The Zimmerman case just won’t go away; it’s an open wound that America will pick at for some time. A second juror has come forward to talk. Unlike the first, she clearly thought that George Zimmerman ought to be guilty of something ... but that the state simply didn’t make that case.

The juror said she had originally been ready to convict Zimmerman on second-degree murder, but after taking a closer look at the way the law was written, thought she had no other choice but to acquit. "As much as we were trying to find this man guilty ... they give you a booklet that basically tells you the truth, and the truth is that there was nothing that we could do about it," she said.

Much of the coverage and social media I saw about this seemed regretful, in keeping with the rash of instant legal reformers who sprang up like mushrooms after the verdict. Much of that commentary, too, seemed to think something was wrong with Florida’s self-defense law. Perhaps Zimmerman should have had a stronger duty to retreat. Or to keep his response proportionate to the injuries he’d incurred. Or maybe self-defense should be an affirmative defense -- something that the defense had to prove beyond a reasonable doubt, instead of the prosecution being forced to show, beyond a reasonable doubt, that the attack had not been self-defense.

Many of the commenters were confused about the duty to retreat issue; because Zimmerman alleges that he was on his back with Trayvon Martin on top of him, retreat wasn’t really an issue. A lot of people were also under the impression that these problems were specific to Florida, a point of view that seems to be completely wrong.

So is the supposition that it's a bad thing if defendants can easily claim self defense. As Will Wilkinson has pointed out, the major problem with American justice is not that it’s too hard to convict defendants, and the major beneficiaries of these kinds of protections are young black men like Martin.

But of course, the rest of us benefit, too. I don’t think that most people who want to get a guilty verdict on future George Zimmermans would actually like the sort of laws that could produce this result. It is perfectly possible to believe that Zimmerman is morally guilty of bad judgment or something worse, and also to believe that the best law is one that would set him free.

Whenever you want to identify something -- a cancer cell, a burglar breaking into your house, a murderer -- you have to accept that you will make some errors. Oh, we can try to reduce the amount of error in the system, but until we can get God to come down from heaven and start monitoring the phone lines for ADT, you’re going to have to accept that sometimes, that alert that caused you to rush home from work will turn out to just be your mom, who stopped by to pick up the clippers she lent you and forgot the alarm code.

Because there will be some level of error, you have to choose what kind of error you want. Statisticians describe the trade-off between Type I and Type II error, which can be loosely translated as false positives and false negatives.

Let me give you an example from personal history. Some years ago, for reasons that I still don’t understand, my doctor decided to test my blood for anti-nuclear antibodies. They’re a biomarker for autoimmune diseases, and are commonly used to diagnose lupus. I came up borderline positive.

As you can imagine, this sent me into a bit of a tailspin. I was practically writing my will. Eventually, I paid a teary and anxious visit to an immunologist who pointed out that the test has a false positive rate of between 5 percent to 10 percent, while the prevalence of lupus among women is somewhere between 1 in 500 to 1 in 1,000. So if you randomly gave ANA tests to 1,000 women, you would get 51 to 102 positive tests -- and of those women, only one or two would actually have lupus.

Because I had no other symptoms, and other markers of inflammation were fine, it was very unlikely that I had lupus. He told me to go home and relax. Which I did. But not before calling my doctor for a spirited discussion of blind testing and Bayes Theorem.

We could probably catch some more cases of lupus early if we gave everyone an ANA test. But for every person we helped that way, we’d have 50 to 100 false positives. The treatment for lupus is unpleasant. So, as I can personally attest, is erroneously believing that you have lupus. And so we allow some false negatives, because the cost of false positives is unacceptable.

The justice system faces the same quandary. The facts are not always clear. Crucial evidence you’d like to have is often missing. Frequently, the only living witness to a murder is the murderer.

And like most doctors (unfortunately not mine), our justice system has decided to allow more false negatives -- acquitting or never trying people who may actually have done something wrong -- because the cost of false positives is too high. I submit that this is the correct choice.

Right now, it feels wrong to many people because a boy who was walking home with Skittles and iced tea has ended up dead. But as lawyers say, “hard cases make bad law.” The law will always have some sad cases that can’t be prosecuted, or some cases where someone doing something understandable gets jail time for breaking the law. Laws written in response to public outcry about those hard cases are usually bad laws. They are the equivalent of deciding to give everyone an ANA test because of the tragic death of someone with undiagnosed lupus.

In both cases, we’re focusing on the emotional impact of the false negative that is right before us, and not all the other cases where false positives could be disastrous. Imagine that someone you have had words with -- your editor, perhaps -- attacks you, and there are no witnesses to the attack. Fearing for your life, you stab him, and he dies. Should the law require you to prove that you acted in self-defense, beyond a reasonable doubt? How could you?

Imagine now that it is a black teenager attacked by a racist 23-year-old looking for a fight, or a woman whose abusive boyfriend finally threatened to kill her. Do you want those people to have to prove that it was self-defense beyond a reasonable doubt? Should the prosecution be able to send them to prison for decades because it’s possible that they were the aggressor -- not even likely, just possible? Should they have to wait until they have suffered life-threatening damage in order to justify protecting themselves?

That seems insane to me. Of course I recognize that this allows some killers to claim self-defense. But -- at least at current low U.S. crime rates -- it seems obviously better for a killer to go free than for us to imprison innocent people for decades.

The weird thing is, under most circumstances, most of the folks who argued for weaker self-defense laws would probably agree that “it is better that 10 guilty persons escape than that one innocent suffer.” They seemed to have just discovered a quandary that readers of true crime stories have known about for decades. One of the most frequent recurring themes in these books and shows is the guy with several (well-insured) wives who died under mysterious circumstances, or who seems always to be around when terrible things happen. The police frequently think the guy’s guilty, but they can’t prove it. So he gets to kill again, until they get enough evidence to actually convict him -- better evidence than, “Gee, Dave’s wives sure seem to be accident prone.”

This is horrible and sad. But it would be even sadder and more horrible if we put people in prison because their wives were accident prone. The justice system has to weigh two true facts: Coincidences do happen; and many crimes leave no witnesses except the criminal.

Prison snuffs out precious, irretrievable years of someone’s life. When that person is a murderer, so be it. But the state should try as hard as it can never to take one minute of an innocent life ... even if that means the guilty sometimes walk.

Of course, when we’re fresh with the memory of a sad case, we want to tear down the obstacles and make justice happen. But we should also call to mind the other kinds of sad errors we might make, like sending an innocent man to prison for 25 years. How many Michael Mortons should we tolerate to convict one George Zimmerman? Because if you want to change the laws the way all these commenters suggested, your answer should be “a lot.”