Apocalypse Early Warning System
In the event of an imminent nuclear apocalypse, we suspect that many people who have access to private jets will immediately take to the skies and escape city centers. This site tracks this indicator in realtime. The current emergency level is reported on a scale of 1 to 5, with 5 being an indicator of a likely imminent apocalypse.
built by Kyle McDonald
Sign up for text message or email notifications.
Emergency level 1/5
502/31,465 planes airborne
5,838 max people airborne
Deviation: -55(-0.7σ)
Last Update: May 9, 9:30 AM CDT
https://ews.kylemcdonald.net/
How This Works
This site watches a fixed cohort of business jets and asks a simple question: is the number currently airborne unusual for this time? It is not tracking all aircraft. The original version used an FAA-only business-jet list. The current tracker builds a broader global aircraft metadata table by merging ADS-B Exchange aircraft records, Mictronics/tar1090 records, and FAA registry data by ICAO hex. The importer classifies metadata into business jets, military aircraft, large airliners, regional airliners, non-jet aircraft, and other known types, then applies a practical business-jet filter. Each tracked aircraft is matched in live data by its ICAO hex identifier.
The flight data comes from ADS-B Exchange heatmap files. Those files are published in half-hour slots and encode recent aircraft positions. The backend downloads the newest available heatmap, parses it, matches the aircraft in the heatmap against the tracked cohort, and stores the latest position, altitude, speed, heading, and airborne state for each match. Military aircraft and non-ICAO addresses are published as separate dashboard snapshots and loaded only when their toggles are enabled.
Historical context comes from the same heatmap format. The backfill job walks through previous half-hour slots, counts how many business jets were airborne, and records those counts in SQLite. The dashboard then compares the current concurrent airborne count with an all-history weekly baseline for the same half-hour of the week. The model also learns local half-hour profiles around U.S. federal holidays, so predictable holiday travel is included in the prediction instead of treated as a generic spike.
The deviation number is the current count minus the expected count. The sigma value puts that difference on the scale of historical model error and combines it with an absolute-excess weighting, so tiny overnight changes do not dominate just because the usual count is low. When multiple aircraft categories are selected, their observed counts, predictions, and variances are combined and the emergency level is recalibrated for the selected total.
The max-people estimate is intentionally rough. It maps known aircraft model labels to published maximum passenger capacities, sums the known matches, and scales missing capacities by the known average. It is a maximum seat estimate, not a passenger manifest.
There are important limits. ADS-B coverage can be incomplete, aircraft may be blocked or misidentified, heatmaps arrive in coarse half-hour windows, and the global cohort is a heuristic rather than a perfect definition of every relevant business jet. The dashboard is best read as an anomaly monitor for public flight signals, not as proof of intent, destination, ownership activity, or who is on board.