The Challenges in the Varying Visibilities of Social Media Data

The post-lunch session at the 2019 AoIR Flashpoint Symposium starts with Christina Neumayer and Luca Rossi, who are interested in invisibilities in social media data. For instance, studying protest movements through social media means studying only what is visible about these movements in specific social media platforms – the data must exist in the various technological layers of these platforms in the first place, and those layers significantly constrict what data are available to the researcher. Additionally, such data must also be perceived as meaningful; this requires a shared understanding in the scholarly community that such data can be used to examine a particular phenomenon.

Data are thus a methodological conundrum that is able either to empower or to bias the research; they are an object of study that is characterised by its contested nature between conflicting actors; and they serve as a research catalyst that is able to make social dynamics visible. To fully understand this, we might need to trace data back to before they become data: this includes the strategies that actual users employ in order to manage their visibility (e.g. through obfuscation, disguises, pseudonyms, or bots); the processes of datafication by the platforms themselves (including business models, policies, and other factors); and our own retrieval and analysis practices as we gather, process, and refactor these data.

All of this involves a range of data visibilities from the invisible through the quasi-visible to the fully visible, and the fully visible tends to be overrepresented in these processes. The visible distracts us from the invisible, while it is the invisibilities that can tell us something about the power relations in activist struggles, and can help us understand the limitations of social media research.