Recognizing that EdTech is increasingly shaping the teaching profession through the datafication and visualization of student learning, this paper advances an analytical framework for eliciting the “politics” of data visualizations. With inspiration from Science and Technology Studies (STS), well-suited for analyzing the co-constitution of technology and society, the paper suggests a framework for analyzing data visualizations’ ‘aesthetics’, ‘agencies’ and ‘alternatives’ as important if we are to understand their implications for teachers’ expertise. Rather than assuming the ‘technical’ and ‘social’ to be separate domains, STS invites an ‘infrastructural inversion’ (cf. Bowker & Star, 2000) where questions of politics, ethics, and knowledge are examined through infrastructural entities. Specifically, the notions of ‘aesthetics’, ‘agencies’ and ‘alternatives’ allow eliciting how the aesthetics of data visualizations also entail an interpretation of data, how visualizations configure expertise across human and machine agencies, and how visualizations themselves are contingent results of ongoing negotiations of their makers (Coopman, 2014; Ratner & Ruppert, 2019; Schaffer, 2017; Suchman, 2007). This approach thus casts light on the power of data visualizations as a device for shaping expertise but also appreciate them as cultural and social artifacts that could be otherwise. The paper demonstrates these analytics through a qualitative case study of a widely used digital mathematics platform for the primary and lower school (“folkeskole”) in Denmark, ‘MathTraining’. Launched as an adaptive and self-correcting platform in 2010, MathTraining today has become one of the most popular Danish platforms in mathematics. The analytical sections show, respectively, 1) how the aesthetics of data visualizations shape expertise by calculating and also interpreting student learning on behalf of the teacher; 2) how data visualizations configure expertise across human and machine agencies, automating student assessments and attracting teachers’ attention towards student engagement and progression, and 3) map out the alternative visualizations that never became part of the platform, demonstrating the contingent aspect of data visualizations, in terms of how different actors in the EdTech company have different ideas about how and which data should be visualized. Examining both the intentions inscribed into the visualizations as well as ongoing mundane negotiations about which data to visualize and how, allow us to better appreciate the normative dimensions of unsettled and ethical questions about the role of automated digital systems in education, including how they reconfigure teachers’ socio-technical way of ‘seeing’ and attending to learning.

Bowker, G., & Star, S. L. (2000). Sorting Things Out: Classification and Its Consequences. MIT Press. Coopmans, C. (2014). Visual analytics as artful revelation. In C. Coopmans, J. Vertesi, M. Lynch, & S. Woolgar (Eds.), Representation in Scientific Practice Revisited (pp. 37–59). MIT Press. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-151377 Ratner, H., & Ruppert, E. (2019). Producing and projecting data: Aesthetic practices of government data portals. Big Data & Society, 6(2), 2053951719853316. Schaffer, S. (2017). Introduction. In S. Schaffer, J. Tresch, & P. Gagliardi (Eds.), Aesthetics of Universal Knowledge. Palgrave Macmillan UK. Suchman, L. (2007). Human-Machine Reconfigurations: Plans and Situated Actions (2nd ed.). Cambridge University Press.