Fetishization Revisited: Faith and Notions of Efficacy in the Making of Algorithms

• In Event: Algorithms in/of Culture: Exploring the Global Reach of Algorithms

Abstract

Algorithms, in recent years, have become a catchword: a focus of academic fascination, a rarified artifact that commands extraordinarily high salaries for those who make them, and "a magic black box" for those who use them. They order news feeds, turn on heaters, and optimize fitness. To explain these near magical powers of algorithms, this paper turns to scholarship of fetishes. Fetishes are not indications of “false thinking,” but instead are “good to think with.” Fetishization, and denouncing accusations of fetishization, occur at particular historical junctures where misrecognizing the agency of human encounters, and investing efficacy and power in an object instead, does important cultural work. Fetishes, be they algorithms or data, act as promises that animate our material worlds when social contracts between people are too slippery or nascent to address directly. This paper will explore these ideas through ethnographic fieldwork with Quantified Self participants and computer vision professionals, where claims about scientific efficacy mingle with faith and hope, alongside occasional accusations of false thinking. Like the artifacts of colonial trade in West Africa, algorithms and data facilitate social exchange in circumstances where shared understandings are thin on the ground. Following David Graeber, the paper argues that fetishization is not the same cultural logic as deification or demonization, which constitute perhaps more profound risks of algorithmic culture.

Authors

• Dawn Nafus, Intel

• Suzanne L. Thomas, Intel Labs

• Jamie Sherman, Intel Labs