Welfarist Control Design

Theory Pathfinder Workshop
14 April 2026

SOCIETAL MANDATE

Society entrusts control engineers to develop methods with specs on 

• efficiency / performance
• fairness
• priorities / guarantees

ENGINEERS

Design control / optimization / decision rules that fulfil the mandate despite

• dynamics and feedback
• stochasticity
• heterogeneous systems
• limited information
• complexity

CERTIFICATES

Individuals and stakeholders can

• access certificates
• test fulfilment of mandate
• obtain explanations
• prequalify solutions ("grid code")

Theory for measurement of social mandate

Theory for certification

Where are we?

scarcity ⮞ tradeoffs ⮞ comparability

Five directions for future research:

1. Social Choice Theory & Axiomatic Frameworks
2. Game Theory & Mechanism Design
3. Optimization & Control Engineering
4. Data-Driven Preference Modeling
5. Explainability & Certification Theory

1. Social Choice Theory & Axiomatic Frameworks

Focus on the "Theory for measurement of social mandate": translating high-level societal values (fairness, priorities) into formal mathematical structures.
 

• Link to Algorithmic Fairness: Connecting algorithmic fairness axioms with social choice axioms.
   
• Dynamic Welfarist Control: Developing the theory to extend social choice to dynamic settings (long-term fairness, ex-ante/ex-post, closed-loop fairness)

#work-in-progress

#work-in-progress

#paper

2. Game Theory & Mechanism Design

Focus on the strategic behaviors in heterogeneous systems.
 

• Equilibrium Analysis: Refining "inefficiency in games" (Price of Anarchy) and Generalized Nash Equilibria for general SCFs.
  
   
• Incentive Compatibility: Designing mechanisms, such as "karma economies", where agents are incentivized to reveal their true priorities/urgencies to achieve a desired SCF in the long run.

#work-in-progress

#paper

#work-in-progress

3. Optimization & Control Engineering

Focus on the implementation of welfarist decision rules that must function despite feedback, dynamics, and constraints.
 

• Economic MPC: Integrating general SCFs into Economic Model Predictive Control.
   
• Online Optimization: Optimize SCFs defined on stochastic trajectories (breaks Markovianity in MDPs).
   
• Co-Design: Integrate societal needs directly into the engineering system design process.
   
• Applications: Fair power curtialment, transportation, train re-scheduling, and more.
   

#work-in-progress

#paper

4. Data-Driven Preference learning

Focus on the "Limited Information" challenge: using data to infer societal or individual preferences.
 

• Preference Learning: Developing theories for learning perceived fairness and individual preferences.
   
• SCF Parameter Tuning: Using data to tune SCF parameters and validate comparability assumptions.
   
• Bounded comparability: Dealing with cases where uncertainty makes it difficult but not impossible to compare utilities.

#work-in-progress

5. Explainability & Certification Theory

Focus on the "Theory for certification", allowing society to test and obtain explanations for the engineered solutions.

• Transparency Tools: Developing theories for "Control/optimization explainability" and counterfactual reasoning to show why certain decisions were made.

#work-in-progress

1. Social Choice Theory & Axiomatic Frameworks
   • Link to Algorithmic Fairness
   • Dynamic Welfarist Control
2. Game Theory & Mechanism Design
   • Equilibrium Analysis
   • Incentive Compatibility
3. Optimization & Control Engineering
   • Economic MPC
   • Online Optimization
   • Co-Design
   • Applications
4. Data-Driven Preference Learning
   • Preference Learning
   • SCF Parameter Tuning
   • Bounded comparability
5. Explainability & Certification Theory
   • Transparency Tools