I'm not sure how much of a problem MIME header dropping is going to
be, but I'm pretty sure that nobody else knows either (not even Ned).
I certainly don't have anything resembling authoritative data on this!!! Thus
far a number of systems that cannot cope with MIME headers have been
identified, and at least some of them have been fixed. I'm confident that more
systems will be found in the future. The LISTSERV problem has been reported
previously; you already described the obvious workaround.
I don't think that tricky embeddings of MIME are the right way to avoid this
problem with LISTSERV; it is quite likely that LISTSERV would then be adapted
to deal with the embedded form as well. LISTSERV is an actively developed piece
of software.
The approach here should not be to try to be trickier than the LISTSERV
maintainer; the simple reality is that he can probably implement things a lot
faster than we can propose and standardize them. I suggest that a better
approach would be to deal with what LISTSERV does now head-on; if a change or
an alternative approach makes sense we should propose it to the author, Eric
Thomas. This may well drag in all the various BITNET-MIME problems, so don't
hold your breath waiting for an answer...
All this begs the question of whether or not such a bypass scheme makes sense
for other reasons. For all I know it may make very good sense. By all means
write it up if you want to.
I will say that in order to be extremely useful such a scheme would have to be
fairly widely deployed; if it isn't widely deployed it will at best be a
point-to-point solution over links that trash headers. Historically speaking,
this is somewhat similar to the RFC1137 restricted address encoding, which
enjoys some popularity in dealing with enclaves of systems that cannot handle
nasty addresses, but has not been deployed on a large scale as far as I know.
Ned