From the issue dated October 15, 1999

For a Scholar of 'Normal Accidents,' Risky Activities Bring Danger and Data By D.W. MILLER It was in the summer of 1979, says Charles Perrow, while he was vacationing in upstate New York, that he began having nightmares about nuclear meltdowns. After agreeing to advise the federal commission investigating the March 28 accident at Three Mile Island, the Yale sociologist recalls, he spent weeks reading ALSO SEE: Sociology, Not Engineering, May Explain Our Vulnerability to Technological Disaster carbon copies of the testimony. "I was blown away by the realization that the operators didn't know what was going on," he says. "First they tried this, and then they tried that. They thought the dials they were looking at were wrong, because the readings were inconsistent. They accepted one and rejected the other." Faced with a complicated series of failures they couldn't possibly have predicted, the operators didn't understand that both readings were correct. As a result, he says, their frantic efforts to keep the right amounts of heat and pressure in the reactor core and turbine system were based on "guesswork and faulty information." The success of Normal Accidents could have led to a lucrative career as a disaster guru and industry consultant. But Mr. Perrow decided instead to pursue other scholarly projects, among them what he hopes will be his magnum opus: a comprehensive history of organizations, still a couple of years away from completion. While fellow scholars advise companies and agencies on system accidents, he has confined his disaster-consulting work to testifying pro bono against the construction of nuclear-power plants in New England and Long Island. (His record: one for two.) He readily trusts, say, airlines to shuttle him safely around the country, because the conditions of systems failure rarely line up perfectly. And many technologies bring benefits that outweigh the consequences of catastrophe. Nuclear power, he has decided, just isn't one of them. He was hardly an ivory-tower naif before Three Mile Island. Described by a colleague as a "'60s Berkeley kind of guy," Mr. Perrow reports that he once studied protest movements "to learn their origins in order to restart them." And he worries that subsequent work in the field of systems accidents has obscured the role that concentrated wealth and power play in shaping organizational failures. In his postscript for the new edition of Normal Accidents, Mr. Perrow trains his sights on the "quintessential normal accident": the Y2K problem. He expects digital disruptions to be minimal. But he is, in a way, hoping for an opportunity to study something he didn't consider in his book: the simultaneous failure of countless systems that interact in unknown ways. On January 1, 2000, which happens to be the date of his retirement from Yale, "I will probably be glued to the TV -- if it's running -- and trying to judge how far off I was in my estimation," says Mr. Perrow. "I'm still intrigued by the idea that it's a test of my social theories about our society. Are we a hard-wired grid or a loosely coupled web that allows resilient opportunities for recovery?" So, as a social scientist, will he be disappointed if that day dawns without incident? He laughs a long laugh, and says, "In a way, no, because it will provide data. But in a way, yes, because it will be poor data if nothing happens." http://chronicle.com Section: Research & Publishing Page: A20 Easy-to-print version E-mail this article