NR has no way to specify situations where "reply to all" is not sent
to a superset of the recipients who would be sent if "reply to author"
were used instead.
Then NR is broken.
Or alternately: NR may not solve the problem you want to solve.
Depending on how you look at it, NR and MRT are both broken (for
different reasons), and the absence of some solution(s) to several
problems associated with replies is also broken. The question is
whether, and how, we can get to a significantly less broken state with a
minimum of pain.
When I send a message to ietf-822(_at_)imc(_dot_)org, I do _not_
want followups to be sent to the original reply address.
Note that even without either MFT or NR, you're free to set Reply-To to
ietf-822(_at_)imc(_dot_)org(_dot_) If the recipient really wants to reply to just the
author, he can always manually override the To field of the reply. (and
if he can't do that, his MUA is broken :) That's no different than many
other valid uses of Reply-To: e.g. when someone sends a message and
wants normal replies to go to his secretary, but a particular recipient
of that message really does need to send a reply to the message author.