Conversational Game Theory (CGT): A Collaborative Consensus-Building Game

©2024 9×3 Narrative Logic, LLC

Conversational Game Theory (CGT) is a novel conversational game and computational system designed to foster collaborative consensus-building around community content creation while filtering misinformation, misunderstanding, and toxicity.

In addition to being a genuinely enjoyable conversational game, CGT introduces a novel method for achieving consensus without relying on a voting algorithm. This approach maintains openness, inclusivity of multiple perspectives, and ensures transparency throughout the process.

The Game can be played human to human, human to AI, and AI to AI.

The game requires a computer interface managed through a computational system which we demonstrate in our closed pilot.

This is a quick outline of the game’s structure, objectives, and potential applications in enhancing digital communication and critical thinking skills.

1. Introduction

In an era of increasing polarization and misinformation, there is a growing need for tools that promote constructive dialogue and consensus-building. Conversational Game Theory (CGT) addresses this need by providing a structured framework for conversations that encourages participants to distinguish between objective ideas and subjective ideas within a para-consistent framework around “the open question” through narrative structure, naturally challenging assumptions, and naturally flowing towards shared understanding.

2. Narrative Game Structure

Conversational Game Theory (CGT) is structured in three distinct acts, each with three narrative chapters. Each chapter requires humans and AI agents to make decisions.

2.1 Act 1: Establishing Positions

  • Participants are presented with an initial idea or topic.
  • Both AI and human participants respond to the initial idea, tagging their responses using a specific system of 0, 1, and 2.
  • This stage sets the foundation for the dialogue by clearly outlining each participant’s initial stance.

2.2 Act 2: Engaging in Dialogue

  • Participants engage in a Socratic method of questioning and a dialectical approach to discussion.
  • AI Agent or Human Agent consistently challenges the tagging and exposes potential contradictions. These sequences of tagging in pairs identify various thresholds.
  • The goal is to align objective and subjective truths through rational and honest discussion around an open question, a threshold which issues powers to make changes in the system.

2.3 Act 3: Editorial Intervention

  • CGT identifies thresholds in conversations, leading players towards convergence or abandonment of editing permission.
  • Participants collaborate to refine the initial idea, distinguishing between objective shared truths and subjective opinions.
  • A decision is made whether to approve the summaries for final edit and publication, rewrite them, or return to Act 2 for further discussion.
  • Open questions are identified and summarized.

3. Tagging System

Conversational Game Theory (CGT) employs a unique tagging system to categorize statements:

  • 0 (Mystery): Used for questions, uncertainty, or idea exploration without committing to truthfulness.
  • 1 (Objective Truth): Represents objectively true statements and related domains.
  • 2 (Subjective/False): Indicates subjective or personal truths, opinions, or potential toxicity.

This system helps participants clearly distinguish between different types of statements, promoting more precise and thoughtful communication.

4. Key Features

4.1 Gamified Computational Socratic Method and Dialectical Approach

The game encourages the use of the Socratic method to uncover assumptions and clarify meanings. The dialectical approach promotes rational discussion to resolve disagreements and align perspectives through the resolution of contradiction.

4.2 Critical Evaluation

Participants are encouraged to critically evaluate each other’s statements and tags, fostering analytical thinking and attention to detail.

4.3 Collaborative Editing as Community Distribution of Power

The Game has no referee, nor admin nor any third party. Who decides which perspectives publish content?

This outcome is decided between the perspectives who actually disagree. CGT distributes permission tokens to pairs of humans or AI that disagree in some way. Only the most collaborative pairs who engage sincerely with the game will decide, ultimately.

The game culminates in this collective and collaborative editing process, where participants work together to create a consensus summary of objective truths, subjective perspectives, and open questions.

The power or influence to write into the Consensus Library itself can only be held by collaborative pairs who disagree.

5. Feedback Loops

Since power is distributed equally to pairs of players who disagree, their decision making is influenced by feedback loops in the system. The feedback loops in the system provide “contextual completeness” to the state of the consensus, providing two distinct types of summaries, Front Page View and Narrative Review. Summaries are written in the “voice of the rational consensus”, which is also the converging voice amongst LLMs.

Front Page View

Front Page View is a summary of the contextual completeness of the topic as discovered within collective conversation, all Front Page View articles are articles in the Global Library of Consensus Articles.

These summaries consistently are; 1.––a summary of the verifiable and shared truths about the topic of conversation. 2.––a summary of the perspectives within the topic, including notable perspectives and influential perspectives, and where common misunderstandings exist between these different perspectives or potentially even deceptions 3.––a summary of the open questions that consensus conversations are deliberating or speculating about.

Narrative View

Narrative View is a “news story” about the consensus conversation itself, a story told about the status of the topic within the consensus reported in real time. This news story provides total transparency while following the arcs of Conflict into Resolution. All of the micro-–agreements and micro––disagreements within a large topic are archived and preserved and continually open to refinement.