Re: Most used and Recently Used [was Re: [Usability] New way of accessing software (WAS: Re: Big Panel menus (32x32))]

[Janne <jan moren lucs lu se>]

On Fri, 2003-06-13 at 15:21, Alan Horkan wrote:
> However I have many programs installed and spend far too much time in
> front of a computer and use lots of different programs at differnt times.
> When I do a bit of graphics work for example I would be using maybe three
> or four differnt programs, and for that day I would want really want the
> Recent Used applications, not my most used application which most
> likely would be Tetris, or some
> other game :)
> 
> Figuring out the recently used application is simple.
> Figuring out the applictions you use most often based on how often you
> launch them is a terrible oversimplification.
> 
> > >> And since we aren't touching them, it's fine to experiment
> > >> with statistics to show the most recently - and heavily - used apps, so
> > >> a one-off usage of some app won't push away an app that is being used
> > >> all the time.
> 
> It is great idea in theory but the implementation requires a vary
> careful balance to keep to the user in control, and offering them
> revelevant choices.

Absolutely. That is why I emphasized that a most/moest recent mechanism
must be a separate thing from the normal menu structure - I would go
nuts if the system rearranged my menus for me.

Since we are drifting off-topic anyway, here's the skeleton of an idea
for a mechanism.

What could maybe work as a mechanism is some kind of finite-horizon
discounted value function. Basically, every time you use something, it
gets some set value added to its total 'score'. And every time step
(likely not actual time, but every time you launch an app, say), all
app's scores are discounted by a fraction. And once an app stays unused
for a set time (a week, perhaps?), its score is set to zero. The apps
shown are those with the highest score.

With this mechanism, stuff you've used quite a bit very recently would
be high on the list, as are apps that aren't that frequent, but
continuously recurring.




-- 

Trust the Computer. The Computer is your friend.

Tel.    +46-046 222 8588             Dr. Janne Morén (mr)
Home:   +46-046 211 4973             Dept. of Cognitive Science
Fax:    +46-046 222 9758             Kungshuset, Lund
                                     S-222 22 Lund, Sweden