Expressiveness, Cost, and Collectivism: How the Design of Preference Languages Shapes Participation in Algorithmic Decision-Making

Title of the Paper

Expressiveness, Cost, and Collectivism: How the Design of Preference Languages Shapes Participation in Algorithmic Decision-Making

Paper Information

• Subject Area: Algorithmic decision-making, preference language design, and participatory design methods in public service applications
• Keywords: Participatory design, preference languages, algorithmic systems, school allocation systems, creative participation, procedural justice, preference aggregation, social choice theory, individualism, distributional fairness

Research Background and Problem Statement

• Identified Issues or Challenges:
  • The design of preference languages directly impacts participants' ability to express their needs and goals, which in turn affects the fairness and effectiveness of algorithmic decision-making.
  • In school allocation systems, some families are unable to effectively express their needs due to resource and information inequalities, leading to inequitable distribution of educational resources.
  • Collective goals (e.g., promoting social equity and educational equality) are difficult to achieve through the aggregation of individual preferences.
• Why It Matters:
  • Algorithmic systems are widely applied in fields such as education, healthcare, and employment, where their fairness and inclusivity are critical to the allocation of key public resources.
  • Existing preference language designs face challenges related to time and resource costs, which may exacerbate inequalities among participants.
• Research Motivation and Related Work:
  • Based on studies conducted in two U.S. public school districts, the authors explore how to optimize preference language design to enhance participants' opportunities to express their needs.
  • Grounded in social choice theory, the research focuses on three core attributes of preference languages: expressiveness, cost, and collectivism.

Proposed Solution

• Methods or Solutions:
  • Propose three pathways to improve preference language design: enhancing basic options, simplifying preference language structures, and providing support to reduce participation costs.
  • Leverage procedural justice theory to explore how participants can collaboratively define collective goals and integrate them into algorithmic design.
  • Suggest using dynamic priority scoring and mathematical programming techniques to achieve collective goals while maintaining flexibility.
• Innovative Aspects:
  • The design of preference languages not only emphasizes participants' expression but also examines the embedded community values and fairness objectives within the system.
  • Applying procedural justice theory makes the participatory process more legitimate, transparent, and equitable.
• Implementation Steps and Techniques:
  1. Investigate the needs and priority expressions of participants and analyze the shortcomings of existing language systems.
  2. Provide choice menus to reduce information asymmetry and recommend matching algorithms based on dynamic priority scoring.
  3. Establish collaborative participation mechanisms to jointly define collective goals and optimize allocation algorithms through mathematical programming.

Research Outcomes

• Specific Findings:
  • Clarified how the attributes of preference languages influence participants' ability to express themselves and the fairness of decision-making.
  • Proposed a series of technical and social interventions to improve the design of preference languages.
  • Highlighted that choice menus and personalized support can effectively balance the costs for participants from diverse backgrounds.
• Advantages Over Existing Solutions:
  • Greater focus on participants' actual needs and social equity objectives, rather than solely satisfying individual preferences.
  • Provides deeper pathways to address algorithmic inequalities by improving language and process design to foster broader participation.
• Experimental and Evaluation Results:
  • Field studies revealed that low-income families provided significantly different feedback on system tiered costs compared to high-income participants.
  • The system framework demonstrated reduced participation barriers, though its performance on collective goals still requires optimization.
• Limitations and Future Directions:
  • Preference languages cannot fully encompass all needs; the conflict between collective and individual interests requires further study.
  • Preference languages themselves may conflict with social distribution goals, necessitating the design of new democratic interaction processes.
  • Future work should consider non-algorithmic alternatives, such as direct redistribution of educational resources or close community collaboration.

Through an in-depth analysis, this paper not only provides a theoretical and technical framework for preference language design but also inspires further reflection on the equitable distribution of educational resources.