Alum Stuart Geiger at the Society for Social Studies of Science in Boston
On August 31st, Stuart Geiger, a computational ethnographer and postdoc at the Berkeley Institute for Data Science, presented a paper about data science research. Stuart presented his paper "Autoethnographic Methods for Studying Data-Driven Knowledge Production," co-written with Charlotte Mazel-Cabesse and Brittany Fiore-Garland.
From the abstract:
This paper is based on a collaborative, multisided ethnography of data science, in which the authors have been embedded in aligned institutes dedicated to data science. In this paper, we focus on autoethnographic methods, which can be powerful and generative ways to conduct empirical investigations into data science practices across many theoretical issues. This paper reviews several different exercises, initiatives, and activities that we have conducted in our fieldwork, reflecting on how they help us better understand different aspects of what it means to do data science. First, this paper discusses an activity in which an ethnographer and a data scientist set out to install all the software infrastructure needed to run a particular machine learning library on a new computer. This proved to be a generative method to explore not only the layered infrastructural dependencies upon which contemporary data science runs. Second, this paper discusses an initiative in which we committed to conduct the analysis of a standard social science survey of data scientists using best practices and platforms for reproducibility, as they were communicated to us by various participants. This provided a rich opportunity to explore what reproducibility means to different people in a particular embedded context. Finally, this paper discusses our use of a website hosted via an open source GitHub repository to coordinate a peer learning group’s activities in one of our fieldsites. This provided opportunities to understand the role that the GitHub platform plays in the coordination of work.
Check out the program here.