Build1 distinct publisher3 min readUpdated
Over 90% of partial-plate searches happen within a week, by Flock's own telemetry. That makes the old 30-day default a capacity finding first and a privacy finding second.
The Engineer · Build desk

Compiled by The EngineerSomething wrong?How this is made
On August 13, Flock Safety CEO Garrett Langley said the company would cut its default license-plate-reader retention from 30 days to seven, make misuse detection mandatory for every law-enforcement customer, and require a case code on every search by the end of the year [1]. In the same announcement he published the sentence that matters to anyone who stores other people's data: "Our analysis shows that over 90% of searches without a full plate are done within a week" [2].
Read that as an operator rather than as a civil-liberties question. Flock set the 30-day default nine years ago and says it believed the number struck a balance [3]. The network now spans more than 120,000 cameras [4]. If over 90% of partial-plate queries land inside seven days, then roughly 23 of every 30 days of retained vehicle-movement data were serving under 10% of that query volume [5]. Nobody decided to keep three extra weeks of national movement data. Somebody typed a defensible number into a config in year one and it was never re-tested against usage [6].
The distinction worth holding onto is that access controls are probabilistic and deletion is deterministic [7]. Sort the announced package by that test. Six of the seven measures are detection, response, or hardening [8]. They work: Savannah PD fired six employees this month after finding unauthorized searches through Flock's own voluntary audit feature [9]. But only the retention change alters how much data exists to be subpoenaed, breached, exfiltrated through a compromised account, misused by an insider, inherited by an acquirer, or shared with a partner under a policy nobody has written yet [10]. Detection keeps working right up until the audit log is the thing that is compromised.
The announcement followed a Washington Post investigation that documented nearly 50 instances of officers charged with or accused of running unauthorized searches [11], and the reaction was not warm. The ACLU called the measures retreads of safeguards that were already inadequate, while allowing that a shorter window may be a step in the right direction [12]. Robert Frommer of the Institute for Justice told the AP it was "window dressing that doesn't address the fundamental problem," his objection being that police still decide who gets searched, a decision he argues belongs to judges and warrants [13]. The Center for Democracy and Technology described it as the same "Just Trust Us" policy with a fresh coat of paint [14]. Andrew Guthrie Ferguson of GWU Law called the shifts better than the opposite while asking for sustained democratic engagement and judicial checks [15].
Underneath the disagreement is a structural point: the retention number is still a setting on a vendor's dashboard [16]. The nearest comparison, the UK's National ANPR Service, runs roughly 12,700 cameras collecting about 60 million reads a day and retains for 12 months, twelve times Flock's old default [17][18]. The UK figure sits under the Protection of Freedoms Act 2012 and published national standards; the American one is a product default announced by a vendor, and can be announced differently later [19].
The transferable exercise is small. Log your age-at-access distribution for a month before anyone argues about policy [20]. If you cannot name the person who chose your retention window and the reason, it is an inheritance, not a decision [21].
Watch whether the case-code requirement actually ships by year end [1], and whether Flock publishes the age-at-access distribution rather than the headline percentile.
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
On August 13, Flock Safety CEO Garrett Langley announced the company is cutting its default retention period for license-plate-reader data from 30 days to seven, making misuse detection mandatory for every law-enforcement customer, and requiring a case code on every search by the end of the year.
Flock's announcement stated: "Our analysis shows that over 90% of searches without a full plate are done within a week."
Flock set the 30-day default nine years ago and, by its own account, believed it struck a balance.
The dev.to author argues nobody deliberately chose to keep three extra weeks of national movement data; a defensible number was set in year one and never revisited against usage.
The dev.to author's framing: access controls are probabilistic, deletion is deterministic, and they are not substitutes.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One self-published account resting on a vendor's own quoted number
The load-bearing datapoint is a verbatim quote from Flock's announcement plus straightforward arithmetic on it, which is checkable within the text. Everything else — the Washington Post misuse count, the Savannah PD firings, the UK ANPR figures, the critic quotes relayed via the AP — is restated secondhand in a single dev.to post with no linked primary sources and no corroborating publisher in the cluster. The vendor telemetry itself is unaudited, and the seven-item package is summarized rather than reproduced.
Fleet-wide default change announced; downstream enforcement partly future-dated
The change is not a proposal: a default applied across a network described as more than 120,000 cameras is being cut for all customers, misuse detection is stated as mandatory for every law-enforcement customer, and there is at least one concrete downstream event (Savannah PD firings via the audit feature). Adoption is held below high because the case-code requirement is promised only by end of year, nothing shows whether agencies can override the seven-day default, and every adoption signal comes from a single secondhand account.
Mildly overstated framing on a real, single-sourced change
The piece is comparatively disciplined — it credits Flock for publishing the number, quotes critics who call the package window dressing, and flags its own commercial positioning. Slight overstatement comes from treating one unaudited vendor statistic as settling nine years of retention practice ('the number that made 30 indefensible'), from generalizing a single sorted package into a universal shrink-versus-watch verdict, and from a conclusion that happens to endorse the author's own local-only product architecture.
Disclosed competing-product author writing about a vendor under investigative pressure
Two incentive layers are visible in the supplied text. The author states outright that he works on Background Camera RemoteStream, a local-only Android camera app, and asks readers to treat the piece as positioning rather than neutrality — the architectural conclusion (no cloud, no retention dial) is his product's selling point. Separately, Flock's announcement is reported as following a Washington Post investigation into officer misuse, giving the vendor a reputational motive to publish favorable telemetry and a tightened default. Score is high but not maximal because the conflict is disclosed rather than concealed and critic voices are included.
Low: single publisher, disclosed positioning, secondhand facts
One source, one publisher, no independent corroboration of any third-party figure, and an author who declares a commercial stake in the conclusion. The core announcement facts and the quoted telemetry are probably reliable given they are attributed to a named CEO announcement covered elsewhere, which keeps confidence from being lower, but nothing in the cluster lets the secondhand counts or the UK comparison be checked.
product
Flock's case-number rule is an honour system, and the company says it never checks2 distinct publishers
product
Flock built the thing it said it could not do, then left the prompts on a public server4 distinct publishers
build
Flock's OS Investigate starts the search before there is a suspect1 distinct publisher
security
A cop, a hot list, and a query log: where ALPR actually fails1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
dev.to
1 article · August 15, 2026