All Guestrin's projects involve information flowing through a network, whether it's temperature data in a forest, forces applied to a chair as a person sits down, or even news of a politician's affair sprinting through the blogosphere. In each case, the algorithm's goal is to get the best data through as few sensors, or with as little effort, as possible. A blog, for example, is a potential "sensor" of a story. Guestrin's algorithm defined the best blogs as those that break big news stories early but don't force users to read through too many posts to find them. As with the forest, it picked one, re-ranked, chose the next one, and so on, narrowing a list of 45,000 possible blogs down to the most efficient 100. (When Guestrin ran the program in 2006, Instapundit topped the list.