palaver-math

Mathematics of consensus-building, inspired by the West African Palaver tradition.

Palaver is the tradition of extended community dialogue — structured conversation that builds consensus through deliberation. This package implements the mathematical foundations of that process.

Install

npm install palaver-math

Usage

import {
  computeCenter, consensusDistance, influenceWeightedCenter,
  PalaverSession, convergenceRate, predictConsensus,
  findCoalitions, coalitionStrength,
  buildDialogueTree, findConsensusPath,
} from 'palaver-math';

// Define participants with positions and influence
const participants = [
  { id: 'elder', position: [7, 3], influence: 5 },
  { id: 'youth', position: [3, 8], influence: 2 },
  { id: 'merchant', position: [5, 5], influence: 3 },
];

// Compute consensus
const center = computeCenter(participants);        // unweighted centroid
const weighted = influenceWeightedCenter(participants);  // influence-weighted
const spread = consensusDistance(participants);     // how far apart they are

// Run a Palaver session
const session = new PalaverSession();
participants.forEach(p => session.addParticipant(p));
session.addProposal('compromise', [5, 5], 'elder');
session.vote('elder', 'compromise', 1.0);
session.vote('youth', 'compromise', 0.6);
const result = session.computeConsensus();

// Find coalitions
const coalitions = findCoalitions(participants, 4.0);
const strength = coalitionStrength(coalitions[0]);

// Build dialogue trees and find consensus paths
const tree = buildDialogueTree([
  { statement: 'We need to decide.', speaker: 'elder' },
  { statement: 'I agree to the compromise.', speaker: 'youth', parentId: 0, isConsensus: true },
]);
const path = findConsensusPath(tree);

API

Types

• Participant { id, position: number[], influence: number }
• Topic { id, dimensions: string[], idealPoint: number[] }
• Proposal { id, position: number[], proposerId, votes: Map }
• ConsensusResult { position: number[], confidence: number, rounds: number }
• DialogueNode { statement, speaker, responses: DialogueNode[], isConsensus? }

Consensus Functions

• computeCenter(participants) → centroid
• consensusDistance(participants) → average distance from center
• influenceWeightedCenter(participants) → influence-weighted centroid

PalaverSession

• addParticipant(p), addProposal(id, position, proposerId)
• vote(participantId, proposalId, weight)
• computeConsensus() → { position, confidence, rounds }

Convergence

• convergenceRate(sessionHistory) → fraction of steps that converge
• predictConsensus(session, maxRounds) → predicted final position

Coalitions

• findCoalitions(participants, threshold) → groups of nearby participants
• coalitionStrength(coalition) → combined influence × cohesion

Dialogue Tree

• buildDialogueTree(statements) → tree structure
• findConsensusPath(tree) → shortest BFS path to consensus node

License

MIT