This is definitely a key area of interest for ed.gov. Ideally, what they want to see is an open source version of a cognitive tutor (http://en.wikipedia.org/wiki/Cognitive_tutor) and/or other intelligent tutoring systems (http://en.wikipedia.org/wiki/Intelligent_tutoring_system). There's a great deal of frustration with the folks at Carnegie Mellon, who did a lot of the cognitive tutor research with grant money, and then turned around and handed the research to Carnegie Learning, who are now putting intellectual property protections around that work. It might be a bit of an uphill climb to replicate that work as open source, but it's precisely that kind of project that ed.gov hopes to fund with their $500m over 10 years.
Anyway. That was the meat of my conversations. ed.gov likes open source, gets open source, and wants to fund open source to help education. I've encouraged them to articulate *particular* problems they'd like to see solutions for, and will continue to press this angle. The more effectively ed.gov can say "gee, it would be great if we could solve X," the more effectively we will be able to drive the geeks of the world to start working on a solution for X.
I refuse to believe they'll do a damned thing for open source until I see some actual licensed code and content. These foundation types talk a good open source game have no track record -- none whatsoever -- delivering the goods in K-12.