How do platforms pay data workers? Insights from Remotasks Plus Venezuela
1 : Yale University
2 : Universidad Icesi
3 : Centre de Recherche en Économie et STatistique (CREST)
CNRS, CNRS : UMR9194
4 : Centre National de la Recherche Scientifique
(CNRS)
* : Corresponding author
CNRS, CNRS : UMR6226
A subsidiary of the technology company Scale AI, Remotasks is a platform designed to crowdsource data work such as image tagging, sound transcription, and other forms of annotation. Between March 2020 and April 2021, it launched Remotasks Plus Venezuela, an ‘experiment' that aimed to attract workers from this country at a time of deep crisis, offering them improved conditions in terms of access to training, minimum hourly wages, and bonuses.
While illuminating the extreme conditions of Venezuela amidst economic collapse and the Covid-19 pandemic, this case offers a tremendous opportunity to address the more general question of the attractiveness of platforms depending on the local socio-economic outlook. Even more importantly, it can provide insight into the management and functioning of a data work platform. Do payments follow productivity and/or working time closely? How frequent are bonuses (resp. sanctions), and do they correlate with drops in engagement time and/or output quality? Are rules transparent, and consistently applied? How do workers respond to the platform's decisions? While some literature on data work hints at irregularities in the management of payments that may even amount to wage theft, it is largely based on anecdotal facts. Our goal is to provide more systematic evidence and to assess its generalizability.
To achieve this, we harness a detailed dataset provided by Remotasks itself, with the hours worked, the quality achieved, the bonuses earned, and the payments made to all the participating Venezuelan workers throughout the duration of the program. To grasp the context and to compare our interpretations to workers' first-hand perceptions, we also leverage qualitative interviews with former Remotasks Plus users.
Results show that the program was indeed attractive in Venezuela. The size of the workforce increased steadily throughout 2020, and many workers transitioned from practicing this activity accessorily to making it their main source of income. We also observe growing heterogeneity in participation over time, with a few contributors doing most of the work, and the majority devoting a limited amount of time to it. Further, the analysis confirms the power of platforms in setting and applying rules in often untransparent ways – in particular, unclear criteria for awarding bonuses, imposing sanctions, and adding hours. Likewise, platforms impose limitations, like a cap on the number of payable hours, without consideration for workers' circumstances. Worsening conditions are one reason why workers' participation declined neatly in 2021, until the final shutdown of the program.
| Subject : | : | Oral presentation |
| Topics | : | Legacy: Algorithmic management and labor control |
| Topics | : | Starting: Clients and employers |
| Keywords | : | AI data work ; platform payment systems ; platform management ; Venezuela |
| PDF version | : |  PDF version |
