We could score comments with sentiment analysis. AI, though not necessarily LLMs, can do it very well. Depending on how child comments reflect on the parent, the parent could be scored for relevance, agreement, helpfulness, or any other metric. This would also fuzz the metric itself (karma), and make it hard to game. The Goodhart's law problem in social media (karma farming, ratio-ing, etc) could be solved.

Of course, that presents many other problems. A corporation using AI to steer what discourse is worthy of promotion and what isn't in social networks has already caused much harm. Many things can be said about it.

But as a thought experiment, I think it's an interesting one. Much good could be done, advertisers and financial interests permitting.

I agree, these days I hate using forums that don't have a tree view. Unfortunately, I don't know of any software like that, except reddit.

Sure, but when everyone else uses the upvote/downvote button that way it doesn’t matter how you personally use it. The end result is anything that goes against the hivemind gets suppressed.

I’ve had fully scientifically sourced rebukes of things (effectively, straight up facts) get downvoted to oblivion on Reddit. Hundreds or thousands of people didn’t see that their preconceived notions were probably wrong. It’s no wonder politics has become more insular.

HN is better than most, thankfully.

Say something minimally negative about macOS in /r/macos. You'll be -1 in no time.

I'd like to see someone post the link of retail macOS 15 is not UNIX, only a bastardized version of macOS 15 configured in-house at Apple passed UNIX 03 cert.

Thread would probably die on the vine given the number of users in that subreddit who reference "it's UNIX"!

Reddit has entirely turned into a downvote == disagree/opinion I don't like.