Adonis Diaries

Posts Tagged ‘fail-safe designs.

Reflecting again: On design errors in human-machine interfaces

Note:  I occasionally edit, translate, and re-publish articles that I deem them worth disseminating: Worthy articles are meant to be read.

Matthew Squair posted this May:  “Having recently bought a new car, I was driving home and noticed that the illuminated lighting controls were reflected in the right hand wing mirror. These sort of reflections are at best annoying, but in the worst case ,they could mask the lights of a car in the right hand lane and lead to a side-swipe during lane changing.

This is one of the classic system design errors that is well understood in domains such as the aerospace field.   Not so much in the car industry apparently.

But what really interests me is the fractured nature of engineering knowledge that this problem illustrates. I guess there is an implicit assumption we make that “we’re all getting smarter”.  But if that’s the case, why are the same errors committed over again?

Henry Petroski points to a study by Silby (1977) of bridge failures:  The study shows a 30 year-cycle between major bridge collapse and posits that, in any technology, we go through a cycle of learning, mastery, overconfidence, and subsequent failure due to over reach.

I’d point to the fragility of corporate memory within organizations and design teams:  I recognize that in the current  environment of rapid organizational change, it’s extremely hard to provide mentoring and oversight for young engineers, who unfortunately “don’t know what they don’t know“!

This remorseless cycle of destruction is exacerbated by codes and standards that record ‘what’ must be done from a compliance standpoint, but not the why”  Without the reason for compliance there is always the temptation…

I do agree with Petroski that failure breeds reflection, insight, and knowledge and that engineers, (especially young engineers), need in many ways to experience failure themselves or learn through the failures of others.

Evaluations of cockpit transparencies for reflections are required as part of the development of a new aircraft. These effects are particularly a problem for fighter aircraft with a large curved canopies and where the pilots’ displays sit comparatively close to the canopy.” (End of quote)

I have published over 30 articles on wordpress.com related to Human factors in design.

Human Factors professionals attempted to establish various error taxonomies, some within a specific context, during their study and analysis of errors that might be committed in the operation of nuclear power plants for example, and other taxonomy that are out of any specific context.

One alternative classification of human errors is based on human behavior and the level of comprehension; mainly, skill-based, or rule-based or knowledge-based behavioral patterns. This taxonomy identifies 13 types of errors and discriminates among the stages and strength of controlled routines in the mind that precipitate the occurrence of an error, whether during execution of a task, omitting steps, changing the order of steps, sequence of steps, timing errors, inadequate analysis or decision making.

With a strong knowledge of the behavior of a system, provided that the mental model is not deficient then, applying the rules consistently most of the errors will be concentrated on the level of skill achieved in performing a job.

Another taxonomy rely on the theory of information processing and it is a literal transcription of the experimental processes; mainly, observation of a system status, choice of hypothesis, testing of hypothesis, choice of goal, choice of procedure and execution of procedure.  Basically, this taxonomy may answer the problems in the rule-based and knowledge–based behavior.

It is useful to specify in the final steps of taxonomy whether an error is of omission or of commission.  I suggest that the errors of commission be also fine tuned to differentiate among errors of sequence, the kind of sequence, and timing of the execution.

There are alternative strategies for reducing human errors by either training, selection of the appropriate applicants, or redesigning a system to fit the capabilities of end users and/or taking care of his limitations by preventive designs, exclusion designs, and fail-safe designs.

You may start with this sample of two posts:

1. https://adonis49.wordpress.com/2008/10/14/whats-that-concept-of-human-factors-in-design-5/, and 2. https://adonis49.wordpress.com/2008/10/26/multidisciplinary-view-of-design/

Note 1: Petroski, H. Success through failure: The paradox of design, Princeton Press, 2008.

Note 2: Sibly, P.G., Walker, A.C., Structural Accidents and their Causes. In: Proc. Inst. Civil Engineers. 62 (May 1977), pp. 191–208 part 1. 1977.

Article #12, April 9, 2005

“What are the error taxonomies in Human Factors?”

There is a tendency to separate errors made by human and those done by machines as if any man made equipment, product, or system has not been designed, tested, evaluated, manufactured, distributed, or operated by a human.  My point is any errors committed by using or operating an artificial implement that causes injuries is ultimately a human errors. 

How Human Factors classifies errors when people operate systems, and what are the types of errors, their frequencies and consequences on the health and safety of operators and systems’ performance?  Human Factors professionals attempted to establish various error taxonomies, some within specific contexts such as nuclear power plants and chemical installations, of deficiencies in design and operation that might be committed, and others that are general in nature and restricted to processes of the mind and the limitations of human capabilities. One alternative classification of human errors is based on human behavior and the level of comprehension; mainly skill-based, rule-based, or knowledge-based behavioral patterns. For example, Rasmussen (1982) developed a decision flow diagram that identifies 13 types of errors; this taxonomy identifies two kinds of errors attributable to skill based behavior such as acts relevant to manual variability or topographic misorientation, four major errors related to rule based behavior such as stereotype takeover, forgetting isolated acts, mistakes alternatives, and other slip of memory, and then seven types of errors that can be attached to knowledge based behavior such as familiar association short cut, information not seen or sought, information assumed but not observed, information misinterpreted, and side effects or conditions not adequately considered. 

These types of errors are the products of activities done in routine situations or when the situation deviates from normal routine and discriminate among the stages and strength of controlled routines in the mind that precipitate the occurrence of an error whether during executing of a task, omitting steps, changing the order of steps, sequence of steps, timing errors, inadequate analysis or decision making.  With a strong knowledge of the behavior of a system, provided that the mental model is not deficient, then applying the rules consistently most of the errors will be concentrated on the level of skill achieved in performing a job.

Another taxonomy rely on the theory of information processing and it is somehow a literal transcription of the experimental processes; mainly observation of the status of a system, choice of hypothesis, testing of hypothesis, choice of goal, choice of procedure and execution of procedure.  Basically, this taxonomy may answer the problems in the rule-based and knowledge–based behavior.

Another alternative taxonomy could be found in measurement errors considered in statistical research such as conceptual, consistent or random errors. Conceptual errors are committed when a proxy is used instead of the variable of interest either because of lack of knowledge of how to measure the latter (i.e., measuring vocabulary ability when mental ability is the object of the research) or because it is less expensive or more convenient.

Consistent errors are represented by systematic errors from respondents whether conscious or not, measuring instruments, research settings, interviewers, raters, and researchers. Consistent errors affect the validity of measures.

Random errors occur as a result of temporary fluctuations in respondents, raters, etc.  Random errors affect the reliability of the measures.

The effects of these measurement errors have different consequences whether committed relative to the dependent or independent variables. It would be interesting to find correspondence among the various error taxonomies as well as assigning every error to either a conscious, predetermined tendency along with the real reasons underlining these errors, or unconscious errors.

It is useful to specify in the final steps of taxonomy whether an error is of omission or of commission.  I suggest that the errors of commission be also fine tuned to differentiate among errors of sequence, the kind of sequence and timing of the execution.

There are alternative strategies for reducing human errors by either training, selection of the appropriate applicants or redesigning a system to fit the capabilities of end users and or taking care of his limitations by preventive designs, exclusion designs and fail-safe designs.

What are error taxonomies, and other taxonomies in Human Factors in Engineering?

Article #12, written in April 9, 2005)

May you allow me just a side explanation on experimentation, to set the foundations first?

Psychologists, sociologists and marketing graduates are trained to apply various experimentation methods and not just cause and effect designs.

There are many statistical packages oriented to providing dimensions and models to the set of data dumped into the experiment, so that a preliminary understanding of the system behavior is comprehended qualitatively.

Every applied science has gone through many qualitative models or schema, using various qualitative methods, before attempting to quantify their models.

Many chairmen of engineering departments, especially those who have no understanding of the disciple of Human Factors in engineering and would never touch this body of knowledge and methods with a long pole, ask me to concentrate my courses on the quantitative aspects.

That hint sends immediate shiver through my rebellious spirit and I am tempted to ask them “what taxonomy of methods are you using in teaching engineering courses?”

What taxonomies Human Factors have to conceive?  How about the classification of human errors when operating a system, their frequencies and consequences on the safety of operators and system performance?

Human Factors professionals attempted to establish various error taxonomies, some within a specific context, during their study and analysis of errors that might be committed in the operation of nuclear power plants for example, and other taxonomy that are out of any specific context.

One alternative classification of human errors is based on human behavior and the level of comprehension. Mainly, skill-based, or rule-based or knowledge-based behavioral patterns.

This taxonomy identifies 13 types of errors and discriminates among the stages and strength of controlled routines in the mind that precipitate the occurrence of an error, whether during execution of a task, omitting steps, changing the order of steps, sequence of steps, timing errors, inadequate analysis or decision-making.

With a strong knowledge of the behavior of a system, provided that the mental model is not deficient, applying the rules consistently most of the errors will be concentrated on the level of skill achieved in performing a job.

Another taxonomy rely on the theory of information processing and it is a literal transcription of the experimental processes; mainly, observation of a system status, choice of hypothesis, testing of hypothesis, choice of goal, choice of procedure and execution of procedure.  Basically, this taxonomy may answer the problems in the rule-based and knowledge–based behavior.

It is useful to specify in the final steps of taxonomy whether an error is of omission or of commission.  I suggest that the errors of commission be also fine tuned to differentiate among errors of sequence, the kind of sequence, and timing of the execution.

There are alternative strategies for reducing human errors by either training, selection of the appropriate applicants, or redesigning a system to fit the capabilities of end users and/or taking care of his limitations by preventive designs, exclusion designs, and fail-safe designs.


adonis49

adonis49

adonis49

September 2020
M T W T F S S
 123456
78910111213
14151617181920
21222324252627
282930  

Blog Stats

  • 1,416,756 hits

Enter your email address to subscribe to this blog and receive notifications of new posts by email.adonisbouh@gmail.com

Join 768 other followers

%d bloggers like this: