danbricklin.com/log

Usability in Safety and Environmental Protection
I received this letter November 13, 2000:

Dan :

I'm finding your discussions on usability quite interesting.  I don't work
in the software or computer hardware field, but instead work with safety
and environmental protection.  Usability is a _big_ issue with us in that
field, too!

If one looks at major industrial accidents in the world over the last fifty
or so years, one finds that between 60% and 90% of the accidents are due to
"human error".  The range is due in part to the running debates on where
designer errors fit into "human error".  Having worked as a fire and
explosion investigator in "another life", I can say that the stats are
pretty accurate.  Reviewing my old accident reports a few years back, I
estimated that the error factor was right around 90%.

The issue of punching out the wrong ballot position ("hanging", "pregnant",
and "dimpled" chad aside) is smaller than picking the wrong valve to open
in an emergency, or to use a real world example, actuating an isolation
valve in reverse during testing, and thereby releasing highly flammable
materials.  That case chalked up more than a score dead, close onto $1
billion in overall losses, and dropped out about 5% of the USA's capacity
in a specific polymer's production.  All from an "error".

The number of "smaller" accidents with fewer dead and smaller dollar losses
is high.  Arguably to some, although not me, a national election is more
important than these cases, unless you're the woman or man who dies in the
accident.

Usability is a complex issue that requires multidisciplinary approaches, a
highly flexible mind to coordinate the approaches, and some old-fashioned
hard nosed thinking about Murphy and Finagle.  It's not something that one
just happens into one day, and then produces marvel after marvel.  As the
level of consequences goes up for failure, the quality of usability must
similarly rise in at least the same slope, and arguably proportionally
higher.  In the case of the federal election results in Florida where the
form was poorly handled, or the results in New Mexico where a specific
windowed box was not "clicked off", the consequences are pretty high right
now - look at the stock market if nothing else... although again, I tend to
consider people's deaths a higher issue.

Which then segues to the next point, most usability issues are decided at
relatively low levels or by people without experience in the area.  The
placement of light switches in rooms, location of critical shutoffs, or
simply where to sign one's name are typically determined by one of the most
junior members of the project team.  Some exceptions exist, to be sure,
such as the relatively recent studies by software companies on usability,
but even those packages fail the usability tests in less "important" areas
like installation.  The stories of installation failures are legion.

All this leads to the final lesson on usability.  Usability only becomes
important when a (relative) catastrophe occurs.  It shouldn't be this way,
but it is.  The issues with Florida's election in terms of recounts are not
new.  The failures of New Mexico in the area of failures to deploy systems
are not recent.  Only when the failures reach a crisis level is any serious
study made to the problems, and at that, is often a "band-aid" approach to
pass through the periodic crisis.

Usability is and will continue to be a serious issue.


John Palmer

© Copyright 1999-2014 by Daniel Bricklin
All Rights Reserved.