I'd say the most important thing for .LRN would be to bring back the "killer features" that the original education package (the one from Randy's Thesis) had. Those are:
1. Web interfaces for managing tasks (assignments, handouts, exams, lecture notes, etc). From my experience using a system derived from ACES 2.0 beta I've found that the file storage is not a good abstraction for most people.
2. Groups of students working on projects together.
3. A grading system that supports the aforementioned groups.
4. Randy's education package had some pages in which professors could see at-a-glance all the grades given to a particular student. This page would display the student's photo, a small blurb written by the student, and a list of all grades received in that class. There was also a page for TA's in which professors could see a photo of the TA, the blurb, and a list of all grades *given* by the TA. This allowed professors to see in one page wether the TA was a high grader or a low grader.
We had also a page in the education module (that we copied and modified from the intranet module) which would display in a portlet information about a random student (the photo, blurb, and courses in which he was enrolled). This was actually the one feature that allowed me to convince professors to use the system. They said the hardest thing about teaching a class was memorizing student's names!
These might be trivial procs or small enhancements, but no matter how robust or sophisticated .LRN is, these small details make all the difference when convincing professors (and the institution) to actually use the system.
Regards,
-Oscar
