Philip Mirowski

The ArgumentWhile there has been muchattention given to experiment in modern science studies, there has been astoundingly little concern spared over the practice ofquanitataivemeasurment.Thus myths about the unreasonable effectiveness of mathematice in science still abound. This paper presents: An explicit mathematical model of the stabilization of quantitative constants in a mathematical science to rival older Bayesian and classical accounts;a framework for writing a history of pracitces with regard to treatment of quantitative measurement erroe; resourece for the comparative sociology of differing discipliness in this regard;and a prolegonmena to a critique of orthodox economics and accounting theories. The key to all these diverse themes is the realization that no one individual alone is capable of fixing the magnitude of a quantitative error estimate, and therefore the social construction of error must be given a more precise meaning, an therefore the social construction of error must be given a more precise menaing, and that this occurs through the istrumentality of meta-analysis.

Recent breakthroughs in the history and sociology of science have begun to help us to appreciate the vast complexity and intricate character of empirical endeavours in the sciences. The days when philosophers could blandly gesture towards “observation statements” or “falsification,” as if they were some readily understood phenomena or set of procedures, are gone, happily. This does not mean we can merely use the Duhem-Quine thesis as a shibboleth, however: we are now much more sensitive to immense difficulties in establishing viable claims to replication or establishing the absence of selectivity bias, or convincing others of adequacy of experimental controls. Yet, even so, I would still assert that modern science studies, with only a very few exceptions, have ignored the problem of the constitution of error in the context of precision quantitative measurement. While rather broad claims have been made concerning measurement by Hacking, Cartwright, Franklin and others, none goes so far as to examine the actual manipulation of the numerical data. More distressingly, in the areas of economics and economic methodology, we have not even got so far as have the researchers in science studies. This fact alone should provide grounds for scepticism about the claim in Cartwright that philosophy of scientific inference has something to learn from the history of econometrics.