I'll probably regret getting back into this thread, but I wanted to
amplify something Keith said a while back about typical users:
I suspect that when people hit "reply to all" to a message from a list
they generally want the reply to go to everyone who received the
original message, not "just" to the list. They don't want to worry
about which of those recipients are on the list and which ones are
not.
I'd go further and claim that the typical user lacks the
interest/attention to deal effectively with more than 2 reply choices,
which the user will naively think of as "the sender" and "everyone."
Trying to get most users to handle a third choice -- e.g. "reply to
all" vs. "followup" -- is asking at least as much of the user as
expecting them to simply edit out the duplicates by hand. (This is why
I ultimately gave up on the three-choices model in Andrew.)
We need to map system behavior onto abstractions that normal users will
be able to comprehend. The present situation is actually be pretty
good in this regard, except for the little matter of duplicate
elimination. But most of us realize that isn't possible in all case --
if I send mail to "foo(_at_)bar(_dot_)com, baz(_at_)ola(_dot_)com" I have no way of telling
that foo(_at_)bar(_dot_)com actually forwards to baz(_at_)ola(_dot_)com(_dot_) In the most general
case, the only duplicate detection you can completely depend on will be
on the receiver's side. This suggests two possible courses of action:
-- Change nothing, and leave most duplicate-eliminateion to the
receiver-side software.
-- Try to optimize some specific common cases, which is the goal of MFT.
However, I question whether MFT might not cause as many problems as it
solves. Consider:
To: List(_at_)example(_dot_)com
CC: Bill(_at_)x(_dot_)com, Bob(_at_)y(_dot_)com, Jim(_at_)z(_dot_)com
This is something I have done in the past in order to bring into a list
conversation a few people who are not on the mailing list. Now, what
should the list processor at example.com do with this? As I understand
it (please correct me if I'm wrong) current implementations would
redistribute the message like this:
Mail-Followup-To: List(_at_)example(_dot_)com
To: List(_at_)example(_dot_)com
CC: Bill(_at_)x(_dot_)com, Bob(_at_)y(_dot_)com, Jim(_at_)z(_dot_)com
Thus people whose UA's respected MFT would end up omitting Bill, Bob,
and Jim from their replies, which was not my expectation as the sender.
Is this really the desired behavior? Or should all the non-list
addresses except the sender in fact be copied into the followup header?
One could argue that the MFT should include everyone that would
normally be in "Reply to All" *except* those who are on the list --
after all, if the list processor is doing the rewriting, it's in a
position to do that, right? -- Nathaniel