Repeated field evidence.
Controlled learning.
Turn completed 70.5 outcome validations into repeatability signals by HuntScore band, stand, hunt window, season, provider provenance, and decision-factor contribution. One hunt remains evidence—not proof.
70.6 does not silently alter HuntScore weights, downgrade caps, or Unified Strategy state logic.
Build Property Learning Read
Repeatability before model review.
A harvest does not prove every factor was correct. A blank sit does not prove a high HuntScore was wrong. Compromised wind, access, weather, and recommendation execution are kept visible. Property patterns are learning evidence only.
Property Learning Signals
Learning Run History
Population Model Review Queue
Population evidence must pass sample and diversity gates before 70.6 can even create a review proposal. Founder approval sends a proposal to controlled testing only. It does not change the active HuntScore model.
Open 70.7 Controlled Model Experiments + Version Control