Sir

I do not agree with Sheila Jasanoff's point in her Millennium Essay “Knowledge élites and class war” about the dangers of technology1. Of the three technological disasters mentioned by Jasanoff — Bhopal, Chernobyl and Challenger — the last two (and perhaps even the first) had a common root cause, in that the advice of expert engineers and risk analysts either was not sought or was overruled by management.

In the Challenger case, for example, the ‘expert assessments’ of the engineers were not incorporated into a proper risk analysis by NASA; engineers' recommendations were overruled by less expert management; and the astronauts were not made aware of the engineers' worries or of the disagreement. (See Richard Feynman's account2.) NASA management procedures have much improved since then, but not to the extent of having an external expert safety review. In 1989, for example, an interagency safety review committee for the Galileo space probe was not informed about concerns expressed by a scientist at a lower level. This disagreement surfaced when the scientist wrote a letter directly to the president.

Chernobyl arose because the design assumed perfect operation. It was the compartmentalization of Soviet society and a politically based management that prevented information about design faults from reaching the operators.

It now seems clear that Bhopal arose from intentional sabotage by a disgruntled employee. Neither Union Carbide management nor the Indian government seems to have required a proper expert safety review of the plant by modern risk-assessment standards. Why, for example, was such an easily sabotaged plant designed and allowed to operate? Why were the operators not trained in the obvious remediation measures? This was a failure of management, not of the experts.

I do agree with Jasanoff that compartmentalizing society is dangerous, but that is becoming less prevalent. The crucial divisions are not those between workers and management or between expert and layperson, but those between management and experts — and between different experts, in societies in which management keeps them apart in order to control them.

The problem is compounded by politicians' refusal to recognize that the knowledge of scientific and technical experts must be at the heart of every scientific and technical decision they make.

There is no simple rule as to what defines an expert. But their role in judicial issues has recently been clarified in three landmark decisions: Daubert3, Joiner, and Kumho4. The first two of these were on admissibility of evidence of causation in medical situations, and the last on causation in engineering.

In Daubert the Supreme Court stated some non-exclusive criteria by which the court may judge the testimony being proffered. Has the theory been tested, or can it be tested? (In other words, is it falsifiable?) Has the theory been peer-reviewed and published? What is the known or potential risk of error? Has the theory been generally accepted in the relevant scientific community? Is the theory based on facts or data of a type reasonably relied upon by experts in the field? And, most importantly, does the testimony have probative value that is greater than, or not outweighed by, a danger of unfair prejudice, confusion of issues, or misleading the jury?

In Kumho the court declared that these criteria apply with greater force to situations in which an expert relies more on experience than on clearly defined theories.

DNA fingerprinting, despite being a relatively new technique, was not successfully challenged during the O. J. Simpson trial — though whether the jury accepted it is, of course, another matter.

It is important to incorporate expert knowledge into societal decisions. But a “conversation between science and society”, as suggested by Jasanoff, is a peculiar recommendation. Science is an integral part of society, whether non-scientists like it or not. Some segments of society (including many non-scientists) understand the role science and technology play within society. That is the understanding that should be enhanced.