November 23, 2016 |
The foucaultblog and the Belgian research group Algorithmic Governmentality have organized an international workshop on digital profiling, to be held at the Université Saint-Louis in Brussels on December 2–3, 2016. The event will take up an ongoing debate whether Foucault's concept of governmentality helps to understand algorithmic practices, especially commercial profiling. The lectures will be published as featured papers on the foucaultblog and edited as special issue in 2017. The program and the abstracts are posted below.
Algorithmic Finance and (Limits to) Governmentality
Christian Borch, Professor of Sociology, Copenhagen Business School
The aim of this presentation is to discuss algorithmic finance, specifically the use of automated trading (including high-frequency trading), in light of Foucault's notion of governmentality. I argue that particular aspects of the predictive models deployed in algorithmic finance can be fruitfully analyzed from a governmentality perspective. However, to fully understand the realm of algorithmic finance, it is nonetheless useful to supplement a governmentality approach with an analytical lexicon which is not primarily centered on productive forms of power. Specifically, I suggest that algorithmic finance often works in ways that are better grasped through, e.g. Roger Caillois's work on mimesis.
Recommender Systems & Relations of Power
Robin Devooght, PhD Student, Université Libre de Bruxelles
Jérémy Grosman, PhD Student, Université de Namur
This paper intends to challenge, both conceptually and empirically, the speculative hypothesis of "Algorithmic Governmentality" put forward by Antoinette Rouvroy and Thomas Berns. The presentation shall thus attempt to articulate, as closely as possible, the making of recommender systems, the power effects ascribed to them and the different concepts of 'power' one can identify in M. Foucault's writings, as well as in some of his predecessors' works (namely, Gabriel Tarde and George Canguilhem). The claim is made that if such articulations exist, they are more likely to be identified by paying close attention to the minute steps where technicians hesitate, in defining their objectives, handling their data, and designing their algorithm.
The Birth of Profiling from the Spirit of Austromarxism
Simon Ganahl, Habilitation Fellow, Austrian Academy of Sciences
Before Paul Lazarsfeld became a major figure in 20th-century American sociology, he tried to make a living as a social researcher in Vienna. In 1931, he founded the Economic-Psychological Research Bureau, which linked statistical studies with in-depth interviews in order to reveal the true motives for purchasing behaviors. This so-called 'motivation research' came to be a key technique in post-war marketing. My talk will show examples of Lazarsfeld's Viennese studies and mainly focus on a survey of Austrian radio listeners. By correlating radio programs with social class, this study broke the mass audience down into specific target groups for the very first time.
Profiling Digital Subjects
Olga Goriunova, Senior Lecturer in Digital Culture, Royal Holloway, University of London
Recently, I worked on exploring the return of the subject within the computational context, which I addressed as a digital subject. This digital subject encompasses a digital identifier, correlations in data or a data profile, alongside a subjectivity performed in social media, moving between biological characteristics and symbolic expression. I focused on the processes through which digital subjects are constructed by matching, correlating, modeling, as well as on how they become enactive. For this workshop, my aim is to enquire into the lines of differentiation that sustain the generation of digital subjects for varying purposes: for instance, commercial profiling as opposed to state surveillance. Using presumably the same machine learning models, the ways of pulling data together into digital subjects and establishing forms of indexicality will vary, depending on how different practices of governmentality enlist complex alignments of economic, juridical and state powers, technical affordances, calls of modernity and pulls of the new neoliberal order.
Recommending as Self-steering
Tyler Reigeluth, PhD Student, Université Libre de Bruxelles
Through techniques such a "collaborative filtering," user-consumers' behaviors on web services are fed back into the recommendations these services provide in an attempt to "personalize" and tailor their supply. While these services' interfaces appear to explicitly address a subjectivity, the effectiveness of their recommendations putatively rely on the user-consumers simply "being themselves" so do they not try to "game" the system. By focusing on algorithmic techniques such as "recommender systems," I will try to sketch out some of the normative dimensions of contemporary practices and techniques of recommendation as a form of government. I will return to some of Foucault's later writings, namely L'herméneutique du sujet, wherein subjectivity is understood as a particular ethical relation to one's self whereby the subject's knowledge and care of self is a precondition for governing others. I will also examine cybernetic theories' conceptions of "self" as being a question of communication and control, and explore some of the articulations with Foucault that he himself opened without pursuing.
The Orders of Power
Bernhard Rieder, Associate Professor, University of Amsterdam
Foucault's The Order of Things is only rarely connected to his later work on governmentality. However, when looking at contemporary arts and procedures of governance, we can observe extensive use of knowledge techniques that rely heavily on data collection and analysis to understand, forecast, decide, and control. Relational database systems, statistical analysis, and machine learning algorithms are some of the computational techniques that introduce particular forms of knowing into the practice of governance and play various roles as engines of order. This talk mobilizes Deleuze's reading of Foucault's work — and in particular the notions of "unlimited finity" (fini-illimité) and the "Superfold" (surpli) — to interrogate both the epistemological character and governmental thrust these forms of knowing imply.