unity-design team mailing list archive

[Celeste Lyn Paul <celeste@xxxxxxx>]

On Monday 01 June 2009 11:19:31 am Vincenzo Ciancia wrote:
> Il giorno lun, 01/06/2009 alle 10.30 -0400, Celeste Lyn Paul ha scritto:
> > kwin can also know when a dropdown or context menu is open. i've been
> > looking
> > into the possibility of delaying notifications which occur during
> > these types
> > of interactions for n seconds. several researcher groups have found
> > significant
> > benefits for these types of "computer-mediated" interruptions.
>
> At an intuitive level can we say that when dropdowns or context menus
> are opened, the user is typically doing something short and that
> requires focus (such as filling in a form or selecting a menu entry) and
> then it is better not to interrupt the user in that precise moment?
>
> I see that when I am chosing a menu entry I wouldn't normally move my
> eyes to look at a notification.

Sometimes you don't have a choice. Screen flickers and animation have a high 
attentional draw and your eye will react by instinct. That's why things like 
advertisements on websites are so annoying. However, the more focused you are, 
the less attentional draw it will have but usually that only matters to tiny 
things like an icon appearing or changing color, not a popup on your screen. 

Cognitively, you can choose to acknowledge that distraction or not and decide 
to interrupt your task. But that only happens after it gets your attention in 
the first place. Researchers aren't sure what makes an interruption a 
disruption because there is conflicting data regarding between time away from 
task, task complexity, task type, etc. The only thing that's been agreed on is 
"context matters".

Keep in mind that "interruption" isn't necessarily bad. Interruption is 
necessary in order to get the user's attention so they can decide if the 
interruption is worth acting on or not. If we make interruptions invisible to 
users, then we shouldnt be wasting resources displaying them in the first 
place. It is "disruption" what we should be aiming to negate.

> Do you have pointers to some document by these research groups?

I have a very long bibliography on the subject, but I'm not sure you want to 
go through a hundred research papers. (Also, it would take a while to compile 
because I don't have all my sources in one spreadsheet yet)

The biggest names are McCrickard at Virginia Tech, Czerwinksi and Horvitz at 
Microsoft Research, and Bailey from UIUC who all have a large body of work 
regarding this topic. Additional work from McFarlane (2002), Fogarty et al. 
(2005), and Gievska et al. (2005) specifically test the effects of computer-
mediated interruptions. I'll have to look through my bib to pull out who 
created useful predictive models from these things.

-- 
Celeste Lyn Paul
KDE Usability Project
usability.kde.org