Alina Mierlus

Autonomous University of Barcelona

Data archives: An epistemic approach

This paper aims to present a detailed description of data archives, arguing that datacan be epistemologically interpreted and that a proper conceptualisation can lead to newtheories to address current issues within data science practices. I follow Derrida's analysis ofthe archive as having two principles: the principle of beginning, which consists of orderingand resembling signs or traces, and the ontological or political principle. Although the two principles are inseparable, I argue in favour of focusing on the firstprinciple, the epistemic one. Doing this allows us to roll out a strategy for conceptuallydefining data as an epistemic entity. I thus switch from the conception of data-as-relata (dataas a relational, proto-epistemic entity, subordinated to the notion of information) todata-as-trace and thus to the proprieties of context and compositionality. This epistemic operation is an intervention into the core of Big Data rhetoric, plantingthe seeds for conceptual and theoretical work that aims to solve the following paradox. Onthe one hand, data can be relevant in tackling many climate change challenges, such asanalysing satellite data to forecast meteorological conditions and make informed decisionson climate change actions. On the other hand, given the current technologies, acquiring,storing, and processing this data requires vast resources and contributes significantly to theproblem of climate change. By drawing on notions such as Rheinberger's epistemology of the concrete, Derrida'sthinking on the archive and context, and other contemporary debates on data science, Ipresent a conceptual framework for data archives. This work aims to contribute directly tocurrent data governance proposals.

Bio

Alina Mierlus is currently a PhD candidate in the Cognitive Science and Languages program at Autonomous University of Barcelona, where her work focuses on Philosophy of Language, Epistemology and the Philosophy of Artificial Intelligence. Her thesis draws on the post-structuralist tradition and its debates with analytic philosophy, proposing a conceptualization of AI Data aimed at tackling current governance and theoretical challenges. Before starting her academic pathway in philosophy, she studied computer science, helped build software, dedicated hours to teaching and digital literacy, and spentmany years contributing to open source.In parallel with her PhD, she obtained a Professional Certificate in Large Language Modelsand is pursuing an online master program in Data Science.