[LINK] Ease of use and Linux for the Rest of Us

[Viveka]

At 7:30 PM +1100 21/11/02, Stilgherrian wrote:
>
>At 19:04 +1100 21/11/02, Bob Bain wrote:
>>  I'm a lazy bugger and don't wish to plough my way through expensive
>>and massive tomes entitled "Red Hat in Easy Steps" or "How to deal
>>with obscure and difficult configuration files."
>>
>>  Computers should be and can be easy to use.  Microsoft have proven
>>this over the years.
>
>The "over the years" is the trick, I suspect.
>
>Whatever tool we use -- application software, operating system, 
>sedan car, light aircraft, power drill, pencil and paper -- it'll 
>feel easy to use if it's been part of your life over an extended 
>period. Conversely, anything new and works differently from what 
>we're used to will be "difficult" and "obscure".

It is a truism that the goal of interaction design is to make systems 
"intuitive". However, when interface designers talk to each other we 
avoid the word, because it will lead to the inevitable response "the 
only intuitive interface is the nipple". [Eudora is telling me not to 
write "nipple", apparently it's offensive]. Now that I have a small 
daughter, I can correct that (from observation, not direct 
experience; in my case Viveka is a boy's name). Even the nipple is a 
learned interface, and it can be quite tricky to work out at first.

I'd used only Macs for some time before I ever had to use a PC. (I 
cut my teeth on a commodore 64, then I went to art school, which is 
of course all Mac, and thence to a design job, which was of course 
all Mac, and then to Optus Vision, which was bizarrely Mac only at 
that time, and I won't tell you the rumours that circulated about 
that. At Optus I also encountered Linux (Slackware) and Solaris. 
Didn't use a PC until I needed something to run Cosmo Worlds and 
couldn't get the IRIX box to boot).

So when I first tried a two-button mouse, it was a nightmare. It 
seemed like the right-hand button should be the main one, since I'm 
right-handed. It took about a week before I consistently remembered 
which button to push most of the time.
When looking for menus, muscle memory would cause me to move the 
mouse upwards sharply so that the cursor would collide with the top 
of the screen, there to seek the menus. Instead menus were all over 
the place - attached to the top of whatever window was in front at 
the time. You could move the cursor *above* a menu, even in a 
maximised window. It took me a long time to learn to slow down enough 
to hit that 16-pixel high menu bar, instead of the infinitely-high 
bar I was used to (do a search on Fitts' Law if you want more about 
this).
The menus themselves were inconsistent between applications. One 
application would have the program options in Edit>Preferences, 
another would be in Tools>Options, or some other bizarre variant 
(what is the Tools menu? Isn't every menu item a tool?). Sometimes 
options would be scattered all over the place, following the 
idiosyncratic ontology of the programmer. I discovered that Windows 
users don't use the main menuing system - most of the options you 
would ever use were in the contextual menu. Then I finally understood 
why the right mouse button was necessary on Windows.

When we say "that's intuitive", we usually mean "that follows the 
standards that I am accustomed to". There are other interface design 
principles that are useful as well - affordances, constraints, 
mapping, conceptual models... but standards trump them all. Even a 
bad standard that you know is easier to use at first than a brilliant 
but unfamiliar interface design. A doorhandle that turns upwards to 
open the door is almost as effective as a lock.

This reliance on standards gives the advantage to the incumbent. It's 
why Linux window managers are slavishly copying the crufty Windows 
interface, instead of using the latest research to transcend the 
WIMP/desktop paradigm. I think it's a shame, but it's a powerful 
reality. New interface designs will need to be built on top of the 
ruins of the old, and carry the weight of them for some time. (The 
way forward is known as "multi-modal interface", and involves 
supplementing WIMP with a combination of voice & gesture recognition, 
computer vision and 3D visualisation. There's plenty of interesting 
stoff out there about this if you're intrigued. And if you want a 
good laugh, check out the example video on Microsoft Research's "Task 
Gallery" website).

V.
-- 
Viveka Weiley, Karmanaut.
{ http://www.karmanaut.com | http://www.planet-earth.org
    http://www.MacWeb3D.org | http://sydney.siggraph.org.au }
Hypermedia, virtual worlds, human interface, truth, beauty.