Theories of rare and extreme events are really inadequate. Neither extreme statistical bounding (9-sigma, 12-sigma…), nor catch-all categories like “black swan” or “long-tail events” really get you anywhere helpful.
Conversation
Replying to
The most salient feature of rare+extreme is ontological explosions. In going from 1 to 3 sigmas of phenomenological coverage, parametric range only increases about 50% but variety can explode orders of magnitude.
1
14
1-sigma of household pets is 2 species (cats and dogs).
3-sigma is maybe 10? You can include goldfish, parrots, hamsters.
But at 6-sigma, you get insane numbers, like Pablo’s hippos, people in Texas with tigers… any exotic species that can plausibly be a pet for some nutjob
1
1
10
This is relevant to extreme weather. “A year’s rain in a day” is “extreme rain” the way a hippo is a horse. It’s an ontologically distinct type of weather event.
The jet stream destabilizing isn’t just seasonal weather, it’s like an Australia-worth of “weather event” species.
2
2
17
I’d like an extreme weather index that tracks variety. The historical frequency fails to convey information past a point.
“Aliens have landed” is an event that fast happened in all 6000-odd years of recorded human history as far as we know.
That’s… not a useful observation.
1
6
* has not happened
2
Replying to
seeing like a state- and hayekian-thinking seem to explain much here
which is to say that the problem isn't the diversity of the world, but the attempt to neuter of the world via goals
1
4
growing corn successfully is really hard and massively destructive (by necessity) because it's only through destruction of environment that we achieve goals
put another way, the only way to control a cow is to demand that it graze and wander
1
2
Show replies
Replying to
I think that a computability framework is better fit to consider those kind of events than probability theory.




