> and it's documented and clear and implementable
As someone who has actually looked at implementing it... no, it's not, by the standards of modern web specs. That's part of the problem with it. The other problem is that its creators ignored a whole bunch of best practices for web API design...
Anyway, the FIDO thing (or an approximation of it, which was the best they could do given the quality of the spec) is what Chrome shipped and what they've been trying to backpedal from since.
If you want the gory details as far as Firefox is concerned, see https://groups.google.com/forum/#!searchin/mozilla.dev.platf... for the discussion about whether to implement FIDO U2F in Firefox. Note that it's about 6 pages of posts, at least the way google groups shows it to me... The posts from Ryan Sleevi (representing Chrome) are particularly interesting.