Normal Accidents: Living with High-Risk Technologies

These are matters of common sense, applied to simple questions of cause and effect. But what happens, asks systems-behavior expert Charles Perrow, when common sense runs up against the complex systems, electrical and mechanical, with which we have surrounded ourselves? Plenty of mayhem can ensue, he replies. The Chernobyl nuclear accident, to name one recent disaster, was partially brought about by the failure of a safety system that was being brought on line, a failure that touched off an unforeseeable and irreversible chain of disruptions; the less severe but still frightening accident at Three Mile Island, similarly, came about as the result of small errors that, taken by themselves, were insignificant, but that snowballed to near-catastrophic result.

Only through such failures, Perrow suggests, can designers improve the safety of complex systems. But, he adds, those improvements may introduce new opportunities for disaster. Looking at an array of real and potential technological mishaps--including the Bhopal chemical-plant accident of 1984, the Challenger explosion of 1986, and the possible disruptions of Y2K and genetic engineering--Perrow concludes that as our technologies become more complex, the odds of tragic results increase. His treatise makes for sobering and provocative reading. --Gregory McNamee

Product Description

Normal Accidents analyzes the social side of technological risk. Charles Perrow argues that the conventional engineering approach to ensuring safety--building in more warnings and safeguards--fails because systems complexity makes failures inevitable. He asserts that typical precautions, by adding to complexity, may help create new categories of accidents. (At Chernobyl, tests of a new safety system helped produce the meltdown and subsequent fire.) By recognizing two dimensions of risk--complex versus linear interactions, and tight versus loose coupling--this book provides a powerful framework for analyzing risks and the organizations that insist we run them.

The first edition fulfilled one reviewer's prediction that it "may mark the beginning of accident research." In the new afterword to this edition Perrow reviews the extensive work on the major accidents of the last fifteen years, including Bhopal, Chernobyl, and the Challenger disaster. The new postscript probes what the author considers to be the "quintessential 'Normal Accident'" of our time: the Y2K computer problem.

Professor Perrow is a PhD in sociology (1960) who has taught at Yale University Department of Sociology since 1981 and whose research focus has been human/technology interactions and the effects of complexity in organizations. (His most recent publication is the The AIDS disaster : the Failure of Organizations in New York and the Nation, 1990.)

In Normal Accidents, he describes the failures that can arise "normally" in systems, ie. those problems that are expected to arise and can be planned for by engineers, but which by virtue of those planned fail-safe devices, immeasurably complicate and endanger the system they are designed to protect. He describes a variety of these interactions, clarifying his definitions by means of a table (p. 88), and a matrix illustration (p. 97). Examples include systems that are linear vs complex, and loosely vs tightly controlled. These generally arise through the interactive nature of the various components the system itself. According to the matrix, an illustration of a highly linear, tightly controlled system would be a dam. A complex, tightly controlled system would be a nuclear plant, etc.

The degree to which failures may occur varies with each type of organization, as does the degree to which a recovery from such a failure is possible. As illustrations, the author describes failures which have, or could have, arisen in a variety of settings: the nuclear industry, maritime activities, the petrochemical industry, space exploration, DNA research and so on.

The exciting character of the stories themselves are worth the reading; my favorite, and one I had heard before, is the loss of an entire lake into a salt mine. More important still is the knowledge that each imparts. Perrow makes abundantly apparent by his illustrations the ease with which complex systems involving humans can fail catastrophically. (And if Per Bak and others are correct, almost inevitably).

Probably the most significant part of the work is the last chapter. After discussing the fallibility of systems that have grown increasingly complex, he discusses living with high risk systems, particularly why we are and why it should change. In a significant statement he writes, "Above all, I will argue, sensible living with risky systems means keeping the controversies alive, listening to the public, and recognizing the essentially political nature of risk assessment. Unfortunately, the issue is not risk, but power; the power to impose risks on the many for the benefit of the few (p. 306)," and further on, "Risks from risky technologies are not borne equally by the different social classes [and I would add, countries]; risk assessments ignore the social class distribution of risk (p. 310)." How true. "Quo Bono?" as the murder mystery writers might say; "Who benefits?" More to the point, and again with that issue in mind, he writes "The risks that made our country great were not industrial risks such as unsafe coal mines or chemical pollution, but social and political risks associated with democratic institutions, decentralized political structures, religious freedom and plurality, and universal suffrage (p. 311)." Again, very true.

Professor Perrow examines the degrees of potential danger from different types of system and suggests ways of deciding which are worth it to society to support and which might not be. These include categorizing the degree and the extent of danger of a given system to society, defining the way these technologies conflict with the values of that society, determining the likelihood that changes can be made to effectively alter the dangerous factors through technology or training of operators, and the possibility of placing the burden of spill-over costs on the shoulders of the institutions responsible. The latter might conceivably lead to corrective changes, either by the institutions themselves in order to remain profitable or by consumers through purchasing decisions.

The bibliography for the book is quite extensive and includes a variety of sources. These include not only popular books and publications on the topics of individual disasters, but government documents, research journals, and industry reports as well. I did not find any reference to the Johnstown flood, my particular favorite dam burst story, but there are a wide variety of references to chose from should someone wish to do their own research on the topic.

Altogether a fascinating and informative book.


Reprint needed   October 27, 1997
5 out of 7 found this review helpful

I specified this book as one of (the better of) two choices for supplementary reading in a university-level engineering course, and I'm dismayed that it's currently in this precarious print status. The book is an excellent--compelling and comprehensible-- explanation of the inherent risk of failure of tightly-coupled complex systems, in other words, the world we have created around ourselves. Engineers particularly need this insight before being unleashed on the world, because engineering as a profession (if not vocation) has taken the obligation to protect humankind from science and technology. If not a reprint or new edition, perhaps a new publisher is in order.


A new way of looking at risk   May 18, 2000
7 out of 10 found this review helpful

I am in the medical field, and the recent attention on medical errors in this country make this a must-read for all those who offer simplistic fixes - there is complexity in most of our modern systems and they are hard to detect. I especially found descriptions of the decision making going on in the various incidents enlightening - certainly has stimulated me to seek out other works in this field. I would like to find a comparable work using this approach in health care.