Funny you should mention this. Benn Jordan, on the heels of his Flock camera research, reverse engineered a non-flock ALRP unit and built his own system using some off the shelf parts, a tablet, and 3D printer. He did it mostly as an experiment but also due to the fact that Flock's algorithm for image detection is astoundingly bad and has a high incident of reading plates incorrectly.

There are a handful of open-source models for license plate detection, I forget exactly which model outperformed the rest, but it was an excellent watch and help me really understand just how inefficient these commercial systems are and how easy they can be to defeat: https://www.youtube.com/watch?v=Pp9MwZkHiMQ

Everyone's got an old Android phone or two sitting in their junk drawer, right? Place it in your window and connect it to the shared, open, public service. Might work, but then if it's open and shared, Flock can simply connect to the open API and add a new category "Public cameras" to their own data.

Any data we make available as an open system will also be available to bad actors.

I’ve toyed with the concept

We should. And not just cars related to law enforcement, nor also the cars of public officials who give cops their marching orders and enact laws, nor also the judges who enforce these laws.

Instead, let's just be indiscriminate and document everyone's cars and make that raw data available.

---

Why everyone? That's easy: It allows organizational roles to be divided up, and dividing these roles promotes operational safety.

Having 4 roles seems like it might be the right number right now (but I haven't had my coffee yet):

Role 1. The camera operators. This is role sees the highest direct risk because these cameras will be associated with real homes and other buildings that the operators control. It is also the most important group because it requires the highest number of participants, and none of this can happen without a large number of them. By recording everyone, they gain some operational safety through plausible deniability. "Yeah, I've got some cameras that see cars on the road and send what they see to a cloud service. So what? I'm allowed to pay attention to the cars in my neighborhood. I'm even allowed to get help with doing that. Everyone else with a fancy doorbell or a Chinese web cam is doing the same thing; I'm not doing anything weirder than what anyone else is doing."

Role 2. The data-mungers. This back-end role collects the data of where cars have been seen. It's also an indiscriminate task; it just collects and sorts data by license plate number. This is less-risky both because it is unfocused and it can happen anywhere on the globe. "Yeah, so these blokes send me a stream of alphanumeric numbers, timestamps, and locations, and I just organize that data for other people to use."

Role 3. The filtermakers. Another back-end role, this one just keeps track of which plates are associated with which government people. This is riskier: It is tightly focused and its role is obvious and undeniable. "I keep a list of license plates numbers and names for others to use. Lots of organizations do that; so what?"

Role 4. The mapmakers. This role operates the presentation layer. It ingests someone else's collected data, filters and labels it based on someone else's filters and labels, and puts it all on a beautiful map for public consumption. This isn't necessarily structurally the most important role, but it's the most public-facing role and likely to be demonized in media. This role will get heat. "Yeah, you're right. Mapping the locations of people in public in the US is exactly the point of what I'm doing; see the FAQ on the website. Anyway, the weather is beautiful here today in Belize; maybe y'all should get to work on your public surveillance legislation."