This page is based on an email I sent to a brand new mailing list after someone accidentally posted an embarassing private email to the whole list.
Header munging is the practice of changing or adding to the headers of an email. Many mailing lists are configured to alter the headers of emails sent to the mailing list, especially by adding a "Reply-To" header. This new header contains not the address of the author but the address of the mailing list itself.
While many consider this mailing list configuration to be a convenience to users who are using poor email applications or to users who are unfamiliar with email, I argue below that the convenience is illusory.
This isn't really an important issue, but it does seem to be a good example of how perfectly good technology can degenerate through its widespread abuse.
From: Ed L CashinSubject: reply-to munging (was Re: Meeting Announcement) Date: 13 May 2001 21:02:26 -0400 XXXXX XXXXXX writes: > On Fri, May 11, 2001 at 08:51:28PM -0400, Ed L Cashin wrote: > > # Yes, I often go on a "don't mess with the email" rant. It goes like > # this: Mail clients should have "reply" and "reply to all" options, and > # since the list didn't write the message, it's wrong-headed to make the > # reply go to the list. > > It seems like it all depends on the list, including the list's audience, > and the list's purpose. The Listserv software can certainly be configured > such that it won't add a special Reply-To: header to messages, but DWEEBS > *is* configured to add a Reply-To: header. This is typically done on > lists whose primary purpose is to be a *discussion* list. It works both > ways: I'll bet if you took off the reply-to feature then many messages > that were intended for the whole list would end up in a single mailbox. Yes, but then you could say that the only reason for that confusion is that people don't know the difference between "reply" and "reply to all", and the reason they don't know is that listserv owners with good intentions have *broken* the distinction in an attempt to hand-hold the users. For instance, I wanted to reply to you personally, since this thread is off topic, but since the Reply-To header was set to the list, I had to manually edit the headers after hitting reply. That's not convenience. If no one messes with the headers, then we have convenience: a reply goes to the author, and a reply-to-all goes to all. It's easy enough to make two different keystrokes or two different buttons to click, and then there's none of this confusion. The confusion comes from the ambiguity that results when the headers are changed. Then "reply" doesn't mean reply anymore, it means reply-to-all, and reply-to-all is meaningless. Then you've lost a helpful, convenient, and realistic distinction. There *is* a distinction; the users cannot really gain anything through ignorance of the distinction, so we aren't helping them out by supporting the ignorance of the distinction. But if instead they recognize the distinction, then they can act intentionally. The way it is now, the distinction is blurred, and people are always doing what they didn't mean to do. Sure it works both ways, if we went back to leaving the headers alone tomorrow, some reply mail intended for the list would go to the original author's instead, but then the distinction would be clear, and the sender would learn that next time all they'd have to do is use reply-to-all instead of reply. That learning experience would empower them, and the experience certainly wouldn't be as negative as having a personal email mistakenly broadcasted to a mailing list. Really, though, I don't have any hope, since it's too late. The practice of breaking the distinction between reply and reply-to-all is so firmly entrenched that it doesn't look like going back is a possibility. There are some other advocates, though. The GNU mailman help on sourceforge for list administrators strongly recommends against setting the "Reply-To: " header. ... Oh, well. :) -- --Ed Cashin PGP public key: ecashin at terry.uga.edu http://www.terry.uga.edu/~ecashin/pgp/