Trying to add a payment method? You can do this on the 
Flock’s automated license-plate recognition system is the wrong answer to a genuine problem.
The system provides tangible benefits that help law enforcement officers do their job. The ability to track cars fleeing the scene of a homicide, shooting, rape or other violent offense makes Eugene safer. That’s important.
What’s also important is protecting our community from overreach. There needs to be strict protection to make it impossible for agencies not endorsed by our community to access the data.
Flock fails this second goal, both theoretically and in practice. The data gathered by Flock cameras are stored on Flock servers and processed by Flock algorithms. This gives the company complete technical control over the data.
With something this serious, contractual restrictions are insufficient. We require technical control locally. It’s entirely believable that Flock would sell surveillance of our community to U.S. Immigration and Customs Enforcement behind our backs, and simply take the risk that we’ll challenge them in court or end the contract if we find out.
It’s also worth noting that the risks of lawsuits aren’t only borne by Flock. Police departments installing this system could be liable if it’s misused. It’s a financial risk I’d rather we not take.
In practice, we see evidence of Flock systems being misused. In Illinois, access to Flock was given to U.S Customs and Border Protection despite it being illegal under state law. The Texas sheriff who searched for a woman who’d left the state to receive an abortion is an example. In Oakland, data was shared with seven federal agencies, including ICE, against state law.
The demonstrable failure of this system to meet the criteria for privacy and local control means we should cancel and remove Flock locally. However, there is a license-plate reading system that meets the criteria of both making our community safer and protecting residents from law enforcement overreach.
With the money we would have spent on Flock in fees and subscriptions, we could contract with the University of Oregon to build a privacy-first license-plate reader. The system I propose would provide our local police with 90 percent of the enforcement benefit and only 10 percent of the privacy risk.
The first thing to know is that license-plate reading is not complicated technology. The cameras send pictures to a computer, which uses image processing to first identify cars, then identify license plates on those cars, and finally read the text on the plate. The computer can then store the license plate number, as well as the date, time, speed and location of the vehicle in a database. There are multiple open-source software libraries that perform this function.
Arguably the most popular of these libraries is OpenALPR (which is free, open source and accurate).
The system as a whole isn’t complicated either. It’s only four simple steps: camera, image reading, storage, access. It’s entirely feasible for Eugene and Springfield to contract with UO to build a system that meets our privacy needs.
The images from the cameras would be encrypted and sent to a secure computer for processing. The computer would process the images without using external resources, thereby protecting the data. It would then store the plate number, location, speed, date and time locally and securely. It could be set to wipe data older than a certain number of days, and we would know the data was destroyed because we would control the server and database. Lastly, the software to access the data could be built with extensive safeguards.
It could require entering the case number of a serious crime to make the request. Every request could be logged and audited. Our civilian review board could perform these audits to ensure data being requested relates to the case under which the request was made.
Limited access to real-time data could be granted in the event of an active situation. This access could be tied to the 911 call for the purpose of auditing it.
The data wouldn’t be accessible to outside agencies, but we could still flag vehicles associated with serious cases in the region. This list of vehicles could be audited as well.
The emphasis is this: With Flock, we don’t have control. With a homemade system, we do. This option is technologically feasible, and better than Flock.
