[lai-yin]

Absolutely agree with this point! You also raised another interesting idea - since we don't know when a giant impact or similar catastrophic event will occur, we can't categorize it as a near-term risk or long-term risk. It could happen anytime within the near to long range. How are risks on that broad time scale defined?

[hotpotamus]

Actually, my understanding is that astronomers have a pretty good idea of where all the dino-killer asteroids are in the solar system and that none of them are a problem for at least a century. Comets are more of a risk, though not much of one. That much is known at least. And similarly, the Yellowstone volcano is a bit over-hyped. The most likely disasters are a pandemic (seems quite likely now.) or a large coronal mass ejection aimed at the Earth, but those would primarily affect humans/technology, so again they might actually be good for the overall ecology.

There have been mass extinctions before, but it's a thing that happens once in awhile in geological time, so not really something to keep us up at night. Really human caused disasters - climate/pollution, nuclear weapons, rogue AI, are the things that are likely to get us.