Certainly some X should be good at Y.
What exactly is stopping you from doing an experiment, instead of asking us to imagine the results of an imaginary test of an imaginary algorithm?
Scrape up a few weeks' worth of comments (perhaps via RSS, if that works; I wouldn't know) and see what you can do with them. Show us side-by-side article pages with the actual order and the Whizbang Algorithm's redisplayed version.