The Wall Street Journal has a really good article today on the topic of science, non-traditional data (non-survey based), and projects/methods for capturing the data as well as beginning to analyze.  I was first introduced to this topic by David Kirsch at the University of Maryland in a meeting with an official from the National Archives.  David, who has training as a historian but deals mostly in entrepreneurship and related topics, has done some very innovative work to secure interesting potential data from being lost.  Indeed, there should be a new call for research proposals coming out soon on some data which David helped to preserve.  I will post the call as soon as I see it officially released.  At some point it might be worth codifying some of what our scholarly community is learning here.  It strikes me that there are actually more people doing work here than I'd realized, including Jon Eckhardt from Wisconsin, and I suspect many scholars more from the industry studies community. 

August 29 ammendment: I neglected to hit one key point in my original post.  A huge amount of potentially helpful electronic data stands to be lost unless proactive steps are taken to preserve it.