This seems like a pretty idiosyncratic use of the word
"negotiation". Is a sign posted on a fence stating:
"Trespassers will be mauled by rabid Pit Bulls"
It is the term of art that is used.
In the past 'negotiation' has meant multiple round trip protocols like
ISAKMP which loosely described is a negotiation of the context for
negotiation of the parameters for a key negotiation.
There was also a disastrous attempt to introduce negotiation into the
base HTTP protocol to make it 'extensible'.
The curent consensus in protocol design is that the only negotiation
that is required is:
* A responder (i.e. a server) should post a notice somewhere that says
'This is what I accept'. The sender can then work out if they can
provide that.
* An initiator (i.e. a client) should post a notice somewhere that says
'This is what I do'. This allows the recipient to prevent downgrade
attack.
In our case we have to ensure that the policy mechanism works and is
capable of supporting a transition to new uses of DKIM. Until we achieve
that we do not get the RFCs approved by the IESG.
This is an area that has traditionally been considered 'difficut'. The
main reason it has been difficult is that people have insisted on using
agenda denial tactics to insist on approaches that are insufficient and
then unsuprisingly fail to work.