The video below comes from the American Civil Liberties Union’s Jay Stanley. It shows the footage from two cameras as five deputies from Marion County, Fla., beat suspected drug dealer Derrick Price in 2014. One video is from an officer’s body camera. You hear the officers yell “Stop resisting!” over the course of the beating. But you don’t see much. The other video comes from a security camera with a much wider view. Here you see that there was no “resisting” to stop. Price had completely given himself up. They beat him anyway.

The difference in the two cameras’ footage is all the difference in the world. As Stanley points out, the body camera was also turned on much too late, and turned off way too early. The police officers have since faced criminal consequences for the beating, but the video was still kept secret for more than 16 months. The officers also lied in their police reports, and their cries of “Stop resisting!” to a clearly submissive suspect are good evidence that the problems in this police department go much deeper than these cops and this incident.

Again from Stanley:

Price was very lucky, and the officers unlucky, that the arrest took place in clear view of a surveillance camera. If the bodycam footage was the only evidence, the deputies’ subsequent lies about Price’s alleged violent resistance to his arrest would almost certainly have been believed.

At the same time, for those skeptical of body cameras it’s worth noting that he would likely have been no worse off with only the bodycamera video than he would have been with no video at all. Or, if every officer at the scene wore a camera, the situation would likely have also been clear — or, with the deputies knowing they were being recorded, wouldn’t have happened at all.

But it did, and the divergence between the two video streams that captured the arrest serves as an important lesson and reminder of the limits of video evidence, as well as the need for skepticism toward police accounts of what such videos portray.

At the same time, this post from author and former cop Peter Moskos includes dash camera footage that shows how quickly a confrontation with a suspect can go awry — a demonstration that those of us who take a more skeptical eye toward when and how police use force should keep in mind. As Moskos points out, if you stop the video just before the suspect lunges at the officer, most of us would have objected if, at that point, the officer had used a Taser on the suspect or taken him to the ground. As Moskos notes, there are other things the officer could have done to protect himself. And the data show that the incidence of violence against police officers during traffic stops isn’t remotely commensurate with the way the threat is played up in police culture. But this is also an example of how a transparency tool — in this case a dash cam — can help the rest of us understand these incidents from the perspective of police officers. That this guy attacked this cop isn’t an excuse to use a Taser on every motorist who talks back or asks questions. Instead, it should be instructive in looking for policies and tactics that protect officers from incidents like this without subjecting non-threatening motorists to unnecessary force.

One of the interesting side debates about body cameras is whether police officers should be allowed to view the footage of a shooting or use-of-force incident before issuing a report. On the one hand, we don’t want cops bending their narrative to fit the footage. The Price beating is a good example. The police narrative fit the footage from the body camera. To many, the body camera likely made the false narrative more credible. And that’s worse than no camera at all.

On the other hand, if we aren’t going to allow cops to view footage before issuing a report, such a policy should come with some appreciation for the fallibility of human memory. Those of us who are critical of the way eyewitness testimony is used in court, for example, know that memories can evolve with the passage of time, that they’re highly susceptible to suggestion and that we tend to remember events in a light that favors us. There should be some allowance for this. Obviously, cops caught swearing to a story that is clearly and incontrovertibly contradicted by video should be held accountable. But there should also be protections from small discrepancies and room for slight deviations from the video, even when those discrepancies favor the police officer, and even when those discrepancies have some legal consequence.

But there may be a better solution. A couple of weeks ago, I spoke to Seth Stoughton about a related story I’m writing. Stoughton is a former police officer and now a law professor at the University of South Carolina. His recommended policy is to ask police officers to write up an initial report before viewing body camera footage, and then to write up a follow-up report after viewing it. The follow-up report would acknowledge and attempt to explain any discrepancies between the video and the initial report. Both reports would be available to the defense (or to the plaintiff in a lawsuit). Police officers would have to account for all discrepancies on the witness stand, but would only be on the hook for disciplinary action or perjury charges for the second report.

This seems like a sensible compromise, though I’d be interested to hear what others think of it. It allows for flaws in human memory, but it also discourages outright lying to conform to the video. It’s a policy that doesn’t aim to “catch” cops with a proclivity to lie so much as one that provides strong incentives for telling the truth. Of course, this policy would need to be accompanied by other appropriate investigative procedures, such as separating officers immediately after the incident in question, and getting those statements as quickly as possible. These procedures, alas, aren’t always followed, and in some places are actually barred by union rules and “police bill of rights” laws.

There’s little question that police body cameras and dashboard cameras offer added transparency. They offer an independent narrative of disputed incidents. They can aid police training, to help prevent incidents like the one in the Moskos post. They can bring accountability, provided that they’re implemented with the appropriate policies and rules to govern their use. But they aren’t the be-all, end-all solution. They aren’t irrefutable. And they can still be used to deceive, both by police and by the policed.