[Sigia-l] Learnability and its impact on testing

[Listera]

"Tania" wrote:

> As such, I am not convinced that the learning variable can be easily
> quantified and the results adjusted accordingly.

Excellent deduction. For not falling for one of the testing industry's
fallacies. :-)

One of the hardest things in interface/workflow design is being able to
anticipate (and design for) the user's 867th interaction with the app. It's
very easy to make a first impression, it's very, very difficult to refine it
for the 867th impression.

Is there a way around this? Partially. There are design patterns that are
commonly used and reused in typical workflows. For example, consider a
log-in process. It's seemingly simple and short, but it has many
permutations and gotchas for usability.  Fortunately, collectively, we know
pretty much all the issues (both design and technical) concerning a typical
log-in process. A good development team usually has a template/framework
that's well-tested for both usability and technical ramifications.

So if you carefully examine an online app, you'll discover that it's the
result of a series of such 'factorable' patterns, call them "capsules of
wisdom." If you're technical, you'd recognize this as reuse of well-tested
OO code/frameworks.

Let's look at an example: the lowly listbox. Not all newcomers will
understand type-ahead facility, but old timers certainly will appreciate it.
Should you have it? If you have it, should you track all keystrokes (with
real-time response but with the attendant overhead) or just the first
character? Should you track for a short idle time between keystrokes, or
favor fast typists? A rapidly changing list order may confuse newcomers, who
may have a hard time making corrections. Should you allow inline editing of
a list item to immediately sort the listbox contents as soon as a tab/enter
is pressed, since the just edited item may disappear from view in a short
listbox? Is a button better than drag&drop for passing items between
listboxes? Is there an optimal ratio between listbox item count and its
height? If the listbox is a hierarchical one, should there be a limit to how
many levels you can nest? Etc.

As you can see, an app with any degree of complexity has an endless list of
such decisions to be made screen after screen, and many of those are
interdependent. It's virtually impossible and commercially infeasible to
"test" all the permutations. If a corporate app has an expected lifespan of,
say, 7 years, with an average user using it daily for 3 years, how exactly
do you quantify and test the learnability of that?

It all comes down to the ability of the designer(s) to recognize and
comprehend the issues, identify patterns, take appropriate risks and make
the necessary decisions. There's no amount of research (academic or
otherwise) that can substitute for experience and wisdom when it comes to
putting it all together as a coherent app.

UI design is not a science.

Ziya
Nullius in Verba