Two hundred years ago, a lawyer named William Blackstone said it’s better for 10 guilty people to go free than for one innocent person to suffer. And for two centuries, legal scholars have considered Blackstone’s pronouncement a profound statement of principle. Apparently, none of those scholars has thought to ask the obvious follow-up question, namely, why 10? Why wasn’t it 12 or eight? The answer, of course, is that Blackstone invented a number out of thin air. That kind of flippancy amounts to a defiant refusal to think seriously about the trade-offs involved in designing a criminal justice system. But for 200 years, legal scholars have cited Blackstone’s refusal to think and mistaken it for an example of a thought.
There’s nothing profound about recognizing a trade-off between convicting the innocent and acquitting the guilty. The hard part is deciding how many false acquittals you’re willing to accept to avoid a false conviction. That number matters. It matters whether it is 10 or 12 or eight, because every time we rewrite a criminal statute or modify the rules of evidence, we are adjusting the terms of the trade-off. So it’s got to be worth it to think about what terms we want to aim for.
Here’s one approach: Imagine how a guilty man going free or a free man getting convicted might affect your life. (Or, so we don’t get too deeply sidetracked into your personal idiosyncrasies, how the guilty going free or the free getting convicted might affect the lives of your neighbors.) On the one hand, your neighbors risk being falsely accused and convicted. On the other hand, they risk being victimized by criminals who have been falsely acquitted (or by others who were emboldened to become criminals because of the frequency of false acquittals). In principle, the cost of either disaster can be measured in dollars. In practice, we can approximate those measures by making a reasonable guess as to how much your typical neighbor would be willing to pay to avoid a year in jail or to avoid being robbed on the way home from work.
After estimating the costs of being either an imprisoned innocent or a crime victim, we can estimate the probability that your neighbor will actually face each of these problems. But once we know the cost and the probability associated with a given risk, we can infer a lot about how undesirable that risk is. We can do this, for example, by observing the way people behave in insurance markets. Suppose you want to know just how unpleasant it is to face a 1 percent chance of a $100,000 loss. Then all you have to do is look at those people who face a 1 percent chance of losing their $100,000 homes in a fire and see how much they are willing to pay for fire insurance.
If you don’t like insurance markets, you can look at labor markets: How much extra must you pay a worker to get him to take a 1 percent risk of, say, losing an arm? If we believe for independent reasons that the value of an arm is $100,000 (no, I don’t mean to say that is the value of an arm; this is a hypothetical example), then we have another way to put a dollar value on the unpleasantness of a 1 percent risk of a $100,000 loss.
Or you can use data from financial markets: How much more interest must you offer an investor to get him to accept a 1 percent risk of a $100,000 financial loss? That’s relatively easy to observe, and it gives yet another measure of how much people dislike this particular level of risk.
False acquittals and false convictions are each associated with certain levels and probabilities of risk. By examining behavior in insurance markets, labor markets, and financial markets, we can make some reasonable guesses about how much people dislike each of these prospects, and also the extent to which people are willing to trade off one kind of risk for the other. That will give an indication of whether we ought to be expanding or restricting the rights of defendants.
It would take quite a bit of work to complete that project, and at the end all you’d have is a rough estimate. Your final number would be suspect in a hundred ways. For example, the data from insurance and labor markets tell a pretty consistent story about people’s aversion to risk, but the data from financial markets make the degree of risk aversion appear much higher. There might be no entirely satisfactory way to resolve such inconsistencies. But until you’ve done some kind of analysis, quoting a number such as “10” is both dishonest and disreputable.
