These are so great! I recognized a couple of them from hacker news as well -- so perhaps they might be usable here in a semi-automated way to detect bad conversations?
(I know there's already a flame-war detector. But other patterns might be useful as well.)
(I know there's already a flame-war detector. But other patterns might be useful as well.)