Note well — I use the term badge-cam for a specific purpose. Today, the symbol of authority for a police officer is his uniform, badge and weapon.
By using the term badge-cam, we change the frame such that a police officer’s authority is derived from his badge, camera, uniform and weapon. That is, his authority and legitimacy should be partly derived from the constraint of being monitored.
I was discussing on reddit some of the issues with police badge-cams.
My interlocutor was demanding 100% 24/7 recording by the cams, and saying that police activated cams were not good enough because police can elect not to activate them when doing misdeeds.
We sat down and calculated what 100% recording for every patrol officer would cost.
The cost we arrived at was around $0.5billion a year for video storage, and $25million for the cameras.
See below for the calculations.
$0.5billion/year is a probably way too much.
I want to propose a way that the amount of recorded video can be reduced substantially, while still capturing the relevant moments.
I propose that the cameras can be activated by the officer AND by the public AND at random times.
Voice recognition chips are relatively inexpensive, and the camera could be activated by the public saying “Start Recording” or some other command. The voice recognition could be tuned to accept a high false-positive rate — its not important if the camera is sometimes mistakenly activated. And, as with current badge-cams, when activated, the X minutes before activation are also stored.
The voice recognition could also be programmed for a number of other sounds: “Freeze”, “Stop Resisting”, <the sound of shots fired>, <the sound of a baton being extended>, <the sound of a gun being drawn from its holster>, <the sound of pepper spray>, <the sound of a taser>, <the sound of raised or angry voices>, and so on and so forth.
Once activated, the badge-cam would signal to the public that it has activated (perhaps by a flashing LED), but not to the officer.
Having the badge-cam also activate at random times, without the officer knowing, would ensure that the officer must behave as if they are being recorded at all times.
This approach also deals with the problem of pervasive surveillance, by ensuring that the recording is NOT activated all the time.
The Calculations
There are an estimated 1 million law enforcement officers in the US. Estimate that half of those are patrol officers, and that half of those are on patrol at any given time. Thats 250K cams recording 24/7.
http://en.wikipedia.org/wiki/Law_enforcement_in_the_United_States
The cost of the badge-cams is relatively small. A badge-cam capable of recording a 12 hour shift should cost not much more than a low end smartphone or about $100.
Cost to equip every active patrol officer with a $100 badge-cam, $25million.
The real cost comes in storing and managing the recorded video.
Well compressed 1920x1080 video takes up about 1GB/hr.
Cloud storage costs $0.02/GB/month.
250K cameras * 24hours * 365days * 0.02/GB/mo * 12months = $525mil/year
So to store full time recording from every patrol officer in the US would cost around $0.5billion per year.
