Abstract:The legitimacy of bottom-up democratic processes for the distribution of public funds by policy-makers is challenging and complex. Participatory budgeting is such a process, where voting outcomes may not always be fair or inclusive. Deliberation for which project ideas to put for voting and choose for implementation lack systematization and do not scale. This paper addresses these grand challenges by introducing a novel and legitimate iterative consensus-based participatory budgeting process. Consensus is designed to be a result of decision support via an innovative multi-agent reinforcement learning approach. Voters are assisted to interact with each other to make viable compromises. Extensive experimental evaluation with real-world participatory budgeting data from Poland reveal striking findings: Consensus is reachable, efficient and robust. Compromise is required, which is though comparable to the one of existing voting aggregation methods that promote fairness and inclusion without though attaining consensus.
What problem does this paper attempt to address?
### Problems Addressed by the Paper
The paper primarily explores the issues of fairness and representativeness in the Participatory Budgeting (PB) process and proposes a new method based on Multi-Agent Reinforcement Learning (MARL) to address these issues.
#### Main Issues
1. **Fairness and Representativeness in the Participatory Budgeting Process**:
- Current participatory budgeting voting results may not fully be fair or inclusive of everyone's interests.
- There is a lack of systematic methods in the project selection phase, making it difficult to scale.
- Citizens' preferences during voting may be based on personal interests, leading to varying satisfaction levels, insufficient representativeness, and lower legitimacy of the voting results.
2. **Lack of Systematic and Scalable Discussion Mechanisms**:
- There is a lack of a systematic and scalable discussion mechanism that allows citizens to exchange, debate, and compromise to reach a consensus.
#### Solutions
The paper introduces a new method based on Multi-Agent Reinforcement Learning (MARL-PB) to address these issues through the following ways:
1. **Consensus-Driven Participatory Budgeting Process**:
- An iterative consensus-driven participatory budgeting process is designed, where consensus is derived through an innovative multi-agent reinforcement learning method for decision support.
- It supports interactions among citizens, helping them make feasible compromises.
2. **Decision Support Framework**:
- A decision support framework is provided, applicable to three different application scenarios and beneficiaries: citizens, policymakers, and researchers.
- Citizens: Digitally assisted communication, discussion, and reaching common opinions, improving participation, satisfaction, and legitimacy.
- Policymakers: Digitally assisted project screening, ensuring rationality and legitimacy in the voting process.
- Researchers: Evaluating fair and inclusive voting aggregation methods and comparing them with consensus-based models.
3. **Experimental Evaluation**:
- Extensive experimental evaluation of MARL-PB using real-world datasets from Poland.
- The quality, efficiency, and robustness of the consensus were evaluated, as well as how these factors are affected by various factors such as the number of projects, the in-degree of the communication network, the number of voters, etc.
#### Main Contributions
1. Proposing an iterative consensus-driven participatory budgeting process based on multi-agent reinforcement learning.
2. Designing a decision support framework applicable to three different application scenarios.
3. Extending reward learning strategies by incorporating decentralized information exchange protocols.
4. Proposing a series of metrics to evaluate the legitimacy of the consensus.
5. Experimental results show that the compromises required for consensus are comparable to existing voting aggregation methods that promote fairness and inclusiveness.
Through the above methods, the paper aims to address the issues of fairness and representativeness in the participatory budgeting process and proposes a new method to achieve a more systematic, scalable, and automated discussion process.