Recent technological developments related to the extraction and processing of data have given rise to concerns about a reduction of privacy in the workplace. For many low-income and subordinated racial minority workforces in the United States, however, on-the-job data collection and algorithmic decisionmaking systems are having a more profound yet overlooked impact: These technologies are fundamentally altering the experience of labor and undermining economic stability and job mobility. Drawing on a multi-year, first-of-its-kind ethnographic study of organizing on-demand workers, this Article examines the historical rupture in wage calculation, coordination, and distribution arising from the logic of informational capitalism: the use of granular data to produce unpredictable, variable, and personalized hourly pay.

The Article constructs a novel framework rooted in worker on-the- job experiences to understand the ascent of digitalized variable pay practices, or the importation of price discrimination from the consumer context to the labor context—what this Article identifies as algorithmic wage discrimination. Across firms, the opaque practices that constitute algorithmic wage discrimination raise fundamental questions about the changing nature of work and its regulation. What makes payment for labor in platform work fair? How does algorithmic wage discrimination affect the experience of work? And how should the law intervene in this moment of rupture? Algorithmic wage discrimination runs afoul of both longstanding precedent on fairness in wage setting and the spirit of equal pay for equal work laws. For workers, these practices produce unsettling moral expectations about work and remuneration. The Article proposes a nonwaivable restriction on these practices.

The full text of this Article can be found by clicking the PDF link to the left.


Over the past two decades, technological developments have ushered in extreme levels of workplace monitoring and surveillance across many sectors. 1 See, e.g., Ifeoma Ajunwa, Kate Crawford & Jason Schultz, Limitless Worker Surveillance, 105 Calif. L. Rev. 735, 738–39 (2017) [hereinafter Ajunwa et al., Limitless Worker Surveillance]; Matthew T. Bodie, The Law of Employee Data: Privacy, Property, Governance, 97 Ind. L.J. 707, 712–17 (2022); Brishen Rogers, The Law and Political Economy of Workplace Technological Change, 55 Harv. C.R.-C.L. L. Rev. 531, 535–36 (2020). These automated systems record and quantify workers’ movement or activities, their personal habits and attributes, and even sensitive biometric information about their stress and health levels. 2 See Ifeoma Ajunwa, Kate Crawford & Joel S. Ford, Health and Big Data: An Ethical Framework for Health Information Collection by Corporate Wellness Programs, 44 J.L. Med. & Ethics 474, 474–75, 477–78 (2016) (describing the comprehensive data collection practices and capacities of worker wellness programs). Employers then feed amassed datasets on workers’ lives into machine learning systems to make hiring determinations, to influence behavior, to increase worker productivity, to intuit potential workplace problems (including worker organizing), and, as this Article highlights, to determine worker pay. 3 See, e.g., Annette Bernhardt, Linda Kresge & Reem Suleiman, Berkeley Lab. Ctr., Data and Algorithms at Work: The Case for Workers’ Technology Rights 6, 15–17 (2021), https://laborcenter.berkeley.edu/wp-content/uploads/2021/11/Data-and-Algorithms-at-Work.pdf [https://perma.cc/TC3U-458E]. As employment law scholar Matthew Bodie has written in reference the role of data extraction at work under systems of informational capitalism:
Workers find themselves on the wrong end of this data revolution. They are the producers of data, but the data flows seamlessly from their work and personal experience to corporate repositories. Employers can capture the data, aggregate it into meaningful pools, analyze it, and use it to further productivity. Individual employees cannot tap into that value, nor can independent contractors. They are trapped: the more data they provide, the more powerful their employers become.
Bodie, supra note 1, at 736.

To date, policy concerns about growing technological surveillance in the workplace have largely mirrored the apprehensions articulated by consumer advocates. Scholars and advocates have raised concerns about the growing limitations on worker privacy and autonomy, the potential for society-level discrimination to seep into machine learning systems, and a general lack of transparency on workplace rules. 4 See generally Bernhardt et al., supra note 3 (arguing that data-driven technologies harm workers through discrimination and work intensification at the expense of safety, depriving workers of their autonomy and dignity); Ajunwa et al., Limitless Worker Surveillance, supra note 1 (“[T]here has been a shift in focus from collecting personally identifying information, such as health records, to wholly acquiring unprotected and largely unregulated proxies and metadata, such as wellness information, search queries, social media activity, and outputs of predictive ‘big data’ analytics.”); Bodie, supra note 1 (“As the data collected in this new environment has become increasingly individualized, the line between person as individual and person as employee has become significantly blurred.”); Rogers, supra note 1 (“[L]abor and employment laws . . . and the broader political economy of work that they help sustain, also encourage employers to use new technologies to exert power over workers.”). Labor law scholars Antonio Aloisi and Valerio De Stefano have argued convincingly in a comprehensive review of technology, law, and work that concerns about the supposed “disappearance of work” lost to algorithmic intelligence are less urgent than the myriad challenges raised by the incipient practices of algorithmic management at work. These nascent practices, they argue, have intensified any number of problems including the devaluation of work, the maldistribution of risks and privileges, the health and safety of workers, the assault on dignity, and of course, the destruction of individual and collective worker privacy. Antonio Aloisi & Valerio De Stefano, Your Boss Is an Algorithm: Artificial Intelligence, Platform Work and Labour 9, 23–24, 98–101, 104–05 (2022). For example, in October 2022, the White House Office of Science and Technology Policy released a non-legally-binding handbook identifying five principles that “should guide the design, use, and deployment of automated systems to protect the American public in the age of artificial intelligence.” 5 White House Off. of Sci. & Tech. Pol’y, Blueprint for an AI Bill of Rights 3 (2022), https://www.whitehouse.gov/wp-content/uploads/2022/10/Blueprint-for-an-AI-Bill-of-Rights.pdf [https://perma.cc/A62C-TV47]. These principles called for automated systems that (1) were safe and effective, (2) protect individuals from discrimination, (3) offer users control over how their data is used, (4) provide notice and explanation that an automated system is being used, and (5) allow users access to a person who can remedy any problems they encounter. 6 Id. at 5–7. The Blueprint for an AI Bill of Rights (hereinafter Blueprint) specified that these enumerated rights extended to “[e]mployment-related systems [such as] . . . workplace algorithms that inform all aspects of the terms and conditions of employment including, but not limited to, pay or promotion, hiring or termination algorithms, virtual or augmented reality workplace training programs, and electronic workplace surveillance and management systems.” 7 Id. at 53 (emphasis added).

Under each principle, the Blueprint provides “illustrative examples” of the kinds of harms that the principle is meant to address. One such example, used to specify what defines unsafe and ineffective automation in the workplace, involves an unnamed company that has installed AI-powered cameras in their delivery vans to monitor workers’ driving habits, ostensibly for “safety reasons.” The Blueprint states that the system “incorrectly penalized drivers when other cars cut them off . . . . As a result, drivers were incorrectly ineligible to receive a bonus.” 8 Id. at 17 (emphasis added) (citing Lauren Kaori Gurley, Amazon’s AI Cameras Are Punishing Drivers for Mistakes They Didn’t Make, Vice (Sept. 20, 2021), https://www.vice.com/en/article/88npjv/amazons-ai-cameras-are-punishing-drivers-for-mistakes-they-didnt-make [https://perma.cc/HSF4-EG4M]). Thus, the specific harm identified is a mistaken calculation by an automated variable pay system developed by the company.

What the Blueprint does not specify, however, is that the company in question—Amazon—does not directly employ the delivery workers. Rather, the company contracts with Delivery Service Providers (DSPs), small businesses that Amazon helps to establish. In this putative nonemployment arrangement, Amazon does not provide to the DSP drivers workers’ compensation, unemployment insurance, health insur-ance, or the protected right to organize. Nor does it guarantee individual DSPs or their workers minimum wage or overtime compensation. 9 As economist Brian Callaci explains, since the DSPs legally employ the delivery drivers, the DSPs, rather than Amazon, bear “liability for accidents or workplace safety,” and DSP drivers, classified as Amazon’s contractors, “do not fall under Amazon’s $15 an hour minimum wage.” Brian Callaci, Entrepreneurship, Amazon Style, Am. Prospect (Sept. 27, 2021), https://prospect.org/api/content/1923a910-1d7c-11ec-8dbf-1244d5f7c7c6/ [https://perma.cc/‌AV2H-59YA]. Meanwhile, Amazon’s contracts with DSPs “[restrict] the wages the DSP can offer” drivers and mandate that drivers remain nonunion by stipulating that “they serve as at-will employees.” Id. If the drivers unionize, “Amazon can terminate the contract and find a new DSP, which is much easier than fighting a union campaign itself.” Id. Instead, DSPs receive a variable hourly rate based on fluctuations in demand and routes, along with “bonuses” based on a quantified digital evaluation of on-the-job behavior, including “service, safety, [and] client experience.” 10 How Are Amazon DSPs Paid?, Route Consultant, https://www.routeconsultant.com/industry-insights/how-are-amazon-dsps-paid [https://perma.cc/684P-WLKB] (last visited Aug. 14, 2023). The scorecards that determine “bonuses” are calculated in constantly changing ways. The DSP scorecards I reviewed include four categories: safety and compliance, reliability, quality, and team. The “scores” for these categories—and for each driver employed by the DSP—are determined algorithmically. See also Peak Delivery Driver, Amazon DSP Scorecard Deep Dive, YouTube, at 1:04–1:58, 2:48–3:10 (Sept. 10, 2021), https://www.youtube.com/watch?v=-mBOYfBZs9I (on file with the Columbia Law Review). The example in the Blueprint, for instance, lowered the score enough to undermine the DSP’s ability to get a bonus. White House Off. of Sci. & Tech. Pol’y, supra note 5, at 17. By contrast, Amazon is guaranteed the data it wants from the DSPs (they cannot reject the use of cameras, for example)—not just while the DSP is servicing Amazon but also for three years afterward. In addition to using such data to calculate bonuses, Amazon can also use it to terminate contracts, terminate specific “underperforming” workers, and punish DSPs with fees. Josh Eidelson & Matt Day, Drivers Don’t Work for Amazon but Company Has Lots of Rules for Them, Det. News (May 5, 2021), https://www.detroitnews.com/story/business/2021/05/05/drivers-dont-work-amazon-but-company-has-lots-rules-them/4955413001/ [https://perma.cc/7REA-NKRU]. DSPs, while completely reliant on Amazon for business, must hire a team of drivers as employees. 11 When a DSP hires other drivers, it may appear more like a company that is legally separate from Amazon. This may protect Amazon from unionization efforts and downstream liability that it may otherwise incur based on allegations that the DSPs are its employees, not contractors. Callaci, supra note 9. It appears FedEx was the first delivery company to use this tactic after redrafting its contracts with drivers in response to Alexander v. FedEx Ground Package Sys., Inc., 765 F.3d 981 (9th Cir. 2014), the Ninth Circuit decision that held that its drivers were employees, not independent contractors. Rather than changing the drivers’ status in response to the decision, FedEx drafted its contracts to make the drivers appear more like independent contractors. V.B. Dubal, Winning the Battle, Losing the War?: Assessing the Impact of Misclassification Litigation on Workers in the Gig Economy, 2017 Wis. L. Rev. 739, 791–92. This included mandating that the drivers purchase more service areas, which in turn made drivers hire others to complete the deliveries. Id. These Amazon-created and -controlled small businesses rely heavily on their automated “bonuses” to pay for support, repairs, and driver wages. 12 Lauren Kaori Gurley, Amazon’s AI Cameras Are Punishing Drivers for Mistakes They Didn’t Make, Vice (Sept. 20, 2021), https://www.vice.com/en/article/88npjv/amazons-ai-cameras-are-punishing-drivers-for-mistakes-they-didnt-make [https://perma.cc/HSF4-EG4M]. As one DSP owner–worker complained to an investigator, “Amazon uses these [AI surveillance] cameras allegedly to make sure they have a safer driving workforce, but they’re actually using them not to pay [us] . . . . They just take our money and expect that to motivate us to figure it out.” 13 Id. (internal quotation marks omitted) (quoting the owner of a Washington-based Amazon delivery company).

Presented with this additional information, we should ask again: What exactly is the harm of this automated system? Is it, as the Blueprint states, the algorithm’s mistake, which prevented the worker from getting his bonus? Or is it the structure of Amazon’s payment system, rooted in evasion of employment law, data extraction from labor, and digitalized control?

Amazon’s automated control structure and payment mechanisms represent an emergent and undertheorized firm technique arising from the logic of informational capitalism: the use of algorithmic wage discrimination to maximize profits and to exert control over worker behavior. 14 “Informational capitalism” or “information capitalism” as a descriptor of the contemporary digital-age world system is generally attributed to sociologist Manuel Castells. Castells first introduced the term in his three-volume study, The Information Age, published between 1996 and 1998. In describing a shift from industrial capitalism to information capitalism, Castells wrote in Volume I, “A technological revolution, centered around information technologies, is reshaping, at accelerated pace, the material basis of society. Economies throughout the world have become globally interdependent, introducing a new form of relationship between economy, state, and society, in a system of variable geometry.” Manuel Castells, The Rise of Network Society 1 (2d ed. 2000). In legal scholarship, Julie Cohen uses the term “informational capitalism” to explore the relationships between political, legal, and economic institutions amidst the propertized expansion of data and information exchange. See generally Julie Cohen, Between Truth and Power: The Legal Constructions of Informational Capitalism (2019). “Algorithmic wage discrimination” refers to a practice in which individual workers are paid different hourly wages—calculated with ever-changing formulas using granular data on location, individual behavior, demand, supply, or other factors—for broadly similar work. As a wage-pricing technique, algorithmic wage discrimination encompasses not only digitalized payment for completed work but, critically, digitalized decisions to allocate work, which are significant determinants of hourly wages and levers of firm control. These methods of wage discrimination have been made possible through dramatic changes in cloud computing and machine learning technologies in the last decade. 15 Zephyr Teachout has created a useful taxonomy of five different forms of “personalized wages” that have recently emerged in the labor market: (1) extreme Taylorism, in which “[h]igh degrees of surveillance [result in] . . . rewarding productivity”; (2) gamification, in which employers use psychological tools to incentivize task completion; (3) behavioral price discrimination, in which workers get paid more if they make certain lifestyle choices, like exercising, which can be tracked through fitness apps; (4) dynamic labor pricing, which, she argues, is based primarily on demand; and (5) experimentation, in which firms test “assumptions about what will lead to the firm gathering the highest output for the wages it pays.” Zephyr Teachout, Algorithmic Personalized Wages, 51 Pol. & Soc’y 436, 437, 442–44 (2023) [hereinafter Teachout, Algorithmic Personalized Wages].
In all these instances, wages are rooted in data extracted from labor. My data indicate the potential to further simplify this taxonomy to two main ways of thinking about algorithmic wage discrimination: (1) wages based on productivity analysis alone (most evident in the employment context), and (2) wages based on productivity, supply, demand, and other personalized data used to minimize labor costs. This second form of algorithmic wage discrimination appears most commonly in on-demand work that treats workers like independent contractors.

Though firms have relied upon performance-based variable pay for some time (e.g., the use of bonuses and commission systems to influence worker behavior), 16 Nonalgorithmic variable payment systems with transparent payment structures are more familiar to many people. See, e.g., United Farm Workers (@UFWupdates), Twitter (Oct. 15, 2022), https://twitter.com/UFWupdates/status/1577795973476220930 (on file with the Columbia Law Review) (showing how California companies use a variable bonus system for some farmers’ pay). They are, nonetheless, controversial. Some critics in the human relations and management literature point to variable pay mechanisms as a contributor to income gaps by gender and race. See, e.g., Emilio J. Castilla, Gender, Race, and Meritocracy in Organizational Careers, 13 Am. J. Socio. 1479, 1502–17 (2008) (finding variable salary bias in salary increases and promotions on the basis of gender, race, and nationality). Others suggest variable pay has psychological costs for workers and other unforeseen consequences. See, e.g., Annette Cox, The Outcomes of Variable Pay Systems: Tales of Multiple Costs and Unforeseen Consequences, 16 Int’l J. Hum. Res. Mgmt. 1475, 1483–93 (2005) (discussing unexpected costs to both employers and employees resulting from variable salary systems). my research on the on-demand ride hail industry suggests that algorithmic wage discrimination raises a new and distinctive set of concerns. In contrast to more traditional forms of variable pay, algorithmic wage discrimination—whether practiced through Amazon’s “bonuses” and scorecards or Uber’s work allocation systems, dynamic pricing, and wage incentives—arises from (and may function akin to) the practice of “price discrimination,” in which individual consumers are charged as much as a firm determines they may be willing to pay. 17 To date, scholars and analysts who have written about what this Article terms “algorithmic wage discrimination” have predominantly adopted the language of pricing, though they describe wage and not product pricing. For example, in her 2021 Enlund Lecture at DePaul University School of Law, Professor Zephyr Teachout referenced some of these practices as “labor price discrimination.” Zephyr Teachout, Professor, Fordham Univ. Sch. of L., Enlund Lecture at DePaul University School of Law (Apr. 15, 2021). Niels van Doorn, in an article analyzing the pay structures of on-demand Deliveroo riders in Berlin, describes “the algorithmic price-setting power of food delivery platforms,” which he understands as a “monopsonistic power that is not only market-making but also potentially livelihood-taking.” Niels van Doorn, At What Price? Labour Politics and Calculative Power Struggles in On-Demand Food Delivery, 14 Work Org. Lab. & Globalisation, no. 1, 2020, at 136, 138. But adopting the language of “pricing” for wage setting is politically and legally consequential. Since at least the rise of neoliberalism, price controls in the United States (and elsewhere) have been highly disfavored as economic interferences in the “free market,” raising conservative critiques of socialism and “planned economies.” See Benjamin C. Waterhouse, Lobbying America: The Politics of Business From Nixon to NAFTA 10623, 13239 (2013) (describing how American businesses rejected government price setting in the Nixon, Ford, and Carter administrations). Wage controls in the form of minimum-wage and overtime laws, on the other hand, have been contested but culturally naturalized as a necessary (or at least, accepted) part of economic regulation. See Amina Dunn, Most Americans Support a $15 Federal Minimum Wage, Pew Rsch. Ctr. (Apr. 22, 2021), https://www.pewresearch.org/short-reads/2021/04/22/most-americans-support-a-15-federal-minimum-wage/ [https://perma.cc/CX5Z-YX9Z] (surveying support for minimum-wage laws across the United States). In this sense, conceptualizing the digitalized wages received by workers not as firm price determinations but as firm wage determinations is a critical political—and legal—corrective. As a labor management practice, algorithmic wage discrimination allows firms to personalize and differentiate wages for workers in ways unknown to them, paying them to behave in ways that the firm desires, perhaps for as little as the system determines that the workers may be willing to accept. 18 See infra Part II. Given the information asymmetry between workers and firms, companies can calculate the exact wage rates necessary to incentivize desired behaviors, while workers can only guess how firms determine their wages. 19 See Aaron Shapiro, Dynamic Exploits: Calculative Asymmetries in the On-Demand Economy, 35 New Tech. Work & Emp. 162, 162–63 (2020) [hereinafter Shapiro, Dynamic Exploits: Calculative Asymmetries] (arguing that “independent service providers” for “on-demand service platforms” are workers and not independent contractors because the platforms set wages and “exhibit substantial information asymmetries”). Uber, for its part, has stated that “suggestions that Uber offers variable pricing based on user-profiling is completely unfounded and factually incorrect.” Cansu Safak & James Farrar, Worker Info Exch., Managed by Bots: Data-Driven Exploitation in the Gig Economy 26 (2021), https://5b88ae42-7f11-4060-85ff-4724bbfed648.usrfiles.com/ugd/5b88ae_8d720d54443543e2a928267d354acd90.pdf [https://perma.cc/TLV3-R2EE] (internal quotation marks omitted) (quoting Letter from Uber Data Protection and Cybersecurity Team to Cansu Safak (Dec. 3, 2021), https://5b88ae42-7f11-4060-85ff-4724bbfed648.usrfiles.com/ugd/5b88ae_f12953beac7e4fd9b6057375cce212b5.pdf [https://perma.cc/LL6M-KVGV]). We have no way to judge the accuracy of this statement.
Since a draft of this Article was posted online, Uber drivers have adopted the term “algorithmic wage discrimination,” testified to how it reflects how they are paid, and documented how they are offered different base pay for the exact same ride when sitting next to each other. See, e.g., The RideShare Guy, The Age of Algorithmic Wage Discrimination for Uber & Lyft Drivers and More?!, YouTube, at 2:16 (Apr. 16, 2023), https://www.youtube.com/watch?v=MfFujB0IY6A (on file with the Columbia Law Review); The RideShare Guy, MORE Algorithmic Wage Discrimination?? Show Me The Money Club, YouTube, at 6:25, 1:01:03 ( June 20, 2023), https://www.youtube.com/watch?v=8mwzsB41-f4 (on file with the Columbia Law Review).

The Blueprint example underscores how algorithmic wage discrimination can be “ineffective” and rife with calculated mistakes that are difficult to ascertain and correct. But algorithmic wage discrimination also creates a labor market in which people who are doing the same work, with the same skill, for the same company, at the same time may receive different hourly pay. 20 See infra Part II. Digitally personalized wages are often determined through obscure, complex systems that make it nearly impossible for workers to predict or understand their constantly changing, and frequently declining, compensation. 21 See infra Part II.

Drawing on anthropologist Karl Polanyi’s notion of embeddedness—the idea that social relations are embedded in economic systems 22 In 1957, Karl Polanyi wrote,
Instead of economy being embedded in social relations, social relations are embedded in the economic system. The vital importance of the economic factor to the existence of society precludes any other result. For once the economic system is organized in separate institutions, based on specific motives and conferring a special status, society must be shaped in such a manner as to allow that system to function according to its own laws.
Karl Polanyi, The Great Transformation: The Political and Economic Origins of Our Time 60 (Beacon Press 2001) (1944). One interpretation of this important excerpt, as used in this Article, is that Polanyi was referring to the ways in which society adapts to and reorganizes itself “by demanding new social institutions that can constrain market forces and compensate for market failures.” Bob Jessop & Ngai-Ling Sum, Polanyi: Classical Moral Economist or Pioneer Cultural Political Economist?, 44 Östereichische Zeitschrift für Soziologie 153, 158 (2019). This, in essence, is what he calls the “embedded economy”: that in order to prevent a “Hobbesian war of all against all,” a market society must limit—through law, politics, and morality—the range of legitimate activities of economic actors motivated by material gain. Fred Block, Karl Polanyi and the Writing of The Great Transformation, 32 Theory & Soc’y 275, 297 (2003).
—this Article excavate the norms around payment that constitute what one might consider a moral economy of work to help situate this contemporary rupture in wages. 23 Various disciplines, including political theory, anthropology, and sociology, have explored the notion of “moral economy” in relationship to labor and work as a way to think about and assess various systems of economic distribution and their impacts on everyday life. See, e.g., William Greider, The Soul of Capitalism 39 (2003) (“The logic of capitalism is ingeniously supple and complete, self-sustaining and forward-looking. Except for one large incapacity: As a matter of principle, it cannot take society’s interests into account.”); James Bernard Murphy, The Moral Economy of Labor 42 (1993) (applying moral reason to the social division of labor and technology); Sharon C. Bolton, Maeve Houlihan & Knut Laaser, Contingent Work and Its Contradictions: Towards a Moral Economy Framework, 111 J. Bus. Ethics 121, 123–124 (2012); Sharon C. Bolton & Knut Laaser, Work, Employment and Society Through the Lens of Moral Economy, 27 Work Emp. & Soc’y 508, 509 (2013) (using a moral economic approach in a sociological inquiry); Marion Fourcade, Philippe Steiner, Wolfgang Streeck & Cornelia Woll, Moral Categories in the Financial Crisis 2 (Max Planck Sciences Po Ctr. on Coping With Instability in Mkt. Societies (MaxPo) Discussion Paper, Working Paper No. 13/1, 2013), https://www.econstor.eu/bitstream/10419/104613/1/757489362.pdf [https://perma.cc/4ZJ4-QYC4] (analyzing the reconfiguration of the moral economy surrounding income inequality in France following the 2008 financial crisis). Although the United States–based system of work is largely regulated through contracts and strongly defers to the managerial prerogative, 24 See Gali Racabi, Abolish the Employer Prerogative, Unleash Work Law, 43 Berkeley J. Emp. & Lab. L. 79, 82 (2022) (“The employer [or managerial] prerogative is the default governance rule in the workplace . . . .”). This legal deference to the managerial prerogative is controversial in the scholarly literature. See, e.g., id. at 138 (“[P]erhaps the employer prerogative’s most sinister effect is convincing work law movements, scholars, and activists that it is a state of nature, a necessary theoretical benchmark for both pragmatic and normative discussions of work law. It is not.”). two restrictions on wages have emerged from social and labor movements: minimum-wage laws and antidiscrimination laws. Respectively, these laws set a price floor for the purchase of labor relative to time and prohibit identity-based discrimination in the terms, con-ditions, and privileges of employment, requiring firms to provide equal pay for equal work. 25 At the federal level, the Fair Labor Standards Act of 1938, 29 U.S.C. §§ 201–219 (2018), establishes a national floor for minimum-wage and overtime. Id. at §§ 203, 206, 207. The central federal laws that prohibit wage discrimination based on protected identities or classes are the Equal Pay Act, 29 U.S.C. § 206(d) (requiring that men and women in the same workplace be given equal pay for equal work); Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e-2 (2018) (prohibiting employment discrimination based on race, color, religion, sex, and national origin); the Age Discrimination in Employment Act, 29 U.S.C. §§ 623, 631 (prohibiting employment discrimination based on age for workers older than forty); and the Americans with Disabilities Act, 42 U.S.C. § 12112 (prohibiting employment discrimination based on disability). Both sets of wage laws can be understood as forming a core moral foundation for most work regulation in the United States. In turn, certain ideals of fairness have become embedded in cultural and legal expectations about work. Part I examines how recently passed laws in California and Washington State, which specifically legalize algorithmic wage discrimination for certain firms, compare with and destabilize more than a century of legal and social norms around fair pay.

Part II draws on first-of-its-kind, long-term ethnographic research to understand the everyday, grounded experience of workers earning through and experiencing algorithmic wage discrimination. Specifically, Part II analyzes the experiences of on-demand ride-hail drivers in California before and after the passage of an important industry-initiated law, Proposition 22, which legalized this form of variable pay. This Part illuminates workers’ experiences under compensation systems that make it difficult for them to predict and ascertain their hourly wages. Then, Part II examines the practice of algorithmic wage discrimination in rela-tionship to workers’ on-the-job meaning making and their moral interpretations of their wage experiences. 26 The social construction of meaning is a central concern of sociologists and anthropologists who seek to account for the variability and diversity of human understandings and experiences. Compare Michèle Lamont, Meaning-Making in Cultural Sociology: Broadening Our Agenda, 29 Contemp. Socio. 602, 603–05 (2000) (offering a detailed taxonomy of sociological literature that takes up how people make sense of their worlds through their experiences of race, ethnicity, immigration, and inequality), with Richard A. Posner, Economic Analysis of Law 3–4 (9th ed. 2014) (describing rationality as grounded within self-interested economic maximization of scarce resources). Though many drivers are attracted to on-demand work because they long to be free from the rigid scheduling structures of the Fordist work model, 27 Philosopher Antonio Gramsci used the term “Fordism” to refer to an emergent system of material production—routine, intensified labor—under the regime of Ford. But due in large part to corresponding political and economic forces, namely the laws and policies passed in response to upheaval during the Great Depression, the Fordist work structure in much of the mid-twentieth century often corresponded to an hourly (living) wage and a forty-hour work week. See Antonio Gramsci, Americanism and Fordism, in Selections From the Prison Notebooks of Antonio Gramsci 561, 56163 (Quentin Hoare & Geoffrey Nowell Smith eds. and trans., 1999). For more on the demise of Fordism, see generally Luc Boltanski & Ève Chiapello, The New Spirit of Capitalism (2007). they still largely conceptualize their labor through the lens of that model’s payment structure: the hourly wage. 28 See Michael Dunn, Making Gigs Work: Digital Platforms, Job Quality and Worker Motivations, 35 New Tech. Work & Emp. 232, 238–39, 241–42 (2020) (discussing the motivations of gig workers, including flexible work hours, despite often needing to maintain the same work structures as traditional employment). It should be noted that nothing about employment status necessitates an inflexible work schedule. This is a business decision associated with, not mandated by, employment. For a discussion of the history of businesses contesting the legal rules defining employment status to avoid legal responsibility for basic employment safeguards, see Veena B. Dubal, Wage Slave or Entrepreneur?: Contesting the Dualism of Legal Worker Identities, 105 Calif. L. Rev. 65, 86–88 (2017) [hereinafter Dubal, Wage Slave or Entrepreneur?]. Notably, the passage of California’s AB5 law made it much harder to misclassify workers in this way. See Hannah Johnston, Ozlem Ergun, Juliet Schor & Lidong Chen, Is Employment Status Compatible With the On-Demand Platform Economy? Evidence From a Natural Experiment 6 (2021) (unpublished report) (on file with the Columbia Law Review). When at least one labor platform company, called Bring Your Package, went on to hire their previously contracted workers in anticipation of AB5 restrictions, this transition did not precipitate any reduction in workers’ desired scheduling flexibility nor in firm efficiency. See id. at 14, 24, 26–27. Workers find that, in contrast to more standard wage dynamics, being directed by and paid through an app involves opacity, deception, and manipulation. 29 These findings comport with research findings from across sociology, communications studies, and media studies literatures on algorithmic management. See, e.g., Antonio Aloisi, Platform Work in Europe: Lessons Learned, Legal Developments and Challenges Ahead, 13 Euro. Lab. L.J. 4, 10–11 (2022) (discussing how platform manage-ment tends to unfold in misleading, opaque ways); Rafael Grohmann, Gabriel Pereira, Abel Guerra, Ludmila Costhek Abilio, Bruno Moreschi & Amanda Jurno, Platform Scams: Brazilian Workers’ Experiences of Dishonest and Uncertain Algorithmic Management, 24 New Media & Soc’y 1611, 1614 tbl.1 (2022) (presenting case studies of the types of dishonesty and deception that workers experience in platform work); Elke Schüßler, Will Attwood-Charles, Stefan Kirchner & Juliet B. Schor, Between Mutuality, Autonomy and Domination: Rethinking Digital Platforms as Contested Relational Structures, 19 Socio-Econ. Rev. 1217, 1224 (2021) (outlining common theories of the position of power that platforms hold over their workers); Steven Vallas & Juliet B. Schor, What Do Platforms Do? Understanding the Gig Economy, 46 Ann. Rev. Socio. 273, 279–81 (2020) (conducting a literature review of the predominant sociological views of platform work, which often conceptualize this work as an extension of existing neoliberal models of work without any of the worker protections); Daniel Susser, Beate Roessler & Helen Nissenbaum, Technology, Autonomy, and Manipulation, 8 Internet Pol’y Rev., no. 2, 2019, at 1, 8 (explaining how gig economy services covertly influence an individual’s decision-making through “online manipulation”). Those who are most economically dependent on income from on-demand work frequently describe their experience of algorithmic wage discrimination through the lens of gambling. 30 See infra section II.B. As a normative matter, this Article contends that workers laboring for firms (especially large, well-financed ones like Uber, Lyft, and Amazon) should not be subject to the kind of risk and uncertainty associated with gambling as a condition of their work. In addition to the salient constraints on autonomy and threats to privacy that accompany the rise of on-the-job data collection, algorithmic wage discrimination poses significant problems for worker mobility, worker security, and worker collectivity, both on the job and outside of it. Because the on-demand workforces that are remunerated through algorithmic wage discrimination are primarily made up of immigrants and racial minority workers, these harmful economic impacts are also necessarily racialized. 31 In the United States, such work is conducted primarily by immigrants and subordinated minorities. Lyft estimates that 73% of their U.S. workforce identify as racial minorities. Lyft, Economic Impact Report 5 (2022), https://s27.q4cdn.com/263799617/
files/doc_downloads/esg/Lyft-Economic-Impact-Report-2022.pdf [https://perma.cc/8BUG-NGAV]. One study estimates that in the San Francisco Bay Area in 2019, immigrants and people of color composed 78% of Uber and Lyft drivers, most of whom relied on these jobs as their primary source of income. Chris Benner, Erin Johansson, Kung Feng & Hays Witt, UC Santa Cruz Inst. for Soc. Transformation, On-Demand and On-The-Edge: Ride-Hailing and Delivery Workers in San Francisco, Executive Summary 2 (2020), https://transform.ucsc.edu/wpcontent/uploads/2020/05/OnDemandOntheEdge_ExecSum.pdf [https://perma.cc/DFH8-7VSY]. In addition to the nationwide Lyft data, we know that in New York City, 90% of ride-hail drivers are immigrants, and in Seattle, ride-hail drivers are 50% Black and “nearly three times more likely to be immigrants than all Kings County workers.” James A. Parrott & Michael Reich, Ctr. on Wage & Emp. Dynamics & New Sch. Ctr. for N.Y.C. Affs., A Minimum Compensation Standard for Seattle TNC Drivers 23 (2020), https://irle.berkeley.edu/files/2020/07/Parrott-Reich-Seattle-Report_July-2020.pdf [https://perma.cc/QA9F-FV47] [hereinafter Parrott & Reich, Minimum Compensation Standard]; Ginia Bellafante, Uber and the False Hopes of the Sharing Economy, N.Y. Times (Aug. 9, 2018), https://www.nytimes.com/2018/08/09/nyregion/uber-nyc-vote-drivers-ride-sharing.html (on file with the Columbia Law Review).

Finally, Part III explores how workers and worker advocates have used existing data privacy laws and cooperative frameworks to address or at least to minimize the harms of algorithmic wage discrimination. In addition to mobilizing against violations of minimum-wage, overtime, and vehicle reimbursement laws, workers in California—drawing on the knowledge and experience of their coworkers in the United Kingdom—have developed a sophisticated understanding of the laws governing data at work. 32 See infra Part III. In the United Kingdom, a self-organized group of drivers, the App Drivers & Couriers Union, has not only successfully sued Uber to establish their worker status 33 Kate Duffy & Theo Golden, Uber Just Lost a Major Legal Battle Over Whether Its UK Drivers Count as Workers and Are Entitled to Minimum Wage, Bus. Insider (Feb. 19, 2021), https://www.businessinsider.com/uber-driver-lost-uk-legal-battle-court-worker-rights
-employment-2021-2 [https://perma.cc/CT27-K2ZP].
but also used the General Data Protection Regulation (GDPR) to lay claim to a set of positive rights concerning the data and algorithms that determine their pay. 34 Jeffrey Brown, In New European Lawsuit, Uber Drivers Claim Company’s Algorithm Fired Them, Geo. L. Tech. Rev. Legal Impressions (Nov. 2020), https://georgetownlawtechreview.org/in-new-european-lawsuit-uber-drivers-claim-companys-algorithm-fired-them/GLTR-11-2020/ [https://perma.cc/887P-RET4] (“The GDPR . . . imposes obligations on companies which collect personal information if that data is related to EU consumers, regardless of the consumer’s physical location in the world. Under Article 22, individuals have ‘the right not to be subject to a decision based solely on automated processing.’” (quoting Council Regulation 2016/679, art. 22, 2016 O.J. (L 119) 1 (EU))). As a GDPR-like law went into effect in California in 2023, drivers there are positioned to do the same. 35 Cal. Civ. Code § 1798.100 (2023) (imposing limits on businesses’ collection of consumer personal information and requiring notice of the purposes behind data collection). Other workers in both the United States and Europe have responded by creating “data cooperatives” to fashion some transparency around the data extracted from their labor, to attempt to understand their wages, and to assert ownership over the data they collect at work. 36 See infra section III.B. In addition to examining both approaches to addressing algorithmic wage discrim-ination, this Article argues that the constantly changing nature of machine learning technologies and the asymmetrical power dynamics of the digitalized workplace minimize the impact of these attempts at trans-parency and may not mitigate the objective or subjective harms of algorithmic wage discrimination. Considering the potential for this form of discrimination to spread into other sectors of work, this Article proposes instead an approach that addresses the harms directly: a narrowly structured, nonwaivable peremptory ban on the practice.

While this Article is focused on algorithmic wage discrimination as a labor management practice in “on-demand” or “gig work” sectors, where workers are commonly treated as “independent contractors” without protections, its significance is not limited to that domain. So long as this practice does not run afoul of minimum-wage or antidiscrimination laws, nothing in the laws of work makes this form of digitalized variable pay illegal. 37 See supra note 25. Antitrust laws, however, are a more promising way to address these practices when and if workers are classified as independent contractors. Part III discusses a California lawsuit filed in 2022 by Rideshare Drivers United workers against Uber alleging that the company’s payment structures amount to price fixing and that it is violating state antifraud laws. As Professor Zephyr Teachout argues, “Uber drivers’ experiences should be understood not as a unique feature of contract work, but as a preview of a new form of wage setting for large employers . . . .” 38 See Teachout, Algorithmic Personalized Wages, supra note 15, at 437. The core motivations of labor platform firms to adopt algorithmic wage discrimination—labor control and wage uncertainty—apply to many other forms of work. Indeed, extant evidence suggests that algorithmic wage discrimination has already seeped into the healthcare and engineering sectors, impacting how porters, nurses, and nurse practitioners are paid. 39 For example, a company that brands itself “Uber for Hospitals” has developed AI staffing software for hospitals. This software uses “smart technology” to allocate work tasks and to judge the performance of porters, nurses, and nurse practitioners. See Nicky Godding, Oxford Tech Raises £9 Million for ‘Uber for Hospitals’ AI Platform, Bus. Mag. (May 21, 2020), https://thebusinessmagazine.co.uk/technology-innovation/oxford-tech-raises-9-million-for-uber-for-hospitals-ai-platform/ [https://perma.cc/8593-M9U7] (“Hospitals can use [this technology] to assign tasks to healthcare teams based on their location. . . . This helps to ensure . . . full visibility of vulnerable patient movement between departments, and connects porters directly with staff . . . .”). The technology company’s “performance analysis” may then be used to determine the pay for these healthcare workers. Id.
IBM Japan is also using digital surveillance systems to help set wages for their workers. In 2019, the company introduced human relations software created by Watson to use as a “compensation advisor.” The Japan Metal, Manufacturing, Information and Telecommunication Workers’ Union ( JMITU), which represents IBM Japan workers, requested disclosure of the data the Watson AI acquired and used, an explanation for how it was evaluating workers, and how these evaluations were involved in the wage-setting process. IBM Japan refused to disclose the information. JMITU subsequently lodged a complaint with the Tokyo Labor Relations Commission. The union argues that the software is being used to unfairly target union members. According to one report, “[i]n awarding summer bonuses in June 2019, the individual performance rate assessed by the company was only 63.6% on average for union members, compared to an average of 100% for all [other] employees. In addition, an exceptional 0% assessment was made for many union members.” Hozumi Masashi (ほづみ まさし), AIによる賃金査定にどう向き合うか: 日本IBM事件(不当労働行為救済申立) の報告 [How to Face AI-Based Wage Assessments: Report on the IBM Japan Case (Unfair Labor Practice Relief Petition)], 338 季刊 労·働者の権利[Worker Rights Quarterly], no. 10, 2020, at 101, 102.
If left unaddressed, the practice will continue to be normalized in other employment sectors, including retail, restaurant, and computer science, producing new cultural norms around compensation for low-wage work. 40 See, e.g., Min Kyung Lee, Daniel Kusbit, Evan Metsky & Laura Dabbish, Working With Machines: The Impact of Algorithmic and Data-Driven Management on Workers, in CHI 15: Proceedings of the 33rd Annual CHI Conference on Human Factors in Computing Systems 1603, 1603–04 (2015) (discussing how algorithms used across industries can produce new norms of allocation of, evaluation of, and compensation for work). Companies across the world use wage algorithms in both contracting and permanent employment settings to incentivize certain behaviors. Technology capitalists have foreshadowed its growth. See, e.g., Shawn Carolan, Opinion, What Proposition 22 Now Makes Possible, The Info. (Nov. 10, 2020), https://www.theinformation.com/articles/what-proposition-22-now-makes-possible (on file with the Columbia Law Review) (predicting increased venture capitalist investment in “all sorts of industries” after the passage of Proposition 22). As Tarleton Gillespie has warned regarding the power of algorithms, “[t]here is a case to be made that the working logics of these algorithms not only shape user practices, but also lead users to internalize their norms and priorities.” Tarleton Gillespie, The Relevance of Algorithms, in Media Technologies: Essays on Communication, Materiality, and Society 167, 187 (Tarleton Gillespie, Pablo J. Boczkowski & Kirsten A. Foot eds., 2014). The on-demand sector thus serves as an important and portentous site of forthcoming conflict over longstanding moral and political ideas about work and wages.