System accident

From Wikipedia, the free encyclopedia

A system accident (or normal accident) is an "unanticipated interaction of multiple failures" in a complex system.[1] This complexity can either be of technology or of human organizations, and is frequently both. A system accident can be easy to see in hindsight, but extremely difficult in foresight because there are simply too many action pathways to seriously consider all of them. Charles Perrow first developed these ideas in the mid-1980s.[2] William Langewiesche in the late 1990s wrote, "the control and operation of some of the riskiest technologies require organizations so complex that serious failures are virtually guaranteed to occur."[3]

Safety systems themselves are sometimes the added complexity which leads to this type of accident.[4] Maintenance problems are common with redundant systems. Maintenance crews can fail to restore a redundant system to active status. They are often overworked or maintenance is deferred due to budget cuts, because managers know that they system will continue to operate without fixing the backup system.[5]

General characterization[edit]

In 2012 Charles Perrow wrote, "A normal accident [system accident] is where everyone tries very hard to play safe, but unexpected interaction of two or more failures (because of interactive complexity), causes a cascade of failures (because of tight coupling)." Charles Perrow uses the term normal accident to emphasize that, given the current level of technology, such accidents are highly likely over a number of years or decades.[6]

James T. Reason [fr] extended this approach with human reliability[7] and the Swiss cheese model, now widely accepted in aviation safety and healthcare.

There is an aspect of an animal devouring its own tail, in that more formality and effort to get it exactly right can actually make the situation worse.[8] For example, the more organizational riga-ma-role involved in adjusting to changing conditions, the more employees will likely delay reporting such changes, "problems," and unexpected conditions.

These accidents often resemble Rube Goldberg devices in the way that small errors of judgment, flaws in technology, and insignificant damages combine to form an emergent disaster.

William Langewiesche writes about, "an entire pretend reality that includes unworkable chains of command, unlearnable training programs, unreadable manuals, and the fiction of regulations, checks, and controls."[8]

An opposing idea is that of the high reliability organization.[9]

Scott Sagan[edit]

Scott Sagan has multiple publications discussing the reliability of complex systems, especially regarding nuclear weapons. The Limits of Safety (1993) provided an extensive review of close calls during the Cold War that could have resulted in a nuclear war by accident.[10]

Possible system accidents[edit]

Apollo 13 space flight, 1970[edit]

Apollo 13 Review Board:

" [Introduction] . . . It was found that the accident was not the result of a chance malfunction in a statistical sense, but rather resulted from an unusual combination of mistakes, coupled with a somewhat deficient and unforgiving design [Emphasis added]. . .

"g. In reviewing these procedures before the flight, officials of NASA, ER, and Beech did not recognize the possibility of damage due to overheating. Many of these officials were not aware of the extended heater operation. In any event, adequate thermostatic switches might have been expected to protect the tank."[11]

Three Mile Island, 1979[edit]

Charles Perrow:

"It resembled other accidents in nuclear plants and in other high risk, complex and highly interdependent operator-machine systems; none of the accidents were caused by management or operator ineptness or by poor government regulation, though these characteristics existed and should have been expected. I maintained that the accident was normal, because in complex systems there are bound to be multiple faults that cannot be avoided by planning and that operators cannot immediately comprehend."[12]

ValuJet (AirTran) 592, Everglades, 1996[edit]

On May 11, 1996, ValuJet Airlines Flight 592, a regularly scheduled flight from Miami International to Hartsfield–Jackson Atlanta, crashed about 10 minutes after taking off as a result of a fire in the cargo compartment caused by improperly stored and labeled hazardous cargo. All 110 people on board died. The airline had a poor safety record before the crash. The accident brought widespread attention to the airline's management problems, including inadequate training of employees in proper handling of hazardous materials. The maintenance manual for the MD-80 aircraft documented the necessary procedures and was "correct" in a sense. However, it was so huge that it was neither helpful nor informative.[8]

2008 financial institution near-meltdown[edit]

In a 2014 monograph, economist Alan Blinder stated that complicated financial instruments made it hard for potential investors to judge whether the price was reasonable. In a section entitled "Lesson # 6: Excessive complexity is not just anti-competitive, it's dangerous," he further stated, "But the greater hazard may come from opacity. When investors don't understand the risks that inhere in the securities they buy (examples: the mezzanine tranche of a CDO-Squared ; a CDS on a synthetic CDO,...), big mistakes can be made--especially if rating agencies tell you they are triple-A, to wit, safe enough for grandma. When the crash comes, losses may therefore be much larger than investors dreamed imaginable. Markets may dry up as no one knows what these securities are really worth. Panic may set in. Thus complexity per se is a source of risk."[13]

Possible future applications of concept[edit]

Five-fold increase in airplane safety since 1980s, but flight systems sometimes switch to unexpected "modes" on their own[edit]

In an article entitled "The Human Factor", William Langewiesche talks the 2009 crash of Air France Flight 447 over the mid-Atlantic. He points out that, since the 1980s when the transition to automated cockpit systems began, safety has improved fivefold. Langwiesche writes, "In the privacy of the cockpit and beyond public view, pilots have been relegated to mundane roles as system managers." He quotes engineer Earl Wiener who takes the humorous statement attributed to the Duchess of Windsor that one can never be too rich or too thin, and adds "or too careful about what you put into a digital flight-guidance system." Wiener says that the effect of automation is typically to reduce the workload when it is light, but to increase it when it's heavy.

Boeing Engineer Delmar Fadden said that once capacities are added to flight management systems, they become impossibly expensive to remove because of certification requirements. But if unused, may in a sense lurk in the depths unseen.[14]

Langewiesche cites industrial engineer Nadine Sarter who writes about "automation surprises," often related to system modes the pilot does not fully understand or that the system switches to on its own. In fact, one of the more common questions asked in cockpits today is, "What's it doing now?" In response to this, Langewiesche again points to the fivefold increase in safety and writes, "No one can rationally advocate a return to the glamour of the past."[14]

Healthier interplay between theory and practice in which safety rules are sometimes changed?[edit]

From the article "A New Accident Model for Engineering Safer Systems," by Nancy Leveson, in Safety Science, April 2004:
"However, instructions and written procedures are almost never followed exactly as operators strive to become more efficient and productive and to deal with time pressures. . . . . even in such highly constrained and high-risk environments as nuclear power plants, modification of instructions is repeatedly found and the violation of rules appears to be quite rational, given the actual workload and timing constraints under which the operators must do their job. In these situations, a basic conflict exists between error as seen as a deviation from the normative procedure and error as seen as a deviation from the rational and normally used effective procedure (Rasmussen and Pejtersen, 1994)."[15]

See also[edit]


  • Charles Perrow (1984). Normal accidents : living with high-risk technologies. ISBN 0-465-05143-X. Wikidata Q114963622.
  • Charles Perrow (1999). Normal accidents : living with high-risk technologies : with a new afterword and a postscript on the Y2K problem. Princeton University Press. ISBN 0-691-00412-9. Wikidata Q114963670.


  1. ^ Perrow (1999, p. 70).
  2. ^ Perrow (1984)
  3. ^ "Charles Perrow's thinking is more difficult for pilots like me to accept. Perrow came unintentionally to his theory about normal accidents after studying the failings of large organizations. His point is not that some technologies are riskier than others, which is obvious, but that the control and operation of some of the riskiest technologies require organizations so complex that serious failures are virtually guaranteed to occur [Emphasis added]. Those failures will occasionally combine in unforeseeable ways, and if they induce further failures in an operating environment of tightly interrelated processes, the failures will spin out of control, defeating all interventions." —from The Lessons of Valujet 592, The Atlantic, William Langewiesche, March 1998, in section entitled A "Normal Accident" which is about two-thirds of the way into the entire article.
  4. ^ The Crash of ValuJet 592: Implications for Health Care, J. Daniel Beckham, January '99. DOC file --> Mr. Beckham runs a health care consulting company, and this article is included on the company website. He writes, "Accidents at both Chernobyl and Three Mile Island were set off by failed safety systems."
  5. ^ Perrow (1999).
  6. ^ GETTING TO CATASTROPHE: CONCENTRATIONS, COMPLEXITY AND COUPLING, Charles Perrow, The Montréal Review, December 2012.
  7. ^ Reason, James (1990-10-26). Human Error. Cambridge University Press. ISBN 0-521-31419-4.
  8. ^ a b c Langewiesche, William (March 1998). The Lessons of Valujet 592, The Atlantic. See especially the last three paragraphs of this long article: “ . . . Understanding why might keep us from making the system even more complex, and therefore perhaps more dangerous, too.”
  9. ^ Becoming a high reliability organization, Critical Care, M. Christianson, K. Sutcliffe, et al., 8 Dec. 2011. Opposing concept. This is a concept which disagrees with that of system accident.
  10. ^ Sagan, Scott D. (1993). The Limits of Safety: Organizations, Accidents, and Nuclear Weapons. Princeton U. Pr. ISBN 0-691-02101-5.
  12. ^ Perrow, C. (1982), Perrow's abstract for his chapter entitled "The President's Commission and the Normal Accident," in Sils, D., Wolf, C. and Shelanski, V. (Eds), Accident at Three Mile Island: The Human Dimensions, Boulder, Colorado, U.S: Westview Press, 1982 pp.173–184.
  13. ^ What Did We Learn from the Financial Crisis, the Great Recession, and the Pathetic Recovery? (PDF file), Alan S. Blinder, Princeton University, Griswold Center for Economic Policy Studies, Working Paper No. 243, November 2014.
  14. ^ a b The Human Factor, Vanity Fair, William Langewiesche, September 17, 2014. " . . . pilots have been relegated to mundane roles as system managers, . . . Since the 1980s, when the shift began, the safety record has improved fivefold, to the current one fatal accident for every five million departures. No one can rationally advocate a return to the glamour of the past."
  15. ^ A New Accident Model for Engineering Safer Systems, Nancy Leveson, Safety Science, Vol. 42, No. 4, April 2004. Paper based on research partially supported by National Science Foundation and NASA. " . . In fact, a common way for workers to apply pressure to management without actually going out on strike is to 'work to rule,' which can lead to a breakdown in productivity and even chaos. . "

Further reading[edit]