Instead we have a company that throws specs over the wall like they're teenaged nerds that have just discovered Pascal's pack function, a company that pushed SPDY on everyone despite it relying on a draft SSL feature that got rejected at the IETF (and in the space of 2/3 years we now have three HTTP variants - and why? Because 3 years ago it shaved 1ms off a maps.google.com request). Did you even take a look at the original WebSocket spec? It worked in terms of comparing the entire start of the HTTP request using memcmp()!
Regardless of my opinion of NaCl, this is stuff that should be getting defined slowly and in the open by bearded old men that have spent their lives in glacial, trusted industries like telecoms, or just about any other mature technology industry. Not a throng of Java graduates who think they're changing the world by continually dumping crap specs over the wall and paying for subsidized articles to promote them.