Dragaera

Holy War of Reply-To Munging

Fri Nov 29 18:19:58 PST 2002

>      In my experience, the more technically adept the members 

So your desire is to restrict the discussion about Brust's fiction to
the technically adept?

> of a list
>      are, the more likely they are to do things *this* way, 
> that is, NOT
>      to munge the reply-to headers.  Yes, some lists of geeks munge
>      headers.  Many do not.

I think you're making huge assumptions about the configurations of
lists.  Without some effort of surveying "all" (which is probably not
worthwhile) lists in existance, I wouldn't be bold enough to say that
either configuration had the "many" or "majority" position. :)

>      Sourceforge.net hosts THOUSANDS of mailing lists- their policy is
>      not to munge.

That's not necessarily an indicator...Yahoo! Groups probably has lists
in the "thousands", and the Reply-To is configurable, I'm sure the other
providers of similar services are similarly configuraable.
Additionally, Sourceforge's intended audience is different from that of
a fan-list.  Hmm.  Also, with sourceforge, you have to be careful not to
confuse "forum" with "mailing list".  They provide _both_, and the
forums have "subscription" functionality, which send you copies of the
posts to those forums you subscribe...which you can't reply to at all
>from your mailer.  Even more obnoxious than the reply-to of this list
(to the point that I only monitor those discussions, because it requires
too much time and effort for me to reply).  

However, I can see why you wanted to comment on "how many do it this way
or that".  Because I appeared to be.  My intent was to indicate not "how
many lists in the world do this way" but how many that *I* have to deal
with.  Since I have to make a "mental exception" for this list, I have
_often_ not replied, or aborted a reply, because a) I'm often in a hurry
when doing e-mail and b) I instinctively used "reply" instead of
"reply-to-all" and simply gave up, because it exceeded my effort/time
level. *chuckle*.  Then again, you may regard that as a "feature",
that'd be less noise from me! ;)
 
>      To those who say that sending replies to sender instead 
> of munging
>      wastes bandwidth: We're supposed to be intelligent, 
> learning beings.
>      We've got opposable thumbs, and can have flame-free 
> discussions on
>      religion.  Remembering something THIS BASIC is NOT HARD, folks.

Why should I have to remember it? The point of computing is to make it
_easier_, not harder.  It's to make it _faster_, not slower.  
 
>      Think of it this way... the mailing list isn't a person, it's a
Umm....I hope no one thinks of the list as a person. :) :) ;)

>      way to facilitate mail delivery within a topic.  As with personal
>      email, when you hit reply, your email will be sent to 
But...it's not personal e-mail.  If I want to have a discussion with you
personally...I would initiate one.   To have the "default" be the what I
do least is counter to usability. :).  It'd be interesting to see the
results of a list that the reply-to was configurable on a per user
basis.., e.g., for you, it'd be sent with the author of the message as a
reply-to, for me, it'd be "dragaera at dragaera.info" (Hmm...Dragaera is
interesting to type on the laptop).  I suspect that it would
dramatically compound the problem, since no one would have any
consistent guide as to how the list was working. :)

> the *person*
>      who sent the email you're replying to, not the delivery system.
>      (Yes, it's more complex than that, and no, I'm not going 
> to go into
>      the workings of SMTP and the like.  Close enough.)
 
>      It's really not hard to remember gang, and DDB's right.  If you
>      accidentally reply to sender instead of the list, it's ONE email
>      inconveniening two people, the sender and the recipient.  The
>      sender has to resend to list, and the recipient will get an
>      extra copy of the message.
.... 
>      Other way around:  You slip for a message meant for DDB alone,
>      and inconvenience the whole list, wasting a lot more bandwidth,
>      and possibly embarrassing yourself or DDB or both with the
>      contents of your hero worship mail.

This is possibly a qualitative vs. quantitive issue.  If you only ever
have an equal number of instances of both "errors" then I would agree
with you.  However, I don't believe that to be the case. It's my belief
that the majority of "replies" are back to the  list.  Of course, since
I largely only have myself to judge acurrately, I could be completely
wrong in that belief.

Assuming my belief is true, having the reply-to set to the sender rather
than the list, adds extra work and effort to the majority to save the
few from "forgetting" and making a mistake.  Which brings me to this
quote from your message (moved out of order for effect):

>      If your memory is REALLY that bad, put a post-it note on your
>      monitor frame.
Good idea. :)
 
>      Go one step further.  Some putz puts a "vacation" autoresponder
>      on their email address, responding to EVERY message they get for
>      the week they're not checking their mail.  Every message goes to
>      the list, and causes a mail loop.  Bad Juju.  (Yes, many smart
>      mailing list packages have patches to avoid this... but just as
>      not all email clients aren't smart enough to handle "reply to
>      list," so are not all mailing lists able to handle loops.
>      If we're catering to mediocrity, we'd best do so on both sides.)

I'm not sure this is relevant.  It's bringing in a completely separate
issue: Do's and Don'ts of auto-responder/vacation replies. 
And nearly in the same breath, you present the correct solution to the
problem.  Either change the list-processor to prevent e-mail loops of
the type you describe, or replace it with one that does. :)

There are other causes of mail loops (misconfigured SMTP servers come to
mind) besides vacation replies. :)

Also, adding a patch or modification to a _server_ to improve
functionality is preferable to forcing users to use your idea of the
"correct mailer" to achieve that same functionality.  If it is possible
to fix a problem at the server, it is likely that it should be. :)