Apocalypse Early Warning System

At a glance

Apocalypse Early Warning System compares public aircraft activity with a historical baseline and gives unusual increases an emergency-style score. It is worth exploring as a data-art project about signals, assumptions and false alarms. The score measures an outlier in the model, not the probability of a nuclear emergency.

Good fit

Readers interested in anomaly detection and how a dramatic framing changes the interpretation of data. The published method uses a selected business-jet cohort, half-hour flight-data snapshots and a baseline adjusted for recurring traffic patterns. Its archive lets you compare the presentation with earlier activity rather than accepting the current score in isolation.

Consider something else when

You need an emergency-warning service, evidence of an evacuation or an account of who is travelling. The site’s own FAQ says an extreme score can arise from ordinary events, data artifacts or cohort mistakes. Aircraft capacity estimates are maximum seats, not a passenger manifest or evidence of actual occupancy.

Start here

Read How This Works and the FAQ before the headline score. Move Traffic Archive to an earlier window and inspect the historical chart, then return to the current end. That worked in the browser trial. Keep the selected aircraft categories consistent when comparing windows, since the method says category changes recalibrate the combined score.

Cost and requirements

The public dashboard and tested archive required no account or payment. The notification signup page advertises email, SMS or both for $5 yearly and asks for contact details. No subscription was started. Its stated use of contact information covers alerts and subscription updates; that page does not establish an exact deletion timetable.

Limits to know

The method acknowledges incomplete coverage, classification errors and coarse update windows. An October 30 update says changing flight-data providers had produced elevated readings and that history was rebuilt to match the new source. That is a useful example of why a spike needs investigation. This trial exercised aggregate history only; it did not validate the model, identify passengers, inspect individual flights or test alert delivery.

Still unclear

• How often does the model issue an extreme score for a harmless event or a data change?