Another type of data that I'd love to be able to get access to is course information. Currently this is sitting in a big composite silo called UCAS - the University and College Admissions Service. 
Of course, like libraries, universities are engaging in a nice big "boil the ocean" type development project, which I can predict now will be thoroughly fruitless. It's called XCRI and has so far managed to get a small handful of institutions to put up a small trickle of information online in a super-complicated XML format.
It's always easier to write the schema than getting people to go along with it. The more complicated the schema, the more likely this is to be true.
A guy called James Mellor has proposed a far more sensible solution - namely some kind of microformat for course information.
I think this is far more sensible. A modification of hListing for the purpose of semantically marking up university course information would seem far more sensible. The thought that universities would sit down and understand XML schema documents is too much to take.
Instead, if they just slap a few DIVs and SPANs marked up with the relevant class-names on their content, someone like me could write an abstraction layer on top of it with outputs available in whatever formats one needs. Then the fun begins.
I think that these "boil the ocean" schemas have a simple goal in mind - to put off actually doing something. Typical academic move. Spend a few years coming up with the most perfect schema possible, then not implement it because it's too complicated.
For all but about 1% of purposes, getting it out in a rough and dirty format that we have to hack around a bit to get working (like microformats) is fine. It has the significant advantage of being something that might actually be implemented sometime in our lifetimes.
