Whenever I ask about the reasoning underlying some legal principle or another, my friend the law professor is always quick to remind me that “there is no such thing as legal reasoning”. So it is with William Blackstone’s famous doctrine that it’s better for ten guilty men to escape than for one innocent to suffer. Why ten? Because that’s the first number that happened to enter Blackstone’s head; that’s why.
Writing 200 years after Blackstone, Emory Law School professor Alexander Volokh surveyed the history of alternatives to “ten” in a charming essay called n Guilty Men. The bottom line is that a great many alternatives have been offered, almost never with anything approaching a justification.
Surely the right value of n must depend on the effectiveness of punishment as a deterrent, and surely this will vary depending on the crime and the punishment. So to focus the discussion, let’s talk about the crime of murder and the punishment of execution. Here the deterrence value has been estimated hundreds of times and most of the estimates are remarkably large. (Caveat: Studies that measure the effect of passing a capital punishment law tend to find no deterrent value at all; studies that measure the effect of actually executing people tend to find a quite considerable effect.) With some notable exceptions, most studies have found that a single execution prevents something on the order of ten murders. If we accept this number, then Blackstone’s wisdom is exactly reversed: It’s better (or at least not worse) for ten innocent men to be executed than for one guilty man to go free. Either way, ten innocent lives are lost.
Except for one thing. This argument might work if the ten innocent men are chosen randomly (by being in the wrong place at the wrong time, or via a passing resemblance to a dimly glimpsed perpetrator). But I guarantee you that if we were to modify our justice system to make convictions this easy, unscrupulous police and prosecutors would make a hobby of conjuring up charges against people they happened to dislike. That’s really not how we want them spending their time.
Yesterday I wrote about Peter Leeson’s paper on the medieval practice of Trial by Ordeal, and why the system might have actually worked pretty well in a world where priests rig the results. The problem, as pointed out by Sierra Black in comments, is that you’ve got to trust the priests to want to do the right thing. I don’t trust the priests, and by the same token I don’t trust the prosecutors, so I’m not prepared to endorse anything close to the reverse-Blackstone one-to-ten ratio. But I doubt that Blackstone’s ten-to-one is anywhere near right either.
Eugene Volokh did not write that piece. His brother, Alexander “Sasha” Volokh did.
This argument might work if the ten innocent men are chosen randomly
… but not if people know that they are chosen randomly! A prospective murderer must expect that he, himself, is going to be killed for deterrence to work.
Unless all prospective murderers are much more stupid than police and prosecutors, it follows that most people sentenced to death must actually be guilty for the system to work.
Even if the 10 were chosen at random, to claim that it is “better (or at least not worse) for ten innocent men to be executed than for one guilty man to go free” you need one more thing:
You have to weight the death of a man at the hands of a murderer equally to a death at the hands of the state.
I always though that the meaning of Blackstone’s saying was that he denied this equivalence. He’s trying to say that saying that for the state to murder an innocent man is (at least) ten times worse than for a murderer to kill someone. Under this moral calculus (ignoring the randomness issue you mention and believing the result about an execution deterring 10 murderers) shouldn’t we relax the conditions for execution until half the men we execute are innocent?
In my book 10 is too low a number for this meaning, we should be more cautious than that. And of course the non-randomness of the innocents we execute should make us still more careful.
Reply to John Jenkins: Thank you for catching this embarrasing error; I’ve fixed it now.
>This argument might work if the ten innocent men are chosen randomly
I really don’t get this. If you just randomly execute 10 people, you don’t deter any murders at all, you just randomly execute 10 people. Surely the deterrent effects are calculated in a world where being executed correlates pretty strongly with having committed a murder, so the calculation can’t possibly ‘work’ – break (or even weaken) the correlation and you have to recalculate the deterrent-per-execution.
John Faben: I could have worded this more clearly. They of course have to be ten men who have been found guilty beyond some threshhold of doubt, but they also have to be ten men who cleared that threshhold for random reasons.
@ improbable : thanks for stating so clearly the issue about the state taking a life vs. an individual murderer doing so.
Improbable and Sierra Black: It seems to me that the *reason* we think it’s worse for the state to take a life than for an individual murderer to take a life is precisely that we’re worried that the state will devote energy to targeting particular groups. This is bad both because that’s energy that could more profitably be directed elsewhere, and because it forces people to take costly measures to hide their membership in such groups. (E.g. if homosexuals are targeted, then homosexuals stay closeted.) These are of course exactly the considerations I already noted in my last paragraph. Are there additional reasons I left out?
@Steve I still think you’re missing the real point. As a society we have decided that murder is morally/ethically wrong. If our justice system is to have any semblance of integrity we cannot have it murdering people ie executing innocent people. The 10 to 1 ratio is not meant to be some numbers you plug into an equation rather it’s an admonishment that if we execute someone, we need to be as sure as possible it is justified, that it is an execution and not a murder by the state. Taking a statistic such as “10 murders prevented for every execution” and plugging it into a formula trying to figure out the correct number of executions to minimize total overall deaths is wrong-headed. It completely ignores the fact that in the real world there are other ways to deter murders other than just executing people. It also ignores the fact that murder is wrong and shouldn’t be used as tool by the justice system.
It’s interesting, is it not, that people are so alert to the dangers of government overstepping its bounds when it comes to punishing murderers. I.e., the state is inherently dangerous. Yet, many of the same people are totally blind to the dangerousness of government when it’s a question of the state interfering in, say, our health care arrangements.
I think there’s another reason why it’s worse to die at the hands of the state than at the hands of a murderer. If a murderer kills you, your family will receive an outpouring of sympathy and support (and possibly benefits) afterwards. (Not that this matters to your after you’re dead, but while you’re alive, this fact might slightly decrease your worrying about being murdered.) On the other hand, if the state kills you, they mount an entire PR campaign to convince the public that you’re a horrible person who deserves to die, which you and your family have to endure while you’re still alive.
Also, even if you accept that one execution can prevent 10 murders, you have to weigh that against other measures that could also reduce the number of murders. Most Western European countries have much lower homicide rates than the U.S., for example, despite having no death penalty:
http://en.wikipedia.org/wiki/List_of_countries_by_intentional_homicide_rate
But then you have to weigh the costs of whatever solutions they’ve come up with (presumably having a lot of social safety nets helps to reduce the poverty and desperation that sometimes leads to murder), against the costs of our (probably cheaper) countermeasure of executing criminals… which means putting a cash value on the lives of murder victims and on the lives of Death Row inmates, so you can weigh that against the cost of the social welfare state.
Stephen Coy: But in the real world we do have to decide on threshhold levels of doubt when we draw the line between convictions and acquittals. Drawing these lines *does* amount to choosing between ratios of 5-to-1 or 10-to-1 or 20-to-1. The only question is whether we’re going to face these hard questions honestly or not.
But a state has a special status– it’s a privileged actor. A state’s legitimacy is what allows its monopoly on the legal use of force. That’s why murder (i.e., use of deadly force by a non-privileged, non-state actor) is illegal. When a state justifies its actions by equating them to an illegal use of force it risks losing its legitimacy.
Correction– I meant only ‘illegal’ to be italicized in the last line of the above comment.
@Steve Yes, I agree that there will always be some level of doubt and that we do have to decide on some acceptable threshold. Where I think we diverge is your statement “If we accept this number, then Blackstone’s wisdom is exactly reversed: It’s better (or at least not worse) for ten innocent men to be executed than for one guilty man to go free. Either way, ten innocent lives are lost.” This statement assumes that nothing else can be done to deter murder. This statement also assumes that there’s no difference between having criminals kill 10 innocent people and having the state kill 10 innocent people. I believe there is.
Steven Coy: Your first objection is that the statement assumes nothing else can be done to deter murder. This seems to me to be not quite right; no matter what else I’ve done to deter murder, if an execution can deter an *additional* ten murders, then that’s the number that matters. Of course, it might be that if we did more of those other things, then the marginal effectiveness of executions would fall.
Your second objection is that having the state kill people is not the same as having individual murderers kill people. I acknowledged this in the original post, pointing out that state authorities with the power to execute might tend to abuse this power, and I emphasized that this is something we’d want to account for in setting our policies. Are you reiterating that, or are you suggesting that there is some *additional* reason why it’s worse for the state to take lives?
Steven, that makes more sense, and I guess I should have figured it out before, but I’m still not quite sure I follow the maths (although I agree with the more general point that *doing* the maths is a good step forward… and I realise it isn’t actually key to the point you’re making)
The figure that one execution prevents 10 murders is presumably calculated using the *current* level of doubt used in murder trials. If you increase this level of doubt, you change two things which affect a potential murderer’s decision to commit a murder: you make it more likely that he’ll be executed if he does commit it (thus reducing the expected value of committing the murder) but you also make it more likely that he’ll be executed if he *doesn’t* commit it (thus increasing the expected value of each murder).
Maybe I’m missing something (I haven’t actually tried a numerical example), but I can’t see any obvious reason why these two should cancel each other out, so I don’t think you can continue to use the 10 murders per execution figure if you start executing more people.
(At the extreme, if you just decree that you’ll execute everyone, execution completely loses its deterrent effect, so there is clearly some tradeoff between the level of doubt you choose the level of deterrence you achieve per execution)
You’re correct that the number 10 was pulled out of the air.
However, it was never intended to be used in an equation. It’s a
moral stance to justify the U.S. model of innocent until proven
guilty beyond a reaonable doubt. That’s in contrast to some
European models where you’re presumed to be guilty as charged,
and it’s up to you to prove your innocence.
@Steve Even if nothing else changes, the 1 execution saves 10 lives statistic doesn’t scale. You’re treating it as if it is a linear relationship. In 2008 there were just over 16 thousand murders in the US. If we had executed every single murderer would we have prevented 160k other murders? Clearly not. The 1-10 data point is just that, a single point of data. (Actually it’s not even a fact, it’s the conclusion of some of the studies you mentioned. I doubt they were able to measure it directly).
But on to the harder question, you ask if there was an additional reason why I thought it was worse for the state to take lives. For me it boils down to a questions of ethics. My reading of Blackstone is that our justice system needs to try as hard as possible to ensure that innocent men go free. Backing off on that effort because some studies suggest that executions have a deterrent effect is not the right thing to do. Each individual needs to be treated as ethically and honestly as possible by the justice system if there’s to be any justice at all. Obviously this veers more toward opinion than argument but there you have it.
John, your math isn’t quite right.
If you raise the likelihood of execution, you raise the cost of murder.
If you raise the likelihood of wrongful execution, you raise the cost of being a citizen in the U.S., regardless of whether or not you murder someone else or not. This doesn’t lower the cost of murdering someone at all.
This raises the marginal cost of committing murder if you value not being executed since you are only adding costs and not taking away any or, as you state, adding expected value. The additional wrongful executions are more like a head tax, they are not dependent on you murdering or not.
Steve, seems to me there’s a countervailing set of incentives that works against increasing the ratio of innocents to guilty executed, especially to as high as 10 to 1. The more innocents are executed the more guilty go free to repeat their crimes, thus undermining deterence. And, not only does the criminal remain free, they also know someone else has been convicted of their crime, so they’re more likely to conclude they can act with impunity the next time. So deterence is further undermined.
Moreover, if it becomes widely known (for example, through the Innocence Project) that the ratio of innocent to guilty executed is high (and therefore, the number of crimes for which the actual perp escaped punishment is high), the general population of criminals, and potential criminals, are more likely to conclude they can act with impunity, and again, deterence is undermined.
Finally, as MattF points out, if it becomes widely known that innocents are being executed in substantial numbers, the legitimacy of law enforcement officals is weakened. This too has implications for deterence. Lower public confidence in the competence or commitment to justice of officials will lead to lower voluntary cooperation with the police, fewer tips, fewer witnesses, lower conviction rates. The loss of state legitimacy could also be expected to lead to a reduction in the number or quality of prosecutors and police willing to serve a state with little commitment to justice, leading to weaker cases and lower conviction rates. Deterence deteriorates.
To Patrick Sullivan: You’re equating the execution of innocent people to providing health care to people who can’t otherwise afford it? Hmmm. I guess that would equate Medicare and veterans’ health care with executing innocents. Sounds warped to me but I’m not a libertarian.
Yes, we worry about the capacity of government to do ill, even evil. That doesn’t negate its ability to do good as well…like, I assume you would concede, catching, prosecuting, and punishing criminals; fighting fires; defending the nation; preventing terrorism; building and maintaining a national transportation system. Maybe even more.
Pete,
>you raise the cost of being a citizen in the U.S., regardless of whether or not you murder someone else or not. This doesn’t lower the cost of murdering someone at all.
Surely it does, since it makes living in the United States without being a murderer less attractive (so the cost of being executed for committing a murder is lower).
Consider taking things to extremes, where you execute everyone who you think has a 1% chance of being a murderer, surely in this case you might as well commit a murder if you see any marginal benefit at all to murdering someone: the punishment has completely lost its deterrent effect.
‘To Patrick Sullivan: You’re equating the execution of innocent people to providing health care to people who can’t otherwise afford it? Hmmm. I guess that would equate Medicare and veterans’ health care with executing innocents. Sounds warped to me but I’m not a libertarian.’
Yes, I though someone would resort to this kind of sophistry, that it’s only ‘providing health care to people who can’t otherwise afford it’. Are you really unaware of the change in the power relationship between citizen and government entailed in that?
About additional reasons to weigh murder by the state more highly than murder by some criminal (apart from the state’s tendency to focus the “error” onto particular groups):
I suppose this is for me a moral thing. The state executes people on _my_ behalf, and I want no part in wrongful execution. I want this even if it comes at some cost in efficiency (I mean in minimising the total number killed).
I think Bennett Haselton’s point is a good one, that the stigma attached to having (say) your husband murdered vs. executed is very different.
When I think about medical care, there is an interesting squiggle which I wonder how many people share. In treating cancer patients I’d be happy to argue for efficiency, in the sense that we should adjust the level of potentially dangerous treatments to minimise the number of deaths, regardless of each one’s proximate cause. But in providing inoculations, I would be tempted to weight a death due to vaccine higher than one due to the illness. So there is some quirk about the number of people involved and how high their risks are, I think. I’d also go with pure efficiency when discussing what weapons to give soldiers, I think. (A helicopter with worse reliability and better armour-plating: just do the numbers.)
Response to Patrick Sullivan. Sure I’m aware of it. How is the change in the relationship between the people and the government involved in hc reform qualitatively different from the state’s current authority to impose a death sentence (including to innocent people as we’re discussing here) as opposed to life sentences, or the authority to go to war, especially nuclear war, or spy (legally; illegal spying is a different matter) on its own citizens, or impose an income tax, or create the Social Security and Medicare programs, among others.
These all have profound effects on the relationship between the state and the people at least as profound as those of hc reform.
By the way, is it really necessary to label the arguments of those who disagree with you “sophisty”?
The reason that it is so much worse for the state to execute a man than for a murderer to do so is because the state has a monopoly on force. Its somewhat similar to the oft-maligned and just as oft-misunderstood canard about people preferring driving to flying even though flying is far safer. That may be true in the aggregate, but *I* am an excellent driver, and *I* can prevent any accident, so for me driving is much safer than putting my life in the hands of some pilot.
Similarly, although less irrationally, it is one thing to put my life at increased risk due to more murderers running around. This is bad. But it is far, far worse to fear the state. *I* am strong, *I* am cautious, so I know that I could prevent a murderer from killing me. Heck, I could probably subdue him with some sweet martial arts moves I saw in a movie last week. I have very little risk! But no amount of roundhouse kicks is going to protect me if the state comes knocking at my door. They have a legitimized monopoly on force.
@ Stephen Coy: “My reading of Blackstone is that our justice system needs to try as hard as possible to ensure that innocent men go free.”
This can’t possibly be what he meant, otherwise he would have chosen infinity rather than 10 (thus making the statement, it is better to let all guilty men go free than to convict one innocent person). Since he didn’t say this, he is acknowledging that there is some tradeoff, and is proposing that 10 is the correct value. Why 10? We still don’t know.
Douglas Bennett: Extremely well put.
@Douglas Bennett You quote my comment about Blackstone and then interpret that as implying that I think Blackstone meant that we should only punish people that are 100% provably guilty with absolutely no error for doubt. Yet you choose to ignore where I wrote “Yes, I agree that there will always be some level of doubt and that we do have to decide on some acceptable threshold.”
You can’t just pick and choose sentences out of context.
Oh well, at least you got a pat on the head form Steve.