| Pillar meanings stay constitutional, while evaluation checks realized adequacy. | E.2 constitutional source plus A.19.ECS evaluation-characteristic construction. | E.2.DA evaluates one scoped FPF object without redefining the Pillars locally. |
| Whole-language adequacy needs aim, evidence, change, and learning. | Model for Improvement, PDSA, and PDCA lineage carried through E.22 and E.23. | The evaluation names declared FPF use, evidence basis, first repair or stop, and reopen condition rather than treating release result as improvement. |
| Feedback needs current state, desired state, next action, and tactics. | Sadler and Hattie and Timperley feedback traditions carried through E.22 and E.23. | Values, short rationales, evidence loci, proposal rows, and checked no-proposal dispositions stay distinct. |
| Local pattern quality is not whole-FPF adequacy. | Pattern-language entry and projection discipline from README, ToC, E.11, and I.2, plus current E.21 and E.9.DA source lines. | E.21 and E.9.DA are evidence loci only when they change Pillar realization; they are not averaged into corpus adequacy. |
| Precision repair can improve wording while damaging use. | E.10, A.6.P, C.2.P, C.16.Q, F.18, and F.19 precision-restoration lines. | Broad cleanup must show pre-repair and post-repair kind, relation, admissible use, and FPF-level Pillar effect; lexical disappearance is not closure. |
| Multi-coordinate improvement needs trade-offs and non-dominated alternatives. | MCDA, Pareto, ATAM, and QD, OEE, and NQD lines carried through E.22 and E.23. | E.2.DA asks what became worse and treats front-like vocabulary as governed semantics, not praise. |
| Pillar-adequacy measures can become targets. | Goodhart and Campbell, management-accounting surrogation, specification-gaming, and reward-hacking lines. | E.2.DA forbids all-5 or 5-defensible repair targeting; values rise only when the scoped FPF object better realizes Pillars for declared use, and E.13 governs any proxy-to-value claim about those values. |