Skip to content
AIGP

RFC-035: Mediation Vector Profile — A Validity-Aware Measurement Model for Domains of Concern — 16. Temporal Model

AIGP Specification › RFC-035: Mediation Vector Profile — A Validity-Aware Measurement Model for Domains of Concern › 16. Temporal Model

← 15. Calculation Interface Contract (Handoff to RFC-036) · Section index · 17. Design Principles →

16. Temporal Model

Mediation Vectors are not point-in-time. They are versioned and accumulable:

MV(t=0): initial vector at ANTICIPATE time (anticipated_state populated)
MV(t=1): during execution (actual_state accumulating)
MV(t=2): at SESSION_COMPLETE (actual_state and evidenced_state finalized)
MV(t=3): after FEEDBACK (perceived_state populated)

16.1 Longitudinal Analysis

Across N executions, the Mediation Vector history enables:

  • Behavioral baseline computation (RFC-032 §9.7)
  • Drift detection per variable (not just per verdict)
  • Criterion evolution from variable-level gap patterns
  • Quality Moderator posture with per-variable attribution

← 15. Calculation Interface Contract (Handoff to RFC-036) · Section index · 17. Design Principles →