THE NETWORKED IMAGE AFTER WEB 2.0: Flickr and the ‘Real- World’ Photography of the Dataset

This chapter considers the networked image in the context of an image-sharing culture informed by the extractive paradigms of the database and dataset. It offers a reading of Flickr as both a photo-sharing community and a standing reserve of visual stock for AI industries, which rely on massive photographic datasets sourced from photographic communities to train machines to see. Central to this account is the position of ‘real-world’ photography in computer science discourse, which is ontologically conflated with the Flickr snapshot. The chapter argues that the Flickr snapshot is able to stand in for a ‘real-world’ scene through its apparent amateurism, in which the automatism of camera and photo-sharing interface secures the snapshot’s position as a naïve, banal and transparent measure of the everyday. From this perspective, the Flickr snapshot, as an ‘image in the wild’ is positioned against the bias of the lab environment and the professional photographer, where social media imagery promises to statistically capture the long tail of the ‘real’. In observing the convergence of snapshot and stock image in AI industries, the chapter concludes that the photographic pipeline of machine learning seeks to stabilise and align the inherent polysemy of the networked image in order to produce actionable insights and, in doing so, renders the photograph transparent.