Hasn't been done to death already
(e.g. ... writing a blog engine is not
what I'd consider as interesting,
although it's technically challenging
- it's been done to death and there are so many ready blog platforms
today)
If this is just a learning exercise, why do you care if its been done to death? More than that, it seems like a blog platform involves a lot of the fundamental skills you'd need to learn anyway to get up to speed on ASP.NET.
You could also try writing a:
messageboard
web-based source-control system.
wiki engine
SO clone
Music/movie management system
Input two celebrities A and A', output a list of movies where A appears B, B appears with C, C appears with D, D appears with A'. See also: Kevin Bacon.
Start your own internet phenomenon. Lolcats, FML, NotAlwaysRight, GraphJam, Passive Agressive Notes, FSTDT, FailBlog, Sh*t Bricks, Keyboard Cat, and JapanWTF have already been done. Find a meme and run with it.
Searchable online taxonomy of species
Decentralized usernames (OpenID), avatars (Gravatar), status updates (Twitter), and currently playing music (Last.fm) have already been done. I predict the next big social network phenomenon will extend the phenomenon by decentralizing another staple of social-networking sites, probably a "current mood" or "signature" that follows you from site to site.
photo gallery engine
a website where people post great ideas for a website.