Anya Bernstein : Works in Progress
Predictive Government Databases. Predictive government databases do not just store information about individuals. They create information by offering predictions about individuals’ future conduct. This article argues that these predictive functions warrant attention and require constraint. I sketch out what distinguishes predictive government databases, which distribute predictive labor over a number of participants. This can smooth out idiosyncrasies of judgment, but it can also cloak predictions in a veneer of objectivity, exacerbating those very idiosyncrasies and giving government a false confidence in its knowledge. We should take these impediments seriously, because predictive government databases can have wide-ranging effects even while they can be subject to perverse incentives that can grant institutional benefits to spurious predictions. By entrenching and emphasizing particular social categories, they can misdirect public policy, skew the distribution of resources, and distort public conceptions of society and its most urgent problems. The legal regime that governs these databases neither recognizes nor regulates their predictive functions. It focuses on descriptive rather than predictive accuracy and protects individual rights instead of social responsibilities. I suggest that requiring such databases to seek feedback, build in regular revision, and acknowledge their subjective moments would mitigate their harms. By bringing their peculiarities to the surface, these constraints would also allow for a public discussion about what governments owe the populations they strive to know.