unity-design team mailing list archive

[Diego Moya <turingt@xxxxxxxxx>]

On 17 September 2010 05:21, Conscious User wrote:
>
> Like I said, it's not necessarily a "much more important action". It
> could be a very mundane action, but whose movement you couldn't stop
> by reflex.
>
> Furthermore, it is one thing to miss one notification because you were
> away or not paying attention. It is another, and much more annoying,
> thing to know *something* has happened and never knowing *what*.
>

Au contraire, they are exactly the same thing. Why should it matter if I
miss a notification because I'm reading a webpage, or because I'm reading a
physical written manual on my physical desk? Or what if I'm away and come
back to see it just in the last half second?

The reason why NotifyOSD does not keeps a log of notifications is because
you will always have a different way to know what happened. Notifications
can be ethereal because they are redundant. If there's no way to know what
happened without reading it in the available 5 seconds, then you really
should keep a log to recover previous notifications.

Also, with my heuristic you wouldn't know something has happened while
you're active, the notification would wait until you made a small pause in
your work - right when you're ready to read it. That's much more humane than
interrupting what you're doing.



>
> I think you are proposing a *very simple* heuristic to guess *very
> complex* thoughts.
>
> Those are the best heuristics! :-)  The important thing is that it provides
a consistent, predictable behavior to notifications.

You won't be afraid that a bubble will appear right over your target on the
screen, which was your original worry. The heuristic doesn't need to
second-guess whether your intentions were important or trivial, it works
well in both cases.



>
> > Actually there's a really simple fix that would solve both your
> > problem and the one of obscuring graphical applications, and it's
> > this: don't ever show a notification while the user is working on
> > something else. The way to detect the user work must be heuristic, but
> > there are some good clues to it:
> >
> I can be typing very fast... to copy a recipe of cake my grandmother
> sent me. I can be constantly moving the cursor... to play a flash
> game. I can be working at a fast pace... because I had one coffee
> too many, not necessarily because the task is important.
>

I though we already established that notifications are even less important
than the least important of the user tasks? That's the only possible
justification for them being ethereal.

I think it was Jef Raskin who said that you should treat user input as
sacred. I concur, and that can be extended to the user focus of attention.

If the notification is important enough to interrupt the user flow then a
transient bubble notification is not the place for it, it should create a
persistent warning in the panel.

In the examples you mention there's not a single reason to show a
notification that couldn't be better  placed as a state change in the menus.
As I said, you're treating notifications as more important that they really
are, and that's creating problems to the design because it conflicts with
the 'ethereal' goal.