Digital Evidence Collection and Analysis by Law Enforcement is moving faster than the Law

[
[
[

]
]
]

On an Oklahoma highway on an ordinary afternoon, a sheriff’s deputy typed a license plate into a database and pulled up roughly a month of a stranger’s movements across several states. A few minutes later he had a reason to pull her over. In the car: bundles of twenty- and hundred-dollar bills, and — by the government’s own account — around forty-five kilograms of methamphetamine, packed in a Betty Boop suitcase lined with dryer sheets to throw off a drug dog.

On October 1, a federal judge threw all this seized evidence out.

Hold that result in your head, because I want to set it next to something the FBI does hundreds of thousands of times, and ask why the two come out so differently. The honest answer is the strangest asymmetry in American surveillance law: right now, you are better protected from a county deputy’s license-plate lookup than from the federal government’s search of its vast database of your communications. The bigger and more intimate the surveillance, the less the law seems to constrain the search. That is exactly backwards, and a case out of the Northern District of Oklahoma just made it impossible to ignore.

What happened in Kyle. The deputy — a drug-interdiction officer — followed a rental car because it had an out-of-state plate. His patrol car feeds Automated License Plate Reader networks (Flock and VehicleManager are the two named), which log plates with time and location and let an officer pull a vehicle’s recent travel history across many states. He ran the plate, saw a quick there-and-back trip toward the Southwest border, and developed a hunch. He stopped the car for following too closely, wrote a warning, returned the license — and detained the driver for a dog sniff anyway.

For the record, this was a serious federal trafficking case: Kyle was initially charged by complaint then indicted for possession with intent to distribute 500 grams or more of methamphetamine — the top federal drug tier, under 21 U.S.C. § 841(b)(1)(A)(viii) — with the complaint putting the actual haul at roughly forty-five kilograms, alongside about $7,000 in cash the government moved to forfeit. This is not a case the government wanted to lose.

Judge Sara Hill suppressed everything, on two independent grounds. First, the stop: once the warning was issued and the license returned, the traffic stop was over, and the deputy needed reasonable suspicion to prolong it. He didn’t have it — unusual travel and a lot of luggage, the court held, describe far too many innocent travelers to add up to suspicion of crime. Second, and the part that matters for this post: the ALPR lookup itself was an unconstitutional, warrantless search. Pulling a month of someone’s multi-state movements out of a networked plate database, the court held, intrudes on the reasonable expectation of privacy in the whole of a person’s physical movements — and it said so by applying Carpenter v. United States (2018) and Chatrie v. United States (2026), consciously departing from other Oklahoma federal decisions that had waved these lookups through under the old 1983 beeper case, Knotts. No warrant, no probable cause, no exception — so the drugs, the cash, and the arrest all fell as fruit of the impermissible search.

Forty-five kilos of meth, suppressed, because the government queried a surveillance database without a warrant. Remember that.

And sit for a second with those dryer sheets, because they capture the whole shift this blog exists to document. Lining a suitcase to defeat a dog is preparation for the search everyone still pictures: an officer and a K-9 at the roadside, sniffing for what’s hidden. But the search that decided this case happened before the deputy ever reached the window — in a database, the moment he pulled a month of her movements out of a plate-reader network. The dryer sheets were fighting the physical war. The search that mattered was digital, done at a keyboard. That is the migration this whole project is about: the decisive government search has moved off the roadside and into the database — and the law is only starting to follow it there.

Now look at the machine. Strip the Oklahoma facts away and describe what actually happened in the abstract: the government maintains a massive, indiscriminate, networked collection system that scoops up data on everyone; it retains that data in a database; an individual officer queries the database using one person’s identifier; and the query returns an aggregated picture — here, the whole of a month’s movements — far richer than any single data point. The constitutional question the court had to answer was not whether collecting one plate-read in public is a search. It was whether the query— the act of reaching into the retained mass database and pulling one person’s aggregated history — is itself a search.

Judge Hill said yes. And if that structure sounds familiar to anyone who has followed this blog, it should, because it is the exact structure of the fight over Section 702 of FISA.

Section 702 is a collection system for communications of non-US persons happening outside the United States: the government acquires enormous volumes of communications and retains them in a raw database. Americans’ communications are also incidentally swept in if they communicated with those under surveillance. And then the government queries that database using Americans’ names, emails, and phone numbers — reaching into the retained pile to pull one person’s aggregated communications. The whole reform fight is about whether that query is its own Fourth Amendment event that needs its own warrant. It is not a new question in the courts, either: in United States v. Hasbajrami, the Second Circuit held that querying the 702 database is a separate Fourth Amendment event from the collection that filled it, and on remand the district court held that a U.S.-person query requires a warrant. Same holding Judge Hill reached about a license plate. Same principle: the query is the search.

So here is the asymmetry, stated plainly. A Tulsa County deputy who queries a plate database for your month of movements needs a warrant — and when he skips it, forty-five kilograms of methamphetamine are suppressed. The FBI, querying the 702 database for the unminimized contents of your communications — a far more intimate thing than where your car has been — has fought for years against any warrant requirement at all, and in one well-documented stretch ran more than 278,000 queries that didn’t even comply with the softer rules it does have. The county drug cop is held to the warrant-for-the-query standard, with the full force of suppression behind it. The national intelligence apparatus, querying a larger and more revealing database, is still arguing it shouldn’t have to be.

One could argue that is upside down. If anything, the bigger, more intimate, less transparent database deserves more constraint on the query, not less. Instead, the protection inverts with the power of the searcher.

Three things sharpen the point. First, this is a full circle to where this blog started. Kyle decides the case by applying Chatrie — the geofence-location decision I wrote about in the very first week of this project — which itself grew out of Carpenter and, before that, the concurrences in Jones. The idea that aggregated location data reveals a protected “whole” even when each piece is public has marched from GPS trackers to cell-site records to geofence warrants to license-plate readers, one technology at a time. The next stop on that line is the 702 database. The theory is already there; the courts just need to follow it.

Second, the plumbing behind that query blurs the public/private line in a way worth pausing on. Flock is a private platform, but most of the cameras feeding it are funded with public money — by police departments and municipalities — while some are paid for by homeowners’ associations and private businesses: the camera at a subdivision gate, or in a store parking lot. Here’s the thing: it almost doesn’t matter who paid. Every one of those cameras pours its reads into the same searchable network, queryable by agencies across jurisdictions, and the deputy running a plate neither knows nor cares whether a given sighting came from a county pole camera or an HOA’s gatehouse. The funding source dissolves at the query. The privately-owned slice is a domestic echo of the “data-broker loophole” that 702 reformers keep trying to close — private hands amassing bulk data the government can then search without the warrant it would otherwise need — except here it’s merged into the same system as the government’s own cameras, so even that distinction disappears. One mesh, mixed money, a single query surface.

Third, the quality of the “justification” is identical, and identically thin. The deputy logged his database query reasons as “traffic violation” and “investigation” — boilerplate, after the fact. Section 702 queries run on a three-part standard (an authorized purpose, a reasonable design, a specific justification) that produced a six-figure pile of violations. Both systems lean on the searcher’s own self-certified reason as the only real check, and both show how little that check means. The Kyle court noted it couldn’t tell, on the record, what the ALPR retention and access rules even were — the same opacity that shrouds 702.

Let me be fair about the differences, because they’re real. Section 702 is a foreign-intelligence authority built around national security, with its own court and its own framework; a highway drug stop is ordinary domestic policing, and the stakes are not the same as a terrorism investigation. Kyle is also a single district-court opinion that openly breaks with other judges in its own district — persuasive, not binding, and very possibly headed for appeal. And suppressing forty-five kilograms of methamphetamine is a genuine cost that a serious person should feel, not shrug off.

I feel it. But notice that my argument never depended on the two regimes being legally interchangeable. It depends on the act being the same — querying a retained mass-surveillance database for one person’s aggregated data — and on a simple principle this whole project is built around: the Fourth Amendment protection should attach to what the government does, not to which badge it wears or which statute it invokes. A model act keyed to the act would treat the deputy’s Flock query and the Bureau’s 702 query the same way, because they are the same search. The citizen’s protection should not evaporate just because the database got bigger and the agency got more powerful. If the stakes argument cuts any way, it cuts toward more care with the larger, more intimate system — not toward exempting it.

The query is the search — on the highway and in the FBI office. One judge in Oklahoma just enforced that against a drug cop with forty-five kilos on the line. The open question is why the same principle, aimed at a far bigger and far more revealing government database, is still treated as negotiable. When the law protects you less as the surveillance grows, the law has the sign wrong.

Leave a Reply