Skip to content

Surveillance Wages: A Taxonomy


Zephyr Teachout (@ZephyrTeachout) is Professor at Law at Fordham Law School.

One notable feature of the so-called gig economy is that workers, such as for-hire drivers and delivery workers, are often paid different amounts for performing the same task. More broadly, as work by the Markup and Veena Dubal shows, the manner in which these wages are calculated is a black box: they are determined by a complicated algorithm to which the workers have no access. At Uber, for instance, the company uses data-rich driver profiles to match the wage to the individual incentives of the driver and the needs of the platform.

Uber drivers’ experiences should not, however, be understood as unique to gig work. The techniques used to extract the maximal gig worker productivity for minimal pay—such as individualized pay, schedules, benefits, and individualized behaviorally-based incentive structures—are on the verge of being imported to the formal employment sector. Moreover, gig work is increasingly the nature of work itself, with surveys showing that between 10 and 30% of Americans have some form of contract work. As a result, wage scales with set pay grades, which have been the norm for blue collar jobs for more than a century, may be on the decline.

Partly as a result of this shift, the market for employee surveillance systems is booming. Monitoring systems include thumb scans, identification badges, closed circuit cameras, geolocation tracking, and sensors on tablets and vehicles. Software flags not only a decrease in productivity but also the expression of “negative attitudes.” Tools record keyboard strokes and conversations, which are then analyzed to rate an employee’s emotional status based on word patterns and content. The company Cogito, for instance, sells software to call centers that records and then analyzes calls between employees and customers, with a real-time behavioral dashboard that tells the employees when to be more empathetic, when to pick up the pace, and when to “exude more confidence and professionalism.” Supervisors have dashboards summarizing these performance metrics, which are used to determine pay and retention.

In addition to making worker’s pay much less predictable, the potential spread of these management techniques has broader democratic implications. They will increase economic and racial inequality, undermine labor solidarity, and put workers in a profoundly humiliating position in relationship to their boss, one where worker speech and autonomy are highly circumscribed.

If we are to have any hope of successfully regulating these new forms of technological control, we first need a clearer understanding of how they operate. To that end, this post offers a brief taxonomy of five different forms of algorithmic wage differentiation, each of which is already visible in the gig work economy: productivity-based wage adjustments, gamified wages shifted through incentive bonuses and demerits, behavioral wages, dynamic wages, and wages shifted to conduct an experiment. After surveying these different techniques, I explain in greater depth why these practices threaten to undermine important democratic values. 

Extreme Taylorism

At the heart of management in the past century was a belief, exemplified by Taylorism, that workers were driven by simple motivations. They wanted pay, and didn’t like to work, and so close monitoring and control were required, lest workers take the pay and avoid the work. Taylorism depended on small, easy to measure tasks, regularized training, and payment based on production.

Extreme Taylorism is a direct outgrowth of these former tools, but with unheard of levels of precision. Employers now measure time in the bathroom and time per unit task, not just the total number of units processed over an hour, and can thus reward or dock employees on a far more ongoing, updated basis. For instance, when Amazon follows workers physical location and tracks multitasking, it can dock pay for long breaks in real time. And with increasingly precise levels of surveillance, companies can monitor and reward productivity down to the head-swivel.


Amazon started using video games in five warehouses from suburban Seattle to near Manchester in Britain, after an initial experiment in a single warehouse in late 2016. With names like MissionRacer, PicksInSpace, Dragon Duel, and CastleCrafter, the games have graphics that mimic Nintendo, according to workers (employees aren’t allowed to take pictures). Success at these games can lead to changes in wages—one worker reported that managers rewarded successful workers with “Swag Bucks,” an internal currency that can be used to make purchases. 

Similarly, Uber tracks millions of metrics every day and then delivers individualized tasks to drivers. At least some of this differentiation appears to be related to the “gamification” of work. The screen used by drivers has point-scoring, levels, competition with others, and ratings, which play on both positive and negative aspects of gaming—they offer engagement and possibly fun, but also pray on irrational tendencies, like gambling. And, like gambling, they appear to depend upon both personalization (differentiation) and some degree of inconsistency.

Behavioral Price Discrimination

Fifteen years ago, social media delivered content based on either requests by users or chronology; today, social media posts are delivered based on a data-based portrait of each individual user. The PII (personally identifiable information) is fed to algorithms which predict behavior and responsiveness. Persuasion technologies are matched with unique features to maximize the profits made by the platforms.

We can expect that a similar shift is on the horizon at the workplace, if not already here. This kind of wage discrimination, which focuses on a worker’s behavior, is the area where we have the least information, and the fewest examples, but we have every reason to think it will be a growth area. Employers can combine the data collected directly on their own sites with personal data bought on the open market from third parties, who collect and aggregate social media activity, credit reports, consumer history, and driving reports. Outside of state law, there is nothing currently stopping employers from using data about, e.g., the indebtedness of a worker, to target a lower wage, knowing the worker has less flexibility to move jobs.

Dynamic Labor Pricing

In the gig economy, dynamic labor pricing—paying more for labor when it is low supply or high demand, and vice versa—is baked into the business model. It is also standard for consumer goods where prices are shifting constantly; according to one estimate, Amazon prices change more than 2.5 million times a day. Outside of gig work, firms—especially those with monopsony power—can write contracts that make bonuses ongoing in the same way they are ongoing for Uber drivers, and in so doing, can engage in real-time labor pricing based on demand. 

It is currently unclear if (or to what extent) this is happening outside of gig work, but opportunity and incentive suggest it could. Within gig work, for instance, caregivers are paid differently depending on the time of day, delivery workers are paid differently depending on demand, and pay can shift on a dime.


In the late 1920s, inspired by Frederick Taylor, a psychologist from Harvard Business School ran a series of experiments on workers at Hawthorne Works, an electric company in Illinois, that became known as “the Hawthorne Studies.” Among other things, these studies explored whether changing pay based on performance impacted productivity. In one group, they gave higher pay to those who were more productive in assembling equipment; in a control group, they did not. They found, perhaps surprisingly to modern readers, that the increased wages did not incentivize productivity. Instead, productivity was more likely to be spurred by group pressures and group standards.

The experiments are interesting not merely for their results, but also as the beginning of a long history in which workers are subjected to experiments and manipulation by their employers. Since Hawthorne, firms have routinely conducted experiments on workers, testing assumptions about what will lead to the firm gathering the highest output for the wages it pays. However, until recently, wage experimentation at a level likely to reveal meaningful results was practically very difficult. This has changed with the rise of surveillance, big data, and electronic contracts with real-time bonuses, which have enabled low-cost wage experiments. For instance, when Instacart was caught using tips to supplement wages, the proceedings revealed that the company was more broadly experimenting with wages. Delivery pay was initially like piecework in farming: a flat rate per item. The company started adding bonuses, and then taking other factors into account, such as product weight and distance. And over the years, with small changes, the pay changed from being clearly tied to productivity to being a black box. Workers are not merely harmed by the end-outcome of these experiments, but also by the manipulation of the experiments themselves, which undermine any sense of stability at work.


With this taxonomy in hand, let’s return to the democratic implications of such wage discrimination. To begin, real time and individually targeted wages will transform the nature of supervision. It undermines the importance of relationships between supervisors and mid-level decisionmakers, instead allowing upper-level management to continuously spy and tinker with low-level workers. Workers are then employed in a state of rational paranoia, where they know that they are being punished and rewarded and experimented upon, but they have no way of knowing whether any given decision they are faced with is a result of a game, an experiment, a punishment, a reward, or changing circumstances on the ground and changing needs at the job. Being dominated, watched, and controlled are destabilizing conditions even when they are occasional, and all the more so if they are the center of work life. Privacy concerns that have long attended the workplace—and never been adequately addressed—are thus even more important today, as the lack of protection from intimate intrusions enables this further harm.

Beyond undermining the liberties of the moderns, intrusive surveillance and experimentation and differentiation also necessarily undermines the liberties of the ancients. The people being surveilled are not just workers but citizens, who must vote, serve on juries, share their experiences with the public, and engage in public debate. Citizens are also subject to some of the same monopoly practices in their role as consumers, but the relationship between the consumer and surveillance capitalism and the worker and the surveilled workplace is different. At work—when labor markets have a handful of dominant players—employees don’t even have the theoretical option of opting out of being watched. Negotiating the terms of surveillance and experimentation simply doesn’t happen. And unlike the consumer, the worker is surveilled for the entire scope of their workday, with no default right of respite.

When you combine personalized labor pricing with the fact that most firms have strong political views, it leads to all kinds of distortions in the political sphere. Sixteen percent of workers surveyed recently reported that they had either personally experienced or witnessed political retaliation on the job. One in eight American workers believes that “someone at their job was treated unfairly, missed out on a promotion, or was fired as a result of political views or actions.” It is unclear the degree to which employers are purposefully tracking the political activities of employees, but the scope of the sweep they conduct on a daily basis means that the conversations and online behavior are necessarily being gathered, whether they are used or not. With both capacity and incentive, we should anticipate that political spying will be a growth area. In the wake of Citizens United, corporations are free to engage in explicitly political activity, to monitor and respond and dissuade and punish. And just as importantly for society, even if an employer refrains from acting on this information, workers, aware of this capacity, will be discouraged from engaging in open, free debate.

I assume that these techniques are not yet the norm, but that the infrastructure to enable them is being put into place as we speak. We know from recent rise of targeted ads just how quickly business practices can change with the arrival of new technological tool. Given that, we should treat these discriminatory wages as a present threat, and think now about how to limit their reach. Key tools involve not only privacy legislation, and absolute bans on the collection of certain data, but using non-discrimination principles to require employers, especially large ones, to pay for the job on a structured pay scale, instead of tailoring pay to individual profiles.