Bureaucracy in Crisis: Three Mile Island, the Shuttle Challenger, and Risk Assessment.

Delving into the organizational events before and after two serious incidents -- the 1979 Three Mile Island partial melt-down and the Challenger Shuttle disaster, Maureen Hogan Casamayou demonstrates that the most persuasive explanation for both accidents is not the more usual political one but rather an endemic organizational failure in assessment. This book explores three reasonable hypotheses in its search for an explanation. The first two, commonly associated with public agencies, are that internal communication blockages and/or distorting pressures from external actors force organizations to neglect important risk vectors. The third hypothesis is that flawed organizational scanning among key leaders produced dangerous misperceptions. The latter can be found in any organization, public or private but especially in those dependent on complex technologies. Casamayou argues that there may be something particularly difficult in dealing with complex technologies with low probability, high consequence risks. Her underlying concern is that these "surprise" accidents may be the only way such organizations are persuaded to change.

In Casamayou's treatment, inaccurate scanning mechanisms seem the most persuasive explanation of the organizations' failures to avert accidents. Although acknowledging NASA was indeed under intense public pressure and the NRC is prone to lean on the nuclear industry for information and support, Casamayou argues these influences do not adequately explain the blindness that led to the two accidents. Key decision-makers at both NASA and NRC did receive adequate warning of the serious risks involved. Only in NASA's case do strong pressures from outside actors seem to also play a significant but arguably secondary role in affecting the sense of urgency of key decisions.

The key difficulty, Casamayou argues, is that neither group was prepared to understand and act upon warning events involving items considered tangential in routine organizational scanning. Presuming good luck in the past dictated the future, NASA shuttle managers downplayed the importance of frequent erosion in an O ring sealing hot gases from the fuel tanks. NASA managers even developed an "erosion standard" by which to justify ignoring corrosion in something never meant to corrode. Similarly, NRC regulators so strongly believed in the robustness of their technology that only large failures or big leaks were given attention. Operator responses to small leak loss-of-coolant accidents (LOCAs) were not part of the licensing process prior to the small leak LOCA at TMI. Operator failures in general were not seen as an industry-wide difficulty or even characteristic of a particular make of nuclear reactors; they were left to the purview of the industry. As a result, while the data was in-house and the external audiences were not overly interfering, the organization misperceived the information and failed to avert the accidents.

In retelling these case studies with a fresh perspective, this book is properly positioned between the technology and organizations literature and the public policy literature. It provides a crosswalk between the more theoretical technological systems analysis of Charles Perrow in Normal Accidents and, for example, the many Brookings Institution publications that offer prescriptive policy alternatives. The two case studies are rich in relevant pre-accident organizational detail and, equally importantly, in the post-accident organizational change so rarely discussed.

The book suggests some thought-provoking questions useful for both classroom and private considerations. First, for the ongoing debate on the distinctions between public and private organizations, Casamayou's work inadvertantly supports those who argue both phenomena are a variation on a theme. Poor scanning is not a function of the organization's nonparticipation in a market or budget constraints but rather due to internal technological enthusiasm and the complexity of the systems involved. She also notes that engineers closest to the actual implementation of the technology were more likely to have had more realistic and apprehensive risk assessments. Can it be that the social use of complex technologies in ever larger scale may be one area in which contracting out functions to private firms would make no difference in either efficiency or safety? Perhaps all organizations run serious risks of failing to appreciate warning signs when complex technologies are involved.

Second, if major accidents are necessary to shake the automatic presumptions and established scanning parameters of large-scale technology-dependent organizations, what then for the policy community to do? What publically legitimate mechanisms exist to substitute or minimize costs, or conceivably direct for maximum benefit, the probably inevitable organizational failures, other than to abandon the technology? Casamayou offers no answer but provides an interesting addition to the literature.

Chris C. Demchak
The Univeristy of Arizona