Physical Reality and the Federal Rules of Evidence

Jamison Koehler has a short post quoting from the Federal Rules of Evidence:

According to Federal Rule of Evidence 401, the test for relevance is whether the evidence has a “tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would be without the evidence.”

I’ve heard that language before, and it still annoys me. Jamison’s explanation from McCormick doesn’t help:

“Probative value” is the “tendency of evidence to establish the proposition that it is offered to prove.” Does learning of this evidence make it more or less likely that the disputed fact is true?

That language bothers me. As I think of them, facts aren’t “more probable or less probable” and nothing can make them “more or less likely.” Facts are things in the real world, and in the context of a trial, they are also things in the past. Either they are true or they are false, and nothing said in a courtroom can travel back in time to change the real world.

Now, statements of fact can be vague, or hard to evaluate. For example, the truth of the statement “John is tall” is somewhat subjective, depending on what height we consider to be tall, so the truth of the statement is open to interpretation. But no matter what is said in the courtroom in arguing whether John is tall, none of it is going to change John’s height.

When considering a fact at trial such as “John shot James,” the jury (or judge) may be asked to consider all kinds of evidence, from eyewitness testimony to forensic analysis to a security video. But no matter what evidence is presented, none of it changes what actually happened. A witness who says “John shot James” doesn’t change the probability that John shot James, because there’s nothing probable about it: Either John shot James or he didn’t.

So rule 401, as written, doesn’t make any sense. And yet it encompasses an idea that is an integral part of our justice system. And really, setting aside my somewhat pedantic objection, we all sort of know what it means, don’t we? So how can we rehabilitate rule 401 to comport with physics and reality?

For a first attempt, we could try to to escape by embracing subjectivity and saying that probative value is something like “the tendency of the evidence to give jurors confidence in their assessment of the truth it is offered to prove.” That certainly comports with the real world, but it raises other questions, such as what manner of reasoning the jurors should use to evaluate their level of confidence. If we’re going to make decisions based on jurors’ predicted mental processes, we have to have some model of those processes.

I can’t say that I know how jurors should reach their decisions as fact finders in a trial, and I certainly don’t know how they do reach their decisions, but I do know of one methodology that has been very successful in other areas, and that’s the scientific method.

The idea is that the jurors should examine the factual assertion in question — “John shot James” for example — and then construct its logical negation (called the “null hypothesis” in science) that John did not shoot James, and assume for the moment that it is true. Now, for all the evidence presented at trial, they should consider how likely it is that that evidence could exist if their null hypothesis “John did not shoot James” was true.

Note that we have eliminated the time travel requirement. We are no longer talking about changing the probability of things that happened in the past. Rather, we are evaluating a proper set of unknown alternatives, where the language of probability makes sense: What are the chances that this evidence could be presented in court if our null hypothesis were true? If John did not shoot James, what are the chances that…

• John’s prints would be on a gun that has a ballistics match to the bullet taken out of James
• A witness would have seen someone he later identified as John driving past James’s house 15 minutes before the shooting
• Blurry security video shows someone, wearing the same clothes as John was that night, enter James’s building at approximately the time of the shooting

…and so on.

If the jurors determine that the evidence would be sufficiently unlikely to arise if their null hypothesis is true — e.g. they are certain beyond a reasonable doubt that it would not be possible for this evidence to exist if John did not shoot James — then they should reject the null hypothesis. And if the null hypothesis is false, then its logical negation must be true: John shot James.

(Scientific experiments are designed to make the evaluation of probabilities clear and convincing, with mathematical precision. For example, an experiment might conclude that if the null hypothesis were true, its results could only be obtained by random chance 5% of the time. This could be expressed as saying it supports the hypothesis at 95% confidence.)

So perhaps it would be more scientifically accurate to say that evidence is probative only if its the probability of its existence would be higher or lower depending on the truth or falsehood of the fact in question.

I can’t help noticing that this bears a similarity to the theoretical definition of the information content of a message. There’s a way to express it mathematically, but the basically the information content of a message is a measure of the degree to which its contents are unpredictable or surprising. In the simplest case, a repetitions message — “The British are coming! The British are coming! The British are coming!” — carries no information beyond the first copy of the message. Parts of the message that can be derived from other parts also carry no information: In the equation message “2 + 2 = 4”, the “4” can be said to carry no information because it can be unambiguously determined from the other parts of the message.

So maybe we could say that evidence is probative only to the extent that it is surprising or unexpected in the context of the facts in dispute. For example, if John lives next door to James, the eyewitness who saw him driving by James’s house is a lot less unexpected, and so has less probative value. And the music selection on John’s iPod is no more or less surprising regardless of whether John killed James, so it’s probably irrelevant.

I think I’ve just boiled the rule down to saying evidence is probative only if it carries information about the fact in question, so I haven’t exactly revolutionized the rules of evidence. But at least I got rid of the damned time travel.