The Unintended Consequences of Automated Scripts in Crowdwork Platforms: A Simulation Study in MTurk

Abstract

Crowdworkers on platforms like Amazon Mechanical Turk face growing competition as a result of the global excess supply of digital labour. As a result, many crowdworkers are turning to automated scripts, which help them locate better tasks faster and to boost their earnings. However, to date, it is not clear whether and to what extent the use of such scripts influence the opportunities for those crowdworkers who do not use them. This an important aspect that warrants further exploration because it can have negative implications for the health of crowdwork platforms. In this study, we use Discrete Event Simulation to identify and quantify the unintended consequences of the excessive use of automated scripts. Our findings show that, while the use of scripts allows some crowdworkers to identify and accept far more tasks than others, in the long run, this behaviour results in their competence persistence and reputational persistence and progressively to detrimental impacts for those workers who do not use scripts, and who may ultimately be forced to exit the platform. As a result, automated scripts have negative consequences, whereby their excessive use leads to a tragedy of the commons for all platform stakeholders, including the crowdworkers, the job requesters and the platform itself.