Scripts and softwares are indispensable components of contemporary research. While the study of scientific instruments has been extensively covered in science and technology studies (STS), as well as aspects related to digital data, software strangely remains absent. I will build on the success of Jupyter computational notebooks to sketch a social study of open source scientific softwares.

The Jupyter project, initiated in 2001 by a physics graduate student interested in scientific programming, has gradually evolved into a global data science infrastructure. I will demonstrate the importance of closely examining its trajectory by identifying the continually negotiated conditions for the transition from local usage to a role as a generic instrument, progressively leading to its integration into a diversity of services beyond scientific communities. The dissemination of computational notebooks has thus not only facilitated a convergence between open code and open science but also between the practices of researchers and those of computer developers, leading to the stabilization of a specific infrastructure for data analysis.