Experimental indices: Situational assemblages of facial recognition

Abstract

Facial recognition technologies are increasingly used outside of constricted, laboratory-like settings. While supporters of the technologies contend that they help in identifying threats by linking specific bodies to hard evidence, we argue that the indexical relations they exhibit are best described as experimental, pointing to specific situational constellations within which they were initially created. By revisiting key moments in the development of (semi-)automated facial recognition technologies from the late 1960s to the present, we identify varying situational assemblages of facial recognition that depend on different understandings of indexicality. These experimental indices rely on historical dynamics, including significant government interest in the development of facial recognition technology, expansion in the scale of experimental settings, and dissolution of the formerly strict boundaries between the social spheres of private image-sharing, commercial image distribution, and institutional image forensics for identification. In coupling experimental indices with the development of facial recognition technologies, we hope to show a way forward to comparing the histories of other evidential technical images too.