Decline Now: A Combinatorial Model for Algorithmic Collective Action

Research Background and Issues

• Issues and Challenges: Workers on delivery platforms (e.g., DoorDash) often incur losses due to low-paying orders, especially under algorithmic management, which places them in a work environment characterized by wage opacity, high stress, and significant uncertainty. The low compensation and the lack of explainability in algorithmic decisions exacerbate workers' sense of powerlessness.
• Significance:
  • Delivery platforms and algorithmic management are transforming labor markets globally, affecting the living conditions of millions of workers.
  • To address this situation, workers need effective resistance strategies, which involve collective action against algorithmic management. However, systematic research on the effectiveness and limitations of such strategies remains scarce.
• Research Motivation and Related Work:
  • Using the #DeclineNow campaign initiated by DoorDash workers as an example, this study examines the effectiveness and impact of a collective order rejection strategy.
  • Existing literature primarily explores the impact of algorithms on workers or the resistance potential of technology itself, with limited focus on strategy design and the cost-benefit analysis for participants.
  • The authors propose a combined model to analyze the strategic interactions between workers and platforms, filling a gap in the research.

Solution

• Methods and Innovations:

  • A combined model is proposed to describe how workers interact with platforms through an order rejection strategy.
  • The model's innovation lies in modeling the dynamic relationships between collective benefits, free-rider gains, and participation cost-effectiveness, while analyzing the impact of supply-demand levels on strategy effectiveness.
  • A theoretical framework is proposed, combined with simulation experiments, to reveal the critical role of supply-demand conditions and participation scale in the success of collective actions.

• Implementation Steps and Key Techniques:

  1. Model Assumptions:
     • A linear wage increase mechanism exists, where unaccepted orders are redistributed with a certain increase in compensation.
     • Workers are divided into participants (using the rejection strategy) and non-participants (accepting any work).
     • The platform allocates orders randomly.
  2. Core Definitions:
     • Utility Function: Measures the average earnings of workers under specific market conditions.
     • Spillover Effect: The additional earnings non-participants receive due to participants rejecting orders.
  3. Supply Conditions: Defines different market supply-demand relationships based on labor surplus or shortage (measured by the ratio of orders to workers).
  4. Simulation Experiments:
     • Based on real-world data (e.g., order volume, average compensation), an experimental environment is established to explore workers' utility functions under varying participation ratios and supply-demand conditions.

Research Findings

• Specific Findings:

  • The overall benefit of the collective order rejection strategy is consistently positive, meaning the average income of workers under the strategy is always higher than without it.
  • In cases of labor shortage, participants consistently earn more than non-participants, and free-riding does not occur.
  • In labor surplus scenarios, non-participants may benefit more due to spillover effects, while participants' earnings are somewhat constrained.
  • A shift work strategy is proposed to reduce the number of workers online simultaneously, alleviating labor surplus and enhancing participants' bargaining power.

• Experimental or Evaluation Results:

  • Simulation results show:
    1. Under slight labor surplus, non-participants benefit significantly from spillover effects.
    2. When surplus levels are high, participants' earnings decline more rapidly, potentially reducing the strategy's attractiveness.
    3. Shift strategies (e.g., scheduling workers in time slots) can significantly mitigate labor surplus and enhance overall utility.

• Comparison with Existing Solutions:

  • Unlike studies focusing on machine learning algorithms, this model emphasizes resistance strategies under fixed allocation rules. Although the strategy's significance is relatively modest, it remains actionable and feasible under appropriate conditions.
  • The innovation of this study lies in treating supply-demand conditions as key parameters for the success of collective actions, offering a data-driven approach to labor collective action.

• Limitations and Future Directions:

  • Limitations:
    • Platform algorithms are often opaque, and the actual ways to counteract worker strategies remain underexplored.
    • Insufficient temporal analysis of variables such as workers' willingness to participate and market supply-demand dynamics.
    • The model construction may underestimate the potential impact of algorithm adjustments on strategy effectiveness.
  • Future Directions:
    1. Incorporate dynamic response models for platform countermeasures.
    2. Explore social behavior mechanisms in worker cooperation and participation dynamics.
    3. Design practical tools, such as collaborative apps, to enable more workers to coordinate actions.
    4. Investigate how platforms can establish allocation mechanisms that balance worker rights and customer experience.

Conclusion

This study, through theoretical modeling and simulation validation, provides a framework and practical significance for designing strategies for labor collective action, particularly in environments of labor surplus. The model effectively reveals micro-level dynamics in labor markets and offers profound insights for improving the strategic collective actions of delivery platform workers.