Antoinette Rouvroy: Contestability in the Big Data era

Abstract

Data driven governance is often presented as « revolutionary » mode of governance emancipated from the yokes of (biased) human representation, (ambiguous) human language, (situated) human point-of-view,... The « attractiveness » of big data and data-driven decision making processes, symptomatic of a crisis of (symbolic, institutional, linguistic, political, artistic,…) representation – enables and presupposes the silencing/repression of everything that would object to the indistinction between the world and the digital reality, or that would object to the « authority », “reality” or “factuality” of the data as « given » and exhaustive, or as the uncontestable state of things. Yet, algorithmic « visions » are not more « adequate »  to the world than human representation but, « governing » what is not (yet) actually present to the sight[1] algorithmic visualization produce new objects and spaces for preemption and optimization and, folding the digital upon itself, tends to immunize itself against the incompressible portion of uncertainty inherent to the physical world. These new potential objects and spaces – foreshadowing opportunities and threats - are neither accurate or erroneous: they are not meant to be true or false, they are meant to immunize the “actual” from the « virtual», the probable from the untameable excess of uncertainty, the inorganic impassibility against the organic events.

These potential objects and spaces (which may take the form of « predictive » profiles), immune from any kind of confrontation with the physical world could appear not more threatening than a poem by Mallarmé - a poetry which also implies the «dissolution of the real and the constitution of radically autonomous objects, that is, of objects which wouldn’t even have as departure point the detection of a divergence with any reality »[2]  - if they were not always already pre-dated by optimization and preemption logics, immunizing them against alteration, hence against life itself.

The questions is thus : how do we invent ways to intervene with(in) these new data-objects and data-spaces in ways that would favour bifurcations rather than optimization of the unsustainable state of affairs? I will argue that recalcitrance to an algorithmic governmentality enclosing the digital upon itself does not imply a rehabilitation of « representation » as much as it requires the preservation and revalorization of inactualizable “heterochronic”[3] scenes, new « drammatisations » allowing enunciation, fabulation, narration of the non-digitalizeable, of projects without predictions (rather than predictions without projects), spaces of critique beyond judgments of accuracy and error.