Navigating Trade-offs in Algorithmic Governance: Agent-Based Simulation of Ride-Hailing Platform

This study examines fundamental trade-offs in algorithmic platform governance through an agent-based simulation of ride-hailing platforms. While platforms increasingly rely on algorithms to manage operations, the opacity of these systems and their impact on stakeholder welfare remain poorly understood. Using a Markov Decision Process framework, we model and compare three algorithmic strategies: profit maximization, driver earnings maximization, and passenger waiting time minimization. Our analysis reveals several key insights. First, extreme profit maximization may be unnecessary, as algorithms balancing multiple stakeholder interests can achieve 98.1% of maximum profit while delivering superior service quality. Second, focusing solely on driver earnings can have unintended consequences, leading to higher income inequality (Gini coefficient 0.39) compared to profit-focused (0.29) and waiting time-focused (0.31) algorithms. Third, service quality optimization provides the most stable foundation for platform operations, benefiting all stakeholders. These findings persist across various parameter settings, suggesting they reflect inherent platform dynamics rather than specific conditions. Our results contribute to platform governance theory and practice by demonstrating how algorithmic choices shape value distribution among stakeholders. We propose policy recommendations for promoting transparency, fairness, and sustainability in algorithmic platform management.