I found the process of coming up with the estimate just as interesting as the conclusion itself. It is described at http://www.niemanlab.org/2009/04/print-is-still-king-only-3-... (linked to by the original article)