Uber's Gig Economy: Psychological Manipulation of Drivers

Labor law faces challenges in regulating platform work because traditional collective bargaining cannot address algorithmic control. Negotiations must now include access to algorithms, not just traditional workplace issues like hours, health risks, and pay. This represents a fundamental challenge for labor law and unions in the digital age. Three potential solutions exist: traditional labor protection through unionization, platform cooperativism where workers become platform owners, and collective ownership of user data generated on platforms. These approaches must be pursued together to address platform labor precarization.

Workers are developing strategies to resist AI impacts through collective organizing. The NBA players successfully negotiated clauses preventing AI performance data from salary negotiations. Hollywood writers' unions secured human supervision in AI-assisted scriptwriting. Platform cooperativism offers an alternative, with worker-owned platforms like Senhoritas Currier in Brazil building their own systems for fairer distribution. The concept of 'algorithmic design justice' involves worker participation in programming and implementation processes.

Algorithms fundamentally control worker hiring, evaluation, and remuneration, determining work allocation and performance ratings across all platform types. Workers report that ratings are decisive for accessing work opportunities, and frequent rejections lead to account deactivation. Despite apparent flexibility, workers are systematically monitored through screenshot requirements and mandatory availability periods, revealing employer-employee dynamics. Platform cooperatives offer alternatives: Smart cooperative's agreement with Take It Easy ensured minimum wages, safety training, and worker compensation upon bankruptcy, though regulatory changes later disrupted this model. Up and Go demonstrates success for domestic workers, with cleaners earning $4-5 more weekly while maintaining only 5% platform fees. These examples show cooperatives can provide better outcomes through collective ownership.

Platform work represents a fundamental transformation of labor relations where traditional distinctions between enterprise and market, subordination and autonomy, and professional and personal life are increasingly blurred through algorithmic management. Unlike traditional salaried work with legal subordination, platform work creates a paradoxical condition where workers appear as independent contractors but face economic dependency and psychological subordination through algorithmic control. This transformation raises critical questions about worker rights, social protections, and the future of employment. In response, platform cooperativism emerges as an alternative model that seeks to democratize digital platforms by enabling workers to collectively own and govern the technology, thereby transforming platforms from instruments of capital accumulation into tools of worker empowerment and democratic control.

Digital platforms represent a new form of labor exploitation similar to historical patterns but potentially more dehumanizing because algorithmic management lacks human judgment. The speaker argues that platform work should be recognized as employment rather than independent contracting. The future of work requires collective bargaining rather than individual negotiations because individual workers cannot achieve equitable relationships with employers regardless of their skills. Collective action provides the power balance necessary for fair labor relations, similar to historical patterns of labor organization.

Algorithmic audits are structured around three main pillars: (1) Technical aspects related to servers and site development, (2) Content aspects including relevance to user needs and information quality, and (3) Popularity aspects examining external links and user behavior. The audit methodology involves three key steps: crawling the site to collect data using various tools, analyzing the collected data objectively, and visiting the site only after data analysis to avoid confirmation bias. Modern audits use slide-based presentations (80-200 slides) with each task assigned to specific team members with deadlines.

Algorithmic auditing is a process to evaluate automated decision systems for regulatory compliance, transparency, bias, and harm, but current practices show significant gaps: most auditors use quantitative methods (70%+) over qualitative approaches (40%), custom-built tools dominate (93%), and public disclosure of audit results is rare despite 82% of auditors supporting mandatory disclosure. Key barriers include lack of organizational buy-in due to liability concerns and high audit costs. The research recommends six policy interventions: requiring independent audits against defined standards, mandating individual notification for algorithmic decisions, requiring disclosure of key audit components for peer review, incorporating real-world harm reporting, prioritizing stakeholder involvement, and establishing formalized evaluation processes for auditors.

Algorithmic auditing is no longer academic due to documented discriminatory impacts. The PPM (Purpose, Process, Monitoring) framework answers three questions: What problem is being solved and how data represents it? How are teams making decisions about data, features, and models? When should monitoring occur since business environments are dynamic? This framework helps identify regulatory, legal, financial, and reputational risks while balancing algorithmic benefits.

AI audits can be understood through a two-axis framework plotting efforts along dimensions of scope (from exploratory to targeted) and independence (from internal to third-party), revealing that no single assessment approach can accomplish all governance goals and that different audit types serve distinct purposes in the ecosystem of accountability.

The White House released a document on big data and civil rights, naming algorithmic auditing as a national priority. Design is proposed as a solution to algorithmic awareness and bias issues. Designers have long faced ethical questions about harm caused by design versus expected benefits. Frameworks for ethical design can be adapted to the online realm, but new approaches are needed. The design approach is in its baby stages and requires collaboration between designers, researchers, and policymakers to address algorithmic bias and discrimination.

Platform companies have invested heavily in developing algorithms that control the work of platform workers, dictating rhythms, times, and movements. These algorithms are designed to maximize extraction of surplus labor while minimizing costs. The Amazon case illustrates this: workers walk 24-25 kilometers per day, pack 200 products per hour, work 55 hours per week, and up to 10 hours per day during peak periods. Amazon stores in the United States have no workers walking through them, as all purchasing is digitalized, yet fulfillment centers require intensive manual labor.

Warehousing has been an unregulated laboratory for capital innovation in worker control through surveillance and algorithmic management. Surveillance focuses on tracking products and boxes rather than workers, with evaluation based solely on goods moved at company-determined rates. Amazon represents a paradigm shift through comprehensive camera networks tracking worker behavior and movement patterns. Workers report skipping bathroom breaks, working through injuries, and facing unpredictable quota increases. This warehouse model is being replicated across industries, making it critical to stop algorithmic exploitation in this sector before it spreads to other workplaces.

Amazon's algorithms constantly monitor delivery workers' performance, measuring their time and efficiency. Workers can be fired automatically by the algorithm if they accumulate too many delivery delays, without human oversight. This system creates a high-pressure environment where workers are treated as data points rather than individuals, with their livelihoods determined by algorithmic decisions rather than human judgment.

The shipbuilding industry provides a specific case study of Taylorism's application, with the transition from coal-fired steam engines to diesel turbines between the 1920s-1930s reducing crew sizes and creating opportunities for new work organization techniques. Taylorist principles have been applied across diverse sectors: in the shoe industry, production was timed and standardized with some work done at home; in the spa industry, contracts specify exact times for client procedures. The Just-in-Time production model (1990s-2000s) eliminated inventory costs by synchronizing production with demand, enabling outsourcing to lower-cost locations. Amazon's warehouse model represents a contemporary form of Taylorism through algorithmic control, using large urban warehouses to shorten supply chains. Workers face strict time controls with contracts specifying exact operation times. The model relies on digitalization to manage logistics and workforce, creating a form of control combining Taylorist time measurement with technological surveillance.

Warehouse management systems (WMS) use algorithms to optimize worker movements and productivity. These systems tell workers exactly where to go and how fast to move, creating a form of control that removes worker autonomy. The systems are designed for robots but are applied to human workers, creating a form of alienation where workers have no control over their own labor process.