How this connects
The layers on this map are usually discussed separately — a camera story here, a sheriff story there, a data-centre zoning fight in the local paper. They are not separate. They are stages of one pipeline, running on ground that was shaped long before any of it was built.
Surveillance
A licence plate reader photographs every vehicle that passes, converts the plate to text, and stores it with a timestamp and a location. One camera is a record. 1,426 of them mapped across Minnesota is a network — enough to reconstruct where a vehicle has been over weeks, not just whether it passed one corner.
Search and sharing
What turns a camera network into an enforcement tool is who can query it. Many ALPR networks let outside agencies run searches across jurisdictions, and audits in several states have found immigration-related searches run against networks that local residents believed were purely local. The camera does not decide who asks it questions.
Enforcement
A 287(g) agreement is where federal immigration authority is delegated to local officers. Minnesota has 10 on ICE's current list, out of 2,179 nationally. The model matters enormously: a Jail Enforcement agreement acts on people already booked into a county jail, while a Task Force agreement pushes immigration enforcement out into ordinary street-level policing — the same streets the cameras watch.
Detention
Enforcement needs somewhere to put people, and that is usually a county jail renting bed space to ICE. 5 Minnesota facilities hold people under ICE contract, every one of them an intergovernmental service agreement with a local jail. This is where the pipeline stops being abstract: a local building, a local budget, a local sheriff.
Adjudication
Immigration court sits at the end of the chain, and it is the part this map deliberately does not render. Court outcomes are made of individual cases, and individual cases are people. Aggregate, de-identified statistics published by researchers such as TRAC describe the system without describing anyone in it — that is the only form this project would ever carry.
The substrate underneath
Two layers on this map are not stages in that pipeline. They are the ground it runs on.
Redlining came first. In the 1930s the federal Home Owners' Loan Corporation graded neighbourhoods for mortgage risk, and the grade turned explicitly on the race, ethnicity and immigration status of the people who lived there. Areas graded D were outlined in red and starved of lending for decades. 168 of those graded areas are on this map, covering the eight Minnesota cities HOLC surveyed. The resulting patterns in wealth, housing, and where police concentrate did not dissolve when the practice was outlawed.
Data centres came last. 20 of them are mapped in Minnesota. They are the physical machinery the rest of this runs on — the storage behind plate-reader networks, records systems, and the analytics products sold to agencies — and they arrive with immediate local consequences of their own: electricity and water demand, tax abatements, and grid costs pushed onto other ratepayers.
Where the layers converge
This is not a hypothetical. 3 Minnesota counties have both a 287(g) agreement with ICE and a jail holding people under ICE contract — the enforcement and detention stages in the same jurisdiction, under the same sheriff, watched by the same cameras.
Sherburne County
- Mapped cameras
- 31
- 287(g) agreement
- Jail Enforcement Model
- ICE-contract jail
- Sherburne County Jail
Kandiyohi County
- Mapped cameras
- 19
- 287(g) agreement
- Warrant Service Officer
- ICE-contract jail
- Kandiyohi County Jail
Freeborn County
- Mapped cameras
- 0
- 287(g) agreement
- Warrant Service Officer
- ICE-contract jail
- Freeborn County Adult Detention Center
Computed from the shipped datasets at build time, joined on Census county identifiers. The camera counts carry the usual caveat: crowd-sourced, incomplete, and a floor rather than a total.
Why look at them together
Because the questions only make sense together. Whether a plate reader on your corner matters depends on who can search it. Whether that search matters depends on whether your sheriff has signed an agreement, and which model. Whether that agreement matters depends on whether there is a jail nearby holding people under contract. And which neighbourhoods sit at the sharp end of all of it is not random — it tracks lines drawn on a map in 1937.
Every one of those facts is public. Held apart, each is a curiosity. Held together, they describe a system, and a system is something a community can actually organise against.
A caution about this page. This describes how these systems connect in general, using documented mechanisms. It is not a claim that any specific camera on this map has fed any specific enforcement action, and nothing here should be read as tracking or predicting what happens to any individual. Check thesources page for what each dataset does and does not establish.