Four Hundred Thursdays

The last item in the first SpotCrime Weekly Reads is filed under “Dumb Criminal of the Week.” It links to a story about a mannequin-challenge video that turned into gun and drug charges. Above it, in the same post, sit roughly forty other links: an Ohio Supreme Court ruling that dash-cam footage is a public record, a MuckRock piece on a South Carolina department stonewalling records requests, the FBI's release of 2015 NIBRS data. That was December 15, 2016. There have been about four hundred more since.

The blog those posts live on is older than almost everything else in American crime-data commentary. Its sitemap indexes 1,251 posts. The first is dated June 1, 2008. The most recent, as of this writing, is dated August 27, 2026.

Eighteen years is long enough that the archive stopped being a blog and became a record. Not of crime — of something nobody else was writing down.

The first two years are a shooting log

On June 28, 2008, the blog published thirteen posts. Each one is a map of recent shootings in a single city. Miami. Austin. Norfolk. Richmond. Houston. Indianapolis. New Orleans. Los Angeles. Las Vegas. Philadelphia. Washington. Oakland. New York. Boston and San Francisco followed within hours.

That month ran thirty-five posts in thirty days. Titles like “Four Drive-By Shootings in Kansas City this Morning” and “Map of 22 Shootings in Baltimore June 14 to June 20, 2008.” There was a London crime map, updated daily. There were instructions for pulling SpotCrime data into Google Earth and Facebook, which in 2008 was a reasonable thing to want.

The first post is dated June 1, 2008 and is titled “SpotCrime Gets Picked UP by Washington Post and BBC.” It opens: “Thanks to a mention in TechCrunch...”

Then it stopped counting and started arguing

Volume drops off a cliff after 2010: 74 posts in 2011, 48 in 2012, and somewhere between thirty and sixty a year ever since. What replaced the volume was argument.

In June 2012 the blog published a post called “Why Public Crime Maps Stink.” A crime-mapping company, saying its own medium is broken. “Maps in general will always have some type of dimensional limitations,” it reads. Zoom the window too tight and a block shows no crime, which reads as safety. Widen it and the pins merge into an unreadable blob. Heat maps look like a fix but only move the arbitrariness into the bandwidth setting, where nobody can see it.

That is the modifiable areal unit problem, described in plain English, by the vendor, in 2012. We wrote a much longer version of the same argument fourteen years later.

Five months after that came “Dots on a Map,” which flips the frame. Every dot means someone called 911, a dispatcher allocated a unit, an officer went, and a records clerk typed it up. The post notes that SpotCrime's database at the time held “well over 15 million crimes.” A map of incidents is also a map of responses.

And in December 2012, “Press Releases v Crime Data Feed,” which may be the most useful thing on the entire blog. Press releases carry narrative — suspect descriptions, what they wore, a step-by-step account — but departments issue them only for the incidents the department “thinks is noteworthy.” Feeds carry coverage but their descriptions are “short and sometimes non-existent.” Neither one alone is a public record. You need both, and almost nowhere gives you both.

October 2012 brought “The Problems with Nomenclature and Crime Classifications.” That is taxonomy normalization, written up nine years before the NIBRS transition made it a national problem.

Four hundred Thursdays

The December 2016 post set a format that has held up with small edits. Section headers: Police Conduct. Police Transparency. ‘Crim-Tech.’ America's Prison System. Crime Rate. Dumb Criminal of the Week. Roughly forty-five links, sorted.

The issue from August 27, 2026 runs Police Conduct, Crime Rate, Crim-Tech, Police Transparency, The Prison System. Same five buckets, reordered, with the possessive trimmed off the prison heading. Crim-tech is still there, nine and a half years on. What is gone is Dumb Criminal of the Week. So are about three-quarters of the links: eleven in the current issue against forty-five in the first.

Since then the series has run essentially without a gap. Counting posts whose URL contains weekly-reads:

Four hundred issues, give or take, at roughly forty a year for a decade. 2026 is at thirty-two through August. The compiler, per her own LinkedIn and her site at spotcrimebrit.com, is Brittany Suszan, who has worked at SpotCrime since September 2009 and whose title is Chief Operations and Data Advocacy. Between 2021 and 2023 she also worked as a research consultant at the University of Florida's Brechner Center for Freedom of Information, doing FOIA work for a podcast with the investigative reporter Sara Ganim. Her college activities list includes public safety dispatcher. (It shows.)

The posts themselves carry a corporate byline. The blog says “SpotCrime.” Four hundred weeks of curation and the author never put her name on it.

What a decade of link dumps turns out to be

Nobody sets out to build a longitudinal dataset by posting links on Thursday. But run the archive forward and something accumulates that no crime dataset contains: a week-by-week record of which American police agencies were publishing their data, and when they stopped.

The original reporting on the blog has thinned to one or two posts a year. It has also gotten much sharper, and every recent one is a records-access story.

December 2024: the Colonial Heights Police Department in Virginia, serving about eighteen thousand people, published monthly crime reports for fourteen straight years and stopped in April 2024. It had migrated to a new records management system two years earlier. The post puts the share of US agencies publishing incident-level data openly at around 27%.

June 2025: Baltimore. In 2024 the department had broken out Shootingas its own incident category, which let anyone track gun violence straight from the feed. The 2025 move to NIBRS dropped that category. What replaced it was “Aggravated Assault – Handgun,” which cannot distinguish a gun fired from a gun shown. The post ran while the city was publicizing a decline in shootings. Its question was whether this was “oversight or obfuscation.”

June 3, 2026: a post asking readers to contact Chief Jim McDonnell, the Board of Police Commissioners, the City Council, and the Inspector General about block-level locations missing from the LAPD feed. Six days later, LAPD published 150,431 backfilled cases. The volume gap in that feed is still unexplained.

And in December 2020, a post whose title is the whole argument: “COVID data is easier to access than crime data.” It is “harder to collect RMS/CAD data from police agencies on a national scale than it is to collect national COVID data.” It notes that when Baltimore County upgraded its records vendor, the upgrade “removed the capability to report crime data to the public in an open and timely format.”

Four separate stories, four different states, eight years apart. Same mechanism every time. A department buys new software and the public feed comes back smaller than it went in, or does not come back. Any one of these is a local curiosity. Twenty of them, logged with dates, is a finding.

The other thing it built: a ranking

Running under the weekly posts since at least 2015 is the SpotCrime Transparency Ranking, which grades agencies on how openly they publish. It has been updated in 2015, 2017, 2020, a university-police edition in 2021, and again in early 2025.

The 2020 revision is the interesting one. It raised the bar: publishing machine-readable data somewhere on a department website no longer earned top marks. You needed a real open data portal, published openly and on time. Under the new rule Boston got downgraded, because a records system upgrade had knocked out public access.

A ranking that penalizes a city for a vendor migration is a ranking that understands what actually breaks a crime data pipeline. It is almost never policy. It is procurement.

Who else has been doing this

The field of people writing seriously about American crime data is small and mostly recent. Scott Henson's Grits for Breakfast launched in 2004 and is the only thing here that is older, though it has taken long breaks, including all of 2023, and its subject is Texas policy rather than data access. Adrian Holovaty's ChicagoCrime.org became EveryBlock in 2007 and has now shut down twice. The Marshall Project started in 2014. The Council on Criminal Justice arrived in 2019. Jeff Asher's Jeff-alytics is about six years old. The Real-Time Crime Index and Andrew Wheeler's crimede-coder are both younger than that.

Every institution the field now treats as canonical is younger than this blog, most of them by a decade.

The more useful distinction is position rather than age. Asher and the Council on Criminal Justice analyze the numbers agencies publish, and they do it with more statistical firepower than a link roundup ever will. Read them for the trend. This archive is doing a different job: it tracks whether the agency publishes at all. That is why it holds things nobody else recorded — the month a small Virginia department went dark, the category Baltimore deleted, the specific vendor migrations that quietly ended public access.

Those events only look like a story in aggregate, across many years. Which means the only way to have the story is to have been writing it down the whole time.

Using an archive like this

For anyone maintaining a crime data pipeline, an eighteen-year change log of agency publishing behavior has direct operational value. Four uses:

1. Structural-break priors.When a feed's volume shifts and you need to know whether the city got safer or the schema changed, search the archive for that agency first. A vendor migration usually left a trace.
2. Category-deletion checks. Baltimore losing its Shooting field is the pattern to watch for. A category that vanishes reads downstream as a decline. Diff your distinct incident types across time, not just your counts. We covered why that matters for small-area comparisons.
3. Coverage expectations. If roughly a quarter of agencies publish incident-level data openly, a national product is a stitched patchwork by construction, and your coverage map is a first-class part of the product.
4. Records-request timing. The archive documents which agencies have historically responded and which have not, which is worth knowing before you spend six months on a request.

The part that does not scale

The reason this archive exists is not a strategy. Nobody built a content engine that runs for eighteen years. Somebody read the week's police news on a Wednesday night and posted the links on Thursday, and then did it again, four hundred times.

The 2012 post about dots on a map ends by asking readers to thank their local police department. It is a strange note for a company that has spent the years since suing the LAPD, grading cities on transparency, and documenting every department that went quiet. Both things are in the same archive, four hundred Thursdays apart.

The last issue before this one carried eleven links, under five headings, with no joke at the end. The first carried forty-five and closed on a mannequin-challenge video. Somewhere in the four hundred Thursdays between them, the bit about dumb criminals stopped being funny enough to keep.