Over the weekend we put a new release into production that significantly enhances the article ranking system. News articles are now ranked by inbound links, tweets, likes (on newsforwhatyoudo), reading frequency (on newsforwhatyoudo) and comments (on the blog itself). Together these elements give a decent sense for how important the community of blog readers thought the article was. News pages display articles first based on importance, and then if there's a batch of articles with similar rank, by date, allowing you to read what's important first. Thanks to the folks at Backtype.com that are providing tweet counts and some of the comment data.
The ranking system is growing in complexity - getting ranking to work on a per-group basis was interesting. There are many knobs that are begging to be turned, such as the relative weights of the various ranking elements - if you see something that just doesn't make sense, give a shout out.
Next on the agenda is a much needed interface upgrade that will allow you to read all the available articles at your own speed - currently we only display the top ranked ones.
0 comments:
Post a Comment