Philipp Brandt, new researcher
Édité le 15 Octobre 2019
Philipp Brandt joined the CSO and Sciences Po as assistant professor. His research is at the intersection of economic sociology and the sociology of professions and expertise.
He received his PhD from Columbia University and spent time at the University of Mannheim as a postdoctoral researcher. His main research analyzes the emergence of the data science profession, using both computational methods and qualitative field observations, to understand the collective construction of a professional identity.
"New York City’s yellow cab industry, which I focus on, is useful case because it combines broad salience with interesting organizational technicalities."
The PhD in sociology you received at Columbia University was on the emergence of the data science profession. How did you get to that topic and what are your main findings?
I began my dissertation on the emergence of data science when it had just become visible in public debates, professional settings and in universities. For a while I feared data science might die before I could finish a dissertation. I also found few examples in the literature on professions that capture this incipient moment. Data science still seemed important and so I started with a few core conceptual themes and took it from there. These themes include formal knowledge and its informal construction, lay recognition of expertise and conflicts between experts. Professional identity construction, my main focus in the end, appeared little in that literature. Data science revealed it.
I combined computational methods with field observations to capture data science’s emergence as best as I could and from the different conceptual perspectives. Analyzing thousands of job descriptions, which reflect outside views of data science, I for example found novel combinations of existing skills in medium-sized organizations. Further evidence of academics who had to teach data science classes revealed practices that deviate from conventional scientific practices while still promising novel insights. Finally, and most importantly, I observed and analyzed how data scientists recognize their novelty and explain their departure from established quantitative work.
This novelty is separate from underlying knowledge. I argue that data scientists draw new boundaries around established quantitative expertise. They integrate technical skills with practices for understanding concrete problems around data. This may sound simple but it is quite an accomplishment because the technical skills come from academic contexts that often dismiss practical problems. I call this process whereby data scientists articulate a novel role “reflexive creativity”.
In your earlier work, you have studied the reconfiguration of institutionalized practices among dissenting preachers in early modern England and technology extension experts in US manufacturing networks. You summarize these publications with reference to the notion of “professional entrepreneurs”. Can you tell us more about it?
Modern data scientists, manufacturing extension agents and dissenting preachers have little in common, it seems. Data scientists first appeared in the Bay Area and New York City and deal with new data problems; our research on the manufacturing extension agents focused on the former industrial regions between the coasts. The two groups could hardly be more different and the preachers were from another time and place entirely.
They all deviate from conventional practices, however, creating new ones. We think of people who create novel organizational or institutional entities as entrepreneurs. The idea of “professional entrepreneurs” aims to take this idea to creative work of all sorts. Data scientists use established quantitative expertise from academia. They use this expertise to solve practical problems in commerce, public administration, health, education and many other fields. The manufacturing extension agents deviate from their politically mandated approach to distribute their services across many establishments. Instead, they build relationships to initiate more fundamental changes. Dissenting preachers are of course named after their deviant practices and readings of the bible. If we think of all these groups as professional entrepreneurs, we can understand the mechanisms that undergird new work.
In a new project, you study the dying profession of yellow cab drivers. Why did you choose this profession far from the «data scientists» ?
New York City’s yellow cab industry, which I focus on, is a useful case because it combines broad salience with interesting organizational technicalities. A diverse landscape of owner-operations, agencies and huge garages with hundreds of vehicles coordinate driving opportunities in this industry. Unlike the iconic yellow cab, they are not very visible, however, which makes it sociologically interesting. I study how taxi drivers learn to move through these organizations, searching for more rewarding driving opportunities.
In the data science project, I am studying the construction of a new kind of expertise. Here I ask how cabbies become experts of social structures. Taxi drivers face fierce competition whereas data scientists can pick and choose their options. Studying both, I hope to shed light on the struggle with new work across different social worlds.
What classes will you teach at Sciences Po ?
I teach about professions, sciences, organizations, political economy and economic activities. Throughout, I try to present sociology as a broad and diverse field of inquiry. I also enjoy teaching how we can use code to study diverse social scientific problems, activities and structures.
Next semester, I offer two undergraduate classes. I teach a substantive seminar on expert work in the digital age. We will discuss issues ranging from the rise of semiconductors and bio technology to randomized controlled trials, genomic sequencing and blockchain.
I also teach a lab that I called “Data science for social good”. We begin developing a sociological understanding of the data transformation and its consequences. Then come practical coding exercises to recognize and access interesting data sources. In the end, we want to sketch out ideas for using data to address social problems.
© Shutterstock/Hans engbers