From an Associated Press report:

In a warning to law enforcement agencies rushing to equip officers with body cameras after killings by police nationwide, a new report says the devices used by Denver officers during a trial period didn’t record most of the use-of-force incidents that occurred.

Denver’s independent police monitor, Nicholas Mitchell, also said police used force more often and citizens’ complaints against officers rose during the cameras’ six-month trial period in the city’s busy downtown district. Police officials repeatedly said they expected the cameras would drive down those numbers.

Experts say the early findings released Tuesday are a reminder that the effectiveness of the increasingly popular technology, billed as a tool to improve police accountability, still depends on the officers using it.

That’s sort of true. It’s more dependent on the policies implemented along with the body cameras. But while the experiment in Denver appears to have had some flaws, I’m not convinced it was the catastrophic failure that the AP article implies. The story first states that according to the Inspector General’s report, the cameras recorded “just 21 of 80 documented uses of force in the downtown district during the trial.”

That seems bad. But then we find that 35 of the 80 cases involved officers working off-duty as security guards, when they aren’t required to wear the cameras. Now, you could argue that it’s a bad idea to let police officers wear their uniforms when working as private security. There are also lots of interesting questions you might ask about why nearly half the use-of-force cases over the six-month period covered by the report involved cops working off-duty, instead of on-duty. But this isn’t really a problem you can attribute to the use of police cameras.

Cutting out those incidents still leaves leaves us with just 21 of 45 incidents that were captured with body cameras. That’s still a troubling statistic. Here’s the most damning passage:

Officers were expected to activate their cameras during a broad range of encounters, including traffic stops and responses to 911 calls. In many cases, officers said situations deteriorated too quickly for them to safely activate their cameras. But Mitchell found that officers often failed to follow policies requiring them to turn on the cameras before initiating an encounter.

Other times, the cameras shifted and were obstructed by officers’ clothes, the batteries died or they shut off in the middle of a scuffle.

The “forgetting” part is pretty easily remedied. You punish officers who don’t turn the cameras on when they’re supposed to. If the police department refuses to do so, a legislative body could pass a “missing video presumption” that instructs courts to interpret cases in a light favorable to those on the receiving end of police force if video should be available but isn’t. (I suspect this will also help officers and departments ensure that cameras have fresh batteries and that they aren’t prone to somehow shutting off at critical times.) There isn’t a whole lot to be done about the occasional obstructed view, though even audio only from a disputed incident is better than no independent narrative at all.

Later in the article, the picture of what actually happened in Denver gets even murkier.

Cmdr. Magen Dodge, who oversees Denver’s body camera program, said Mitchell’s numbers were skewed because he looked beyond the scope of the pilot program. She said officers involved in the pilot used force 53 times, and there was at least some footage associated with 46 of those cases.

If this is true, why lead an article about the effectiveness of the program with figures that include incidents that weren’t actually part of the program? I mean, 46 of 53 is quite a bit more encouraging than 21 of 80.

Of course “some footage” doesn’t necessarily mean good footage. But if you really want to evaluate the pilot program, you should start with those 53 incidents that were actually covered by it, then ask questions from there. What happened in those seven cases that didn’t produce footage? How many of the other 46 produced relevant footage? Of those that didn’t, why didn’t they?

The article contrasts the pilot program in Denver with the well-publicized program in Rialto, Calif., in which use-of-force incidents and complaints about police misconduct dropped dramatically. But there’s a good reason for that, too.

That city’s success could be due in part to policies requiring officers to tell citizens they are being recorded, Mitchell said.

Well, yes. That’s pretty important. Not only that, but in Rialto, every patrol officer was outfitted with a camera. The Denver program included just 125 of the city’s 1,400 cops. I’m not sure why you’d expect to see significant, citywide results when using such a small percentage of the police force over such a relatively short period of time. You certainly shouldn’t expect to get results that compare to a program that included the entire police department.

If the goal of body cameras is to deter police misconduct, reduce use-of-force incidents, discourage citizen misconduct and generally encourage both police and citizens to behave better when the two interact, there are two primary things you need to do:

1) Let citizens know that they’re being recorded when they interact with police. This seems pretty self-explanatory. If you expect cameras to improve behavior, you first need to let people know about the cameras. Why this wasn’t part of the Denver program is rather perplexing.

2) Require police to actually use the cameras and impose real consequences when they don’t. Real transparency will encourage best practices. “Transparency” that can be switched on and off at the discretion of a police officer is useless at best and may actually benefit those officers already prone to misconduct. It certainly won’t foster any trust with the public.

Even when accompanied by ideal policies, police cameras aren’t a panacea, and that shouldn’t be the standard for gauging their success. (There are also some issues with privacy, storage and security that need careful attention.) The Rialto results could well be a fluke. But even a fraction of a 50 percent reduction in use-of-force incidents or an 89 percent reduction in citizen complaints would be well worth pursuing. When it comes to government, more transparency is almost always better than less. This is especially true when we’re talking about the government employees entrusted to use force against us, up to and including lethal force.

It isn’t at all surprising that implementing police cameras with poorly written policies governing their use would produce less than spectacular results. It also doesn’t tell us much of anything about how cameras will work when accompanied by policies that actually make some sense.