A.19 typing vs adjacent substrate requirement | A.19 already declares CharacteristicSpace, but source-set and publication-form semantics still need a separate substrate declaration. |
| Precision vs over-typing | The line needs explicit ref positions, an explicit ref-to-ref relation kind, and explicit relation posture, but it should not fabricate composition, derivation, metrics, or transition qualifier when the case does not need them. |
| Reuse vs semantic collapse | DescriptorMapRef, DistanceDefRef, OutcomeMapRef, or BridgeDistortionNote are useful qualifiers, but they must not silently become the whole substrate. |
| User readability vs architectural honesty | Cold readers need a first-minute explanation, while specialist readers still need exact boundaries and docking rules. |
| Interpretive views vs substrate core | Atlas or interpretive-view lines can be valuable, but they should remain optional derived help rather than the default meaning of the substrate. |
| Uncertainty honesty vs fake closure | Many current lines use learned, adaptive, unstructured, or distribution-valued spaces or relations; the pattern must expose that posture without pretending the heaviest qualification posture is already settled. |