Gig Workers Are Demanding To See the Algorithms That Manage Them

Gig workers are not objecting to being managed by algorithms, per se. 

But they understand that unless they have access to the data and algorithms by which they are evaluated and compensated, they will always be at a disadvantage. JL 

Karen Gregory reports in Wired:

Gig workers have been fighting for formal recognition of employment (which would allow  sick pay, holiday pay, pension benefits, and the right to unionize), basic safety standards, wage increases, and stable scheduling, as well as an end to unfair dismissals from platforms. (It) is a demand for transparency for platforms to offer insights into how they collect and analyze data. (They want) to understand how their labor generates value for the company, a demand to be recognized and compensated. (Transparent) data enables workers to challenge automated decision processes such as unfair dismissals and bias

AROUND THE GLOBE, gig workers are waging some of the most visible and vocal campaigns for worker rights. Across multiple platforms and countries, gig workers have been fighting for formal recognition of employment (which would allow access to benefits such as sick pay, holiday pay, pension benefits, and the right to unionize), basic safety standards, wage increases, and stable scheduling, as well as an end to processes that lead to unfair blocks and dismissals from platforms. At the heart of many of these campaigns is a demand for transparency and for platforms to offer reliable and meaningful insights into how they collect and analyze data. Gig workers are asking to be shown the algorithms that define, manage, and control the nature of the on-demand work they do.

Interest in worker data and inquiries into the “black box of the platform” stem from two key issues. First, gig and platform workers know they generate vast amounts of valuable data. Platforms engage in what has been called “dual value production,” where any profit the company makes through its service is “augmented by the use and speculative value of the data” produced before, during, and after. In effect, by demanding to be shown the algorithmic processes that shape their work experiences, gig workers are asking to understand how their labor generates value for the company. This is a demand to be recognized and compensated.

However, research with gig workers has shown that their interest in their algorithmic bosses is more nuanced than a simple desire for higher wages. Absent employment status, gig work is a form of self-employment, and workers should enjoy autonomy, flexibility, and choice about when and how to work, as well as have clear information about how to stay safe while working and how to mitigate the risks associated with self-employment. 

Currently, gig workers do not enjoy these benefits. Rather, gig and platform work are forms of risky work, where working individuals themselves must absorb the myriad financial, physical, and emotional costs of doing work. In response to these risks, workers are arguing that access to platform data and clearer explanations about how their data is gathered and analyzed by the platform can help them make better-informed choices about when and how to work. Worker interest in platform data is fundamentally driven by the immediate need to make gig work livable and safe.

While regulation of the platform economy and strong employment rights are fundamentally necessary in the long term, gig workers have been clear that they also need information about their working conditions to be more readily available. They are teaching us that the legal fight for strong labor protections is also a fight for worker data rights. Yet, for workers, demands for algorithmic transparency and accountability raise as many challenges as they do opportunities.

DEMANDS FOR DATA immediately reveal the power imbalance in the platform economy. Data, as it is conceived of now, simply flows away from workers and to the platform, where it becomes proprietary, valuable, and “big.” While platforms enjoy the advantages of gathering and analyzing big data, current data protection laws function at a “smaller” scale and are based around individual rights. 

Under GDPR and the UK’s Data Protection Act of 2018 (although the latter is under government review and consultation), individual workers do have the right to request their personal data from platforms and the right to an explanation of how their data is involved in automated decisionmaking. However, while the process of requesting personal data is relatively simple for individuals, meaningfully aggregating and analyzing that data requires resources and skills. It also speaks to issues about the long-term maintenance of worker data, as aggregating worker data raises the question of where that data should be stored, how it will be secured and maintained, and who will have access to it in the end. Furthermore, to generate a useful and robust database of worker data, workers must encourage one another to make access requests and contribute their data to a collective project. That project can be onerous, incomplete, and ineffective.

Still, as Uber drivers in London have shown, GDPR does provide powerful rights for workers to exercise. With the support of Worker Info Exchange, Uber drivers in London are not only requesting but pooling their data into a worker-owned data trust, which in turn allows workers to ask and answer their own questions about working conditions—questions that can be particularly valuable when they are about the number of hours worked or when attempting to calculate wages over time. (For example, such data would enable workers to determine whether they earn a minimum wage.) This collectivized data has also enabled workers to challenge automated decision processes such as unfair terminations and dismissals, and to raise the alarm on issues of bias in the deployment of facial recognition technologies. Thus far, the gig economy has operated as an unregulated testing ground for managerial and logistics data science, but the challenges posed by automated decision processes are by no means limited to the gig economy. The data harms currently faced by Uber drivers should be seen as harbingers for workers more broadly.

AS GIG WORKERS raise the need for more robust data rights and draw attention to data harms, a range of tools and apps designed to offer insight into the algorithmic functioning of platforms has also emerged. The options that exist can be powerfully combined to generate what has been called a form of “digital worker inquiry.” A recent conference at the University of Edinburgh, which I organized, brought several of these projects together to explore the possibilities and challenges of these tools. Drawing inspiration from predecessors like the Turkopticon browser extension, which allows crowdworkers to share and access reviews of employers that use the Amazon Turk platform, developers have built apps to track working time, identify and combat wage theft, track underpayment, track living wages, harvest and port data, illustrate and visualize working conditions, and build solidarity and organize. These tools support gig, platform, and precarious workers by offering data-driven and measurable insights into working conditions.

 

For example, We Clock, which is free and open source, helps workers track their working time and quantify their workday. This can be used to understand how many working hours are unpaid—a key concern for workers who are paid “per gig” but can spend hours a day waiting for work. Projects such as RooParse use the PDF invoices that Deliveroo couriers receive in their email to extract and aggregate weekly earnings, which can help workers understand how their wages rise and fall over time. Deliveroo Unwrapped reveals hourly pay and can show that riders earn much less than minimum wage. 

Beyond these inquiries, projects such as Contrate Quem Luta (“hire who struggles”) allow marginalized and homeless workers to circumvent platforms altogether and more directly access gig work through a WhatsApp chat bot. Up and Go draws workers directly into questions about who should own this technology and its attendant data. Broadly, all of these projects start conversations between workers, and they can be powerful mechanisms for drawing media attention to workers’ concerns.

However, these projects also raise serious questions about best ethical and technical practices for building and maintaining worker data; about the kinds of collaborations and funding necessary to conduct this form of worker data science; and about the distributions of power among workers, researchers, and organizers. While workers may want to request and build with their data, many will lack the technical skills and financial resources to actually create a tool or app. This means that workers will need ethical collaborators who are willing to invest in the full scope of a data maintenance project. (Here, universities and researchers should play a stronger role, along with trade unions, to support worker-led projects and help workers steward their data and establish ethical and secure data practices. A good example of this is happening at Northwestern’s Civic AI Lab under the direction of Saiph Savage.)

Still, even projects that draw workers into their development and are open source and built to protect privacy raise questions about relying on technological solutions in lieu of organizing, or of what researcher Danny Spitzberg has called “solidarity as a service.” These projects risk replicating power imbalances already entrenched in the gig economy, a result of a platform or service being accountable to investors and not workers, as in a democratic union or cooperative. As a result, for some workers, new spaces such as worker-led observatories are needed so that workers themselves remain in control of the inquiry and data-gathering process.

WHAT IS AT stake in the process of building with worker data transcends the ultimate use of an app or a tool. As James Farrar has suggested, data rights, data-driven projects, and data trusts should be seen as imperfect tools that workers take up in the process of raising awareness and reforming and regulating working conditions on platforms. Fundamentally, these tools should be used in the service of organizing and building worker leverage. They cannot, however, replace the labor necessary to organize.

Still, while apps and tools cannot provide a quick technological fix, they can be used to add measurements and evidence to worker demands, and these projects can be starting points for essential conversations in and across unions. As Roz Foyer, the general secretary for the Scottish Trade Union Congress, has noted, unions have long been data-driven institutions, but if they are to “fight fire with fire” in the digital economy, they will have to grapple with the complexities of worker data via a renewed capacity for research.

For Christina Colclough, the founder of the Why Not Lab, unions must specifically build capacity to understand the “ins and outs of data and algorithms” and develop their own teams of data analysts. As Colclough has argued, trade unions have a fundamental role to play in protecting workers’ collective digital rights. While digital inquiry tools may offer new forms of data, it is essential that these projects help build union strength, rather than fracture or privatize worker interests. Any long-term change that might be made possible through these tools will come through drawing unions into larger political conversations about data governance.

Unions will need to do the work of connecting the dots between the challenges workers currently face, the future of work, and the central role that data and data rights will play. Some unions, such as Prospect, are putting resources into this area and engaging in what Lina Dencik has called “data justice unionism,”—a “form of social justice unionism that engages with data-centric technologies as firmly situated within a workers’ rights agenda.” While worker inquiry apps and tools cannot immediately give rise to a data justice agenda, they do offer tangible case studies capable of bringing workers, organizers, unions, and researchers together to develop the field of worker data science. This field is the future of work.