Interview (mp3, 33:38, 30.8MB, December 2007)
When Chris Johnson of the Glasgow Accident Analysis Group suggests engineers should focus more on past failures than future success, he's not encouraging defeatism. Accidents, he tells Namahn, are ideal opportunities for systems designers to learn from failures. While accidents are rare events for most Western industries, Johnson believes that an increasingly interconnected world must share its lessons or be condemned to repeat them with more far-reaching consequences.
Johnson's "Mission Unaccomplished" is to get more institutions interested in learning from complex systems failure. That includes dealing with the often-blurred lines between human and technical error. What proportion of accidents are due to human error? Many agencies have responded to accidents by trying to remove people from a system altogether, says Johnson, but you can't take humans out of the equation.
That's not to say he agrees with some human factors experts who use errors traced to, say, management decisions to argue that nearly 90% of accidents are people-related. Johnson and his colleagues at NASA are working on a method that would make it scientifically feasible to figure out the proportion of human error in critical safety system failures by standardising processes so that the study can be replicated.
However, Johnson emphasises that human factors play only a part in complex system errors. Accidents can occur because of a combination of equipment failure, bad weather and high operational loads; to focus only on one area is to lose the complexity of the failures we encounter.
"Often we don't deal with the truth of what happened in the accident but with what investigators consider important from previous failures," he notes. For example, while software is being used increasingly in automobile breaking systems and engine management systems, investigators almost never cite software problems as the cause of car accidents. However, in the aviation industry, where more lives are at stake, more data have been gathered on software-related failure and investigators are more likely to consider it in their assessment. Johnson's appreciation of the growing complexity of systems has led his research towards the area of contingency planning--preparing us for our increasing vulnerability to the escalation of accidents, including terrorist action.