Hi all,
Summer of code has always been a good way for us to get some important work, that no one has time to do, done. I encourage everyone to come up with good summer of code projects so we have a good number when the time for students to apply comes around. Empirically projects that focus on existing infrastructure (e.g. Cabal) work well. They limit the scope enough for students to 1) make something that's impacts lots of people and 2) doesn't give them too much rope*.
* Building something good from scratch requires lots of experience, something most students don't have, almost per definition.