On the Disassembly Line

Many jobs​ involve work we would prefer not to think about. When you recycle your rubbish, you know that someone somewhere will have to sort through it. When you eat meat, you know that someone did the killing. We overlook much of the work done to meet our needs, even when it’s closer to home: cleaning the streets, unblocking the sewers, digging graves. Someone always has to do the dirty work.

The internet was supposed to be different. The new technologies central to contemporary capitalism offer the possibility of improving our working lives, even if they do so partly by eliminating our jobs. But badly paid and repetitive work has not diminished, it has proliferated. Machine learning, it transpires, depends on humans doing boring and unpleasant things – above all, cleaning and annotating data. Twenty million people around the world earn a living performing tasks on ‘microwork’ sites such as Amazon Mechanical Turk, Appen, Scale and Clickworker. ‘Microwork’ refers to any small task carried out on the virtual assembly line that an algorithm can’t reliably perform. Microtaskers work as data taggers for companies such as Google, Amazon and Tesla, adding, fixing or removing labels by identifying apples, cars, houses, traffic lights and so on across digital media. This is mind-numbing work. It can also be worse than that: Facebook needs a small army of people to remove images of violence and child pornography.

‘Tech work’ still connotes well-paid Silicon Valley computer engineers and entrepreneurs. But the low-wage workers who keep much of the internet going live in India, Uganda, Palestine, Venezuela, the Dadaab refugee camp in Kenya and the Shatila refugee camp in Lebanon. In Work without the Worker, Phil Jones calls this the ‘hidden abode of automation’, making an analogy with Marx’s factory. According to his account, the big contractors, as well as smaller companies and individual researchers, act as ‘requesters’, soliciting ‘taskers’ who don’t need much more than an internet connection to do their work. They are paid by the task, often at a few cents per survey filled or piece of data tagged. The average wage is thought to be less than $2 an hour. If tasks are completed imperfectly or too slowly, the microworker loses out – Jones claims that up to 30 per cent of them ‘regularly go unpaid’. Sometimes they are paid in ‘points’ or gift cards, which they can only use on specific websites or in particular locations, thus undermining the principle of compensating labour with wages. In countries including Botswana, Qatar and South Africa, Amazon acts as what Jones calls a ‘digital company town’, paying workers in tokens that must be spent on its goods and services.

Microwork is the latest proof that technological development doesn’t end work, but only produces new forms of labour – and new ways of concealing it. Facebook’s user-friendly façade masks the poorly paid and precarious work that goes on behind the scenes, replicating, rather than departing from, institutions in the physical world like the care home, the garment factory, the abattoir, the prison. In Dirty Work, his account of ‘low-status jobs of last resort’, Eyal Press interviews Harriet Krzykowski, a psychology graduate who needed a job so she could fund her postgraduate degree in counselling. The only steady work she could find (this was in 2008) was with the Florida Department of Corrections, providing mental health support to the third largest prison population in the US. In recent decades, the number of incarcerated people with mental health needs in Florida’s prisons has grown by 153 per cent. The number of people employed to support them has not grown at the same rate, and Krzykowski had few illusions about the work she would be doing. But she wasn’t prepared for the brutality. At Dade Correctional Facility, inmates were denied food and beaten by guards, rape victims were taunted, wheelchair users pushed over and abused. Mentally ill inmates were routinely scalded in boiling hot showers. She tried to report what she saw, but the guards retaliated: they abandoned protocol, leaving her alone to manage violent situations. She wanted to speak out about what was happening or to leave. But she needed the job and feared for her own safety.

It’s probably easier to feel sympathy for the microworkers of the Global South than for Florida’s prison workers. For Press, that lack of sympathy is a function of our ethical relationship to what he calls ‘dirty work’: work that is morally compromised, as he sees it, yet necessary to the maintenance of the status quo. As well as prison psychologists, he interviews drone operators, slaughterhouse workers and engineers in the fossil fuel industry. These are jobs that sustain the political and social order in the US, but are pushed to the margins. Such work exists in all neoliberal economies; the tendency is to ignore it.

This argument holds up better for some of Press’s examples than others. Take slaughterhouse work. By the early 1990s, a quarter of America’s meatpackers were undocumented immigrants. In the slaughterhouse hub of Texas, a large proportion of the state’s 1.6 million undocumented people have worked, or still work, on the ‘disassembly line’, earning between $11 and $13 an hour. Hanger workers are responsible for hoisting livestock out of crates and shackling their feet to the conveyor belt (at a rate of 65 chickens a minute for poultry workers). Line workers separate the carcasses. Both types of work exact a physical and psychic toll, what Jonathan Cobb and Richard Sennett call ‘the hidden injuries of class’. On the kill floors, most workers are ‘at-will’ employees and can be fired at any time. In 2019, the annual turnover in many Texan slaughterhouses was 100 per cent.

The ethnographer Timothy Pachirat describes industrial slaughterhouses – along with prisons, hospitals, nursing homes and refugee camps – as ‘zones of confinement’ (the phrase comes from Zygmunt Bauman). Workers at such places are often segregated and isolated, their work hidden from society at large. This invisibility isn’t an accident. Industry lobbyists have shrouded slaughterhouses in secrecy with ‘ag-gag laws’, which in many states prohibit or severely restrict photography or recording on site. Although the industry has come under greater scrutiny following campaigns by animal rights groups, the focus has been on animal welfare rather than labour practices. People who buy organic meat aren’t paying over the odds to improve the lot of workers on the disassembly line. Researching her book Labour and the Locavore (2013), the political scientist Margaret Gray asked a butcher in the Hudson Valley why customers weren’t more concerned about the working conditions on the farms. ‘They don’t eat the workers,’ he replied.

Press​ doesn’t deny that the workers in his study cause harm just by doing their jobs. Some do this kind of work because they have few other options. All of them operate under some sort of mandate. The data analysts who launch drones for the state are public servants, as are the officers in public prisons. Other forms of morally suspect work are mandated by consumer choice: we want to eat meat, so chickens must be killed. These workers get their hands dirty so the rest of us don’t have to.

In The Civilising Process (1939), the sociologist Norbert Elias described the acts of ‘concealment’, ‘segregation’ and ‘removing out of sight’ as crucial to modern social organisation. They protect the consumer or citizen from understanding the true cost of their choices and enable the outsourcing of morally injurious work. For Press, this makes the rest of us complicit. His argument makes sense for some forms of work. If you eat meat, you owe something to the slaughterhouse worker. But if your government embarks on war, do you owe anything to the drone operator? At its most extreme, this way of thinking risks dispersing responsibility and diluting agency. We are all responsible – therefore none of us is. Decisions made by individual consumers are a poor basis for politics. It might be better to say, as Iris Marion Young argued, that political responsibility, unlike blame or liability, is borne both by those who perform an injustice – those who do the dirty work or exploit workers – and by those who participate in the social processes that allow the injustice to take place. It is not enough to acknowledge our complicity; we should participate in changing those processes.

It may be that we owe it to these workers to abolish their work – making it visible isn’t enough. The first step is for workers to try to change their conditions, but for those in ‘dirty’ sectors, means of resistance are limited: filling in forms (tedious additional labour), whistleblowing (risky) and exit (most probable). American slaughterhouse workers, who perform essential jobs that few want to do, are more likely to take collective action than drone operators. Meatpacking was a steady job until the industry shut down unionised slaughterhouses, rebuilt them in regions with weak unions, and employed an immigrant workforce less likely to engage in collective action. Though there have been some recent gains, most workers in the industry don’t belong to a union and the associations that organise undocumented workers have as yet little power to improve their conditions. Prison workers, too, face retaliation if they resist abuses, but at least they are often unionised: ‘protective service occupations’ (cops, firefighters, prison officers) have the highest unionisation rates among public sector workers in the US. In America’s enormous private prison industry, however, and in the 28 so-called ‘right-to-work’ states, unionisation is all but impossible. (Press doesn’t make the point, but it is in any case hard to imagine those employed by repressive arms of the state, even as psychologists and doctors, proving to be allies in transforming the structure and status of dirty work.)

Microwork is hidden and fragmented in a way that makes collective action particularly difficult. Like other employers in the gig economy, those who employ microworkers have gone to great lengths to avoid classifying their workers as employees and to prevent them acting in concert. Jones describes the result as a new form of ‘economic blindness’. The lack of a workplace or colleagues, and the absence of laws and protocols governing terms of employment, leave workers vulnerable. The purpose of the individual tasks they perform is often unclear (‘why am I labelling an apple?’), let alone any sense of their amounting to a meaningful whole.

This has real benefits for certain industries. When the US military uses taskers, or Google hires them for a Defence Department initiative, the workers themselves don’t know that what they are doing – tagging features of a territory, say – is for the purposes of war. (Some of this work is being done unknowingly by refugees created by America’s wars.) The delegation of tasks by software – usually referred to as ‘vendor management systems’ – makes these chains of outsourcing even harder to untangle. Alongside the non-disclosure agreements taskers are sometimes forced to sign, these allow companies to keep their employment quiet. The ideology of automation as liberation has masked the fact that the tech industry is just as exploitative as any other.

Press entertains and rejects the notion that forms of tech work might count as dirty work: he sees it as elite employment that doesn’t involve the same kind of moral injury. But it is badly paid, insecure and just as dependent on concealment. Indeed, often it is more so, since the nature of the work can be elusive to the workers themselves. Some microwork, especially in China, takes place on data farms, but taskers are usually geographically dispersed and have little, if any, communication with one another. They have, despite this, found ways to challenge their conditions: workers for Amazon Mechanical Turk, for instance, have developed a piece of software called Turkopticon, which piggybacks on the Amazon service to warn others away from abusive contractors. But these are technological fixes that moderate the behaviour of requesters, not strategies for wholesale change. Jones calls microwork the ‘apex of neoliberal fantasy’ – it represents an economic system without unions, work culture or institutions, in which the wage contract is broken, workers have no specialist knowledge and no opportunity for advancement.

New forms of digital organisation and militancy are developing, but may struggle to keep up with the proliferation of jobs that could be classed as ‘subemployment’. The number of informal workers whose labour has an unstable relation to wages is increasing in all parts of the world, while the dirty work that used to be synonymous with stable employment is subject to the same dynamic of immiseration. Press describes these jobs as ‘necessary’, but they aren’t in any real sense of the word. They are symptoms of social dysfunction, and we would do better to take their prevalence not as evidence of moral complicity but of social decay.