Lilly Irani

Bio

Lilly is a PhD candidate in Informatics with a Graduate Feminist Emphasis. Her current work looks at circulation of design methods and knowledge, particularly between the U.S., Europe, and India.She works at the intersection between feminist and postcolonial STS, human-computer interaction, and design.

Prone to side projects, she has been haunted my Amazon Mechanical Turk since summer 2008. In response to workers' own complaints about AMT, she co-authored Turkopticon with Six Silberman. Turkopticon is a Firefox extension that allows Turk workers to review employers and avoid ones with bad records, but it also stands witness to Amazon's neglect of worker welfare.

Previously, she worked as a User Interface Designer at Google, quite implicated in the digital economies she critiques.

Abstract

Killing Time on Mturk
Amazon Mechanical Turk is internet "marketplace" in which employers can have large volumes of digital tasks performed by workers at very low cost, often less than a dollar per hour. Yet several surveys report that the vast majority of AMT workers participate for fun or to kill time, quite thrilled to get paid for their entertainment rather than paying for it. Eleven percent of workers, however, always or sometimes rely on the AMT earnings to make ends meet. The experiences of exploitation are dramatically varying in degree.

This talk will describe ways of thinking about modes of exploitation in AMT: social thinness, response-ability, and precarity. First, social thinness is James Ferguson's account of what made 1990s African mineral extraction practices particularly exploitative; I describe social thinness and ways in which AMT is -- and is not -- socially thin. Second, Donna Haraway offers response-ability as an aesthetic of ethical labor; I explain how AMT has been designed to reduce the burden of response-ability. And third, I situate AMT in broader trends in labor precarity.

These different ways of seeing labor configurations in AMT suggest different kinds of activist responses and affinities.