Re: Different "replying" modes in a MUA

On 01/07/2013 11:22 PM, Dave Crocker wrote:
> It appears that you intend Private Reply to mean what Reply originally 
> meant:  Send this reply /only/ to the author.

Indeed, it's useful to understand that back in ancient history, "Reply" 
was assumed to only go to the author, and this helps to explain why the 
"Reply-To" field seems to have a peculiar meaning. But user agents 
capable of some variant of "reply all" have been around for a very long 
time.  (since the early 1980s, IIRC)

> The topic really is user interface design, not protocol design. Hence, 
> the knowledge, intent, psychology and, ultimately, expectations of the 
> author and the recipient are the controlling factors.  The message 
> generated for the recipient should naturally reflect the recipients 
> /and/ the original author's intent.

The problem is, of course, that the author's intent and the recipient's 
intent may not be the same.  But since the author cannot reliably 
anticipate the nature of a recipient's response, and also since the 
recipient responding is responsible for any replies that he sends, 
clearly the recipient who is composing the reply must be the one to 
ultimately choose the recipient(s) of a reply.

> The cited Principle of Least Astonishment is a usefully concise 
> reference to the question of expectation, which arguably synthesizes 
> the other human factors.  One of the problems with citing the 
> Principle is the implication that expectations can be readily or 
> completely discerned.
>
> By contrast, the rule I've observed among skilled UI designers is that 
> the application of the best cognitive, usability, etc. modeling, for 
> doing UI, merely produces a good /start/ and never is sufficient to 
> produce a final result.  For any interesting innovation in the UI, 
> it's necessary to experiment with real usage and modify as experience 
> is gained.

I absolutely agree with this.

> Two natural assumptions drive many people's thinking about UI design:
>
>    1)  Users always benefit from additional information
>
>    2)  Users always benefit from having more control
>
> For interactive interface (UI) design, both of these assumptions are 
> more often wrong than right.  Too much information or being forced to 
> make too many decisions both create an excessive cognitive burden.
>
> The challenge are to get the right /balance/ of the right information 
> and at the right time.  Users don't pay attention to all information 
> equally or even consistently.  What I might notice today, I might not 
> tomorrow.
>
> And users essentially never do the kind of diligent address field 
> review that some of the notes on this thread seems to expect.

Users do, on the other hand, suffer embarrassment or worse (sometimes 
much worse) when their replies go to a different set of recipients than 
they expected or intended.

To the extent that users "essentially never" review addresses in 
replies, this might in part be because most user agents aren't designed 
to facilitate effective review of such addresses.   If they let you edit 
reply recipients at all, it's usually via some "character-oriented text 
editing mode"  that seems cumbersome and antiquated in comparison to the 
way the GUI facilities specification of recipients of an original 
message: with name lookup and completion (for better or worse, often 
automagically populated from previously received messages) and automatic 
filling-in of the address in proper Internet Mail syntax.   With the 
"character oriented text editing mode" of editing reply addresses, you 
typically have to be responsible for getting the syntax right yourself.  
If ordinary users don't make use of this facility very often, it's 
little wonder.

Also, it seems to be current fashion in UA design to abbreviate the list 
of recipients of a message, hide their addresses, etc. as if that 
information were somehow superfluous.   This makes it even more 
difficult for a recipient to see to whom his reply will go.

(Those of us who remember paper memos also remember that it could be 
very important to read the listed set of recipients of a memo in order 
to understand the context of a message, and to choose carefully to whom 
you'd send a reply.   This is in principle no less true for email, 
though it's possible that poor MUA UI design has encouraged email users 
to be careless.)

Keith

_______________________________________________
ietf-822 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ietf-822