Jump to text
On this page
Copy and source
Published source confirmed at last check

Source changed 2026-11-18 21:11:39 UTC · snapshot created 2026-11-18 21:12:35 UTC · last check 2026-11-18 22:20:20 UTC

B.5.2.1:5 - Implementation & Binding into B.5.2 (two injection points)

Step 2 — Generate candidates. Before generation. Establish the Method’s applicable input/seed conditions. For actual generation Work, satisfy its independently applicable authority, assignment, scope, window, budget and permission conditions, including USM coverage and an enactable RSG state when the governing use requires them. A theoretical candidate comparison does not acquire a fictional performer or permission gate. A generated candidate’s ConstraintFit is assessed after construction and controls its admission to the specified comparison.

When the pattern is imported, use NQD-Generate for the selected generation contribution. Its candidate package carries conjectural content and the constituted coordinate results and provenance used. Report diversity to the extent supported by its declared reading; generation assigns no assurance level.

Step 3 — Plausibility filters. Apply B.5.2’s plausibility criteria:

  • Falsifiability / probeability → require an interpretable implication or possible discriminating contrast under B.5.2. Keep a meaningful conjecture distinguishable from the present availability or worth of its check.
  • Explanatory power → prioritize candidates whose Q-improvements and attached rationales address the stated explanatory question.

Apply B.5.2 to return a qualified prime conjecture when its comparison warrants one, or an explicit abort, defer or split result. Deduction, evidence acquisition and action choice remain separate questions.

Primary dominance test: compute the (ε-)Pareto front over the declared DominanceSet. When the Context consumes the ordinary default, that means the declared Q components. For tie-breaking, use only the constituted Novelty, ΔDiversity_P, Surprise or IlluminationSummary results named by the active C.19 policy. Optional archive or telemetry readings retain their declared use; promotion into dominance requires an explicit policy. Use-Value remains non-dominant unless the active Q tuple includes it.

Fallback when no narrower local policy is specified

Use the ordinary default DefaultId.DominanceRegime from G.Core/G.5 together with the active C.19 policy-side defaults; in ordinary Q-front use this means the declared Q components, with ConstraintFit=pass as the candidate admission condition for that front, assessed after generation. Tie-breakers: use only the constituted Novelty@context, ΔDiversity_P or Surprise results named by the active C.19 policy. IlluminationSummary remains report-only unless that policy names a tie-break or promoted use. Unused optional readings need not be produced. Archive: K=1, ε=0, deduplication in CharacteristicSpace. Policy family: one uncertainty-aware explore policy family with one declared regime key; UCB-class with moderate temperature and explore_share ≈ 0.3–0.5 is one didactic starter profile, not the semantic default family. Provenance (minimum): record DescriptorMapRef.edition, EmitterPolicyRef, TimeWindow, and the seeds used; include DistanceDefRef.edition when the run uses a distance definition.

Scope of the coordinate claims. Record the BoundedContext and TimeWindow that qualify each coordinate result.

Surprise: scope and default role. Use Surprise as a secondary tie-break only when the active C.19 policy names that use, among candidates otherwise Pareto-equivalent on the declared primary characteristics. A policy may explicitly promote Surprise into dominance with its constituted basis. If the policy does not use Surprise, omit that reading.

B.5.2.1:5.1 - Distinguish the returned results

Keep three results distinct: the front records non-domination under the declared DominanceSet; the ExplorationArchive records retention under its policy; a shortlist records the set selected for a named receiving use. Label each result supplied by the generator and preserve the complete front before any thinning.