Quoting Standlee, Kevin ([EMAIL PROTECTED]):

> On the other hand, many people object strongly to "reply-to munging,"
> and indeed, one of the lists to which I'm subscribed (one of the L.A.con
> IV staff mailing lists) had someone recently had a spate of messages
> triggered by a member dropping off because as a matter of principle he
> refuses to be a member of a list that sets the default reply-to as the
> list, rather than the sender.

A melodramatic overreaction, in my view:  Selecting a good mail client
allows one (i.e., those objecting to the practice) to nullify almost
100% (with one exception, detailed below) some mailing lists' forcing of
an add-on Reply-To header to all posts.


My own effort to shed more light and less heat on the subject:

The Reply-To header is a device intended for _users_ to set in their own 
messages, indicating an alternate address where any direct reply should
be sent.  E.g., if I know my [EMAIL PROTECTED] mail will soon be
inaccessible for a while, I might prepend "Reply-To: [EMAIL PROTECTED]"
to signal to correspondents' mail clients that their subsequent replies,
if any, should go to my alternate address.

"Munging" refers to the practice of summarily stripping and discarding
all such _senders' own_ Reply-To headers, on mail arriving at a mailing list
address, and substituting a "forcing" Reply-To header pointing directly
back to the mailing list address.  Thus, one of munging's effects is to
silently overrule and ignore the sender's directives, if the sender has
employed Reply-To for its actual intended purpose.

> [Munging] does occassionally lead to someone who meant to reply
> privately sending an embarassing message to the entire list.

This is easily munging's most destructive side-effect, and is notorious
for causing both recurring trivial annoyances and rarer but serious
personal unpleasantness.  Common but trivial:  Pretty much all those
ubiquitous "Drat, I meant to send you off-list private mail.  I wonder
what happened?" messages you see daily on mailing lists aimed at
non-technical users (where munging is most common) owe to this effect.
Rare but serious:  Job terminations, broken friendships, and divorces
are not unknown.

Why?  Because the _other_ immediate effect of munging (aside from
silently overruling and discarding the header's _legitimate_ use) is to
make direct replies (as opposed to group replies) artificially difficult
and the mistake of mis-addressing highly private mail to the public very
easy -- except for people like me who use really good mail clients
capable of detecting and _removing_ munging.  Thus, technically
sophisticated users (who tend to strongly oppose munging) suffer no ill
effects; by contrast, it's the less technical users (who tend to favour
munging) who (overwhelmingly) are hurt by it.

(Additionally, sometimes the munging triggers huge storms of looping
mail, when someone's badly written "I'm on vacation until [date]"
mailbot keeps responding to its own list postings a couple of times 
per second.  Authors of mailing list software attempt to intercept and
short-circuit this catastrophe by attempting to recognise all varieties
of mailbots ever written -- past, present, and future.  As you can
imagine, this mitigation tactic is only partial successful.  The
disasters are by now infrequent, but still gruesome.)


Technical users also aren't hit by what John David Galt cited:

> On the gripping hand, some of us despise the habit of using "Reply
> All" because it sets the To line of the reply to "To: original-sender,
> list-alias", thus causing the person who sent the first message to get
> two copies of every reply.

Notice how my response stripped all addresses other than the mailing
list itself?  That's because I use a well-written mail client, that
comes configured to ensure that:  Kevin and John will each receive only
one copy of my reply.

I might still get two copies of _other_ people's replies, who are using
less capable mail clients.  This is, as John notes, the disadvantage of
_not_ configuring list-management software to force (munge) Reply-To:
You get a direct reply because the later poster replied to one of your
mails, plus a separate copy from the mailing list software because
you're a subscriber.

Technical users can remedy that situation easily with a simple
"duplicate" filter, if it actually bothers them.  Less-technical users,
of course, tend to get up in arms about it -- and cite it as a reason to
"munge".

So, as a user of a really good mail client, I (basically) inherently
sidestep all of the harmful effects of munging except one, the
"exception" I mentioned, above:  If any poster has tried to use Reply-To
for its intended purpose, I'll never see (or benefit from) that use,
because the mailing list software autodiscarded the user's header before
I got his/her mail.

Kevin wrote:

> My theory is that [non-announce-only] lists are _discussions_ and
> therefore by default you want your reply to go to the
> whole community, not just the original sender.

This is the usual chief justification.  

In the broad picture, any mailing list is an effort to wrangle SMTP
e-mail -- which is structurally a one-to-one communications mechanism -- 
into behaving like Usenet netnews, which is structurally a many-to-many 
medium.  Because the technology is a poor fit to the intended medium,
you will inevitably get unwanted side-effects from any configuration.

Without munging, you get "duplicate" e-mails between participants in
many reply threads.

With munging, you get a continual stream of "Drat, now why didn't that
go to private mail as I intended?  Sorry, guys" postings, plus the
occasional odd job termination, divorce, or bloody nose -- and other
artifacts of crippling the mechanisms for direct as opposed to group
replies.

In netnews, there's a clear and universally understood distinction
between a "followup" (which is a response back to the newsgroup) and a
"reply" (which is a direct response to the sender via e-mail instead of 
netnews, to the other guy only).


The technical user's traditional response to Kevin's traditional
pro-munging justification (above) is "The _correct_ way to get default
replies back to mailing lists is to point out to users that there are
always _two_ reply modes, not one, in their mail clients.  Please don't
cripple one of those two modes just because some users are so clueless
they can't simultaneously hold two related user-interface concepts in 
their tiny little heads."

That usually concludes the ritualised exchange, at which point, someone
puts the question up to a vote, and the technical novices inevitably
prevail (except on techical mailing lists) -- and then those same
less-technical users suffer (as described) for their preference, while
the more-technical users are able to eliminate munging's ill effects
from their own mailing list experience.  

Heartless bastard sysadmins tend to regard this dual outcome as poetic
justice.  I personally haven't decided exactly how heartless I am, today.  ;->


Reply via email to