The reason is simple: the W3C has a long history of technically flawed and unreasonable specifications, and the WHATWG was born as a reaction to that madness and to bring the web forward after many years of stagnation while the W3C was masturbating with things like XHTML2.
When the W3C noticed that they were going to become irrelevant, they had a brief moment of sanity and accepted to collaborate with the WHATWG, but now seems like they are back to their old habit. Politics trumps solid engineering and decisions are heavily influenced by loud people who will never implement or use the technologies that they are proposing.
See e.g.: "Remove Section 5. Microdata". Hint if you don't understand what's going on: RDFa sucks, its proponents know it and so they are acting destructively on other technologies that may actually work and solve the problems that RDFa is supposed to solve. Hixie is the good guy here and Shelley Powers... well I really don't understand why she acts the way she does, ignoring any technical argument that doesn't fit inside her predetermined "solution".
Other references:
- http://www.w3.org/2001/tag/2009/11/02-minutes.html#item06 (the data-* attributes are so useful and widely used that even I use them in one of my websites);
- http://hsivonen.iki.fi/rdf-competition/;
- http://hsivonen.iki.fi/ (some of this is obsolete, scroll down to "Blogish Notes" for the juicy new stuff).
