If you didn't come from my two sides of the story page, you should probably go there first

Reply-To Munging Considered Useful

An Earnest Plea to Mailing List Administrators

Last revised: 3 January 2000

An email message requires some amount of processing when it is redistributed to a mailing list. At the very least, the envelope must be rewritten to redirect bounces directly to the list administrator.

While the message is being processed, the list administrator might take advantage of the opportunity to munge some of the message headers. Many list administrators want to add a Reply-To header that points back to the list. This transformation is also one of the most useful.

Some administrators claim that Reply-To munging can have harmful -- even dangerous -- effects. I assert the opposite, that not adding a Reply-To header has even more harmful effects. If you think Reply-To munging is a bad idea, I hope I can change your mind.

RFC 822 and "Text Message Teleconferencing"

The first thing to consider is that RFC 822, the document which defines the standards and usages for email, specifically mentions this usage in section 4.4.3: 

        A somewhat different use may be of some help to "text message
        teleconferencing" groups equipped with automatic distribution
        services: include the address of that service in the "Reply-To"
        field of all messages submitted to the teleconference; then 
        participants can "reply" to conference submissions to guarantee
        the correct distribution of any submission of their own.

Aside from this official sanction, there are a number of reasons for munging the Reply-To header. The arguments which follow are my own. They may not be comprehensive, but I think they are compelling.

The Principle of Minimal Bandwidth

The ``Principle of Minimal Bandwidth'' is a good rule that will keep you out of trouble. It says that you should make any changes which will reduce the amount of email traffic on the Internet. The ``Principle of Minimal Bandwidth'' will help you avoid the sorts of problems we are about to discuss. This principle is a rule designed to be broken, but you can avoid some significant heartache by thinking hard and long before you do so.

Reply-To Munging Adds Something

Reply-To gives the respondant an option which would not otherwise exist: namely the ability to reply only to the list. Despite the fact that many (though not all) email clients have the ability to "reply to sender" or "reply to all recipients", many list subscribers want to reply only to the list, which is not the result of selecting either of these options. So, to ensure that the reply goes to the list, they select "reply to all recipients", which generally results in the sending of at least two email messages, one to the list, and one to the original sender.

This is frequently quite annoying to the original sender, who now receives two copies of the reply. Furthermore, in many cases the original sender has added additional recipients. Not only does "reply to all recipients" send the reply to each of these additional recipients (who are frequently also members of the list), it also propagates this list of recipients onto the reply to the list.

The effects of this snowball, as each additional person replies to the messages using "reply to all recipients", they become the sender, and thus get added to the list of recipients with the next reply. Thus the list of recipients grows and grows. Frequently, as the subject matter changes, members of the list find themselves receiving multiple copies of messages which have strayed from the topic in which they were originally interested, even after they have unsubscribed from the list.

Many people have pointed out that it is relatively easy to implement a procmail filter to remove duplicates. This attitude merely reveals a Unix-centric and US-centric viewpoint. Many users of inferior operating systems do not have a tool powerful enough to ensure the removal of duplicate messages. Furthermore, in many European countries, connect time is charged by the minute. Even with procmail, the duplicates have to be downloaded before they can be filtered, resulting in unnecessary additional expenses for some of our European list mates.

This last fact reveals that the issue is really related to bandwidth. By applying the ``Principle of Minimal Bandwidth'', we conclude that it is necessary to add a Reply-To header that points back to the list.

It Doesn't Break Reasonable Mailers

If you use a reasonable mailer, Reply-To munging does provide new functionality, namely the ability to reply only to the list. Furthermore, it does not decrease functionality. In Pine, for example, when there is a Reply-To header, Pine will ask, ``Use "Reply-To:" address instead of "From:" address?'', easily allowing one to reply only to the original author. In KMail, it is even easier. One merely right-clicks on the hyperlinked From address.

If your mailer doesn't have this option, you should request it from its development team. Any mailer, whose development team refuses this simple request due to some ideological position, cannot be said to be reasonable.

Freedom of Choice

Since Reply-To munging adds additional functionality, it actually increases freedom of choice. Not only can you now reply only to the list, you still have the option to reply to the original author, or to all recipients, easily and conveniently.

Some Mailers are Broken

There are, unfortunately, some poorly implemented mail programs that lack separate "reply-to-author" and "reply-to-group" functions. A user saddled with such a mailer can benefit from Reply-To munging. It makes it easier for him or her to send responses directly to the list.

Furthermore, this change does not penalize the conscientious person that uses a reasonable mailer. Reasonable mailers give one the ability to reply to the From address. Therefore, it would be unkind to further penalize those with poorly implemented mail programs, since munging the Reply-To header causes no harm to those with reasonable mailers.

Principle of Least Total Work

For discussion type lists, I would estimate that ninety percent of the time, people want to reply to the list. Without munging, they either have to break the ``Principle of Least Bandwidth'', or type in the list address. Many people, being lazy, will choose the former, sending unnecessary copies of emails to people who will either have to delete them, or take the time to set up a filter (if they are lucky enough to be running an operating system which facilitates this).

On the other hand, about ten percent of the time, replying to the sender might be more appropriate. Even if the respondant has an unreasonable mailer (a decision for which they are probably responsible), the worst case scenario is that they have to type in an address ten percent of the time. Of course, if they took the time to add this recipient to their address book, they could reduce the amount of typing to a minimum.

So, which produces least total work: typing in the list address ninety percent of the time (plus possibly taking the time to set up a filter), or typing in an individual's address ten percent of the time?

I'll take munged Reply-To headers every time, thanks.

People are Responsible for Their Own Mistakes

Some administrators claim that munging Reply-To headers is harmful because it surprises people, and can cause damage when things go awry. They assert that administrators should prevent the possibility of a private message being mistakenly broadcast to the entire list.

This is simply not the responsibility of the administrator. People are responsible for their own mistakes. If someone is sending a private email which is derogatory, or otherwise embarrassing were it to be made public, they should probably be sending it directly, rather than as a reply to a public message. They should also pause and think about whether they should be sending it at all. This pause should be quite sufficient for a conscientious person using a reasonable mailer to catch any mistake that they might be about to make.

In any case, it is an entirely trivial matter for the list administrator to provide an obvious clue in the subject line of every message that the message was received from a mailing list. If your Mailing List Manager doesn't provide an option to prepend "[listname]" to the subject, then switch to one that does ( e.g. GNU MailMan or Majordomo ).

And in the End...

If you are not convinced yet, then allow me one final plea. Most mailing lists are intended to facilitate discussion on a given topic. If this is indeed the primary purpose of your list, then you really should add a Reply-To header which directs replies to the list. This helps to ensure that the entire thread of the conversation is available to all who might be interested.

I can't count the number of times I have searched the archives of a list for a solution to a problem, only to find the question asked, but no solution. Yet, when I subsequently post the question to the list, the long-time members insist that it has already been discussed, and that I should search the archives. If I'm lucky, a newer member forwards to me the private reply which answered the question.

Thus, munging the Reply-To header benefits those lists which are intended for serious discussion. If your list is intended primarily for announcements or other one-way mailings, you may safely ignore these arguments.

It's What People Want

I have been and am subscribed to both munging and non-munging mailing lists. On the non-munging lists, there are regular requests to change the list so that Reply-To replies to the list. On the munging lists which already do this, there are hardly any requests for change.

Summary

Many people want to munge Reply-To headers. They believe it makes reply-to-list easier, and it encourages more list traffic. It really does both of these things, and is a very good idea. To reiterate:

 

Addendum

There are, of course, a few details that need to be addressed to make Reply-To munging more pleasant and productive for everyone. One potentially serious problem with Reply-To munging is the possibility of mail loops. It should be possible for the list server to detect and prevent this. If anyone has any patches to implement this feature, I would be happy to provide a link to the patch on your ftp server, or to make it available on my own ftp server.

Also, patches are needed for any mailers that do not implement the ability to reply to the From address. Please send links or patches to me at sdhill at metasystema.net. Thanks.

Patches

A patch for Emacs rmail is available at: http://www.metasystema.net/pub/patches/emacs/rmail-query-reply-to.el.

Dissenting Opinion

I originally wrote this essay as a response to Chip Rosenthal's Reply-To Munging Considered Harmful.

Simon Hill
sdhill at metasystema.net