Data jurisdictions and rival regimes of algorithmic regulation

Abstract

This article aims to characterize and compare some approaches to regulation manifest in distinct yet intersecting domains of data assemblage and algorithmic development, and to explore some implications of their operating in concert. We focus on three such types of domain, each oriented towards different purposes: market jurisdictions; public science jurisdictions; and jurisdictions of humanitarianism. These domains we characterize as data jurisdictions because they tend to propagate distinct normative claims and concerns, and authorize particular types of speech and action, through algorithmic operations and data formatting. In this paper, we focus on the intersection of these archetypal data jurisdictions in two, related initiatives of the United Nations (UN): Haze Gazer and CycloMon. In the context of these projects, the market domain is represented by their incorporation of Twitter and social media data; the public science domain by their use of NASA Earth Observatory data, US National Oceanic and Atmospheric Administration (NOAA) data, and Air Quality Index China (AQICN) air quality data; and the humanitarian domain by their status as UN projects designed to serve the aims and enlarge the capacities of development and humanitarian professionals. We analyse how, and with what ramifications, these domains of algorithmic regulation intersect in Haze Gazer and CycloMon. \n \nIn so doing, we advance two main arguments. First, we argue that certain normative commitments regarding data, data use, and data users circulate and gain ground through their embeddedness in seemingly benign infrastructures and formats of data handling and representation. Particular (contentious) norms are prioritised, spread and imbibed as much through day-to-day data usage as through explicit argument or endorsement. Second, we argue that blind spots tend to emerge from the intersection of different jurisdictions over, or approaches to, the challenge of responsible algorithmic regulation. The data jurisdictions that we analyse in this article demand quite divergent normative commitments, but the conflicts among these are hard for users to discern in day-to-day interaction with the platforms that we describe. We contend that jurisdictional analysis of projects in operation may help data contributors and users to take account of, and potentially take a stand on, these important differences.