What Changed

This update ships a fix, not new data: a bug in how 287(g) agreement records were matched to county FIPS codes was silently dropping signal from some of the counties we most need to see clearly. Place-name mismatches and typos in source data meant a subset of 287(g) agreements never resolved to a county at all — they just vanished from the heat model instead of erroring loudly.

The fix adds place-name and typo-tolerant fallback matching, and we’ve now applied it against the live knowledge base and rebuilt every heat score.

Notable Signals

The recompute moved several already-hot counties further up the board, because they now show their true signal count:

• Maricopa County, AZ: 124 → 139 (+15)
• Wayne County, MI: 147 → 161 (+14)
• Orange County, FL: 157 → 169 (+12), also gained a 10th signal type
• Hidalgo County, TX: 122 → 134 (+12), now spans 8 signal types
• Allegheny County, PA: 143 → 150 (+7)

Broward County, FL remains the highest-heat county in the country at 202, now with all 10 signal types represented — it was previously showing 9.

A handful of Texas border and interior counties (Williamson, Harris, Frio, El Paso, Bexar, Webb, Cameron) each picked up their previously-missing 287(g) signal, nudging their scores up by 1 point apiece — small individually, but it confirms the fix is working as intended across the board rather than just on the highest-profile counties.

Coverage Updates

No new counties entered the dataset this cycle — this was a correction pass, not new ingestion. 2,032 counties remain scored, same as before. The timeline (timeline.json) was regenerated alongside the heat scores and is back in sync — it had been stale since early July.

Numbers

Data current as of August 11, 2026. Contribute what you know — local knowledge is what converts signals into stories.