Predicting the future configuration of a network is pretty straightforward. Whether you're talking about social networks like Last.fm or biological networks like those associated with protein interaction within cells (you know the type), many of them share common growth patterns, making it possible to predict the future of a network being studied. Unfortunately, you can't really work backwards using the same process, so until recently it wasn't really possible to determine a past state. That's why a paper recently published on arXiv.org is so promising. According to a light beach read called Network Archaeology: Uncovering Ancient Networks from Present-day Interactions, researchers Saket Navlakha and Carl Kingsford have been able to do things like "decompose" actual observed networks to correctly guesstimate when Last.fm users signed up, and determine the age of proteins and how they evolved by looking at how they duplicated and mutated in the past. Rumor has it that Mark Zuckerberg has a keen interest in developing this technology, if only to spare current social networks the fate of Friendster.