Proof Unburdened

Talk about the burden of proof is usually dumb

In online debates, discussion of the burden of proof is ubiquitous. There seems to be widespread agreement that the burden of proof—who is tasked with proving a claim—is on the one who asserts the claim. Endless debates are had over who shoulders the burden of proof.

I think this is mostly confused. Talk about the burden of proof is largely nonsense. We should just be Bayesians.

Bayesianism sometimes gets a bit complicated, and has a special formula, but the essence is pretty simple. You have priors—which are how likely you regard claims to be before you’ve considered the evidence for them. Then you have evidence in the form of facts that are likelier if some view is true than if it’s false. A fact is evidence for a hypothesis if it’s likelier if the hypothesis is true than if it’s false. For example, if my DNA is at the crime scene, that’s evidence that I committed the crime, because it’s likelier my DNA would be at the crime scene if I committed the crime than if I didn’t.

It may not be sufficient evidence to establish the claim, but it’s evidence nonetheless.

To a Bayesian, the burden of proof looks a bit weird. Your evaluation of a claim shouldn’t depend on who asserted it. You should just have a credence attached to a claim, depending on how intrinsically unlikely it is and how much evidence there is for it. It doesn’t matter if something is a positive claim or not—you evaluate positive claims the way you evaluate every other class of claim.

Generally, people say that in the context of a debate, if a claim hasn’t been demonstrated, you should treat it as false. But this is totally unworkable so long as you affirm the law of non-contradiction. Suppose there is a claim P. I should think P is false, because it hasn’t been proved. But then suppose there’s another claim: the claim “P is false.” Well, I should believe that’s false, because it hasn’t been demonstrated. So I should believe…that P is false and that P is false is false???

No, this is all ridiculous. You should just assign a probability to P! It doesn’t even have to be an exact probability.

There are, of course, some cases where taking into account the burden of proof makes some sense. In a courtroom, for instance, the burden of proof is on the prosecution. But this isn’t because of some fact about epistemology, but about justice. We don’t want people to be convicted unless there’s highly convincing evidence for their guilt. So for that reason, we make the bar for establishing guilt quite high. This isn’t because a person who suspects someone committed the crime even if it hasn’t been proved would be making an error—it’s because we don’t want people to be convicted unless there’s very decisive evidence of their criminality.

Laypeople are also bad at thinking about probabilities. If juries were told to only convict if the odds of the person convicting the crime were above some probabilistic threshold, that could go horribly wrong. Asking juries to think about probability is like asking turtles to think about the ontological argument.

There’s another context in which something similar to the burden of proof is appropriate. Consider:

Me: I think the self-indication assumption is definitely true.

You: Oh really, why?

Me: It’s not my job to educate you. Do the work. Don’t force self-indication assumption understanders to do the emotional labor of teaching you, which is often traumatic and triggering for POSIAA (persons of self-indication assumption understanding).

In this case, even if we think that the self-indication assumption has a high prior, I’m clearly doing something wrong. But this isn’t an epistemic norm, it’s a conversational norm. If you assert something in conversation, you better be able to explain why you think it. If you can’t, that’s not an epistemic problem, but it’s probably just being weird for no reason. You may have perfectly good reason to believe the SIA, but if you bring it up, and then refuse to explain it, you’re failing to live up to ordinary conversational norms.

Treating the burden of proof as a mere conversational norm is fine. But the ubiquitous talk in public debates, along the lines of, “you have made a positive claim, so you shoulder the burden of proof,” is quite silly. Often people think you need some other explicit argument for every proposition you believe, even if the proposition is intuitive and plausible. That obviously can’t be right on pain of regress—you can always ask for a deeper argument for any argument given.

Trapezoids keep zoids trapped. This is a very nasty thing! Let the zoids be free! Similarly, the burden of proof keeps proof burdened. It should stop. Let the proof be free of its shackles!