HTTP isn't really very distributed
It identifies a file on a server, not the data itself.
A better protocol would be one which identifies data that you own (e.g. some kind of Facebook-like feed, photo album, piece of music you composed, whatever). You publish that data and it is automatically encrypted and replicated to servers around the world. Anyone with the right URL (and credentials if necessary) would be able to view it. Or you could withdraw publication and it would be deleted from servers. If a server went down it wouldn't affect you as the data could be retrieved from elsewhere.
That's my great idea, all I need to do is code the daemon, get everyone with a backbone server to run it, get them to cough up for storage too, probably change DNS a bit as well, then become a billionaire.
Biting the hand that feeds IT
