Accepted Paper:

Gender politics of algorithms: On biases and 'facts'  


Corinna Bath (Technische Universität Braunschweig)

Paper short abstract:

This paper explores the gender politics of algorithms in two different ways. First, it aims at conceptualizing biases in algorithms with a new materialist approach. Secondly, it discusses a recent case of a struggle against gender studies and critical thought on the same theoretical grounds.

Paper long abstract:

Algorithms can be used and exploited in unethical ways. The recent case of the Volkswagen crisis, where software engineers manipulated code in order to satisfy certain thresholds values, for instance, shows that a specific historic and economic context can make them political. This understanding of the politics of artifacts, however, does not question the generally assumed neutrality of computational artifacts, whereas those examples that demonstrate effects of algorithms which were not intended by the designers, particularly sexist or racist effects, provoke common sense thinking even more.

My paper argues that the theoretical approach of the new materialism provides suitable concepts for addressing the gender politics of algorithms in the latter sense and its increasing relevance in times of big data and automated decision making. I will furthermore draw the attention to a specific case: an article published in a serious German newspaper that discussed these issues and stimulated contradicting reactions. Gender studies researchers invited me for talks, radio stations asked me for interviews, whereas a demagogue working against the institutionalization of gender studies at German universities aimed at undermining my academic expertise.

In order to explore this case on basis of new materialist thinking I will analyze intra-actions and entanglements of various subjects, objects and materializations: gender studies proponents and opponents, university representatives and legal administrators as well as the backlash against gender studies in Germany, dominating strands of epistemology in computer science, particularly naïve realism and struggles between layperson's and expert's knowledge.

Panel T159
Political Objects. Prescriptions, Injustices and Promises of Material Agents