http://www.vandenoever.info/blog/2015/11/15/ideas-for-odf.html

(for next version, 1.3)

Interesting, normalization so that document versions can be diffed without lots of noise seems like a big/obvious win.