IMNSHO, a protocol extension will succeed only if implementing it brings
value to the user. If implementing e.g the SMTP MUMBLE extension
doesn't benefit either the implementor or the user, and does benefit
someone else greatly, then that extension isn't going to be implemented
very much.
I may be wrong, of course. But let's suppose I'm right. In that case:
1. Mobile MUAs get considerable benefit from compressing messages as
they go from/to the server, so an SMTP/POP/IMAP extension to provide
compression for all messages or all attachments ought to be fairly
widely implemented by authors of such MUAs (and likely seldom by
others).
2. Even mobile MUAs don't get much benefit from c-t-e: deflate, since
that works only when both the message sender and all receivers support
it, which'll be fairly seldom. Authors of clients who assume ethernet
to the server almost never get any benefit.
Arnt