> How can you identify this systematically (vs it being just your opinion), but not identify hate speech?
What I think distinguishes malice is the prosecution example. Where if someone makes an argument about you personally, and they are unable to abstract the idea from you personally as an individual, that is acting with malintent toward you.
Antagonizing someone personally by trying to scandalize them in the eyes of others (performatively, as though in front of a judge or jury, or public opinion) is not discourse, it's just persecution and animus. The modern internet version in the form of, "hey everybody, this person is an X!" is probably malice in its more pure form, with a spectrum of dilutions that are variations of, "surely you aren't tainting yourself with this taboo!" after that. It's a purity game, but really a proxy for the 'deadly sin' of wrath and the secular concept of animus. In this sense, it's something that with reflection we recognize ourselves as capable of and learn to moderate it within ourselves.
Hate is something we only recently started accusing others of. I do think it's part of a substitute, secular, belief that persuades people they cannot ever be good, so instead we can repent of our recieved worthlessness by giving up our moral agency and becoming _anti-bad_, all while accepting poor treatment and even demanding it for others because the people we have animus for are also equally bad. The only hope is that by redemption through criticism, some are less worthless than others, but that's the upside. It's like a profoundly false religion whose core tenant is that you believe in the universality of hate, and then be against it. Things are shaped by the forces they oppose, and I think indexing on hate has had the effect of moralizing malice. When we start with the premise that we are all interchangeable group elements without moral agency, and we must redeem ourselves by becoming above criticism, that is a system of slavery with the reins in the hands of the most zealous critics - who happen to be other living people, and not a relationship with a self or an ideal. This whole cycle is based on this _anti-bad_ negative definition that mostly seems to moralize malice in its followers. Back in the day they had to call them sins because they felt good, but really weren't.
Anyway this isn't your outgroup either, it's us. In terms of detecting it, I think ML is just on the cusp of doing it as well or better than most people and that's the future of forum moderation. This question of whether we can observe intent is going to be a big one. It's a huge topic.