Estelle Becker Costanzo has worked in Pittsburgh restaurants since age 15. Now 56, she is a server at The Capital Grille, a position she is proud of. “This is a good job,” she says—relative to the rest of the industry. Still, because her base wage has stayed put at $2.83 per hour for 25 years, she struggles to cover her basic expenses. “Originally, [tips] were supposed to be 50 percent of our income. Now it’s more like 100 percent.”
As tips became her primary income, the demands on that income increased. As is standard in the industry, The Capital Grille charges Costanzo a fixed percentage of every bill as a tip for bussers and bartenders. The charge assumes she receives a 20-percent tip. This was workable when large parties were charged a tip by default—a policy known by the awkward Americanism “automatic gratuity” or “autograt.” Recently, however, The Capital Grille ended autograt, leaving Costanzo’s income up to her customers’ whims. When large parties reserve private rooms, they’ll often tip only a fraction of what she needs to cover her obligations. As a result, she might work for hours only to pay for the privilege.
For Costanzo, the loss of autograt is just the latest of a series of attacks on her security. At this rate, even if autograt is reinstated, she doesn’t see an end to struggle. “Retirement? I’ll be working ’til I’m 80,” she says. “We don’t really think about the future until it finally comes up on us.”
Toronto’s Kristy Milland thinks a lot about the future. Like Costanzo, Milland’s income long depended on how much the people she served decided to pay. But she didn’t work in restaurants or get paid in tips. In fact, for years, she never even met a single one of the many customers she served. Milland is one among the possibly half-million people who currently work for Amazon’s Mechanical Turk (“MTurk”), one of the largest employers in the gig economy of network platforms, like Uber or Handy, that link workers to employers one job at a time.
“Turkers,” as they call themselves, connect with “requesters,” employers who advertise fixed payment in exchange for a single task. Amazon touts Turkers as “artificial artificial intelligence” that can skillfully perform jobs that computers can’t do well: transcribe audio, categorize images, or serve as subjects in academic experiments. Pay ranges from pennies to dollars for tasks that require seconds to hours. Though wages are tiny, competition can be fierce. “Sometimes I’d get a text in the middle of the night and get out of bed to start,” Milland says. To put together a reasonable day’s pay, she would sometimes find herself working 17 hours at a stretch. Other times she might go a week without working, but that week was no vacation. “That’s still 17 hours a day looking,” she says.
Even when work was steady, Milland couldn’t be sure she’d get paid. Requesters can reject work with no explanation—and minimal consequences: Another Turker will take the returned task, often in seconds.
What links gig workers like Milland with tipped employees like Costanzo is that both workers’ situations can fairly be called precarious—their income, employment, and ability to meet basic needs are insecure and, increasingly, unprotected. This lack of security is self-perpetuating. Workers laser-focused on finding—and keeping—work can be unwilling to risk attempts to organize for better conditions.
"We don't really think about the future until it finally comes up on us."
Rochelle LaPlante, a moderator for Turker discussion forum “MTurk Crowd,” has publicly criticized the platform and faced recrimination—from other Turkers. She says that even some members of her own forum would prefer she stay silent. Privately, they tell LaPlante that they worry requesters might leave the platform if its poor treatment of workers is exposed. Others fear more direct reprisal: LaPlante says at least one person’s account was permanently suspended after they spoke publicly about worker conditions. That sort of retaliation is explicitly outlawed against employees, but because the work that Turkers perform—and from which Amazon and requesters derive revenue—is considered independent contracting, the company doesn’t have to comply with those labor laws.
Milland believes that with the advent of these platforms, precarity comes for us all in the end. “Teachers, doctors, lawyers, accountants, programmers, designers, authors, journalists—we’ll all be going to some platform every morning and searching for work 17 hours a day,” she says. She has lived the future that Costanzo sees coming. “We are working in the labor world of the future, and we’re coming back with bad news.”
To the platform owners, precarity is central to the future they want to create. In an open letter to the European Union, a consortium of 47 platform corporations argued against regulations for their industry. They described themselves as “innovators” who are “remodelling whole value chains.” To the Millands of the world, they promise “new sources of income, micro-entrepreneurship, and flexible working.” In other words: more and deeper precarity.
Corporations have long tried to create precariousness. What’s different about the current moment is that the technologies that undermine old kinds of employment also enable new kinds of empowerment.
“Part of what we’re doing is expanding the sense of what’s possible,” says Michelle Miller, co-founder of CoWorker, a platform for workers to build labor rights campaigns. CoWorker’s users tend to be employees of large corporations or chains who are, like Turkers, physically isolated from both their employers and each other. The users tend to be non-unionized. But even when workers have unions, their unions can’t fight every battle. They must allocate limited resources toward campaigns that have a high chance of success. This means that some of the issues most important to workers are left unaddressed.
CoWorker started from a simple idea: Let the workers lead. Its tools help people like Costanzo start their own workers’ rights campaigns and build online networks to overcome isolation. Among the thousands who have pledged support for Costanzo’s campaign are dozens of employees in the same restaurant group, from cities around the nation.
Miller thinks campaigns like Costanzo’s, which use these new digital networks to regain lost protections, are absolutely critical. “We’re in a place where we have some rights reclaiming to do,” she says. But the fact that these campaigns are needed also reflects how far we have yet to go. “Our culture and our economy have reinforced for many years [that] when you go to work, you take what you get,” she says. “You’re ‘lucky to have a job.’” When struggle is needed just to maintain the status quo, there’s little hope for radically improving working conditions.
Costanzo agrees. She knows her campaign is important, but she sees the limits of the approach. “I’m not sure what recourse we have at this point,” she says. “More media attention? What happens is we do this thing, and then it just sits stagnant. We have to get stronger and more vocal, or we’ll just disappear. Somebody has to take the next step.”
Struggle is needed just to maintain the status quo.
Miller hopes these campaigns will be only the beginning of the beginning. She knows that, as employment becomes more precarious, issue advocacy alone isn’t going to build lasting change. “With the growth of the contingent workforce … we do need new forms,” she says. She hopes that the networks built via CoWorker will eventually aim toward universal guarantees not tied to an individual workplace, employer, industry, or even employment at all. “We need to have a much more imaginative, robust set of options than just what we have right now, which is you have to be an ‘employee’ to even have a single right.”
She thinks the path to this future begins by confronting the conditions of the present. “It is our civic responsibility to be active participants in our workplaces,” she says.
How can gig workers become “active participants” when their workplace is virtual?
Turkopticon is one answer. The service, built by Lilly Irani, an assistant professor at the University of California, San Diego, and Six Silberman, a researcher and programmer, allows Turkers to collectively track requesters. Smoothly integrated with the regular MTurk interface, it allows Turkers to see whether their peers have reported a requester for rejecting valid work. In this way, Turkers created a vital check on precariousness: consequences for wage theft and worker abuse. Importantly, they did it without appealing to the platform owners or waiting for regulators to step in. Turkopticon was the first time many Turkers realized they could fight back.
For bigger fights, though, they’ll need greater cohesion. Although Turkers readily collaborated on Turkopticon, coordination at greater scale was hindered by fractured, isolated communities.
“Dynamo was a place for us to come together,” says Milland. She is speaking about the website Dynamo, a forum in which Turkers choose actions to take collectively. I talked to Niloufar Salehi, the driving force behind Dynamo. Niloufar is a third-year Ph.D. student in computer science at Stanford University. In the heart of Silicon Valley, alongside peers who build the algorithms that power precarity-generating platform services, Salehi has different designs. In early 2013, she heard Irani present on Turkopticon and was immediately intrigued. “I had never seen research used that way before.” That fall, she enlisted collaborators, including Irani, Milland, and one anonymous Turker, known by the nom-de-Turk “Clickhappier,” to build Dynamo.
Salehi’s view of Turkers is exactly opposite of Amazon’s. She believes treating humans like algorithms leaves out most of what makes us human. “Human beings want to work with other people. They want to be able to grow,” she says. “They want to have meaning in what they’re doing. There’s a lot of benefit in looking at human beings and trying to understand them as human beings.”
"It is our civic responsibility to be active participants in our workplaces."
Following that ethos, the Dynamo project started with a listening tour. Salehi visited Turker forums to learn what features they needed. Dynamo is built with their concerns in mind. For instance, to ensure the forum wasn’t usable only by Turkers with spare time to spend organizing, Dynamo action proposals are limited to a single tweet-length message, and participation can involve a simple “up” or “down” vote. In this way, the community has inclusion at its core.
Two campaigns illustrate the promise—and the limits—of their efforts to date.
In one, a team of Turkers collaboratively wrote guidelines for academic requesters. The guidelines establish basic standards for decent behavior—“identify yourself,” “provide reasonable time estimates”—and, importantly, fair wages. That campaign was a success: The guidelines are widely embraced and even adopted as official standards at some universities.
In a second campaign, Turkers aimed to change the platform itself and found visibility has limits. They organized a letter-writing campaign to Jeff Bezos, Amazon’s CEO, hoping he would see they were “human beings, not algorithms.” Despite media attention, the plan backfired. Soon after the letters, Amazon raised fees dramatically—effectively reducing wages—and, Milland says, appears to have banned communication between internal employees and Turkers. She says the last time anyone openly identifying as an Amazon employee posted to her forum, Turker Nation, was May 2015, just before fees were raised. The timing of the changes was suspicious, and Milland believes the letters were a major factor. “I think we pissed [Bezos] off.”
Amazon’s stonewalling prompted many in the Turker community to look for new ideas that go beyond appealing to platform owners to change. As Milland puts it, “We’ve got to look for alternative points of pushing back. This is just the start.”
Part of what drives Turkers forward is that, for many, there is no going back. Because of medical issues and family responsibilities, “I can’t have a [full-time office job],” Milland says. She thinks the same is true for the vast majority of Turkers, and LaPlante agrees. “Some care for children or elderly relatives at home. Some cannot find employment in their rural areas. Some have felonies on their record. Some do not have transportation.”
Like Milland and LaPlante, some Turkers hope the decentralized, inclusive communities they are building can create something entirely new, something that reflects their principles. Milland has a very clear goal in mind: “We need worker-run platforms.”
Trebor Scholz, associate professor of culture and media at the New School in New York City, agrees. He thinks platform services reveal a flaw in the way we measure progress. “How is something innovation if it benefits 50 people in Silicon Valley?” he asks. Scholz sees the move of work to platforms as an opportunity to build new services owned not just by the designers, but by everyone who uses them. With ownership, he says, “you have more control.” Such platform cooperatives should be more likely to treat workers well, be good members of their surrounding communities, and generate shared wealth.
For many, there is no going back.
The idea is based on worker-owned cooperatives, which have found success throughout history. Generally, cooperatives succeed when they are protected—by geography, regulation, or culture—from competition with corporations. For instance, several taxi cooperatives have existed in cities around the United States for decades. But while these services may carve out local niches, they can’t aspire to the resources and visibility of a globalized platform like Uber.
At least, not alone. The promise of platforms for cooperatives is that they can provide increased scale without taking away worker ownership. Arcade City is one such service that promises to cut out the middleman and connect drivers directly to riders via a single, cooperatively owned app.
Beyond duplicating existing platform success stories, Scholz’s broader aim is to link many different kinds of cooperatives into a networked community, an ecosystem in which food co-ops work together with startups and with cooperatively owned social networks. Critically, he says, this linking-up has to be part of a political movement organized enough to turn its collective influence into regulatory and legislative power. In Scholz’s vision, this would entail a whole different kind of economy, in which resources are allocated and shared via platforms that participate in markets but also function as a firewall against them by, for instance, providing mutual guarantees of material support. For example, workers on a cooperative gig platform could earn shares at the food co-op. In this world, everyone would be an owner, and the benefits of innovation would always be shared.
The gig economy is often seen as the dystopian future of work. But there’s not much new about the underlying economics. What is new is that the global network enabling the current wave of dispossession also allows workers unprecedented connectivity and reach. While connectedness alone won’t bring revolutionary change, movements can now coordinate and build at unprecedented scale while remaining inclusive and democratic.
CoWorker and Dynamo are two early examples of tools for building democratic, worker-led network communities. Right now they focus on issue advocacy, but Miller sees them as a stepping stone to a new way of working together: “I think the internet is not a set of tools … It’s a place and a culture. It requires us to behave differently. And if we can do that, we can do amazing and incredible things.”
The nascent networks of worker movements that come together via these tools could be the first adopters of new platform cooperatives. What kinds of services can thrive on the cooperative platforms and whether they can create a new kind of economy are questions that can only be answered through practice. Scholz agrees that no one yet knows where platforms will take us. “I’m as curious as you are as to what will happen.”
Having spent a decade pioneering the future of work, there might not be anyone more curious—or more ready—to create something new than Milland. “What does it look like if we build it today? I don’t know. We’ve got to try it.”
Meet the Turkers
Amazon’s Mechanical Turk labor force is always online and hungry to work. Some fit it in as a side job. Some do it as their main job. Members of this “artificial” artificial-intelligence network work in isolation for as little as pennies to carry out tasks from “requesters.” It’s a situation that sidesteps minimum-wage laws and can leave workers hanging when “employers” don’t pay. Turkers, as they’re called, have forged online communities that empower them to share advice, boycott bad requests, and build the kind of relationships found in traditional workplaces. We hired a few Turkers to tell us about their work.