That's not a Flock error. That's a law enforcement error. Flock isn't responsible for a bogus warrant in the system.
The woman probably has a good case for damages from the police department for doing it twice. That's sloppy police work.
The woman probably has a good case for damages from the police department for doing it twice. That's sloppy police work.
Sometimes that means a tradeoff in terms of efficiency, cost, or something else. Often this tradeoff is not one that people in charge of the system are willing to make, unless they are forced to do so by a larger institution or force.
To really drive this home:
1. What steps did Flock Safety take to ensure this warrant was valid and current, and that the plate number was correctly entered?
2. What steps did the police officers take to validate the warrant against the APB hit from the camera?
3. What process is in place to ensure incorrect warrants are not entered or mistakenly kept live for longer than intended?
My guess is that the answer to all three of these is basically "none".
I’ve seen too many cases where people intuitively trust the magic technology instead of verifying it. The tool shouldn’t not present itself as magic or AI or whatever the current term is.
2. The vehicle was correctly identified by Flock as being the one flagged, and verified by the local PD that it was the one that was flagged. That said, they could have verified that the flag was still meaningful before converging on the vehicle.
3. This is a real problem and this is really the core of the oversight.
When the lawsuit over this is filed, it seems like 80% of the liability should attach to #3, and 20% to #2, and essentially none to #1. It's hard for me to see why Flock is even relevant to this discussion except to drum up outrage.
Furthermore, it's a new evolution of technology. Sometimes quantity has a quality all its own.
It was one thing when tracing a call or tapping a line required a technician to run around the electromechanical switch with a volt meter or a human to listen to your calls 24/7. It's an entirely different thing when it can be done automatically and cheaply at scale.
Before, you could hope to hide in the noise, and unless you were doing something "really bad" (or you had powerful enemies), you probably wouldn't be targeted. We're on the precipice of being technically able to have cheap, total enforcement of most laws.
Should we actually do that though? Will the societal changes be broadly good or bad? That's a question we should be asking, not just jumping in with both feet and pretending that the technology is neutral in the outcome.
For example - the introduction of radios in police cars had a huge impact on public safety and played a huge part in bringing an end to the era of the public enemies. No longer did you just need to outrun the posse chasing you and get out of town for a few weeks until the heat was off. Now you needed to slip out of the dragnet being coordinated over radio around you. That seems like it was probably broadly a good thing ... but even then, it required cost and human risk to accomplish. As we move further into the era of AI policing, we'll leave all of that behind.
(Actually yeah. Chicago Police uses an on-prem 3rd party blackbox fax service to avoid those records being FOIA'able. But that's beside the point.)
Qualified immunity makes accountability a distant mirage.
The police operate exactly as they are allowed and encouraged, by the system they labor within, to behave.
Your words 'sloppy' and 'error' imply that a 'careful, correct' approach fixes these ills. History and experience shows us that this hypothetical good will not be produced until laws and incentives more broadly, change.
Product, technology, implementation, police department, policing method, workflow. These aren't discreet. Incidents involve all of them.
Responsibility is on people.Technologies cannot be "found guilty." But... this isnt' a trial. There are consequences to doing policing any particular way, and that's the subject.
Doing it this way... using baby skynet for command and control has advantages, like response time. It has disadvantages, like false positives. "Law enforcement error" is technically correct, but not informative about the matter at hand. The error
Saying that the responsibility is only on the user is extremely myopic. It's also dismissing the role Flock has in how they make and design their software.
If they can't handle this maybe they should be disbanded as a company or nationalized for better public control.
I am not a fan of flock, nor police surveillance state, but what could they have done better here?
"Milwaukee police later admitted the alert should have been cleared out of the system days earlier, but an employee simply forgot to remove it. "
The system seemed to work as intended - that numberplate was flagged for investigation with homicide - and gave the alarm as intented, nothing more. The problem was wrong data given to them by humans.
The human in the loop did not properly delete an invalid license plate number from one of the databases the flock system checks against.
The issue issues are with:
1. Human in the loop
2. Third party database police use for homicide info
3. Police response
I don't see what the license plate reading technology has to do with any of this. It seems like the only thing working reliably.
"Flock amplifies existing bugs. Sounds like Flock is the error."
We'd need to take a look at what they have access to and how they're ingesting that info. I'd start with caring greatly about false positives and not automating department onboarding.
Working with various counties I've seen situations where someone needs to tell you a specific table dump exists to get an accurate count because that's how they published deletions back in 1983. There are no standards and the people who created these systems are all dead.
> The problem was wrong data given to them by humans.
This is always a problem. It's never not a problem. The question is how Flock chooses to address the problem.
Flock should stop existing. I hope people help them with that, enthusiastically.
There's a lot of money in cyberstalking and real stalking. see: Facebook's valuation
There are plenty of ways if you are creative enough.
Somebody could start putting Flock executive license plates in with erroneous police data.
Which means they wouldn't have flagged the car and pulled it over.
A couple of days go by and the error is out of the system.
Not exist. That is the only societally acceptable answer.
Saying there is nothing we can do is pathetic behavior and just shows that maybe you (nor Flock) should be in charge of these things if these are the outcomes they find acceptable.
If you don't know how to design a system to prevent this, DON'T MAKE IT!
Good grief our industry is absolutely pathetic with these excuses.
Just because you have zero standards or ethics does not make this acceptable at all.
This was a purely created system that was purposely designed exactly for these outcomes. You do not have to create such a system. There can be more accountability.
Have standards please. This is impacting real human lives here because you refuse to have standards.
If your point is that pulling her over at gunpoint is a flawed procedure that's valid. But that has nothing to do with anything Flock cameras are doing.
"Flock amplifies existing bugs. Sounds like Flock is the error."
What should the cops have done? Ignored it?
I'd argue Flock making that incorrect association is similar to a false accusation and the consequences should be on Flock.
There's enough blame to go around to hit Flock and the PD both.