It would be interesting to analyze HN comments at scale to see if there are predictable genres or features of comments that are upvoted. Same for down voting, b

It would be interesting to analyze HN comments at scale to see if there are predictable genres or features of comments that are upvoted. Same for down voting, but it seems easy to predict downvotes. What pockets of the HN readership are rewarding what genres of comment? I am sure this is not a fresh idea... Readership could choose to reward TLDR summaries of submissions. In general, discussion would then be better informed about the content submission. Personally, I have found interesting discussion to sometimes be not so much about the content of a submission, but about the topic of the submission. A tedious pattern for content-focused stuff is that "what the author didn't do ..." post; a more engaging one is essentially "yes, and here is related or parallel work..." Elaboration is more interesting than negation for this reader. Up or down voting is a big signal of relevance for the reader but not the only one. Really what I would like is some agent that can read HN submissions and comments for me, extract topics from submissions and interesting comments and links from discussions, and allow me to browse that digest first. Again, not a fresh idea.