As algorithms have come to dominate the lives of gig workers, treating many like poorly performing robots, community structures are emerging that allow workers to spontaneously organize – and create a potentially powerful counterforce.
Piecework and the labor/management divide
Frederick Winslow Taylor was the very first management guru, indeed the first management consultant. His ideas, which were radical at the time, posited that for any physical task, there was one best way to get it done. To get to the heart of the one best way, he did time-and-motion studies, compared how different workers stacked up and experimented with different techniques. Once the best way had been found, it would be codified in manuals, used to train workers and ultimately to deliver higher productivity across the board. Indeed, at one of his early employers, Midvale Steel Company, he managed to double productivity.
This did not come without tension between the bosses and their labor force. Workers, paid on a piecework system, received a specific amount of compensation for each “piece” produced or action performed. The disadvantage is that both sides of the production equation are incented to game the system. Employers want prices as low as possible. Workers want to produce as much as possible, which can have terrible impacts on both quality and learning.
Taylor’s own career put him smack in the middle of these tensions. As he writes in The Principles of Scientific Management, at the age of 22 at Midvale, he “was given work as a machinist in running one of the lathes, and, as he turned out rather more work than other machinists were doing on similar lathes, after several months was made gang-boss over the lathes.” (p. 48). He observed that “the shop was really run by the workmen, and not by the bosses. The workmen together had carefully planned just how fast each job should be done, and they had set a pace for each machine throughout the shop, which was limited to about one-third of a good day’s work. Every new workman who came into the shop was told at once by the other men exactly how much of each kind of work he was to do, and unless he obeyed these instructions, he was sure before long to be driven out of the place by the men.”
Bitter, bitter struggles ensued. The men resisted stubbornly, including sabotaging the machines (and blaming him for trying to run them too hard), exerting social pressure and engaging in physical threats. Taylor felt that there were two factors that allowed him to eventually overcome this resistance. The first was that he was a member of the upper classes, giving him credibility with management. When the workers complained that it was his supervision causing the machines to become inoperable, management sided with him (eventually putting the burden on the men operating the machine to pay for its repair). The second was that he did not live amongst the men, and therefore could withstand the social pressure of the community. As we shall see, social ties can create powerful communities to advocate for workers’ interests.
Taylor, misunderstood
Ironically, Taylor himself believed that management and the workforce should each benefit by implementing the principles of scientific management. In an ideal system, workers would generate an optimal level of output, and management would reward them by sharing in the gains, paying them more than if they operated in the old, “unscientific” way. I find it fascinating that when we criticize Taylorism, we have largely forgotten this point.
As he himself wrote, “to work according to scientific laws…almost every act of the workman should be preceded by one or more preparatory acts of the management which enable him to do his work better and quicker than he otherwise could. And each man should daily be taught by and receive the most friendly help from those who are over him…this close, intimate, personal cooperation between the management and the men is of the essence of modern scientific management.”
Shortly before Taylor’s death in 1915, Henry Ford implemented the moving assembly line, which led to the simplification and routinization of tasks. Under this system, the pace of work was up to the design of the production system. Workers hated it, and just as they had in Taylor’s case, resisted by engaging in absenteeism and withholding their best efforts. Ford’s response was to introduce his famous $5/day rate of pay. We give him a lot of credit for that – but it is fascinating that the $5/day was not the actual worker’s wage. The wage was divided into two parts – the wage, which every worker was paid (about half the $5). The remainder was paid in “profits” and was paid only to those workers deemed “worthy.” As one observer notes, “the Ford profit-sharing plan was a unique experiment in the social engineering of immigrant auto workers to inculcate the personal habits and work discipline suitable for assembly line production. In effect, the standardization of production required the standardization of labor.”
Today’s piecework: Welcome to the gig economy
As Wall Street Journal columnist Christopher Mims and I discuss as we chat about his book Arriving Today, Taylorism might just be the most successful management idea ever implemented.
There is, however, a dark side. Work intermediated by algorithms – such as making deliveries or ride-hailing – subject workers to algorithmic control. It’s a lot closer to the assembly line model than to Taylor’s vision of profits being shared in a mutually beneficial way between workers and company management.
The dissatisfactions are many. Platforms can gamify work encouraging workers to work just a bit more. The algorithms are relentless, leading workers to forgo bathroom breaks. Platforms can arbitrarily change their algorithms in a way that disadvantages workers, including deleting accounts unexpectedly. Pay for given amounts of work can be set at unfair levels. Platforms can collect personal information about workers and use it to their disadvantage. Workers are subject to covert monitoring and surveillance. The criteria used by algorithms to evaluate work are often opaque. Even worse, when workers respond, they can find themselves in an “iron cage” as Kellogg’s Hatim Rahman found when he looked into the topic. Just as Ford’s “profit sharing” sought to control worker behavior, the platform Rahman investigated allows “employers to constrain behavior and set the standards of success.”
The rating criteria for freelancers on the platform he studied were clear and transparent. Enter the problem: most workers were getting four and five stars, which the platform’s executives decided made it less useful for buyers. The decision? Implement a deliberately opaque algorithm which resulted in only 5 percent of the freelancers receiving a grade of 90 or above. Workers, feeling both paranoid and frustrated, resorted to two main tactics: experimenting to see what would increase their scores, or moving off the platform for some activities, shielding them from the algorithm.
When workers’ skills improve, their employers’ benefit. In an unfortunate twist, these incentives don’t apply to gig workers, who are not afforded the protections offered to employees. Minimum wages and overtime, a set number of hours, benefits such as access to healthcare, protections against discrimination and harassment, access to workers’ compensation, sick leave and unemployment insurance. And gig workers face the requirement to operate as the algorithm demands – or else. Part of why the platforms get away with this is that it pits workers against each other and geographically scatters them, decreasing their power.
The community empowering workers (at least a little)
Given all this, I was intrigued to run across an MIT Technology Review article that went to an unlikely place – Indonesia – to find several ways in which worker community groups were banding together to take back some control from the algorithms. This is particularly intriguing given the work of Damon Centola, who has found that a change in behavior – social change – becomes far more likely to the extent that 25% of a community group’s members pushed for a change in norms.
One form of worker unity takes the form of on-line communities like Gojek on Twitt, in which drivers can find each other. These communities are supplemented with other circles on platforms such as Whatsapp and Telegram. The circles help drivers find resources and connect with one another. They can also unit against their algorithmic overlords. In what scholars who have studied this call “everyday resistance” thousands of drivers are engaging in small hacks that improve their jobs.
A practice called “account therapy” allows those who are more knowledgeable about the way the algorithm works to tweak it for others – for instance, if a driver wants fewer food delivery orders, the “therapist” might take over his phone and refuse such orders for a week until the platform ‘gets it.’ A more sophisticated form of resistance are so-called “grey market apps” that do things like spoof the GPS system in the phone. This can allow drivers to wait in more convenient locations than those near a prospective order, or convince the algorithm that they are working when they are actually resting.
The rules always lag the reality
While such small tweaks to the system make drivers’ jobs a bit better, we are far from achieving Taylor’s vision of scientific management driving fairly shared productivity gains. What is likely to be needed is a new standard for workers’ technology rights. As this report from the UC Berkeley Labor Center proposes, policy standards need to “give workers rights with respect to their data; hold employers responsible for any harms caused by their systems; regulate how employers use algorithms and electronic monitoring; ensure the right to organize around technology; guard against discrimination; and establish a strong enforcement regime.”
It seems clear that we are once more heading into the selfsame struggles Taylor observed between workers and those who profit from their labor; this time with a digital dimension.