Why we exist

Nobody had counted.

Automated license plate readers have been installed across American communities for more than a decade. Cameras went up on poles, at intersections, on the approaches to bridges. Contracts were signed, sometimes voted on and sometimes not. Data was collected, retained, and shared.

And when you ask a straightforward question — how many are there, where are they, who operates them — no one can answer.

Not the sheriff's office, which in one Florida county of nearly 800,000 people said it owns none of the cameras it has access to and could not say how many exist. Not the cities, which often did not purchase them. Not the vendors, who have no obligation to say. Not the federal government, which does not require a registry. The information exists in fragments, in procurement records and volunteer datasets and local reporting, and nobody had put it together.

So we did.

The record now holds 339,136 surveillance devices across 156 countries — every plate reader, camera and gunshot detector documented in open data. It is the largest public accounting we are aware of, and it should not be. This is the kind of thing a government keeps, or a university, or an industry regulator. It took a volunteer effort and a database, and the fact that it did is itself the finding.

Three things stand out

Plate readers concentrate in one country. Roughly 96 percent of documented plate readers worldwide are in the United States. In America, 72 percent of documented surveillance devices are vehicle trackers. Everywhere else the figure is 2.5 percent. Other countries watch places. This country watches cars.

Most devices have no recorded owner. About 75 percent carry no operator in any public source — not an agency, not a company. Nobody can tell you who runs them.

One manufacturer dominates. Among devices with a named maker, Flock Safety accounts for 82 percent of the record.

We do not tell you what to conclude from that. Every figure on this site is computed from the record when the page loads, and every one can be checked down to your own county. If you think we have it wrong, the map is right there, and the registry lists every row.

What we are not doing

We are not campaigning against plate recognition. Law enforcement has real uses for it and we say so plainly in The Two Systems. A stolen vehicle, an Amber Alert, a specific car wanted in a specific crime — those are legitimate, and none of them requires a record of people who are not suspected of anything.

What concerns us is the retained database: the continuous logging of ordinary movement, kept and searchable afterward, in case some of it matters later. That is a design decision, not a technical necessity, and it can be made differently.

What we are doing instead

Publishing the record, so the scale is checkable rather than asserted.

Publishing an open standard for plate recognition that keeps the legitimate uses and drops the retention. Free to implement, free for any jurisdiction to require.

Building a camera network that points the other way — at government facilities rather than residential streets. Because in a record of 339,136 devices watching the public, almost none is watching the institutions.