I'm not sure, but I think I may have just taken a huge step toward solving one of the toughest problems in criminal law: The meaning of reasonable doubt. It is famously difficult to define this standard for members of the jury, which is a problem, because they're supposed to use it to decide if someone is guilty.

Scott Greenfield brought this up again a couple of months ago in connection with a California court ruling:

Yet, even this resort to once-pop culture doesn't overcome the utterly amazing reality that we persist in using a phrase to decide whether a person goes to prison, gets executed, that defies definition. We don't know what it means.  Jurors don't know what it means. Every time a judge or lawyer makes an effort to explain it, he ends up making things worse, obscuring it more.

...

That's the problem with vagaries, that they end up becoming whatever a juror ultimately decides to make of them. We can't articulate a meaningful definition, and they can't conceive of what exactly we expect of them. It's not their fault. Not in the slightest. It's our fault for acquiescing in the perpetual use of this meaningless phrase because we can't seem to figure out a definition that means what we intend it to mean.

The irony is that our inability to arrive at a viable definition reflects our own inability to agree on what this phrase means, or how one could possibly explain it to others.

As you'd imagine, leaving one of the most important parts of a juror's duty completely undefined tends to make the whole rendering-a-verdict process very unpredictable.

(I've often wondered if that was the intent, if the unpredictability of individual cases somehow improves the system as a whole. There are biological mechanisms and scienctific processes that make good use of the effects of randomness. However, I haven't been able to form concrete theory of why randomness would be good for our justice system.)

Jeff Gamso dropped by in the comments and wrote this:

Back when I was in Texas, judges weren't even supposed to define reasonable doubt. That left prosecutors free to tell jurors something like "reasonable doubt is a doubt for which you can give a reason," which is horrible. More than one criminal defense lawyer gave what's always seemed to me to be the clearest definition. "You've got to be real sure."

Gamso knows a hell of a lot more about these things than I do, but the definition that "reasonable doubt is a doubt for which you can give a reason" has great appeal for me because it's an operational definition. That is, it doesn't just define reasonable doubt, it defines a test procedure that jurors can perform to determine reasonable doubt. Operational definitions tend to make it much easier to reproduce a result. Perhaps this definition leads to injustice, but it's the right kind of definition for an important decision making process.

A little later, my Nobody's Business co-blogger Rick Horowitz posted on his blog in attempt to exhume the original meaning:

In any event, I disagree with Scott Greenfield that the problem is the phrase's inherently undefinable quality. The phrase is not indefinable; it's just that most of us have lost touch with the historical roots that could lead to a clear understanding of the phrase.

...

Our problem today, contra Greenfield, is not, as he declares, that the concept "defies definition." While it's true that "we" (I assume he means, at least, "most lawyers") don't know what it means; juries don't know what it means; and judges are too fucking stupid and ignorant of history to explain it, the concept itself no more defies definition than does any other concept from the past that ordinary students make sense of every day.

Rick then goes on to link "beyond a reasonable doubt" with "to a moral certainty" and quotes Barbara J. Shapiro:

Initially, there had been little need to construct a rationale for the truth-finding capacities of juries who reached verdicts based on their own common sense and knowledge of the facts. As the role of witnesses increased in the late medieval and early Roman period, the problem of the credibility of second-hand reports of facts that had become central to theologians, naturalists, and historians became central to legal theorists who borrowed conceptual elements from the new empirical philosophy.

Ah, "the new empirical philosophy." I assume that would be science. This appeals to me. I have some knowledge of the history and current meaning of scientific knowledge.

Throughout [the development of the standard of proof for criminal trials], two ideas to be conveyed to the jury have remained central. The first idea is that there are two realms of human knowledge. In one it is possible to obtain the absolute certainty of mathematical demonstration, as when we say that the square of the hypotenuse of a right triangle is equal to the sum of the squares of the other two sides. In the other, which is the empirical realm of events, absolute certainty of this kind is not possible. The second idea is that, in this realm of events, just because absolute certainty is not possible, we ought not to treat everything as merely a guess or a matter of opinion. [...] The highest level of certainty in this realm in which no absolute certainty is possible is what traditionally has been called moral certainty.

Shapiro goes on to give a jury instruction of sorts:

We can be absolutely certain that two plus two equals four. In the real world of human actions we can never be absolutely certain of anything. When we say that the prosecution must prove the defendant's guilt beyond a reasonable doubt, we do not mean that you, the jury, must be absolutely certain of the defendant's guilt before finding the defendant guilty. Instead, we mean that you should not find the defendant guilty unless you have reached the highest level of certainty of the defendant's guilt that it is possible to have about things that happen in the real world and that you must learn about by evidence presented in the courtroom.

That sounds fascinating, but I wish Rick he had provided some more explanation of what Shapiro means, or perhaps even a few examples of how jurors could apply an instruction like this. (Shapiro's paper itself does not appear to be available online for free, which means it might as well not exist for me.) As it is, however, I'm going to have to guess. And that may be a problem, because as stated, Shapiro's standard seems impossibly high.

That's because in the new empirical philosophy of science, there is no "highest" level of certainty in a statement of truth. What happens is that as more and more supporting evidence piles on, our level of confidence improves, but never quite reaches absolute certainty. So there is no highest level of certainty because more supporting evidence will always make us more certain.

But perhaps I am abusing the argument with too much modern scientific analysis. The statistical distributions that are used to analyze scientific certainty weren't derived until the late 1800's and weren't put to use until the early 20th century, long after the concepts of reasonable doubt and moral certainty entered use in law. Moreover, no one actually expects jurors to think about guilt in mathematical terms. Perhaps instead of interpreting the "highest level of certainty" as a precise mathematical test, I should interpret it on the scale and resolution of normal human reasoning, using the normal human meaning of "highest level of certainty."

That's a bit vague, but there is an interesting implication we can make use of: If we are at the highest level of certainty, then no higher level is possible. The highest level of certainty is that level of certainty which cannot be improved upon, not even by the introduction of more evidence.

This leads us to a nice operational definition that jurors can use: If a juror believes the defendant is guilty, he can then conduct a thought experiment by imagining that the prosecution was allowed to re-open the case and produce additional evidence. If he can conceive of no possible evidence that would further increase his level of certainty of the defendant's guilt, then his current certainty the highest level of certainty, and he can confidently vote Guilty. But if the case fails this test, if the juror can imagine additional evidence that the prosecution could resonably produce but did not, then he has to return a verdict of Not Guilty.

That sounds like it would work, but it still seems to set the bar awfully high. Would more than a tiny fraction of criminals ever be convicted? And what about DNA? The introduction of DNA evidence would make a lot of cases tighter, so does that mean that any case without DNA should automatically lead to a Not Guilty verdict? Again, that would make it awfully hard to convict anyone. I know I tend to root for the defense, but that's an awfully tough standard for the prosecution to meet.

I imagine jurors could be instructed to consider only the kind of imaginary evidence that would be pertinent (e.g. DNA not needed when identity is not in doubt) and reasonable (e.g. no new DNA evidence in a 30-year-old cold case) but that would seem to require juries to have unrealistically comprehensive knowledge of crimes and criminal investigations. Then again, don't we expect that of them now?

It feels like I've gotten somewhere with the concept of reasonable doubt, but I'm not sure if I've gotten anywhere useful.

And there it would have remained, if Scott Greenfield hadn't posted about the acquittal of two cops for raping a woman in their custody.

For those who balk at the fact that the system always seems to work better when it's a cop in the dock, another unfortunate reality, the answer isn't to be unfair to cops, but to be more fair to all.

Suddenly, I saw the light and had my answer:

Proposed model jury instruction on reasonable doubt:

In considering your level of doubt, imagine that the the defendant is a police officer who has served the public for many years. And further imagine that this courtroom is filled with his fellow officers, in full uniform, who are here to support him. If convicted, he could be imprisoned and forced to spend months or even years locked in a cage with the same kind of villainous scum he has been arresting his whole career. Even if he receives no jail time, your finding of even the slightest bit of guilt will likely bring his police career to an ignominous end. You will have declared him a disgrace to the uniform. He will be kicked out of the best job he has ever known and never allowed to return. The brotherhood of police officers will turn their back on him. His friends will leave him, and he will forever lose the respect of his family. To find a police officer guilty is to ruin him.

Now, as it happens, this defendant is not a police officer, but the standard of reasonable doubt is exactly the same. If the evidence is not good enough for you to convict a police officer, it's not good enough to convict this defendant either.

Of course, in a bench trial, this could be shortened to a procedure in which the defense attorney moves that his client be "tried as a cop."