The root of the problem we're facing is that there is a difference between
using IP and using the Internet model. IMS uses IP, but uses a different
model entirely (Whatever you think of that model, I don't think anyone
who looked at how it works could argue that it is the same as the Internet
model.)
The argument that Steven and others have made is that the document set
needs to have an explicit acknowledgement of the other model, so that
there is a realization by those reading them that just because they have
IP, doesn't mean they are dealing with the Internet model. You could
argue that everyone knows that, but the document itself muddies it
far enough that it is far from evident.