–
High, active now
–
Medium, active now
–
Info, active now
–
Events logged
–
RED incidents
Active now
Statistical flags: a statistical observation, not evidence of misconduct.
Incidents
Each item is the source's own headline or text as published (cut to 240 characters), with a link. The app adds no wording of its own. Keyword-based labels can be wrong; social posts are unverified.
Reports collected tonight
Each mark is a report our sources published, not a confirmed incident. Volume follows how many sources cover a state and how often they post.
Event log
Statistical flags: a statistical observation, not evidence of misconduct.
What gets flagged
- Vote decrease
- A candidate's count went down between updates, for a race or a county. Usually a correction, but worth a look.
- Reporting backwards
- The share of expected vote counted fell by more than a point.
- Call changed
- A race call was changed or withdrawn.
- Large batch
- One update added 15% or more of a race's expected vote. Shows how that batch leaned.
- Lopsided batch
- A county batch of 5,000+ votes landed 35+ points away from the model's projection for that county. Early mail and Election Day batches often do this, so it's informational.
- County swing outlier
- A county's swing from its 2022/2024 baseline is far from the other counties' (robust z ≥ 3.5).
- Turnout outlier
- A county's implied turnout relative to its baseline is far out of line with the rest of the state.
- More votes than expected
- A county has counted more than 105% of the feed's own expected total vote.
- Result vs. expectation
- A race is running 2.5σ+ away from its pre-election prior after accounting for the national environment.
- Senate/Governor mismatch
- In an Ohio county the Senate and Governor two-party totals differ by more than 8%, which usually means one race's results updated without the other.
- Stale feed / feed error
- No new results for 15 minutes even though some unfinished race was updating within the last hour, or the feed failed.
- Feed source changed
- The app switched between NYT and CNN. Their counts differ, so comparisons restart instead of flagging the difference.
- Sources disagree
- The main feed and a second source (CNN, or the Ohio Secretary of State's official feed) differ on a race's call, its two-party total (by more than 2%), or its margin (by more than 2 points with similar totals), on two reads in a row. Usually one source is behind or has a data-entry error.
- App and markets far apart
- The app's chance of control and the Kalshi/Polymarket average have differed by more than 25 points for 10 minutes. A prompt to check the results data; it doesn't say which is right.
- Poll closing
- "Result vs. expectation" and "stale feed" wait until a state's polls close, including court-ordered extensions listed in
data/extended_hours.json.