Rogier De Langhe

I will clarify disagreement about Thomas Kuhn’s Structure of Scientific Revolutions by combining recent developments in philosophy of science, economics and scientometrics in an agent-based model 1) to separate what is essential to Kuhn’s mechanism of scientific change from what is peripheral, 2) to provide a stricter interpretation of its essential concepts and 3) to make testable predictions. The model will then be tested against scientometric data. As such this project will provide definite results about “the most influential English-language philosophy book of the last half-century”(Rorty, 1995). 

The three recent developments that will clarify Kuhn’s view are 1) the reinterpretation of the notion of incommensurability in philosophy of science as incommensurability of standards, 2) the development of agent-based models of dynamic allocation between standards in network economics and 3) the improvement of scientometric data. Using modeling techniques for standards in economics, I will model the dynamics of Kuhnian standards and test the resulting patterns against scientometric data.

A breakthrough that was especially relevant for my purpose was the pioneering work by Arthur, Ermoliev & Kaniovski (1983, 1984, 1987), demonstrating that the adoption of technological standards can be generally described as a nonlinear Polya process. Based on this process I will build an agent-based model: a model in which autonomous agents interact in a shared environment with an observed aggregate, emergent outcome.

Construction of the model will be based on nonlinear probability theory. Dynamic distribution models in network economics using nonlinear probability theory typically take the form of urn models. These models consist of a collection of various colored balls placed in an urn. In each period, a ball is selected from the urn and, depending on the color of the ball selected, other balls may be added or removed from the urn.

Testing of the model will focus on two important observable consequences detectable through scientometric analysis.

(i) Hurst exponent: Occasional elevated Hurst-exponents in scientometric time-series are expected to occur during revolutions and could not be explained from Kitcher’s one-process view.

(ii) Power law distribution: It has long been established (Lotka 1926) that there is a power law distribution of citations which is very stable, even across disciplines, with 20% of scientists receiving 80% of citations. During a revolution, the non-ergodicity of the system will temporarily disrupt the stable distribution, again a factor that can only be explained by assuming the existence of multiple, incommensurable standards.