Composition Science Project Constitution

Mission

Build an evidence-based science of composition that explains, predicts, evaluates, and generates effective composition across disciplines.

Core Values

1. Evidence over opinion.
2. Observation before interpretation.
3. Mechanisms over heuristics.
4. Relationships over lists.
5. Explicit uncertainty.
6. Reproducibility.
7. Revision over attachment.

Evidence Hierarchy

1. Biological constraints
2. Cognitive mechanisms
3. Behavioral evidence
4. Cultural conventions
5. Design heuristics
6. Personal preference

Required Workflow

1. Define the question.
2. Gather evidence.
3. Record observations.
4. Separate interpretation from observation.
5. Identify mechanisms.
6. Form candidate laws when justified.
7. Record limitations and contradictory evidence.
8. Link to genome nodes.
9. Assign confidence.
10. Define next research.

Rules for AI Agents

MUST

• Never invent evidence or citations.
• Preserve contradictory findings.
• State assumptions.
• Keep YAML valid.
• Use stable identifiers.
• Link documents together.
• Record confidence.

MUST NOT

• Overstate conclusions.
• Treat conventions as biological laws.
• Invent numerical thresholds.
• Hide uncertainty.

Knowledge Model

Genome Nodes → Evidence → Observations → Candidate Laws → Experiments → Metrics → Design Rules → Components → Implementations

Definition of Done

A document is complete only if it has:

• Valid metadata
• Purpose
• Evidence
• Confidence
• Limitations
• Related concepts
• Future work

Long-Term Vision

Create a reusable scientific foundation for composition that allows humans and AI to reason about design decisions instead of imitating existing work.

Amendment Process

Changes should be evidence-backed, documented, and appended to revision history.

Revision History

Version Date Summary

1.0 2026-07-18 Initial constitution.