Web lists-archives.com

Re: Reply semantics (was Re: policy around 'wontfix' bug tag)




On 2018-02-06 at 10:00, Vincent Lefevre wrote:

> On 2018-02-06 08:49:01 -0500, The Wanderer wrote:
> 
>> On 2018-02-06 at 08:18, Vincent Lefevre wrote:

>>> This is not contradictory with the setting of
>>> "Mail-Followup-To:".
>> 
>> Arguably, if the mailing list does not default replies back to it
>> normally, a responder who wishes to send a private reply may not
>> be expecting that Reply will go to the mailing list, and so (when
>> Mail-Followup-To is set to the list address) may fail to notice
>> that adjusting the addressee list is needed.
> 
> If this is a private reply, the user should just use the "reply"
> feature of his mailer. The reply will never go to the list. There is
> no need to adjust anything.

That conflicts with the idea that the "reply" feature should always go
to the correct place by default.

When replying to a message from a mailing list, unless someone has taken
actions to cause a different result, "reply" should direct the message
to that mailing list. That is what "replies should go to the list by
default" means.

In fact, I've been assuming in this discussion that setting
Mail-Followup-To will cause the "reply" feature to direct the reply to
the address specified in that header. If that is the case, then using
"reply" will not always produce a private reply; if it is not the case,
then I'm not sure what good that header would be.

>>> But the other users cannot know what you want if you do not set 
>>> "Mail-Followup-To:".
>> 
>> They should not have to.
>> 
>> They should be able to assume that you have taken all necessary
>> actions to ensure that you see any replies you want to see.
>> 
>> That could involve setting the header, or it could involve
>> subscribing to the mailing list, or it could involve requesting
>> explicitly (in the body of the mail or in your signature) to be
>> CCed, or it could involve something else.
> 
> Users don't want to have to look at the signature. They don't want
> to add addresses manually. If a user requests that he wants to be
> Cc'ed and I notice it, then I will tend to do a group-reply instead
> of a list-reply.

Using "Reply to All" on a mailing list is usually the wrong thing to do,
indeed.

> This means that other users of the discussion will also be Cc'ed,

Unless you then go in and delete from your reply's To/CC list the
addresses of anyone who has not requested CCs, which admittedly is a bit
of a pain.

> unless the Mail-Followup-To is set up correctly, in which case
> everything is fine: the users who do not want to be Cc'ed won't be
> Cc'ed; the other users will be Cc'ed. Unfortunately some mailers
> lose Mail-Followup-To information, but in general, this is better
> than nothing.

Can you define what "correctly" would be, in this context, in your view,
for someone who wants to receive replies only through the list unless
the person replying is specifically attempting to draw that someone's
attention?


(Warning: wall of text incoming.)

The thing is: no matter what the configuration is, there will sometimes
be cases when some people have to do manual work to get the right result.

In the overwhelming majority of cases, the correct destination for a
reply to a message posted on a public discussion mailing list is that
mailing list. Cases which require private replies (either CCed or
exclusive) are the exception, rather than the rule.


If replies go to the poster by default, then achieving this correct
destination requires special action by either every single reply-er
(adjusting To:/Cc: by hand, or using a special reply function, different
from what would be used in replying to non-list mail), or potentially
even every single poster (configuring things so that Mail-Followup-To
gets set correctly, whatever "correctly" may be).

The effort involved in doing this may be minor, and spread out across
large numbers of people, but it still adds up to be a lot.

When this does not happen, there are two failure states. The less severe
is that people who are subscribed to the mailing list will get two
copies of the mail; the more severe is that replies which were intended
to be posted publicly will instead be sent privately. How bad either of
these things are is probably a matter of opinion.

(There's also the failure state when Mail-Followup-To settings get lost
or not respected, because one of the participants in the discussion is
using a mail client which does not handle them correctly, but that's
beyond my ability to properly assess.)


If replies go to the mailing list by default, then only a subset of
posters need to take special action: people who want private copies of
replies, people who reply to the first kind of person, and people who
want to reply privately.

The effort involved in doing this is in any given instance is
noticeable, and potentially irritating, but it is limited to only the
instances where any special action is needed. In the overwhelming
majority of cases, no effort will be required.

When the first type of person fails to take the needed action, they will
not receive the private reply they want, and may miss the reply
entirely; that's a bad thing.

When the second and third type of person fails to take the needed
action, a message which was intended to be private will be posted
publicly. That is also a bad thing; depending on the details, it may be
anywhere from relatively minor to utterly catastrophic.


The question is how the effort requirements and the failure states of
the two default policies stack up against one another.

I concede that the failure states of "replies go to the mailing list by
default" are more severe than those of "replies go to the poster by
default".

And I concede that the case-by-case effort of "replies go to the mailing
list by default" is greater than that of "replies go to the poster by
default".

But I maintain that a tiny bit of effort and annoyance in all cases is
likely to outweigh a slightly larger amount of effort and annoyance in a
relatively small subset of cases.

And I suspect that failures in both cases are sufficiently infrequent
that the costs of of the worse failure states in the former case are
outweighed by the reduced frequency with which those failure states will
have the opportunity to occur.

By the principle of optimizing for the common case, it seems clear to me
that the model under which replies go to the mailing list by default is
the better and more sensible option.

-- 
   The Wanderer

The reasonable man adapts himself to the world; the unreasonable one
persists in trying to adapt the world to himself. Therefore all
progress depends on the unreasonable man.         -- George Bernard Shaw

Attachment: signature.asc
Description: OpenPGP digital signature