Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

Digital Pantheon: Simulating and Auditing Coalition Formation with LLM Agents

Source

arXiv

arxiv.org/abs/2607.15095v1

Read the source

Source

Plain-Language Brief

Researchers have developed a multi-agent framework using Large Language Models (LLMs) to simulate political coalition formation. The framework combines Supervised Fine-Tuning, Direct Preference Optimization, and Retrieval-Augmented Generation to create party-specific personas that adhere to their official manifestos. The study uses this framework to analyze the 2019 Flemish election and introduces tools for making the negotiation interpretable. Results show a stable winner and ranking, and the proposed method accurately predicts materialization of policy provisions compared to hallucinated content.

What the Document Says

  • The researchers present a multi-agent framework for political coalition formation using LLMs.
  • The framework combines Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Retrieval-Augmented Generation (RAG).
  • DPO instills aggressive party-specific personas, while RAG keeps each agent bounded to its official manifesto.
  • The framework was operationalized on the 2019 Flemish election.
  • A hub-and-spoke negotiation was arbitrated by a formateur.
  • A Multi-Layered Information Lineage Topology (MILT) was introduced to trace every clause in the final agreement back to its manifesto origin and classify it into five provenance states.
  • A Coalition Influence Score (CIS) was also introduced to aggregate contributions from each party and identify which shaped the agreement.
  • The study simulated negotiations three times, yielding a stable winner and ranking (N-VA ahead of CD&V and Open Vld).
  • Manifesto-anchored lineage reliably predicts real-world materialization whereas hallucinated content does not.

Key Concepts Mentioned

  • political coalition formation
  • Large Language Models (LLMs)
  • Supervised Fine-Tuning (SFT)
  • Direct Preference Optimization (DPO)
  • Retrieval-Augmented Generation (RAG)
  • 2019 Flemish election
  • hub-and-spoke negotiation
  • Multi-Layered Information Lineage Topology (MILT)
  • Coalition Influence Score (CIS)

Claims Made by the Document

  • The proposed framework yields a stable winner and ranking.
  • Manifesto-anchored lineage reliably predicts real-world materialization.

Evidence Used

  • Across three independent simulations the framework yields a stable winner and ranking (N-VA ahead of CD&V and Open Vld).
  • The result is a transparent, scalable testbed for the ex-ante exploration of party compatibility and formateur-mediated compromise.
  • Manifesto-anchored lineage reliably predicts real-world materialization whereas hallucinated content does not.

Methods / Approach

  • multi-agent framework
  • Supervised Fine-Tuning (SFT)
  • Direct Preference Optimization (DPO)
  • Retrieval-Augmented Generation (RAG)

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

The formation of political coalitions is a complex negotiation driven by both concrete policy objectives and deep-seated ideological convictions. While Large Language Models (LLMs) open new avenues for computational political science, the neutrality and helpfulness biases instilled by Reinforcement Learning from Human Feedback (RLHF) prevent them from sustaining steadfast partisan behaviour. We present a multi-agent framework that reconciles factual grounding with ideological alignment by combining Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Retrieval-Augmented Generation (RAG): DPO instils aggressive party-specific personas, while a per-party RAG pipeline keeps each agent bounded to its official manifesto. We operationalize the framework on the 2019 Flemish election, deploying the partisan agents in a hub-and-spoke negotiation arbitrated by a formateur. To make the emergent negotiation interpretable, we introduce a Multi-Layered Information Lineage Topology (MILT) that traces every clause in the final agreement back to its manifesto origin and classifies it into five provenance states, a Coalition Influence Score (CIS) that aggregates these traceable contributions to identify which party shaped the agreement, and a real-world grounding pass that benchmarks each simulated provision against the historically adopted coalition agreement. Across three independent simulations the framework yields a stable winner and ranking (N-VA ahead of CD&V and Open Vld), and manifesto-anchored lineage reliably predicts real-world materialization whereas hallucinated content does not. The result is a transparent, scalable testbed for the ex-ante exploration of party compatibility and formateur-mediated compromise.