A warrant for the mosaic
On June 29 the Supreme Court held, 6–3, that police need a warrant to pull your location out of a company's database — even when the company is Google and the data is something you supposedly gave away by carrying a phone. The case is Chatrie v. United States. Justice Kagan wrote for a majority that ignored the usual lines, three of the Court's conservatives in dissent. The sentence that matters: "an individual has a legitimate expectation of privacy in his cellphone location data."
I think this is good news, and not for the reason most people will reach for. It is good because the law is finally catching up to something my old profession treated as a first principle: surveillance was never about the single observation. It is about the aggregation.
One idea, three cases
This did not arrive out of nowhere. Three decisions, the same idea sharpening each time.
In 2012, United States v. Jones, the Court held that attaching a GPS tracker to a suspect's car and following it for a month was a search. The majority leaned on the physical trespass, but two concurrences saw the deeper problem: it was the duration — the unbroken record of everywhere the car went — that turned ordinary public movements into something private.
In 2018, Carpenter v. United States, a 5–4 Court required a warrant for months of historical cell-site records. It refused to extend the old third-party doctrine — the rule that anything you hand a company loses its protection — to a device that quietly logs your location every few minutes. Sharing your number with the phone company, the Court reasoned, is not the same as surrendering a map of your life.
And now Chatrie, the geofence case. Police drew a box on a map around a Virginia credit-union robbery and ordered Google to hand over every device that had been inside it. The Court's most important move was to reject the government's fallback: that the sweep was harmless because investigators only pulled a thin, time-limited slice out of an enormous dataset. Once the Fourth Amendment applies, the majority answered, it does not matter how small a bite you take out of an all-encompassing database.
Hold onto that line. It is the whole game.
The mosaic
There is a term of art for this in the intelligence world: pattern of life. No single data point indicts you. A car on a road, a phone near a tower, a plate read at an intersection — each is nothing. But collect enough of them, across enough time, and they resolve into a portrait: where you sleep, who you visit, which doctor, which lawyer, which church, which lover. The analyst's craft was never in the single ping. It was in the mosaic.
For most of history the mosaic was expensive. Tailing one person around the clock took a team. That cost was itself a kind of privacy — a natural friction that meant the state reserved its attention for people it already had reason to watch. Digital surveillance erased the friction. The mosaic that once took a squad now takes a query.
What the Court has been doing, case by case, is noticing this — writing the mosaic back into the Fourth Amendment, recognizing that a thousand innocuous observations can add up to a search even when no single one of them would.
Why a warrant is the right answer
A warrant is not a ban. This is the part critics miss. Requiring one does not put location data off limits; it puts a neutral magistrate and the word particularly between your movements and the state. Show a judge probable cause, name what you are after, and you can have it. What you cannot do is vacuum up everyone and sort it out later.
That is precisely the abuse the Fourth Amendment was written against. The founders' grievance was the general warrant and the writ of assistance — the open-ended license to search anyone, anywhere, for anything. A geofence order that captures every phone near a crime scene, guilty and innocent alike, is a general warrant with a software license. The Court just said so.
Are Flock cameras next?
I think they are — and I think that is where this line was always heading.
Flock Safety sells automated license-plate readers, and it has quietly blanketed American towns with them. The cameras log every plate that passes: number, timestamp, location, direction. The records are retained for weeks and shared across a national network of thousands of agencies. It is Carpenter for cars — a continuous, retrospective record of where nearly every vehicle in a town has been, searchable after the fact, on anyone.
The Institute for Justice is litigating precisely this in Norfolk, Virginia. In January a federal judge dismissed the challenge, reasoning that the network was fine because it does not track "the whole of a person's movements." That sentence is now living on borrowed time. Chatrie's rejection of the "small slice" defense is aimed straight at it: if the Fourth Amendment attaches once the database is comprehensive, it does not save the government to say it only looked at part of it. The plaintiffs are appealing. The reasoning they need is now on the books.
Whether license-plate dragnets are next is no longer really a question of principle. It is a question of when a case reaches the right court.
The good thing
The through-line of these cases is a single, slow realization: the danger was never the data point. It was the portrait you can assemble from enough of them — cheaply, after the fact, on anyone at all. For twenty years the technology ran ahead of the doctrine, and the friction that used to protect us simply evaporated. What the Court is doing now is putting a measured amount of it back, in the oldest form the law has: go ask a judge first.
That is a good thing. And notice what it is not: the Constitution did not change, and it did not relearn anything. The principle has been right since 1791 — the same rule that condemned the general warrant condemns the geofence and the plate-reader dragnet. What took three centuries was us, finally applying it correctly to a world of phones and cameras.