Performance Is a Moving Target, Reliably

Short of money from home, one of the nicest things that can happen in this business is to be taken seriously, and to have one's work rigorously criticized by smart and accomplished people. I'm deeply complimented by the care and attention with which Philip Scranton and John Law and Vicky Singleton have critiqued my "Reliable Knowledge" article.

I do believe, however, that both critiques see my portrayal of reliable knowledge as more hostile to social constructivism, and more positivist, than I think it is, or certainly was meant to be. The intent of the whole exercise was to try to provide an account of translocal reliable knowledge that both emerges from and is used recursively in local practice--an account that gets beyond the conundrums of "truth," "objectivity," and "relativism." Philip Scranton's meticulous reading and deconstruction of the argument raises a variety of issues, while John Law and Vicky Singleton present a more evocative challenge. In this short rejoinder, let me first try to address what I think are Scranton's more central objections--the ones that seem to animate his others--before turning to John's and Vicky's alternative construction.

Most of Scranton's discontent seems to derive from my use of "spatiotemporal universal." In David Hull's argument, only spatiotemporally universal entities can enter into spatiotemporally universal scientific laws. (Being spatiotemporally unrestricted, or universal, is his criterion for being a scientific law.) Hull was trying to solve the problem of biological species--indisputably historical entities and therefore ineligible to participate in scientific laws--on his way to a general theory of selection applicable both to biology and to science--and, by implication, to technological knowledge as well. ¹ While I ultimately would want to nest any Bayesian account of reliable knowledge within a more general "selection theory of competent reference," in Donald Campbell's felicitous phrase, such talk may have muddied the waters here. ²

The whole point of the Bayesian exercise, which Scranton finds uninteresting, is to argue that technological practice is based upon a hierarchy, or at least a rank-ordering (according to degrees of Bayesian confirmation), of rational beliefs. Some of these beliefs, those "high up" in the hierarchy, such as conservation of energy, are foundational for such an astonishing array of practice that they enjoy a nearly unbounded preponderance of confirming instances. Others, such as Ohm's law, or Darcy's law, are more domain specific, but since they too undergird so much practice they also enjoy a superabundance of confirming instances. These beliefs, because they are expressed in spatiotemporally unrestricted terms, can count as spatiotemporally unrestricted laws of nature. But being spatiotemporally unrestricted, or universal, or a "law of nature," is not the defining property of a rational belief: the mathematical relationship between numbers of confirming versus disconfirming instances is. That the most strongly supported beliefs tend to be spatiotemporally unrestricted is a contingent fact, not a necessary condition, and does not imply, in my argument, scientific "truth"--witness "the conservation of energy." Thus ad hoc practices, such as the parameter estimation used in evaluating high-relief hydrocarbon reservoirs, if they are judged successful and are replicated (or nearly replicated in similar situations), can cumulate substantial numbers of confirming instances and also become rational foundations for further domain-specific practice and, indeed, its extension. But such practices are unlikely to attain the magnitude of confirming instances associated with, say, the conservation laws. The same logic extends right down the hierarchy: to communal belief in specific designs--V35B Bonanza airplanes, for example--or even to individual belief in specific artifacts and their behavior. I have rational confidence that my Miata, with its six-year-old battery (it's become something of a challenge to see exactly how long it will last) will start and that, as soon as the oil warms up, the awful clatter from its sticky hydraulic valve-lifters will go away. But the paltry few...