The evidence gap

Two of the five yes votes left a clear public reason. Council member Scott Benson: “This is not just about law enforcement, this is about human lives.” Council member Renata Miller: “We need a deterrent, and we do not accept that gunshots is a normalcy in Detroit, especially on the east side of Detroit.” Both are values-based arguments, not data arguments — and neither engaged the 0.03 percent figure on the table that day.

The other three yes votes are harder to trace. Council member Angela Whitfield-Calloway voted no on ShotSpotter's 2022 expansion — “I cannot support spending $8 million on in my opinion, failed technology” — and yes in 2026, with no public explanation of the change found in available reporting. Council President Pro Tem Coleman Young II and Council member Latisha Johnson voted yes without a specific justification traceable in local coverage reviewed for this story (Johnson had pressed Chief Bettison on how DPD measures effectiveness at a March budget briefing, weeks before voting yes). To be precise about what that gap means: some coverage — including the Detroit News — sits behind a paywall this project can't access, so “no explanation found” is not the same claim as “no explanation given.”

The no votes, by contrast, engaged the data directly. Council member Denzel McCampbell, voting no: “arrests only follow two to 3% of the alerts, that aid is rendered to victims that are less than 1% of the calls or cases.” Council member Mary Waters, voting no: “They have not been able to demonstrate the data or otherwise that ShotSpotter actually works and that it actually saves lives.”

Meanwhile the arrest count itself kept moving. A slide deck Chief Bettison presented to the Board of Police Commissioners on March 6, 2025 put “arrests connected to ShotSpotter alerts” at 401 for 2023 and 373 for 2024. At a March 23, 2026 budget briefing, he cited 256. At the May 18, 2026 Council meeting, weeks before the vote, he cited 78. Four numbers, four different rooms, no published reconciliation between any of them — and a different metric and time window than the 0.03 percent figure from the underlying performance data. That's not evidence of dishonesty; the honest reading is that “arrests,” “arrests connected to alerts,” and “alerts resulting in an arrest” aren't the same measurement. But if the numbers can't be reconciled, they can't be the basis for a vote either.

The city's own next step is a DPD–University of Michigan “academic review” of ShotSpotter, which Deputy Chief Mark Bliss told a Council committee on June 22, 2026 was underway. Its funder, scope, and reporting date are still undisclosed. That review — done independently and published before the March 2027 deadline — is the plainest test of whether the next vote gets a real number or a fifth one.

Project Green Light is nine years old, 850-plus locations deep, and rated “ineffective” at reducing violent crime by the Department of Justice's own research arm. It has never received the independent audit that rating begs for. The city's freeway license-plate readers run roughly $6.25 million through June 2029 while 30-plus localities nationwide have canceled the same vendor's systems. Police drones are, in the chief's words, “on the road map” — studied against other cities, pitched in part for drag-racing enforcement — before any cost, scope, or rules have been published.

Add it up and Detroit has spent more than $34 million on surveillance technology in a decade. Across that same decade, the purchases have followed a consistent sequence: the technology arrives, the evidence question comes later, and the public review — where it happens at all — happens after the practical decision has been made. A court has already found the review step was skipped once. The sequence is the story.

Here is the uncomfortable comparison. Detroit's homicides fell to 165 in 2025 — the fewest since 1965 — and the interventions with the strongest evidence behind that kind of drop are not sensors. A randomized trial across 541 Philadelphia lots found cleaning and greening vacant land cut gun assaults 29 percent in poorer neighborhoods. Community violence intervention and summer youth employment carry real, published effect sizes. Detroit's own ShotStoppers CVI program reports large reductions in its zones — though the city grades its own homework there, and independent evaluation is overdue.

The point isn't that surveillance can never work. The point is that one column of the budget carries randomized trials and sixty-year lows, and the other carries a court ruling, an “ineffective” rating, and 0.03 percent — and the second column is the one that keeps getting funded without conditions.

When the renewal came up, the loudest community testimony in favor came from Live In Peace, a violence-intervention group whose pastor-director spoke passionately for keeping the technology. What was never stated at the podium: Live In Peace has been a ShotStoppers grantee since June 2025 — eligible for up to $1.4 million a year through the very program created as the alternative to surveillance spending.

The facts here deserve precision, because they cut in more than one direction. The grants are not conditioned on supporting any technology. Most ShotStoppers grantees took no position on ShotSpotter at all. There is no evidence of a quid pro quo, and we do not allege one. But disclosure is not optional in public comment for a reason: Council members weighing “community support” deserve to know when the community speaking is also a city vendor — and so does everyone watching from the gallery. The tie was real, it was material, and it went unsaid. That is a process failure the city can fix with one rule: paid partners disclose the relationship when they testify.

None of this requires believing anyone acted in bad faith. It requires noticing that Detroit has no working evidence gate between a vendor's pitch and a Council vote. Other cities have built one: an independent performance audit before any renewal; published effectiveness data in the 14-day CIOGS review; a sunset default so contracts expire unless the data justifies them; and a disclosure rule for paid city partners testifying in public comment.

Detroit already owns the legal machinery — the surveillance ordinance exists, the court has enforced it once, and a new administration has every incentive to be the one that made the money follow the math. The question for Council President and the members who voted yes is simple, respectful, and overdue: what evidence standard will the next surveillance dollar have to meet?