Spec spec_003
SPEC-003: Gravity Score Formula
v4 · draft · Owner: Q
If gravity = retrieval frequency → rich-get-richer. If gravity = validation count → "validation" undefined. If gravity = LLM-assigned → only as reliable as extraction. The entire compounding proposition depends on this.

Definition

source_type * 0.35 + validation * 0.30 + corroboration * 0.35

Inputs

NameWeightSourceNotes
confidence_score0.35extraction pipeline
validation_count0.3outcome tracking
corroboration0.35edge analysis

Anam #01: "Gravity score appears in the schema, emission threshold, confirmation machine risk, and novelty scoring. It is never defined." Escalated from OPEN to BLOCKING.

Problem: If gravity = retrieval frequency → rich-get-richer. If gravity = validation count → "validation" undefined. If gravity = LLM-assigned → only as reliable as extraction. The entire compounding proposition depends on this.

Proposed: gravity = (validation_count × 0.5) + (cross_collection_edges × 0.3) + (recency_decay × 0.2). ANAM recommends this as starting point. Must define: what counts as "validation"? How is cross_collection_edges normalized? What's the recency decay function?

Reconcile: Build eval set: 50 queries where we know the "right" top-5 results. Test gravity formula against confidence-only ranking. The formula that produces better retrieval quality wins. All three founders score the eval set.

Curve: non-linear

Decay: time-based, 90-day half-life

Whitepaper References

Sovereign Intelligence Infrastructure (Whitepaper v5) > 05_knowledge_architecture
Gravity replaces confidence as primary retrieval signal
View in whitepaper

Changelog

v4 2026-04-05 Q Mapped to whitepaper sections

v3 2026-04-05 Q Merged Will gravity formula research into SPEC-003

v2 2026-04-05 Q Imported SPEC-003 from model_specifications_v2.html

v1 2026-04-05 Q Created spec: SPEC-003: Gravity Score Formula