norman@cogsci.EDU (Donald A Norman-UCSD Cog Sci Dept) (12/25/89)
This note is inspired by the repeating of Bill Buxton's observations
of the complexity of the auto interface, but with what I thought was
exactly the wrong conclusion (that the complexity of autos is easily
learned and OK, so why should we be so concerned about a few buttons
on a mouse).
Let's not get carried away with the apparent ease of use of
automobiles and other household goods. People have severe problems
with them.
The auto works only because:
1. The critical control number about 6 (gas pedal, brake, steering,
shift lever, and lights. Plus knowing where to insert the key and how
to start the car.) (Manual transmission adds one more, but remember
why we have automatics -- many people never comfortably manage to
synchronize brake, gas, clutch, gear lever and turning, signalling,
and driving).
2. The other controls DO cause great confusion and difficulty. Many
people complain to me (because I collect these things) that they make
errors and are <<extremely>> frustrated by failure to find the horns,
turn signals, windshield wipers (and washers), or to figure out how to
work the (electric) windows, radios, heater, air conditioner, etc.
Especially at high speed while driving. And I haven't even touched
upon the cruise control yet.
3. With the computer, we need to use the minor controls and functions
repeatedly and as part of the actions toward our goal. With the auto,
all these other items are secondary, and although they do cause
frustration, they are not required for the primary goal and so if we
do have trouble, we can still go from point A to B, which is the goal.
4. Moreover, it takes a few months to learn to drive a car safely,
about a year to feel comfortable. I rent automobiles tens of times a
year, and I consider myself lucky if I can learn how to find a decent
radio station on the radio. I often can't find the horn and I almost
never can learn to set the clock correctly. Driving is so complex
that we have special schools to teach it, special license exams, and
special police to enforce the procedures. With road signs and markers
(you couldn't drive safely without the white line down the middle of
the highway). A large industry exists to keep driving within
acceptably safe parameters (see item 5),
5. And, unlike computer interfaces, we accept 40,000 + deaths /year
(in the US alone), plus hundreds of thousands of injuries. Each year.
None of this was Bill Buxton's fault: he was being quoted, and
unfairly I think. Buxton also has a nice story about the weirdness
and lack of standards of the first automobiles. Go to a car museum
and look. And for your holiday amusement, I append his story.
----------------------------------------------------------------------
Here is a note I got from Thomas Erickson (of Apple), on Bill Buxton's
observations on early auto interfaces (written while Bill was in
Cambridge, England, working for Xerox EuroParc -- he is now at
Toronto).
Unindented material is from Erickson: indented from Buxton:
Date: Tue, 3 Jan 89 11:04:20 PST
From: Tom Erickson <thomas@apple.com>
Subject: auto quote
Interface Design, 1884-1904.
Bill Buxton writes:
On a number of occasions, I've heard speakers compare the operation of
computers unfavorably to that of automobiles. "Why," they ask, "can't
we design computers like cars, where one can move from one to the
other, despite different manufacturers, and still be able to drive?"
The line of argument usually proceeds along lines like, "What if cars
were designed the way computer user interfaces are? What if the
clutch was on the right foot and the accelerator on the left?" etc.,
leaving visions of death, destruction and all kinds of havoc on the
highway, and by implication, on the computer-way.
Buxton then goes on to describe his attendance at the finish of the
London-Brighton [antique] car race, and his discovery that early
automobiles (1884-1904) were designed much like current user
interfaces. The programme of the race provides more detail:
"The driver of an early Lanchester, for example,steers it by laying
his right arm along a side-lever, while his right hand flashes from
vapour regulator (mixture control) to petrol pump to compound gear
trigger, between two engine governor levers and two gear levers, one
of which is always a brake. His left foot operates the accelerator
and his right foot is merely used to blow the horn, by stamping hard
on a floor-mounted rubber bulb. By contrast, the Locomobile steam car
has a transverse steering lever which is gripped in the left hand.
The right hand operates the throttle lever, the link-motion control
lever, the water-feed bypass lever, and all the mysterious stopcocks
associated with steam power. Like the Lanchester, the early Delahaye
had an accelerator pedal on the left, but you pressed it to slow down
and released it to go faster; not much chance of catching that one if
you happened to fall out. On the popular little Dion Bouton, the left
pedal is not only a decelerator but also a transmission brake, and the
right pedal (operated with the heel) engages the reverse. The
Brotherhood-Crocker had a flanged accelerator pedal, because it was
controlled by swiveling the foot from side to side, instead of up and
down. And Denmarks pioneering car, the remarkable Hammel, turned left
when the steering wheel was turned to the right, and vice-versa, so
your Hammel driver wore an earnest, rather worried expression at all
times."
This sheds some light on why early market researchers predicted that
the American demand for cars would peak at one million because of a
limitation in the number of available chauffeurs.
End of quoted section
----------------------------------------------------------------------
I am glad that we don't make computers as complex as the modern auto.
Don Norman INTERNET: dnorman@ucsd.edu
Department of Cognitive Science D-015 BITNET: dnorman@ucsd
University of California, San Diego AppleLink: d.norman
La Jolla, California 92093 USA FAX: (619) 534-1128