The Economist reports on some research undertaken by Fang Wu and Bernardo Huberman, a pair of researchers at Hewlett-Packard's laboratory in Palo Alto, California, to compare two strategies used on news sites: one based on the previous popularity of a story, the other on its novelty. The researchers based their study on analysis of articles posted to digg, and the article provides some insight into the digg algorithm and the half-life of an entry (about 69 minutes before an entry gets demoted from the home page).
They summarize with some editorial advice:
[W]hat it means in practice is that if you run a website, you would be wise to learn more about exactly how interest in your stories cools off, if you want to display those stories in a way that will entice the largest number of people to read them.
http://www.economist.com/science/displaystory.cfm?story_id=10804121