The problem is, number of comments doesn't seem to be highly correlated with how interesting a story is. Sometimes a story has a lot of comments because it raises a lot of interesting issues, but sometimes it happens because people get onto an interesting topic only tangentially related to the story. Without AI it's hard to distinguish between the two cases.
I remember reading a great post somewhere about using the visual shape of usenet discussion threads (as rendered by common news readers) to spot patterns like trolling, political tangents, etc. This was only proposed as an NI (Natural Intelligence) method, not AI, but I wonder...
http://news.ycombinator.com/active
The problem is, number of comments doesn't seem to be highly correlated with how interesting a story is. Sometimes a story has a lot of comments because it raises a lot of interesting issues, but sometimes it happens because people get onto an interesting topic only tangentially related to the story. Without AI it's hard to distinguish between the two cases.