good design basics case study

[sungod <as387 yfn ysu edu>]

here's a (rather long, sorry) passage which embodies several concepts
about design we must keep in mind while working on the gnome gui. anyone
who is interested in real, effective design should take a few minutes to
read it.

I recently spent six months at the Applied Psychology Unit in Cambridge,
England. Just before I arrived the British Telecom Company had installed
a new telephone system. It had lots and lots of features. The telephone
instrument itself was unremarkable (figure 1.11). It was the standard
twelve-button, push-button phone, except that it had an extra key
labeled "R" off on the side. (I never did find out what that key did.)

The telephone system was a standing joke. Nobody could use all the
features. One person even started a small research project to record
people's confusions. Another person wrote a small "expert systems"
computer program, one of the new toys of the field of artificial
intelligence; the program can reason through complex situations. If you
wanted to use the phone system, perhaps to make a conference call among
three people, you asked the expert system and it would explain how to do
it. So, you're on the line with someone and you need to add a third
person to the call. First turn on your computer. Then load the expert
system. After three or four minutes (needed for loading the program),
type in what you want to accomplish. Eventually the computer will tell
you what to do--if you can remember why you want to do it, and if the
person on the other end of the line is still around. But, as it happens,
using the expert system is a lot easier than reading and understanding
the manual provided with the telephone (figure 1.12).

Why is that telephone system so hard to understand? Nothing in it is
conceptually difficult. Each of the operations is actually quite simple.
A few digits to dial, that's all. The telephone doesn't even look
complicated. There are only fifteen controls: the usual twelve
buttons--ten labeled 0 through 9, #, an *--plus the handset itself, the
handset button, and the mysterious "R" button. All except the "r" are
the everyday parts of a normal modern telephone. Why was the system so
difficult?

A designer who works for a telephone company told me the following
story:

"I was involved in designing the faceplate of some of those new
multifunction phones, some of which have buttons labeled "R." The "R"
button is kind of a vestigial feature. It is very hard to remove
features of a newly designed product that had existed in an earlier
version. It's kind of like physical evolution. If a feature is in the
genome, and if that feature is not associated with any negativity (i.e.,
no customers gripe about it), then the feature hangs on for generations.

"It is interesting that things like the "R" button are largely
determined through examples. Somebody asks, 'What is the "R" button used
for?' and the answer is to give an example: 'You can push "R" to access
loudspeaker paging.' If nobody can think of an example, the feature is
dropped. Designers are pretty bright people, however. They can come up
up with a plausible-sounding example for almost anything. Hence, you get
features, many many features, and these features hang on for a long
time. The end result is complex interfaces for essentially simple
things."(*)

As I pondered this problem, I decided it would make sense to compare the
phone system with something that was of equal or greater complexity but
easier to use. So let us temporarily leave the difficult telephone
system and take a look at my automobile. I bought a car in Europe. When
I picked up the new car at the factory, a man from the company sat in
the car with me and went over each control, explaining its function.
When he had gone through the controls once, I said fine, thanked him,
and drove away. That was all the instruction it took. There are 112
controls inside the car. This isn't quite as bad as it sounds.
Twenty-five of them are on the radio. Another 7 are the temperature
control system, and 11 work the windows and sunroof. The trip computer
has 14 buttons, each matched with a specific function. So four
devices--the radio, temperature controls, windows, and trip
computer--have together 57 controls, or just over 50 percent of the ones
available.

Why is the automobile, with all its varied functions and numerous
controls, so much easier to learn and to use than the telephone system,
with its much smaller set of functions and controls? What is good about
the design of the car? Things are visible. There are good mappings,
natural relationships, between the controls and the things controlled.
Single controls often have single functions. There is good feedback. The
system is understandable. In general, the relationships among the user's
intentions, the required actions, and the results are sensible,
nonarbitrary, and meaningful.

What is bad about the design of the telephone? There is no visible
structure. Mappings are arbitrary: there is no rhyme or reason to the
relationship between the actions the user must perform and the results
to be accomplished. The controls have multiple functions. There isn't
good feedback, so the user is never sure whether the desired result has
been obtained. The system, in general, is not understandable; its
capabilities aren't apparant. In general, the relationships among the
user's intentions, the required actions, and the results are completely
arbitrary.

Norman, 1988

(*): I thank Mike King for this example, and others.
-- 
 ______(sungod)_____________________________________
| To ensure privacy and data integrity this message |
| has been encrypted by using dual rounds of ROT-13 |
 --------------------------(as387@yfn.ysu.edu)------