The idea of p(doom) is really convincing me that pure probability cannot sufficiently express uncertainty.
The Bayesians say that if you're uncertain about p(x), you just work that uncertainty into the probability and you get a new p(x).
But for p(doom) that's really unsatisfying. Saying my p(doom) = 0.5 doesn't really differentiate between having no clue what's going to happen and being really sure that there's some pivotal event that could change everything (like an election).
I don't know if something like Dempster-Schafer theory is the answer here. I think it's more frequentist than Bayesian and it's kind of too complex for everyday conversation.
https://en.wikipedia.org/wiki/Dempster%E2%80%93Shafer_theory
Dempster–Shafer theory