Digital Apprehensions: Policing, Child Pornography, and the Algorithmic Management of Innocence

Abstract

In this paper, I discuss the embodied labor of policing child pornography through the ways in which algorithms and human reviewers like Linda see abuse images. I employ the concept of “apprehension” to suggest that the ways that reviewers “see” child pornography is always already oriented toward the capture and arrest of suspected offenders. As I have argued elsewhere (Author 2017; Forthcoming), the use of new digital techniques to find child pornography has fundamentally transformed and expanded policing into a distributed network of labor increasingly done by computer scientists and technology companies. Rather than suggest new software is the cause of these transformations, I draw attention to the constitutive and mutually defining relation between computing and corporeality, or how image detection algorithms need the work of human perception to put their detective skills to work. I argue further still that the case study of child pornography detection offers an entry point into examining the algorithmic management of race. I suggest that childhood innocence is coded as whiteness, and whiteness as innocence, in the algorithmic detection of victims and abusers. By taking ‘detection’ as a dynamic practice between human and machine, I make an intervention into critical algorithm studies that have tended to focus solely on the programming of racial bias into software. The algorithmic detection of child pornography hinges, crucially, upon practice and the tacit observation of human reviewers, whose instinctual feelings about child protection and offender apprehension become embedded within the reviewing and reporting process as cases escalate for law enforcement.