I made something like this in like 2007 called Apocalypse Feed. It took in a few factors and aggregated them into a 0-to-100 number that updated and published over RSS. First it pinged debian mirrors around the world and made a map based on mirror city's lat/long: green for online, red for offline. If there was a cluster of red, that part of the world was considered gone. Then it checked space weather data and nearest asteroid, increasing the value if it was looking bad. It scraped news headlines looking for key words like zombie, pandemic, virus, war, bomb, etc. These fed into a pie graph showing what "type" of apocalypse was most likely at any given time.

It was all fun and games until my VPS host banned me for pinging too many people every few mins.

https://web.archive.org/web/20110516084503/http://www.apocal...

> 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.

1. I think if nuclear war is actually immanent, your best bet of an early warning is an EAS National/Presidential alert (https://en.wikipedia.org/wiki/Emergency_Alert_System), because I'd hope people with access to actual early-warning sensors would cause one to be sent (while they're getting ready for a second-strike attack).

2. I think the logic behind this is flawed. What's the flight time for an ICBM? 20 minutes if from Russia, and less than that from a submarine? I don't think a billionaire could get to his jet in time, unless he lives on an airstrip like John Travolta. Some might get some early notice if their country planned a first strike (but I doubt it, as loose-lips like that would probably give the enemy notice, too).

This has the same issue as many other types of event warning systems based on noisy, incomplete data.

The latency of constructing a semi-reliable warning signal from the data sources described significantly exceeds the latency of event onset. You can modify the algorithms to reduce latency but then the false positive rate skyrockets. Not what you want for an "apocalypse" early warning system.

To mitigate this you need more data from more diverse sources and lower latency feeds.

I'll ask the obvious: wouldn't the aircraft just take to the skies directly, without bothering with the formality of setting their transponder, if they were knowingly escaping an apocalypse scenario?

Fun idea of a metric, but if I'm reading this correctly, we get roughly one apocalypse warning per year?

> Level 5 is calibrated so only the highest daily peak in the trailing year should exceed it.

Do you think that rich people are on some sort of private 'end of the world' mailing list?

There was a Sci-Fi book I read where this was a service provided to rich people. Basically you signed up for it, and you'd get a text when everything was about to go down. Time to drop everything and fly to your bunker.

I have something somewhat similar at <https://blog.sentinel-team.org/>, tracking events that could kill over a million people.

>>we suspect that many people who have access to private jets will immediately take to the skies and escape city centers

Why would that be true? There would never be enough warning to get to the airport and take off anywhere, even if everything else was still working perfectly.

in case of Apocalypse you think they're all filing flight plans?