Sorry but I think what you posted may be misinformed rubbish. Steve stated himself that STUN is in use when using Facetime. This means that there is a server out there that handles part of the call setup, even if it is only NAT translation part.
I suggest you read up on how VoIP and NAT work before saying that the phones do it all themselves.
Sorry, that seemed overly harsh. To clarify, if you and the other caller are sitting behind NAT routers, and you both have private IPs, how do the phones communicate? They can't because, assuming that you are sane, you aren't port forwarding any ports and even if you were, how could you know which ports you need to forward?
This is where the STUN server comes in. It works out what your public IP is and what ports the given application is using. Without this you just have two applications that are able to shout out to the Internet but have no way of answering back because your firewall will block them.
At least that is my understanding of it. There are many articles out there that describe how it works and since Apple want to make Facetime an open standard, and they are already using existing open standards to facilitate its use, then I'm sure more detailed documentation is available, if not now then shortly, so that developers can get to grips with it.