Sociotechnical Audits for Targeted Advertising Evaluation

This document is a research article that presents the concept of sociotechnical audits (STAs) as a method for evaluating algorithmic systems, specifically in the context of targeted advertising. It outlines the limitations of traditional algorithm audits, which focus solely on technical components without considering user interactions and behaviors. The authors introduce a platform called Intervenr, designed for conducting longitudinal, browser-based STAs with compensated participants. The article details a case study involving 244 participants over two weeks, where the first week involved observing ad targeting behaviors and the second week implemented an intervention that altered the ads seen by participants. The findings indicate that while targeted ads perform better initially, users adapt to different ads quickly, which raises questions about the effectiveness of personalized ad targeting over time. The research emphasizes the importance of understanding the interplay between algorithms and users in sociotechnical systems.