[horde] [imp] default reply behavior in dimp/imp

[D G Teed]

On Tue, Sep 6, 2011 at 4:53 AM, Michael M Slusarz <slusarz at horde.org> wrote:
>
> Again, this is wonderful news.  We *automatically* pick the proper reply
> type.  Less work for the users.

There is one button to push either way.  It is the same "work" to start
with, but more work for users to fix as it is not working as designed.

In reality, REPLY does a REPLY TO ALL in most situations, against the
wishes of our Mailman mailing lists, which were configured
with Reply-To: of the sender, not the list.

Let's make this very clear how the sequence is actually behaving.
I'm referring to actual behaviour, not design.  Please test/verify.

How to reproduce:
1. Mailman is configured so replies go to sender, not list.
2. User in dimp selects the only visible button, Reply
3. Compose window appears with both original sender and
    mailing list set to receive the reply.
4. If user clicks on the checkbox beside the reply all warning,
    to change the default and reply only to sender,
    it does nothing to change the recipients.

There are actually 3 bugs here :
1. Missing button for Reply options: should have Reply and Reply All,
just as in traditional
2. Reply, as designed to be dynamic, fails to honor Mailman
configuration for Reply-To config
3. Warning within compose screen, that mail is composed to reply to
all, when clicked,
    does not remove any recipients

I urge you to try actually using the software, not the design document, to
see what it does.

> Your argument seems to be that giving the users *more* obvious reply buttons
> in the UI is *less* confusing.  But this is exactly opposite of general UI
> principles.  Normally, you want to provide a basic interface with
> reasonable/sane defaults so that the normal users don't have to think: they
> click a button and it just works.  And then you give the power user a way to
> do what they need to do, that doesn't interfere with a normal user's UI
> experience.  I believe our reply button achieves this goal (could they
> graphical display of the button be improved?  Certainly.  But that is
> cosmetic; the underlying theory of a single reply button is sound).

The proof is in my users.  I let dimp (dynamic mode) be the default
and it backfired with the users.  They reply to everyone
by default sometimes.  It causes major problems if even only 1% of our
replies are accidents like this which could have been avoided.
It really doesn't matter what you believe or I believe, *because it is
the damage caused by the default, on our user community, which matters*.

We will use a mailer which provides the user with a clear choice on how
they will direct their reply before seeing the compose window.

Have you driven a standard transmission car?  If so you might be
familiar with the way the clutch works in the new cars in contrast with
those from the 80's or earlier.  In the old days you could use the
starter without depressing the clutch.  Even experienced drivers have
bad days when they don't push the clutch before using the ignition,
and on those days, they could lurch the car into another car.  Now, today,
car designs forbid this.  The ignition of the car is disabled until
the clutch is depressed, as a safety.

This is the same for Reply buttons.  The Reply button can be
made less accident prone by design.  The buttons are the initial
engagement with sending the reply.  It is much safer if users think
about who they are replying to before being shown all of the
detail in their compose window where the recipients are less
visible, especially when one is tired, emotional, in a rush, etc.

> Not to mention, your suggested approach defeats the whole purpose of the
> automatic reply button.  Namely, that the majority of messages sent to
> multiple recipients SHOULD BE REPLIED TO ALL.  A reply to these messages
> SHOULD NOT be sent only to the recipient, absent a conscious decision by the
> sender to do so.

Why are you married to the automation when many users don't want it?

It might be smart (in design perhaps) 95% of the time, but it is unsafe
and dangerous to the lives of the users the other 5% of the time,
and for that reason, unusable. You are assuming far too much about
the clarity of mind when people use email, and forgetting there
are personal lives impacted by what happens when people send
email to the wrong people.

Can you demonstrate that your defaults help to prevent people from being
unfriended, prevent people from being fired, or from losing confidence
of others due to a simple mistake of failing to keep information confidential?
Perhaps on your tech lists, reply to all and open democracy is great.
In the world of business and academics, it is different.
Email can affect lives, and you should think of this first, not last.

I don't know if you guys have a good picture of who your end users are.
In our case they are mostly people who don't know what open source is,
and probably don't care.  They have likely never seen something like
Kontact and probably only 2% of them have used Linux.  Many
of them think email is something that is on their cell phone.
Others have email clients native to their Mac or Windows system
but refuse to use them, preferring the online solution they have learned from
prior exposure to gmail, hotmail, yahoo and cell phone providers.

We are seeing more use of Horde webmail at our institution than
three or four years ago, even though all of our users have their
own laptop or desktop system. At this minute there are over
360 active sessions on our webmail.

As for the drop down window...  You know about it, but very few
people will find that tiny arrow.  Again, an interface
people are not tuned to see as it is not present in other
email interfaces.

In Traditional mode, the drop down choice takes place as we get
ready to click on the Reply button.  It pops up on mouseover
and offers Reply choices. This similar choice is not likely
to be noticed in dimp with the tiny arrow beside the Reply button
and the lack of the mouseover event trigger.

There are several ways to fix this problem, but as long as users are clearly
presented with Reply and Reply to All as clear choices before composing
the reply, we are going to ensure as much as possible that misdirected
replies and the damage they cause can be prevented.  On the other hand,
there is no way I can defend the choice of a webmail interface
which breaks with the common email interface design and helps
people to have accidents with confidential information.