Unfortunately, the production Internet (ie, since 1983) has never been
fully end-to-end at the IP layer. Never.
this depends largely on what you call "the Internet".
It's fine to create a clean architecture, but not very helpful to ignore or
complain about market-driven extensions (or work-arounds, or...) to it.
so we should pretend that everything is rosy then?
(I can see it now...
"Doctor Strangenet, or How I Learned to Stop Worrying and Love the NAT")
Folks -- people would not be making those extensions unless they
experienced benefit in them.
understood. there is no question that NATs provide some localized,
short-term benefit. just like burning rainforests, or producing toxic waste.
We claim to believe that the market is the ultimate venue for resolving
choice among standards. We need to acknowledge that that applies to
missing standards, as well as competing standards.
The market may be the ultimate venue, but it isn't exactly blessed with
foresight. We most certainly need to acknowledge that when markets
develop "solutions" which damage the net, that it's often due to a lack
of facilities in what the net provides. But that doesn't mean that we
should axiomatically accept "market-driven 'solutions'" as viable in
the long term.
Keith