Retracing Algorithms: How Digital Social Research Methods Can Track Algorithmic Functioning

The expanding use of algorithms in society has called for the emergence of “critical algorithm studies” across several fields, ranging from media studies to geography and from sociology to the humanities. In the past 5 years, a consistent literature on the subject has developed. Inspired by these studies, we explored the ways digital traces may be employed for auditing algorithms and find evidence about algorithmic functioning. We focus on the analysis of digital traces through search engines and Application Programming Interfaces (APIs). We present four cases of how digital traces may be used for auditing algorithms and testing their quality in terms of data, model, and outcomes. The first example is taken from Noble’s (2018) book Algorithms of Oppression. The other three examples are very recent, two of them related to COVID-19 pandemic and about the most controversial type of algorithms: image recognition. Search as research and the analysis of digital traces and footprints within quasi-experimental research designs are useful methods for testing the quality of data, the codes, and the outcomes of algorithms.