Platform Mediated Labor Management: Upwork, Fiverr and Amazon Mechanical Turk Workers’ Experience
Nga Than, Sociology
Faculty Advisor: Jessie Daniels
NML Award: The Data Analysis and Visualization Award (October 2019)
Increasingly capital has been adopting new technology in managing labor. Visceli (2016) and Synder (2016) show that both low-skilled and high-skilled workers such as truckers, and finance workers are now managed by algorithmic systems such as GPS, or high frequency trading, which requires more of their mental and physical labor. Workers in the service sectors are scheduled in a manner that they don’t have control over their time (Reich & Bearman, 2018). Recently the discourse about promises of the sharing economy to worker’s lives, and solving unemployment have been challenged by scholars (Schor, 2017; Rosenblat, 2018). Ravenelle (2017, 2019) argues that the sharing economy platforms have undermined various legal protections that we have achieved in many markets such as in taxi service industry, and hospitality industry. However, most research in the sharing economy has framed this algorithm-mediated labor management as a new way to find jobs for workers (Shor, 2017). Others have attributed the disorientating, and changing nature of work to the increase in using of algorithms in labor management (Rosenblat, 2018; Srnicek, 2017). This project follows the latter line of inquiry that more workers participate in the digital economy, and more workers are being managed via various platforms. It questions how workers navigate the control of platforms, algorithms, and whether they form a community both online and offline to talk about their labor conditions.
The project uses digital trace data collected from online forums of three different groups of workers: Amazon Mechanical Turk workers (or Mturkers), Upwork freelancers, and Fiverr freelancers. These three platforms provide workers with easy ways to find jobs in the algorithmic economy. The project utilizes machine learning tools such as structural topic modeling, and sentiment analysis to analyze a large corpus of text data. Preliminary results indicate that workers use these virtual spaces to share information when the platforms fail to communicate directly with them. For example they discuss issues such as how to get a new gig, payment difficulty, and abusive customers. We find that workers train each other how to start a gig, how to sustain their stream of income, and how to educate themselves continuously in a knowledge economy. Contrary to platforms’ intention to individualize workers’ experience, workers find a collective voice on these forums. This raises questions about occupational identity, and labor organizing in a digital age.