Digital economic justice and worker-driven futures
This piece was commissioned by JRF as part of their work on Social Justice in a Digital Age. It responds to James Plunkett’s essay, The Invidious Hand.
Digitally-mediated and managed work might look unfamiliar, with its new terminology and ‘disruptor’ businesses, but it is actually the latest frontier for the replication of age-old inequities. Rather than being dazzled and confused by the unfamiliar, we need to recognise the replication of existing problems, and highlight what is new: how digital spaces obscure and make it harder to address those inequities. We must also seize the opportunity to innovate, to take the technologies which underpin the problems and use them for new, radical approaches to worker rights and power.
To imagine a sharp boundary between the digital and physical worlds obscures the reality of people’s working and social lives. Every day we see real-world prejudices, inequalities and long-standing unequal power dynamics being replicated in digital spaces. Facial recognition software tends to misidentify people of colour more than white faces. Women are subject to torrents of online abuse and harassment. People in poverty are locked out of opportunities and support because they can’t afford good quality, reliable digital access.
These are not entirely new, digital problems, but extensions of generations-old offline struggles.
We also see long-standing problems in our working lives taking root in digital spaces: low paid workers lack rights, those in charge of workplaces have disproportionate access to power and information, workers’ autonomy is in short supply. The settings may be unfamiliar, and in some cases may exist outside the reach of our existing laws and norms, but the challenges are rooted in familiar dynamics.
There may now be digital means of assigning, monitoring, and paying for work, but we must not overlook the physical reality of people doing these jobs in the real world. Being part of creating and maintaining a digital space does not guarantee good quality employment; many of those whose labour is essential to the functioning of the most ostensibly high tech and automated operations are well aware of their downsides.
Despite appearances, spaces like Facebook do not automatically review and moderate content;
algorithms designed to control self-driving vehicles need to be taught by humans to recognise cars, people, street signs and trees.
This work, intrinsic to the creation and maintenance of many digital spaces, is often done by low-paid, insecurely employed workers, as first brought to our attention by Mary L. Gray and Siddharth Suri in Ghost Work.
As Phil Jones notes in Work Without the Worker, “this automated dreamworld is more fantasy than reality… Behind the cargo cult rituals of Silicon Valley is the gruelling labour of sifting hate speech, annotating images and showing algorithms how to spot a cat”. Jones describes ‘microworkers’, informally employed, usually living in South America, East Asia or the Indian subcontinent, in poverty, refugee camps or countries in dire economic straits. Without them, many of the big tech giants would grind to a halt, yet they are not only undervalued economically, they are actively hidden from view.
Plunkett is right to identify in The Invidious Hand that definitions of employment and self-employment and associated rights and laws have not kept up with the new situations of those who work through and for digital spaces. Our institutions are trying to respond, for example the legal case in the UK to reclassify Uber drivers as workers rather than self-employed contractors, and a new proposal from the US Department of Labor to avoid the ‘misclassification’ of gig workers.
But catching up is not as good as anticipating and avoiding problems before they take root, or, even better, building entirely new ways of organising work that shift the dynamic.
What then can we do? I believe it needs a mix of adaptation, evolution and invention. Digital spaces do not have to be disempowering for workers; if we can find ways to undo the replication of existing inequities we have a chance to make change.
In some cases existing institutions need to work hard to adapt. We probably don’t need to rebuild employment law from the ground up, but as with the Uber case, try to ensure our existing laws and regulations can adapt to better reflect and address new challenges as they emerge.
In other cases, we do need new tools, institutions and rights, and some of these are already emerging. They may take their cue from digital tools, repurposing them and their approaches to rebalance or shift power towards workers; using the exact techniques that replicate inequities to undo them. They may be evolutions of existing things like unions or cooperatives, designed to proactively take control.
In Scotland, the Edinburgh Workers’ Observatory is encouraging gig workers to take control over their own data. Rather than be at the mercy of opaque algorithms that decide which jobs they get and how much they get paid, delivery riders are collating their own data about their working patterns to understand more about how decisions are made that impact their work and income.
In the US, driver-owned Drivers’ Seat allows Uber, Lyft and other platform drivers to see when and where they are likely to earn most; information that would normally only be held by the platforms. As Dr Karen Gregory describes, many more worker-driven and owned tools are springing up globally.
These examples demonstrate a significant shift: in both ownership of data, and the power and insight to be gained from it.
Rather than passively having their data extracted from them by platforms, these workers are actively generating and retaining their own data, and creating ways to derive value from it.
This is not easy however. The rights used by gig workers to obtain their data are individual, and have to be known and understood in order to be used. There is the additional challenge of collating and analysing data from multiple individual requests, and turning it into something useable, which requires time and expertise not available to many. A collective right to request data from platforms could enable workers to speed up and simplify this, but unions or other worker collectives must take care not to replicate the same surveillance and privacy-invasion as the platforms in their efforts to assert their rights.
Microworkers on the Amazon Mechanical Turk (AMT) platform, a low-paid, ‘on-demand’ workforce, have minimal protections or rights. Clients who decide not to pay for work are difficult to challenge and can leave negative reviews of workers without clear reasons, affecting workers’ ability to secure more tasks. Worker-led non-profit Turkopticon, set up by two students who recognised the power imbalance between those getting tasks done for them and those doing the tasks, allows AMT workers to share their ratings of the clients they work for, building collective knowledge and agency.
Ultimately, while these efforts improve transparency and workers’ power to negotiate and exercise choice, they might not fundamentally shift the underlying structures of how work is distributed, monitored and paid for through digital spaces. For this, we need to look to models of work that are actually owned and steered by workers, where they set the terms and conditions, ensure that the use of data is not extractive, that workers are not surveilled for hostile or invasive reasons, and that the value created is shared fairly by all involved.
Platform cooperatives, in which a digital business is owned and run by its workers, are one such model that offers a different vision of the future of work.
Of course, cooperatives are not new; rather, this is an evolution of an existing model, adapting to the digital environment. There are high barriers to entry for new platform coops however, as UK cooperatives are hampered by complex rules, more so than a for-profit or shareholder owned business (see Autonomy, 2021, Delivering Rights: Alternatives in the Online Food Economy). This restricts the growth of platform coops, but could be remedied through pragmatic policy change to simplify requirements around set-up and management.
As well as calling for these simplified rules, think tank Autonomy also argues for municipal spaces for the wellbeing of delivery riders, incorporating facilities for rest, storage and meetings. The need for physical space to improve delivery workers’ conditions once again highlights the porous edges of the digital and the physical spaces of work, and the need for solutions that consider both.
The other critical aspect to consider in a fairer future of work is what regime those out of work encounter.
What should a welfare state look like that supports those who cannot participate in paid work?
We can see existing inequalities and power imbalances play out in the digitisation of the welfare state just as we see them in digitised labour. Automated social security systems replicate existing biases: in the Netherlands, the SyRI system disproportionately targeted low income communities and breached citizens’ right to privacy. Mistakes are made by digitised services: the Australian ‘robodebt’ scandal wrongly pursued thousands of benefit recipients for inaccurate debts, leading to financial hardship and significant psychological impacts.
These mistakes are hard to unpick and correct because automated welfare systems are typically so opaque and secretive it is hard to see where mistakes arise, and because they are sold as efficient, accurate and foolproof, and are deliberately difficult to build a credible case against. The ‘pervasive rhetoric of objectivity’ described by Amironesei et al (see Amironesei, Denton, Hanna, Nicole and Smart, 2021, The case for interpretive techniques in machine learning, in Fake AI, ed. Frederike Kaltheuner) puts the onus on the individual suffering the negative impacts of flawed systems to prove their case, rather than on their creators or deployers to prove their efficacy.
Social security is an institution unlikely to be entirely scrapped and rebuilt: even the radical and lengthy move to Universal Credit builds on existing policy approaches and does not replace all ‘legacy’ benefits for all people. What we should be aiming for as its next evolution remains very much up for grabs.
Adopting key principles of a Universal Basic Income such as removing conditionality would significantly change the relationship between claimants and the state, away from the current mode of suspicion and blame. A UBI-like model could also mitigate some of the risks of increasing automation, as it would reduce the need for surveillance. The Danish model of Flexsecurity balances a highly flexibilised labour market with far more generous entitlements than UC. It may be well suited to the precarious nature of much digitally-mediated work and would provide a far stronger safety net than UC. This will be a site of long-term struggle to rectify the risks and harms already embedded in our social security system, let alone new ones introduced by increasing automation.
Platform coops demonstrate the potential for workers to be in the driving seat of radical change. Other approaches may rely on a wider partnership approach, bringing local authorities, community organisations and civic partners together. Some change will come from lobbying, campaigning, perhaps taking targeted legal action, or investigative journalism.
Many acts of resistance and reinvention are already underway at a small scale;
ideas that are close enough to our current reality to be relevant and achievable, but also thinking beyond the restrictions and inequalities so many work under.
This is a challenge we need to work on together, to resist the individualisation of work, welfare and poverty: to say, we are not all individually responsible for our fates. By working together, using our collective imagination and innovative approaches to creating and repurposing technology, we can create and influence a new future of work.