What I’m most interested in is knowing how moderation works behind the scenes. It can’t be that dang and sctb (and other mods I don’t know about) read every single comment. Yet they rely to comments that violate the guidelines and warn users.
There must be some software analyzing comments based on different parameters (length, downvotes, scans of problematic words/phrases, and more...machine learning/AI too?) and flagging them in a queue for attention.
I think that it would be entirely practical to skim every new comment on HN given a simple chronological list without threading. HN isn’t actually that big.
Metafilter is another example of a small-ish, highly moderated community that does a thorough job with a small handful of dedicated moderators.
> I think that it would be entirely practical to skim every new comment on HN given a simple chronological list without threading. HN isn’t actually that big.
The oldest comment on the first page of https://news.ycombinator.com/newcomments was posted 15 minutes ago. Reading through all of them took me 5 minutes. Ignoring the fact that posting frequency varies throughout the day, reading every comment would require a single person to spend 8 hours a day every day. So it seems possible to moderate HN like that, but I really wouldn't want to be in dang's shoes monitoring the contextless comment stream all day.
Well the article mentions that a thread with more comments than upvotes is ranked lower. In this case it was the top thread for an hour or so before being thrown to #26. This is interesting and I personally never upvote or downvote individual threads, but maybe I'll start.
There must be some software analyzing comments based on different parameters (length, downvotes, scans of problematic words/phrases, and more...machine learning/AI too?) and flagging them in a queue for attention.