This is more hyperbole. How have NATs created a mess out of the network?
Yes, I understand that they've made the network environment more
complicated which makes life hard on protocols designers. So what?
If the customers are getting what they want, that seems to me that it
can hardly be characterized as a "mess". And you have yet to establish
that they're not getting what they want.
NAT (and, to a less extent, firewalls) are definitely
interfering with the deployment of some new services,
including VoIP and multimedia conferencing, of course, but
large-scale distributed scientific computing is a problem as
well.
I'm highly uncomfortable with your argument. I don't think
that the IETF should be architecturally fundamentalist (that
is to say, it's not that I think that we need to roll over
and play dead for every popular bad idea that's out there,
but that I think we should learn to be smart about where and
when to compromise) but I really take exception to the
notion that as long as users are happy there's really no
problem. As I said before, the workarounds that are being
used to help facilitate application traversal of NATs are
definitely introducing new security problems that wouldn't
exist if the NAT weren't there. There are other problems
around robustness and routing. I think this is a bad thing,
myself.
Melinda