Epistemology After the Algorithm: A second-order cybernetic intervention

Arantzazu Saratxaga

Epistemology After the Algorithm examines the assumptions that allow algorithmic systems to acquire epistemic authority in contemporary institutions. The track challenges the idea that reality is objectively given, fully representable, and manageable through computation, and questions the related claim that algorithmic outputs can function as neutral or comprehensive knowledge. Drawing on second-order cybernetics, it redirects attention from observation as transparent access to the world toward observation as a selective operation performed by observers who necessarily have blind spots.

The contributions develop this intervention across education, architecture, engineering, urban governance, clinical judgement, machine intelligence, and computational design. They consider how algorithmic systems reorganise learning, decision-making, professional judgement, and design practice; how claims of objectivity conceal the distinctions embedded in models and datasets; and how automation can narrow the spaces in which uncertainty, interpretation, moral discernment, and responsibility are exercised. Several papers use design and pedagogy as sites for testing alternatives, including conversational and reflexive approaches to AI, parametric and generative processes, and forms of learning that treat error, difference, and interpretation as constitutive rather than as noise to be eliminated.

Second-order cybernetics provides the connective tissue for these inquiries by insisting that the observer cannot be removed from the observed system. From this position, transparency cannot mean a view from nowhere, and control cannot mean the elimination of uncertainty. The track instead foregrounds tensions between observation and representation, self-organisation and algorithmic determinism, communication and control, reflexivity and automation, and uncertainty and total calculability. Its aim is not simply to critique algorithms, but to reopen spaces for thought and action in which knowledge remains situated, contestable, and accountable to the relations through which it is produced. In practical terms, this means designing institutions and learning environments that keep interpretation visible, preserve room for judgement, and make the conditions of algorithmic observation themselves available for scrutiny.