Human computing via online labor markets. The perils and promises of crowdsourcing in data-rich ecosystems

In this article, we focus on the need for “human computing” in data-rich ecosystems, notably as a consequence of data variety and typically for name disambiguation, and explore ways to manage it via online platforms for paid crowdsourcing. Based on several studies of Amazon Mechanical Turk, a well-established platform for matching data treatment tasks to human beings willing to carry them out, we illustrate the difficulties involved as requesters compete for the attention of workers. We suggest that researchers should shift from a technical analysis and tentative resolution of human computing perils and pitfalls, towards a more economic and managerial analysis of human computing platforms understood as online labor markets, notably in their multi-sided nature and with respect to how they manage the attention of online workers.