While I sympathize with the difficulty of writing a document for an
unfamiliar audience, I generally agree with Mark's recommendations
for improving the document.
Some notes:
1. Even if you've had to argue long and hard for model/*, the RFC itself
doesn't have to contain that argument. Also the argument that
"Nicholas Negroponte said X" isn't likely to have a lot of weight
with IETF types anyway ("we reject kings, presidents, and voting...")
2. It isn't necessary or useful to include large sections from other
documents. The rules for defining new MIME types will undoubtedly
(hopefully!) evolve in light of experience, as will other parts
of the MIME specifications. There should be one definitive source
for these rules, and (as I understand your proposal) the rules
for registering model/* subtypes aren't expected to be different
from other types anyway.
3. I don't think it's necessary or feasible to exhaustively list
security considerations for all possible subtypes of model.
If there are general security risks associated with using
model/* subtypes, you should describe them, but the real
security analysis will have to be done on a per-subtype basis
anyway. I would expect your security analysis section to
say something to that effect and be relatively brief.
Regards,
Keith