully hash out. But I think that there's probably an important disanalogy between the two cases. It's not clear to me that rapists have any particular sort of prior claim to a right in the way that the perpetrator of a "hate crime" has a prior claim to freedom of thought. Or at least a parallel right isn't occurring to me.

IOW, I think that there's room for someone to consistently stake out a principled deontological opposition to hate crime laws while still maintaining that, in other sorts of cases, CE is something that merits additional punishment.

3. I think you're maybe on dangerous ground just as an empirical matter here. You'd have to claim that the victim will suffer more harm than the good that will be done by demonstrating that no rapists are above the law. These sorts of comparisons are pretty tough to do. But, really, isn't that a pretty extraordinary claim? That one person (not the one who is going to jail, mind you, and one who has already been thrust back into the spotlight and forced to relive experiences no one should ever have to endure) will suffer so much anguish as the result of her rapist going to jail that it completely balances out any deterrence effect plus the message to all other American women that no rapists are above the law. I think that the burden of proof generally falls upon those who advance extraordinary hypotheses.

There's also the problem that making RV negative has pretty unsavory implications. I mean, RV represents the cost to make the victim whole. If you posit that RV is a negative number, then you are, ipso facto, positing that the victim has been made better off by the crime. Are you suggesting, then, that maybe she should pay Polanski?

I'm only suggesting, based

I'm only suggesting, based on the victim's own words, that further prosecution represents a cost to her; i.e. it is not costless.

I want my protector to be a doomsday machine

Deterrence has to be credible to be effective. But a lot of people are softies. They don't want to be hurt by other people, but if they are hurt, they are liable to prefer to let bygones be bygones for a variety of reasons. The problem with this is that criminals know this, and this encourages the criminals. If only people were less soft then they would be hurt less often in the first place. But they're soft!

What to do, what to do? Here's the solution: get yourself a doomsday machine. If someone hurts you, then the doomsday machine will smash them. The important thing is that you can't stop the doomsday machine once it's been triggered. You're a softie, but the doomsday machine can't be stopped by you.

That's an effective deterrent. So, how to build a doomsday machine? Do this: hire a security company that will be a doomsday machine for you. If someone hurts you, then the security company will pursue the person who hurt you. If you change your mind (which you will because you're a softie), then they will ignore you. You agree to all this ahead of time. It's all in the contract that you sign when you sign up with the security company.

This security company will protect your ass from Polanski's dick.

So there you have it. No need to invoke "society", no need to invoke global utility.


This is a neater response than mine, and I'm inclined to think that it might be a more plausible argument. Interestingly, it's also pretty similar to the strategy of Locke's Second Treatise (though there's it the opposite problem; individually, we're too hard on those who violate our rights). Structurally, though, it's a similar strategy: individuals are bad at protecting violations of their own natural rights, so they should sell those rights to someone else (PPA or nightwatchman state).

I was mostly trying to clarify the logic of the view that (I think) econoholic and Curunir war articulating and to explain why I think that a certain strain of objection is missing the point.

Personally, however, I'm more inclined toward your explanation. I'd probably quibble that deterrence generally requires invoking global utility. But I suspect that that's just a matter of our using the word in different ways. Once we agreed on a set of technical terms, I think we'd find ourselves pretty much in agreement.

President Merkin Muffley:

President Merkin Muffley: But this is absolute madness, Ambassador! Why should you *build* such a thing?

Ambassador de Sadesky: There were those of us who fought against it, but in the end we could not keep up with the expense involved in the arms race, the space race, and the peace race. At the same time our people grumbled for more nylons and washing machines. Our doomsday scheme cost us just a small fraction of what we had been spending on defense in a single year. The deciding factor was when we learned that your country was working along similar lines, and we were afraid of a doomsday gap. [...]

Dr. Strangelove: Of course, the whole point of a Doomsday Machine is lost, if you *keep* it a *secret*! Why didn't you tell the world, EH?

Ambassador de Sadesky: It was to be announced at the Party Congress on Monday. As you know, the Premier loves surprises.

Are there really externalities?

There must be a lot of literature addressing this question--maybe some of the more economically literate out there can point me towards it.

Externalities seem to me to be just an excuse for representatives of the State to claim that they are acting on behalf of society and should tax negative externalities and subsidize positive externalities. Thus, members of the State conveniently have an excuse to redistribute wealth.

Is it too much to ask all victims to make a case for the harm they have suffered, including indirect harm? So your equation becomes:

S = RV1 + RV2 + ... + RVn

where we enumerate the victims.

For example, in a drive-by shooting, V1 is the target of the shooting, and someone prosecuting on his behalf demands restitution from the suspect for his maiming. V2 is the neighbor living behind V1 whose home was damaged by stray bullets; her prosecutor demands restitution for damage to the home plus endangerment. V3 through Vn are a group of nearby property owners whose prosecutors argue that their peace was disturbed, and property values diminished by the suspect's act.

Of course, in the AnCap world, the insurance companies representing V3 through Vn may decide the claim is too nebulous to justify the cost of prosecution and either pay their customers' claims directly or not depending on the policies in force, the premiums charged, whether representatives of V1 and V2 have already borne most of the cost of capture and prosecution of the suspect, and the competitive pressure within the insurance industry.

Victimless Crimes

I think you pose here a good framework for the punishment of so-called "victimless crimes". The people who want these crimes punished would argue that while RV = 0, CE is not 0.

"There's also the problem

"There's also the problem that making RV negative has pretty unsavory implications. I mean, RV represents the cost to make the victim whole. If you posit that RV is a negative number, then you are, ipso facto, positing that the victim has been made better off by the crime. Are you suggesting, then, that maybe she should pay Polanski?"

This is clearly fallacious. Just because she doesn't want restitution NOW doesn't mean the act helped her in the past, and the restitution could clearly make her worse off today due to various costs. You might rethink your equation measuring the externality. If we are talking about externalities, the correct procedure for analysis would be to measure the social marginal costs and the social marginal benefits, and to take action if SMB > SMC. We can clearly see from the victims actions that she does not want restitution, so I would include her on the side of social costs. If she wanted to take action, then she would go on the other side.

So the inequality I would use states that if:

Victim's utility loss (somehow measured in $$$ I guess) + Cost of Sanctions <= Social benefit

...then we take action. This should avoid the issue that arises from your equation.