Editor’s note: Today’s Guest Opinion is written by Michael Maddox, PhD, a senior scientist at HumanCentric Technologies.Lately, you’ve no doubt been reading and hearing a lot about human errors. Medical errors have certainly been in the news, along with stories of the Columbia space shuttle investigation, friendly-fire incidents in Iraq, and other less-grievous occurrences.
I work in a specialty knows as “human factors,” which is the science of designing systems so they correspond to the capabilities and limitations of human users. My intent is to introduce you to some ideas related to the often-misunderstood area of human error and to describe how systems (both simple and technologically sophisticated) can be designed to minimize the effects of human error.
To err really is human
Error is one of the truly universal facets of human performance. Each of us commits many errors each day. Fortunately, most of these errors have trivial consequences, such as having to repeat a word or re-dial a phone number. However, on some occasions, a human error combines with other events to cause an accident. Combinations of human errors and other events can lead to serious injuries and, in some situations, even death. Estimates of the contribution of human error to accident causation range from a low of at least 50 percent to a high of 90+ percent.
I’m sure you’ve heard a lot of talk about eliminating errors. The truth of the matter is that it is virtually impossible to eliminate human errors.
They are simply part of the human condition. Yes, we can reduce certain types of errors and we can most assuredly reduce the effects of errors. But we cannot eliminate all errors. The good news is that it is entirely feasible and practical to design products and systems to be “error-tolerant.”
The “systems” view
To understand why human errors are so difficult to reduce (and virtually impossible to completely eliminate), you have to adopt the so-called “systems” view of the world. In this way of looking at things, humans are a component of an overall system intended to perform a particular set of functions, such as medical care, public transportation, banking, etc. Some components in these systems are people. Some components, such as computer applications and other business tools, are designed and built by people. Still other components, such as geography or the weather, are not under our influence.
We can reduce the error potential of certain system components. For example, we can make certain the people in the system are well-trained, highly motivated, well-rested, etc. Likewise, we can use the established principles of user-centered design to make certain our tools, software applications, and other devices are easy to use correctly (and difficult to use incorrectly). Other system components, such as the weather, are not under our control. The best we can do is account for the range of weather conditions we will encounter and design the system accordingly.
In a system, failures generally occur at the weakest points. We should view accidents related to human error as system failures. If we find a specific cause for an error and fix it, then some other system component becomes its weakest point and the most likely locus of future errors. Actually, almost all accidents are caused by a series of errors or failures, not just a single element, so we typically wind up trying to fix several pieces of the system. Nonetheless, after these “fixes” the weakest links will simply be located elsewhere–waiting for the next alignment of circumstances to enable another accident.
How do we fix things?
We know a lot about what doesn’t work to reduce errors. For example, re-training people who commit errors is not very effective. The problem is that errors are unintentional and usually caused by a confluence of circumstances that are not directly under the control of those who ultimately commit the error. Training, and most other common approaches to error reduction, affects only intentional behavior.
The way we reduce errors and, probably more to the point, the effects of errors is to take a layered approach. First, we design the system elements so they can be easily used by the humans in the system. The methods of “user-centered design” are well known (if not always used). Products, procedures, and processes that are well-designed will, by definition, exhibit lower error rates than the same poorly-designed elements.
Second, we examine the system to determine the most likely places errors will occur — and then try to identify the consequences of particular types of errors. There are several established techniques for performing these types of analyses — typically known as “hazard” or “risk” analysis. When we identify the most likely candidates, we have to put into place whatever mechanism or process is needed to (a) reduce the likelihood of those errors occurring and, (b), mitigating the effects of errors when they occur.
Third, we need to inculcate into the culture of the group (or organization) the idea that everyone is responsible for recognizing and preventing errors. This is really the most difficult part of error reduction. The idea that error management is not someone else’s job must be driven by management. I’m not talking about just lip service, but a real commitment to identify error-causing conditions and change them so errors are actually prevented.
Perhaps as important as preventing errors is the ability to recognize errors when they occur and to take immediate steps to prevent their effects from propagating to cause an accident. Often, this requires people to question authority. In some domains, such as medicine, individuals are often not rewarded for questioning the perception or judgment of authority figures. The aviation industry found through hard experience that, in order to prevent errors from causing accidents, every crew member must feel empowered to question information and the perception of other crew members.
When these steps are taken, then we have the best practical situation in terms of errors:
- Each component is designed to minimize errors.
- We know where errors are most likely to occur and their most probable effects–and we’ve taken steps to deal with them.
- Everyone is on the lookout for errors and is motivated to prevent them from cascading into an accident.
This is the practical definition of an error-tolerant system.
Note: Guest Opinions are welcome. Please address yours to LTW’s managing editor, Rick Smith, at firstname.lastname@example.org.
HumanCentric Technologies: www.humancentrictech.com