An exploration of the consequences of the gig economy for autonomy and freedom at work.
Ben Wray is a freelance journalist and researcher based in the Basque Country, Spain. He is co-author of ‘Scotland after Britain: The two souls of independence’ (Verso, 2022).
Cross-posted from Green European Journal
“The gig economy is zero rights and zero protection,” says Shaf Hussain, a food delivery courier in London. “But I don’t think I’ll ever be able to work for someone again.”
“Even though we are being controlled by these companies, I like that small degree of independence you have. I’ve not got any human boss on me.”
Hussain’s thoughts about the gig economy sum up well the contradictory impulses most of the hundreds of gig workers I have spoken to in the past five years have about their job.
On the one hand, gig workers want autonomy. They have typically worked in standard low-paid jobs like those in supermarkets or call centres and have not enjoyed the experience. Hussain came to the gig economy after falling out with his line manager at Waitrose, a British supermarket. He had been on a one-month temporary contract there, so the gig economy was simply a transition from one form of precarious work to another. At least with the latter he could work outside, ride his bike, and not have to deal with an annoying supervisor pestering him all day.
On the other hand, workers who have spent any length of time in the gig economy know that they don’t have real autonomy. Everything about their work is micro-managed, not by a supervisor, but by an algorithm that collects enormous amounts of data on workers and uses that data to maximise efficiency: to serve customers as fast as possible while keeping labour costs as low as possible. Hussain, a trade union leader, knows this only too well, which is why he specifies that he doesn’t have a “human” boss rather than a boss per se.
This tension between what gig workers want and the reality of their working lives is at the heart of a major debate in Europe today, with significant implications for the future of work. Are workers who are algorithmically managed by global corporations like Uber and work “on-demand” – paid per task, rather than per hour – independent contractors, or are they just employees with an algorithmic, rather than a human, boss?
Old wine in new bottles
Algorithms existed long before digital technology. A Persian scientist, Muhammad ibn Mūsā al-Khwārizmī, invented algorithms over 1000 years ago. The word “algorithm” comes from the Latinised version of al-Khwārizmī’s name, “algorithmi”. What al-Khwārizmī did for the first time was to create algebraic rules for solving problems.
That is the same function of modern algorithms in the gig economy – to solve problems such as which courier should be allocated the task of picking up food from which restaurant. The only difference is that the process of algorithmic problem-solving is now computerised and conducted through automated processes. In these, thousands of data calculations that would have taken al-Khwārizmī weeks to carry out can be done almost instantly.
What digitised algorithms do is reduce the friction of space and time when it comes to information. We have all experienced this when using Google Maps to direct us; compared to paper maps, we can now get to where we need to go a lot more efficiently. When algorithms as powerful as those used by Google Maps are applied to managing workers, this has a profound effect on the way work is organised and the power relations between bosses and workers. Digital labour platforms like Uber provide instructions for workers via algorithms, but they also collect mountains of data on those same workers. One study found that Uber collects 80 different data points on each ride-hail driver.
One of the most important data points is time; platforms know to the second how long a worker has taken to execute a task. Being able to time not just how long a worker is at work but also the exact amount of time it takes them to carry out a specific task means it is possible to monetise their work on a pay-pertask basis, rather than per-hour.
Again, pay-per-task is not new. Over 150 years ago, during the white heat of the Industrial Revolution, Karl Marx lamented in his magnum opus Capitalthat “piece wages” – being paid for each piece (of cloth) produced rather than for your time – were “a lever for the lengthening of the working-day, and the lowering of wages”. Marx argued that under a piece wage system, workers believe that if they intensify their labour to produce more pieces in a shorter amount of time, they will earn more money. However, because the boss still decides how much each worker earns per piece, the company can take a greater share of the worker’s increased productivity, forcing the worker to lengthen their working day to earn the same amount.
The same dynamic is at play in the modern gig economy. The commission that platforms take on each task – known as the “take rate” – has risen in recent years as platforms have faced increased pressure from investors to achieve sustainable profits.
The mechanism that big ride-hail and food delivery platforms like Uber and Wolt have used to raise their take rates is known as “dynamic pricing”, whereby pay rates are detached from objective criteria like time and distance travelled. Instead, they are algorithmically determined based on a variety of data inputs unknown to the driver such as real-time supply and demand conditions but also their personal characteristics, including their history of acceptance rates for tasks.
This personalisation of pay has been used to deliver pay cuts by stealth: drivers are not informed that the take rate is going up because there is no objective pay rate per trip. Just like in Marx’s day, this fall in the labour share is forcing drivers to work longer hours to make the same amount of money. “When I go to work, I don’t know how much I’m going to make,” Pascal, a Wolt food delivery courier in the Finnish city of Jyväskylä, said after the platform introduced dynamic pricing. “We don’t even know how they determine the payment amounts. It feels like we have been betrayed.”
Dynamic pricing creates a perverse situation in which driver data is used against drivers. If the company knows that courier X never accepts pay rate offers that are lower than 10 euros per trip, but that courier Y does, it can use this data history to offer courier Y lower rates than courier X for the exact same trip. Drivers have conducted experiments to test this, putting their phones side by side and assessing the trip offers they get. One dynamic pricing experiment found that drivers were offered different rates for the exact same trip 63 per cent of the time. Gig economy professor Veena Dubal has described this practice as “algorithmic wage discrimination”.
Mūsā al-Khwārizmī and Karl Marx would have had no problem recognising how algorithms and piece wages operate in the modern gig economy. They may have been more surprised to learn that this organisation of labour could be presented as a form of freedom from the strictures of the employer-employee relationship.
Is gig work really flexible?
“My son never gets to see me,” says Barbara, a Just Eat food delivery courier in Belfast. “I sometimes work 15 hours a day because the price of everything is going up but our pay is going down. We can’t continue to live like this.” At the heart of the gig economy model is a grand bargain: while gig workers are only paid per task, rather than their whole time at work, they can pick and choose when they want to work. It is this aspect of the gig economy that digital labour platforms promote. It is this aspect of the gig economy that digital labour platforms promote relentlessly to workers, arguing that it gives them the flexibility they want to organise their work around the rest of their lives.
There is little doubt that many gig workers value having flexibility over their work schedule, especially those who do gig work as a side hustle, a way of topping up their income, the majority of which is derived from another source. However, for people like Barbara who fully rely on gig work to survive, the flexibility of the gig economy is largely illusory.
In service-based jobs like food delivery and ride-hail, working time is largely dictated by consumer demand. In both of these jobs, peak demand is on Friday and Saturday nights, when most workers in standard jobs are off work. Couriers and drivers can choose not to work these hours because they want to socialise with their partners and friends, but they will miss out on the strong incentives the algorithm provides.
“Surge pricing” is an algorithmically determined pay methodology that responds to times of high customer demand by raising driver pay, sometimes by up to three or four times the normal amount. Surge pricing typically happens around big concerts or football matches, but it can also happen during heatwaves and torrential rain: whenever customer demand spikes. Pictures of food delivery couriers slogging through waist-high water have gone viral on social media. Others have suffered from sunstroke and dehydration while working through heatwaves.
“They spend 12 hours a day in a row on the street without having access to a bathroom or a chair because the platform rules,” Fernando García, a former food delivery courier in Madrid who now works as an organiser for the UGT union, says during one heatwave. “The algorithm does not understand labour rights.” But as gig workers are not paid for waiting time, deciding to work during the hours when demand is low can mean not getting paid at all: there are not many customers for take-away at 8 in the morning.
There is evidence that a demand-led working life hurts female workers, because they tend to have more caring responsibilities at home and therefore less ability to work during times of peak demand. As a result, in a sector like food delivery, they earn less per task than men on average, despite doing the exact same work.
In the “cloud work” gig economy – working on your computer from home to perform tasks like transcribing an article or coding a website – researchers have found that many women leave standard jobs in search of more flexibility so they can combine work with caring responsibilities. However, once on cloud work platforms like freelancer.com and UpWork, these women find clients to be extremely demanding, requiring them to respond to messages straight away and work anti-social hours to complete tasks as soon as possible, undermining the balance they were seeking.
The reality is that there is far greater flexibility for platforms in the gig economy than there is for workers. It is the platform that has the flexibility to not pay workers when customer demand is low or when the worker is sick or on holiday. This points to the need for a more holistic concept of what we mean by a flexible working life, one which goes beyond simply the hours of the day we choose to work. Flexibility must entail rights as workers, including some degree of income security and workplace power. Otherwise, it is just a pretty word for precarity.
The EU Platform Work Directive
Finally passed in December 2024 after years of wrangling between and within the EU institutions, the Platform Work Directive (PWD) is an attempt to provide a unified answer in all 27 member states to the question of employment classification in the gig economy. “A young person riding a bike with an app or a mobile device is not an entrepreneur,” Yolanda Díaz, Spain’s labour minister, was fond of saying during the three years in which the EU debated the directive.
Spain, the first country to establish a law that considered all app-based food delivery couriers to be employees (the “Riders’ Law”), stood at the head of a group of about seven member states that supported strong employment rights in the platform economy. A similar-sized group led by France, and supported by far right-led governments in Italy and Hungary, opposed employment classification, arguing that it would undermine jobs and tech innovation. A slightly bigger group sat between the two, being tugged one way and the other, while Germany refused point blank to take any position at all.
The end result was a compromise PWD that no one was entirely happy with and that failed to achieve the original aim of creating a unified set of standards across the EU. The law does establish a legal presumption of employment in the platform economy, but it is up to each member state to decide on the criteria and implementation rules for how this is applied, meaning it’s highly likely that, on either side of the border between social-democratic Spain and ultra-liberal France, there will be highly different applications of PWD.
Uber responded that the compromise directive would “maintain the status quo”, indicating that they will continue to fight member state by member state to resist employing their workers. This makes it very likely that we will see platforms and unions, which have been at the forefront of the campaign for employment rights despite levels of union membership still being low in the gig economy, battling it out in courtrooms across Europe to establish legal precedents over the application of the PWD. With member states required to transpose the directive into national law by December 2026, the debate on employment classification in Europe’s gig economy is far from over.
AI and gigification
The outcome of this battle will have implications well beyond the workers who currently make up the gig economy. (The most high-quality survey evidence has found this to be between about 5 and 15 million workers in the EU.) The number of workers who could be subject to gigification is likely to be far greater than this. One study has found that 20 per cent of workers in Germany and 35 per cent in Spain are subject to at least one form of algorithmic management. The OECD has found that 79 per cent of companies in Europe now use algorithmic management. While not all of these workers will be “Uberised”, the same technologies ultimately come with the same risks to workers’ rights and protections.
And technological change is not slowing down. The rapid roll-out of generative AI technologies opens up the prospect that anyone who creates any form of content today could join the rapidly expanding “data annotation” workforce tomorrow. Data annotators are hired on a gig work basis to test, train, and fix AI systems on platforms like Amazon Mechanical Turk. They have, until recently, been typically low-skill jobs that can be done by almost anyone. But as generative AI becomes more complex, it increasingly needs enhanced forms of data annotation, which require specialised expertise from specific sectors.
One data annotation company, Surge AI, promotes on its website a mathematics reasoning project it completed for Open AI, the company that created ChatGPT. Data annotators it hired for the project include Joe, who has a mechanical engineering degree and 25 years of experience developing software, and Maria, who has a postgraduate degree in biochemistry. If it becomes more efficient for an AI system to develop content in professions like mechanical engineering and biochemistry and for humans to test and finesse AI, all these jobs could potentially be organised via algorithms on a piece-rate basis and thus gigified.
The current battle over PWD is an early skirmish in a much bigger conflict over the future of work. It is not for nothing that Sam Altman, CEO of OpenAI, has said that “the social contract” is up for “some degree of debate and reconfiguration”. What the social contract will look like in the future will to a large degree be determined by what happens to the gig workers who are at the forefront of technological change today.
Be the first to comment