KB-5473

PD Runtime Observation — 06 KG/dot-kg Rescoring

3 min read Revision 1

06 — KG / dot-kg Rescoring After Observation (Workstream E)

Before → after (the only candidate that moved is dot:kg)

candidate v2 readiness_class v3 candidate_status v3 gate moved?
job:cut verified_candidate (conf 100) verified_candidate VERIFIED_PENDING_OWNER no (already real)
dot:kg runtime_missing (conf 50) simulated_observed SIMULATED_PROOF_ONLY_NOT_VERIFIED yes
dot:doc / dot:kb / dot:nrm runtime_missing (50) structural_candidate BLOCKED_NEEDS_RUNTIME_AND_CORRELATION no
dot:gov structural_weak (30) structural_candidate BLOCKED_NEEDS_EVIDENCE no
dot:schema (34) / collection / migration / birth not_a_process structural_candidate BLOCKED_NEEDS_EVIDENCE no
dot:api/field/script/seed/sync, wf:WF-001/002 (1 member) not_a_process not_a_process NOT_A_PROCESS no

dot:kg flag delta (honest)

flag before after note
runtime_missing true true no REAL_RUN; simulated does not satisfy runtime
correlation_missing true simulated correlation present (1 corr_id over 2 components) proves wiring, not the process
event_missing true true process.* still inactive
owner_missing true true no owner registered
confidence (verified scale) 50 50 unchanged — simulated must not inflate
evidence tier none SIMULATED_DRY_RUN new

The 18 KG pairs / 10 law processes A–J

  • All 18 KG producer/verifier pairs remain structural_candidate at the family level; only the DOT_KG_EXPLAIN pair carries a simulated observation, and that rolls up to the single dot:kg candidate as simulated_observed. The other 17 pairs have no observation.
  • The 10 KG law processes (A–J) are not individually born candidates in this discovery model — dot:kg is one rolled-up candidate (36 members). They remain at simulated_observed collectively via the one pilot run; none is verified. No per-law promotion performed.
  • Overall dot-kg family verdict: simulated_observed — NOT verified, NOT birth-ready. The simulated run is a substrate/correlation proof, explicitly not process proof.

Honest summary

The macro raised dot:kg exactly one rung — from "structurally a candidate with no runtime evidence" to "the ledger + correlation path is proven via a clearly-labelled simulation." It did not make dot:kg verified, did not raise its confidence, and did not move it toward birth. That requires a REAL_RUN (or approved DRY_RUN) through an execution wrapper.

Back to Knowledge Hub knowledge/dev/reports/architecture/process-discovery-runtime-observation-apply-safe-dryrun-2026-06-04/06-kg-dot-kg-rescoring-after-observation.md