Jason Dowdell gave me an exclusive on a new patent filed by Become.com named "AIR" - affinity index ranking. Jason tells me that its going to provide much better results because humans can identify certain sites / pages that are high in rich content and then the machine learning algorithm takes that feedback and scales. He goes on to explain that "they're able to add the human touch to their algo in a way that scales because they're in a specific vertical - none of the other engines can do that." I believe they actually call some of these online stores and then add them to a special "reservoir". Jason adds that "google / yahoo / msn use the random link method".