Models, Parameterization, and Software: Epistemic Opacity in Computational Chemistry

Abstract

Computational chemistry grew in a new era of “desktop modeling,” which coincided with a growing demand for modeling software, especially from the pharmaceutical industry. Parameterization of models in computational chemistry is an arduous enterprise, and we argue that this activity leads, in this specific context, to tensions among scientists regarding the epistemic opacity transparency of parameterized methods and the software implementing them. We relate one flame war from the Computational Chemistry mailing List in order to assess in detail the relationships between modeling methods, parameterization, software and the various forms of their enclosure or disclosure. Our claim is that parameterization issues are an important and often neglected source of epistemic opacity and that this opacity is entangled in methods and software alike. Models and software must be addressed together to understand the epistemological tensions at stake.

Epistemic issues in computational reproducibility: software as the elephant in the room

“…In another publication (Wieber & Hocquet, 2020), we have argued that opacity in computational chemistry models lies in different layers: in the parameters of the mathematical formalization of the model itself, in the code into which these models are translated, in the packages where these implementations are embedded, in the licensing policies that define how software is shared and used, and in the communities of developers and users. We have also argued that these layers are entangled and not easily separated.…”

“…The industrial license thus allows more transparency that the academic one, because the user can actually check and benchmark parameters (the black box is open). Yet, actors argue that published calculations from industrial licensees may have used altered parameters, thus revealing a problem of consistency in the model being used for the calculations, because of the then possible obfuscation of which modelling parameters have been actually used in a calculation (Wieber & Hocquet, 2020).…”

Scientific understanding through big data: From ignorance to insights to understanding

“… 3. For an analysis of black box methods and interpretable machine learning in heterogeneous catalysis see Esterhuizen et al (2022) and for an extensive discussion on epistemic opacity in computational chemistry see Wieber and Hocquet (2020). It is important to stress that not all black box models use ML nor all of them are subject of essential epistemic opacity. …”