Send message to Authors

Short abstract

Contemporary data infrastructures increasingly detach data representations from social reality. This paper introduces digital identity epistemic fragility to analyse how this detachment emerges and stabilises, drawing on an in-depth empirical study of South Africa’s identity data infrastructure

Long abstract

Contemporary data infrastructures operate as epistemic anchors, mediating how social reality is rendered knowable, actionable, and comparable across organisational and institutional contexts. While growing scholarship has examined bias, discrimination, and opacity in data-driven systems, these concepts insufficiently capture a more subtle and structural transformation: the progressive detachment of data-based representations from the social realities they are meant to describe. This paper introduces digital identity epistemic fragility as a conceptual framework to analyse how such detachment emerges and stabilises within large-scale data infrastructures. The concept is developed through an empirical analysis of South African identity data infrastructures. The empirical analysis draws on over forty semi-structured interviews and informal conversations with government officials, industry representatives, standardisation bodies, and civil society actors, complemented by policy documents, technical standards, patents, and archival material tracing the historical evolution of South Africa’s identity infrastructure. This project first clarifies the conceptual foundations of digital identity epistemic fragility and distinguishes it from adjacent notions such as bias, infrastructural failure, and algorithmic opacity, situating the contribution within the broader tradition of data infrastructure studies. It then outlines how it can be treated as an empirically tractable process through qualitative analyses of data infrastructures, including the tracing of referent displacement across data lifecycles, the reconstruction of infrastructural decision logics, and the examination of how organisational actors interpret and rely upon system outputs over time.