Position
Where plate reader data should live
This page is an argument, not a finding. Everything else on this site is sourced and dated. This is my judgment about where the technology is going and where it ought to go.
Opinion · Reviewed 26 July 2026
The problem with the current model
Nearly every commercial plate reader deployment in the country works the same way. Cameras belong to the agency. The data goes to the vendor. Searching happens in the vendor's system, against a pool that may include scans from thousands of other agencies, governed by settings that live in a dashboard.
That design has a specific weakness, and it is not storage. It is that no single agency can see the whole picture of who is querying its data, and the audit tooling belongs to the party with the least incentive to surface a problem. Several cities have now discovered, through audits rather than notification, that outside agencies were searching their scans under settings nobody at the city had knowingly enabled.
Note what that failure is not. It is not a data ownership problem in the contractual sense. Those cities owned their data the entire time. Ownership on paper did not help, because control and visibility sat somewhere else.
What a federated model looks like
Scans stay on infrastructure the collecting agency controls. There is no central pool. A federation operator maintains the directory, the identity layer, and the query routing, but holds no plate data of its own.
When Agency A needs records from Agency B, the query goes through the operator, gets checked against a standing agreement between those two agencies, and is executed against Agency B's system. Agency B's log records who asked, when, under what authority, and what was returned. If there is no agreement, there is no query.
This is not novel. Nlets has brokered state to state criminal justice and motor vehicle queries this way for decades. The pattern is proven. It simply has not been the commercial default in this particular market.
What it fixes and what it does not
It fixes access governance, structurally rather than by policy. A sharing setting cannot be quietly flipped on for data that was never pooled in the first place, and the audit record lives with the agency that has every reason to keep it accurate.
It does not fix hotlist integrity. A bad entry is a bad entry regardless of architecture. A recovered vehicle nobody cleared will produce the same wrongful stop in a federated system as in a pooled one. That failure is a records discipline problem and it needs a records discipline answer.
Anyone selling federation as a complete solution is overselling. It solves one of the two big problems in this field, and it solves that one well.
Why it has not won
Infrastructure burden, and this objection is serious. The centralized model won because the agency does not have to do anything. No server, no database administrator, no uptime obligation, no storage bill. Federation asks a forty officer department to run and maintain a queryable node, and most cannot.
There are also real operational costs. Query latency is higher. A node that is down returns nothing, which feels worse to a working officer than a slightly stale central result.
Those are honest tradeoffs and I am not going to pretend otherwise.
What changed
The law started moving toward requirements that a pooled model cannot satisfy cleanly. Proposals now in circulation would require that no default access be granted to any national database, that scans be inaccessible to other agencies unless deliberately shared, and that every search carry a case number or task force name.
Read those requirements without the labels and they describe a federated architecture. Not because anyone drafted them that way, but because those are the properties federation has natively and a central pool has to simulate.
The infrastructure burden has not gone away. What has changed is that compliance is now a reason to carry it, which it was not two years ago. Shared regional nodes among neighboring agencies are one way to spread that cost, and county or state level hosting is another.
Disclosure
I have no financial interest in this outcome. I do not sell a federated platform, hold a stake in one, or receive anything if an agency moves in this direction. This is a position, published because the question keeps coming up in council meetings and nobody in the room is offering a third answer between keep it as it is and shut it off.