I like Keith's
From: =?0?Q?Keith_Moore?= <moore(_at_)cs(_dot_)utk(_dot_)edu>
method since it goes further than the Smart/Freed proposal by making
the old From, To, etc headers completely self-contained, and also
because it allows multiple charsets in one message.
While proposals based on Nathaniel's Encoded-Variable method, such as
my
From: Yasushi_Nakahara <ynk(_at_)ome(_dot_)toshiba(_dot_)co(_dot_)jp>
are undoubtedly more "beautiful" when viewed in unextended UAs,
Keith's method makes sure that all the info is in the From, To, etc
headers, even if an old UA is used to reply to such a message, etc.
One might also say that Keith's method is likely to get people to use
new UAs since the encoded names are so ugly.
I wouldn't mind if something like Keith's method is included in
RFC-XXXX, even if it was mandated.
An alternative approach would be to simply write up two experimental
RFCs based on Nathaniel's and Keith's proposals, and "wait and see"
which one catches on.
Regards,
EvdP