ABSTRACT
Science has traditionally been justified by pointing to what might be called its “crown jewels”—widely accepted achievements such as the Michelson-Morley experiment, the discovery of the germ theory of disease, the discovery of the equations of motion of the planets, and so forth. Close examination of crucial experiments that supported these discoveries shows that they were generally much less decisive than hindsight accounts would suggest but, in any case, science in the public domain is generally more like economic forecasting or epidemiology—far from exact and secure. Therefore, the crown-jewels justification is likely to lead to disappointment. Disappointment among the public is dangerous because it reduces the power of scientific expertise as a “check and balance” in pluralist democracies, instead opening the way for populist dictators to decide whatever they want in respect of the technological questions. We need to justify science even when it is uncertain and fallible; we need to love it as “craftwork with integrity” even when it fails to give an exact and correct answer to problems in the public domain. I explain why meta-analysis is not a solution except in special cases.
