By Brandioch Conner
Novell recently released some video of “usability” testing
featuring 11 people who were familiar with Windows trying to
accomplish various tasks under the Linux Desktop.
Now, to me, that doesn’t sound really effective. The reason is
that you’ll have only two groups of test subjects:
- The Ignorant
- The Tainted
The Ignorant are the ones that have not used any other systems.
Which, in this case, pretty much means that they haven’t used
any system. They won’t be much use beyond the very basics
of the interface. I’ve had to teach people who have never used a
computer before how to use Windows and it is extremely difficult.
One woman just did not have the muscle control necessary to hold
the mouse steady and double click with it. She had to use two hands
to accomplish it. After a week of playing solitaire, she had no
problems with the mouse.
The Tainted are the ones that have used at least one other
system. Novell chose 11 people who were familiar with Windows. So,
they will be trying to find the same menus/commands that they used
before (in Windows).
And Novell’s findings seem to support this because their test
subjects had trouble with such things as knowing that “Evolution”
was the app to launch for email. If they named it “Outlook” people
would have had an easier time. What they are actually “discovering”
is that people are most comfortable with the interface they’ve
already learned. You really cannot compare two different interfaces
with a class of Tainted subjects.
Rather than test whether someone who has never heard of “Linux”
before can figure out what the name of the web browser app is, why
don’t we just re-think the entire concept of “interface.” An
interface that is easy for the Ignorant class of users to work with
is probably not as useful for the experts (and all experts did, at
one time, belong to the class of Ignorant users).
How about an interface that is cartoonishly simple for the
beginners (Level 1 users)… but… offers all the functionality
possible with the system for the experts (Level 10 users)?
Would the “best” interfaces for each of these 10 Levels have the
same appearance? I don’t think it would. It would have the same
commands, preferably in the same location, but it wouldn’t be the
same. Think back to the books you first learned to read with. Then
compare them to the books you read now. They both have words, and
the advanced books may have the same words the beginning books do,
but the sentence structure is more complex and there are longer
words in the sentences.
So, rather than focusing on how quickly a user who is unfamiliar
with your interface can hunt-and-click to find the applications and
use them, why not measure how quickly a new user can progress
through the various levels until s/he is considered “proficient”
with that system and can complete his/her regular tasks?
And I’m not just talking OS/desktop environment. This also
applies to the applications.
- The Level 1 email interface would have very few buttons. Level
1’s are not expected to use encryption or digitally sign their
messages. - The Level 1 word processor interface would have open, save,
print, bold. - At Level 2, they get underline and italics and size.
- At Level 3, they get a selection of fonts.
- And so forth up to Level 10.
Also, as the level decreases, the protections increase. At Level
1, no data files are ever deleted if it can be avoided (temp
browser files and such can be cleaned). This will take more than a
bit of coding effort. Whether this is done on a per-file basis or
using a database as a file system or some other approach doesn’t
matter. What matters is that the beginner’s data is as safe as
possible. As the level increases, the user is given the option to
actually delete these files. The user can stop whenever s/he is
comfortable with the level of protection versus the degree of
customization/functionality.
Unless the interface is paired with education, and our standards
for “success” change, we’ll be “discovering” that the users are
most comfortable with the interface they’ve used before over and
over and over again.