> Anything that more than two "real user" entities are going to want to use, > and *might want to inter-operate*, is probably standards-track material. I disagree, and that's not the metric that has traditionally been applied. A protocol that doesn't work over the Internet in general, but which only works in limited environments and for limited groups of users, has not generally considered suitable for IETF standardization, and this is entirely appropriate. IETF's endorsement of a protocol is taken to mean that the protocol has broad applicability. IETF doesn't have the resources to standardize every protocol that some small group of people might want to use. Nor would it really serve IETF's purposes for it to do so. Small groups of people can agree among themselves how to solve a problem, and they don't need to require that their solutions be universally applicable; it's only when you have large numbers of users that you need to impose the criteria that IETF expects for standards. > About the only reason not to do it would be if it was decided that the > design chosen was technologically flawed. However, I think NAT should have > taught the IETF (as bridges taught me) that sometimes what seems > technologically flawed is in fact just what the market needs and wants - > usually for some reason one hasn't sufficiently appreciated. I agree that the existence of NAT should have helped us recognize legitimate needs and wants, but that doesn't argue for standardizing NAT either.