[blaufast]

I think HN readers have a strong desire to speak the inconvenient or untouchable topics. I think this is a healthy impulse.

This topic is hard because so many commenters are triggered by it, for lack of a better word.

Could AI help here? What if a comment was given an automatic 'emotional temperature' score? We might be able to address the fact that a lot of the comments will be basically pure emotion, and we can discuss the emotions instead of being distracted by the content.

[dang]

Sentiment analysis tools haven't yet proved accurate enough to be useful for HN moderation; traditionally they misclassify too many things because they don't have access to intent. It does seem like LLMs have a chance at doing this better, though, and if anyone wanted to work on that, I'd certainly be interested in what they find.