“Track every move”: Analyzing developers’ privacy discourse in GitHub README files

Abstract

We adopt a socio-material perspective to examine how developers translate privacy, as a social value, into user applications. Our comprehensive survey of the research on developers’ privacy highlights their key position as privacy mediators and their forums as productive settings for unobtrusive studies of their discourse. The open-source code-sharing platform GitHub contains both discourse and code; by focusing on GitHub, we analyzed nearly 60,000 README files created between 2008 and 2020 that include the term “privacy,” studying quantitatively and qualitatively how discourse is translated into code. Using VOSviewer.com, we identified two main word clusters: “security” and “privacy policy.” Voyant-tools.org confirmed these findings, suggesting that some references elaborate on practices that safeguard privacy, while others discuss policy as a means of complying with both public and, ironically, commercial regulations. A closer reading of the files reveals that even privacy enthusiasts may inadvertently promote code that poses threats to privacy.