Ask HN: Why doesn't Twitter/X use semantic ranking for their reply system?

I love Twitter/X, but I can’t stand the current system they have for replies. It boggles my mind when I see completely irrelevant content in the first replies to some of the most powerful people in the world, and most importantly creates a subpar user experience.

Why don't they switch to some type of dynamic ranking? By dynamic ranking, I’m referring to something like an authority index + semantic relevance scoring. Think Google search/SEO, but for Twitter. Replace PageRank with ElonRank and instead of backlinks, the authority could come from follower count/retweets/etc. Semantic relevance would be on a “per tweet” level.

When I see a tweet by the world's leading Neurosurgeon, I don’t want to read random peoples replies first, I want highly accredited and “authority” responses first. You can easily test this. Just go to any famous person posting a tweet and you'll see gold within the replies... but usually way down. A semantic mapping would VERY easily pick out garbage replies and push the more relevant content to the top.

The system makes sense to me since I have a ton of SEO and embedding search experience, but I'd be curious to hear HN's thoughts on this, or another way they could improve their reply system.