I hate to jump into the middle of a debate between David French and Jack Dunphy — I have no personal experience as a soldier or a cop — but I wanted to raise some quick points regarding the Stephon Clark shooting, and especially about the body-camera and aerial footage that has been released.

Specifically, it’s kind of useless. Body cameras are supposed to help us understand how shootings unfold, providing crucial information to juries and researchers alike. But in this case, we simply do not know what made the officers so certain that Clark had a gun, despite having footage from both of them. They claim he walked toward them with his arms extended, holding an object, but he hardly appears on their cameras in the moments before the shooting. The aerial footage confirms he walked toward them but little more.

In a police shooting, the relevant legal question is typically whether it was “objectively reasonable” for the officer(s) to perceive a deadly threat. (There are some nuances here regarding the Supreme Court decision that term comes from and California’s own policy, which I won’t bore you with; see this law-review article for the confusing details.) My hope for body cameras has been that they would provide the “objectively” part: They would show exactly what the cop saw and heard, and the jury could then decide whether the action was reasonable from the cop’s vantage point. The footage doesn’t do that here, and I hope improvements to the technology can make body cameras better witnesses in the future — less jerking, more coverage of the surrounding area, better sensitivity in low light.

A few extra thoughts. First, in some cases the “objectively reasonable” standard does mean acquitting cops who shoot unarmed people, particularly if the suspect failed to follow instructions and made movements that suggested he was drawing or wielding a gun. There’s a good reason for this: If the suspect is in fact armed and intent on harming the cop, the officer will be dead before he can positively identify a firearm, simply because it’s easy to produce a weapon and discharge it quickly. By a similar token, French defended the shooting of a man who refused to follow police instructions and reached into his car, even though it turned out there was no gun in the vehicle.

Still, I’m sympathetic to French’s argument that cops should, out of a duty to minimize shootings, take the risk in some of these cases even when legal justification is present — though this is an easier point to make when someone is confused and makes an innocent mistake than when he’s broken the window to a home and flees when the cops approach him. And that brings us to Bayesian statistics, which believe it or not seems to be the concept at the heart of the French vs. Dunphy debate.

The 18th-century English statistician Thomas Bayes is famous for his eponymous theorem explaining how to integrate new information into one’s prior beliefs. Here, the key question is how strong those prior beliefs should be relative to the new information.

As French notes, police officers are very rarely shot — this is the “background level of risk” he says should have informed the officers’ decision. It’s clearly the relevant fact for determining the risk that a randomly selected officer faces on a randomly selected day. But what are the chances of being shot conditional on the fact that you’re chasing a suspect who was seen breaking into a home and ran when you confronted him? Conditional also on the fact that he turned around to walk toward you? Conditional also on the fact that he’s holding something that (apparently, given the cops’ reaction on the videos) seems to be a gun? These are very powerful pieces of new information, powerful enough to drive that background information awfully far into the background.

And this, in turn, puts us back where we started: Did Clark really behave in a way that strongly suggested he had a gun, and/or did his phone look like one from the cops’ vantage point? If so, we should evaluate the officers’ actions the way we would if he’d actually been armed, because the cops had no way of telling the difference. In that case, I think their actions were plainly legally justified and arguably the right course of action subjectively (given what they knew) as well. If not, not.

And we don’t have footage that can tell us which is true, despite both officers’ having worn body cameras.