> > if people understand that NATs allow them to run web and mail > > clients from multiple machines but prevent them from running > > most other apps, then I don't have any problem with it. > > And if your objection to NATs ended there, I wouldn't have a problem > with it. But instead of then working to change the protocols that break > with NATs, you continue to insist, Canute-like, that you can turn back > the tides and move the world back to a pre-NAT world. A Canute-like approach would be trying to fix protocols that break with NATs. It's easier to turn back the tides than to recover the functionality lost due to NATs. To the extent that you can recover it, you lose scalability, efficency, reliability, and increase deployment barriers. I'm not trying to do that, because it's clearly futile. Nor do I pretend that NATs will go away without a better solution in place to problems that caused people to buy NATs. Rather, I'm concentrating my efforts in areas that might actually produce useful results. Part of that effort is to improve IPv6. Part of that effort is to try to dispel some of the widespread disinformation/delusion/fantasy about NATs. As for a genuine effort to understand how to build applications that can operate (to some degree) in the presence of NATs, see draft-moore-nat-tolerance-recommendations-00 Keith