How this is counted

Short version: nobody knows how many cars hit buildings in Columbus. Not this site, not the police, not the guy who has tracked it since 2022. Here is what this board actually does and does not know.

Where the incidents come from

Every row in the log links to the story it came from. Three ways they get there:

Daily feed: 614NOW publishes a category feed for exactly this phenomenon. A job checks it each morning and appends anything new.
Archive walk: That feed only exposes the ten most recent posts, so a separate pass reads the full category archive and pulls the history out of the articles themselves.
By hand: Incidents from NBC4, ABC6 and Columbus Underground, entered with the details the reporting states — including a 2024 NBC4 roundup that listed six crashes with addresses in one go.

What the machine reads, and what it guesses

The scraper reads English prose, not a database. It pulls a street address, a time of day and a neighborhood out of the article text when they are stated. When an article says "on April 30," that is the crash date. When it does not, the row falls back to the date the story ran — which is usually a day or two late.

This is why there is no day-of-week chart. Roughly half the entries carry a publish date rather than a crash date, and 614NOW tends to publish on Mondays. A weekday chart would mostly measure their editorial calendar. Hour-of-day is charted because those times are quoted in the articles, not inferred.

Why the totals disagree

The best count does not come from the news at all. Columbus Division of Fire tracks what it gets dispatched to, and Battalion Chief Jeffrey Geitter gave NBC4 the numbers: 313 vehicle-into-structure runs in 2022, 279 in 2023, and 102 through June 1 of 2024. That is a crash about every 1.3 days — and by that data the trend is going down, not up. Through June 1, the department ran 132 of them in 2022, 115 in 2023 and 102 in 2024.

Everything else is a fraction of that. Paul Meara tracked 71 in 2025 as @cbuscarikaze; Columbus police documented 67 through June 2025. This board has 60. Fire sees every call. Reporters see the photogenic ones. So the counter on the front page is measuring how long since a crash made the news — roughly one per 1.3 days keeps happening either way.

The counter runs from the most recent crash this board could source. Columbus has very likely beaten it since you loaded the page.

The math

Crashcon measures days since a crash last made the news, which is the only thing this board can see. The fire-run cadence is printed beside it for contrast.
The weekly odds treat crashes as a Poisson process: at one every 1.3 days, the chance of at least one somewhere in Columbus in the next seven is 1 − e^−7/1.3, about 99.5%. The interesting number is the second one — the odds any of them gets written up.
Quiet stretches are only measured from January 2025 on. Coverage before that is too patchy for a gap to mean anything.
Implied odds on what gets hit next are just the historical mix of venue types. There is no market. Please do not bet on this.

The map

Addresses are geocoded through OpenStreetMap's Nominatim and checked against a bounding box around Franklin County, because "830 Bethel Rd" exists in other Ohio towns too. A filled dot was pinned to a street address; a hollow dot was pinned to a neighborhood and is approximate. Incidents too vague to place are counted under the map rather than dropped.

Corrections

If a row is wrong — wrong date, wrong business, wrong street, or your building is on here and it should not be — say so and it gets fixed. The underlying data is a single JSON file, so corrections are cheap.