Our counts next to his, as of Nov 16
Default view: every event in our history file, starting in 2016, all labels. That is the closest like-for-like period, even though he used 224 records and we have far fewer.
| Window | Rohit current | Our current | Rohit earlier high | Our earlier high |
|---|
Trailing counts, day by day
The shaded band is 2006 to 2015, the decade Rohit added later. The dashed line is our earlier high for windows ending by Oct 31, 2026. The orange tick is the current window.
A sensitivity check on size
Turn on "known scale of 1 million or more." Most of this week's events then drop out, because we do not have a counted impact for them, or the official scope is smaller. The recent spike is many mid-size incidents close together, not a spike in historically huge events.
Coverage bias
Plain English
- Rohit is right that a run of headlines is not the same as a record.
- On our thinner list, this week is still a high, especially if we add the FAA and United items.
- We searched 2026 harder than 2016, so recent bars have an unfair boost.
Data: cluster.json, built from history.json, list_b, and wire.json. Rebuild with scripts/cluster_build.py.