Plenty of Very Important IETF initiatives fail. Look at IP multicast. Clearly, 6 billion endpoints don't need globally routable addresses for the Internet, in its current service model, to work. Moreover, it's unclear whether once the addressing problem is solved, the scalability problems will be tenable.

I don't know why the Kuro5shin comments are so harsh. I think the article makes a decent point. Here's a credible backup (although, from someone who is ultimately an IPv6 supporter):

http://cr.yp.to/djbdns/ipv6mess.html