I recently watched a YouTube short about police officers pulling over a car based on a misreading from their surveillance technology:

https://youtube.com/shorts/9mP1Ssp1jnc?is=T8AYF8yP8Cu0GsG6

When I was in school I learned about the classic “off-by-one errors”. It was a lot more common in lower level programming. An example found in Wikipedia’s page is the “fencepost error”

https://en.wikipedia.org/wiki/Off-by-one_error

“A fence requires a post every 3 feet. You have a 12 feet fence. How many posts do you need?”

--- --- --- ---

At first glance most people may guess 4 posts by dividing 12 by 3, the result being a fence looking like the following:

|---|---|---|---

The actual answer should be 5 so you can support the end:

|---|---|---|---|

Programs still contain some rendition of the above bug. Modern programming languages try to avoid the issue by designing with this classic problem in mind but programmers can still hit the issue.

The Flock incident from the YouTube clip is one version of the bug. The police was looking for a car with a license plate “APX 55Y” and they pulled over a car with a plate “APX 55X” after their Flock camera misread the car. If you didn’t see the video: The driver was handcuffed, as was the owner of the car, and they had her 1 month old in the vehicle as well. One of the officers drew their gun and pointed it toward the vehicle during the arrest.

There were multiple failing points here:

  1. The Flock camera’s OCR, or optical character recognition, technology misread the plate
  2. The police officers in the vehicle failed to double check the plate and assumed the Flock camera reading was correct

The car occupants became victims of an off-by-one error. This incident alone should have led to the cancellation of the municipal contract for Flock, and severely discredit them from other contracts.

If that’s not enough to convince you that Flock and similar surveillance technology is bad news, check out 404media’s recent podcast that interviewed a woman who was jailed for 13 days based on bad Flock data:

https://www.youtube.com/watch?v=Xc_6Xap8Yy4

Flock’s main technology is ALPR, or Automatic License Plate Readers, and aggregates the data they gather to serve to their clientele which is law enforcement. If you are curious what ALPRs are around your neighborhood you can check out this map:

https://maps.deflock.org

Finally, if you see cameras where you live or hear about your local government trying to pick up ALPR contracts, I strongly encourage you to attend the next meeting and submitting public comment on this issue using the above as examples of what has gone wrong and what can go wrong.