With their intensive use of data to classify and price risk, insurers have often been confronted with data-related issues of fairness and discrimination. This paper provides a comparative review of discrimination issues raised by traditional statistics versus machine learning in the context of insurance. We first examine historical contestations of insurance classification, showing that it was organized along three types of bias: pure stereotypes, non-causal correlations, or causal effects that a society chooses to protect against, are thus the main sources (...) of dispute. The lens of this typology then allows us to look anew at the potential biases in insurance pricing implied by big data and machine learning, showing that despite utopic claims, social stereotypes continue to plague data, thus threaten to unconsciously reproduce these discriminations in insurance. To counter these effects, algorithmic fairness attempts to define mathematical indicators of non-bias. We argue that this may prove insufficient, since as it assumes the existence of specific protected groups, which could only be made visible through public debate and contestation. These are less likely if the right to explanation is realized through personalized algorithms, which could reinforce the individualized perception of the social that blocks rather than encourages collective mobilization. (shrink)

The image of Max Weber as an “interpretivist” cultural theorist of webs of significance that people use to cope with a meaningless world reigns largely unquestioned today. This article presents a different image of Weber’s sociology, where meaning does not transport actors over an abyss of meaninglessness but rather helps them navigate a world of Chance. Retrieving this concept from Weber’s late writings, we argue that the fundamental basis of the orders sociologists seek to understand is not chaos. Action is (...) rendered interpretable, rather, to the extent persons orient themselves to possibilities and probabilities, which remain real but unknowable. In this framework, interpretation and probability are allies, not antagonists. Weber’s systematic use of the probabilistic notion of Chance as a central resource for concept formation and sociological explanation seriously challenges current understandings of probability as a purely statistical, atheoretical concern. We outline the conceptual difference it makes when basic categories of sociology are understood as rooted in Chance, and we point to the larger implications of Weber’s probabilism for contemporary debates around issues of prediction, action, and interpretation. (shrink)

As a reply to the commentary (Humphreys in Found Sci, 2012), we explore the methodological implications of seeing artificial neural networks as generic classification tools, we show in which sense the use of descriptions and models in data analysis is not equivalent to the original empirical use of epicycles in describing planetary motion, and we argue that agnostic science is essentially related to the type of problems we ask about a phenomenon and to the processes used to find answers.

In this paper we argue that data science is a coherent and novel approach to empirical problems that, in its most general form, does not build understanding about phenomena. Within the new type of mathematization at work in data science, mathematical methods are not selected because of any relevance for a problem at hand; mathematical methods are applied to a specific problem only by `forcing’, i.e. on the basis of their ability to reorganize the data for further analysis and the (...) intrinsic richness of their mathematical structure. In particular, we argue that deep learning neural networks are best understood within the context of forcing optimization methods. We finally explore the broader question of the appropriateness of data science methods in solving problems. We argue that this question should not be interpreted as a search for a correspondence between phenomena and specific solutions found by data science methods; rather, it is the internal structure of data science methods that is open to precise forms of understanding. (shrink)

As a reply to the commentary (Lenhard in Found Sci, 2012), we stress here that structural understanding of data analysis techniques is the natural counterpart to the lack of understanding of phenomena in agnostic science. We suggest moreover that the dynamics of computational processes, and their parallels with the dynamics of natural processes, will increasingly be, possibly, the driving force of the development of data analysis.

Computer models and simulations have become, since the 1960s, an essential instrument for scientific inquiry and political decision making in several fields, from climate to life and social sciences. Philosophical reflection has mainly focused on the ontological status of the computational modeling, on its epistemological validity and on the research practices it entails. But in computational sciences, the work on models and simulations are only two steps of a longer and richer process where operations on data are as important as, (...) and even more time and energy-consuming than modeling itself. Drawing on two study cases—computational embryology and computational epidemiology—this article contributes to filling the gap by focusing on the operations of producing and re-using data in computational sciences. The different phases of the scientific and artisanal work of modelers include data collection, aggregation, homogenization, assemblage, analysis and visualization. The article deconstructs the ideas that data are self-evident informational aggregates and that data-driven approaches are exempted from theoretical work. More importantly, the paper stresses the fact that data are constructed and theory laden not only in their fabrication, but also in their reusing. (shrink)

In this commentary to Napoletani et al. (Foundations of Science 16:1–20, 2011), we put agnostic science in a wider historical context of philosophy of mathematics. Secondly, the parallel to Tukey’s “exploratory data analysis” will be discussed. Thirdly, it will be argued that what is new is the mutually interdependent dynamics of data (on which Napoletani et al. focus) and of computational modeling—which puts science closer to engineering and vice versa.

In this commentary to Napoletani et al. (Found Sci 16:1–20, 2011), we argue that the approach the authors adopt suggests that neural nets are mathematical techniques rather than models of cognitive processing, that the general approach dates as far back as Ptolemy, and that applied mathematics is more than simply applying results from pure mathematics.