The claim
Normative half, A38 §11.5: “Examine whether recreating original speaker footing is always ethically or socially appropriate. Recognize that the choice of preserved invariants is itself interpretive and potentially normative.” A40 §14.3: “Preserving the original’s social effect is not always desirable or possible. Translators may deliberately change footing, register, or communicative function. Distinguish fidelity as one evaluative goal from adaptation, critique, parody, and transformation.” A40 §14.2: “Partial equivalence may be more appropriate than global matching.”
Formal half, A40 §7.3: “Consider one-way simulation where target audiences can reproduce the source utterance’s relevant communicative possibilities. Discuss why asymmetric simulation may be appropriate when translation is deliberately adaptive.”
Where it went
The bisimulation section was demoted deliberately (QUESTION-008), and the one-way case went with it. The ethics section did not reach the manuscript. What survives is directedness: transport “is intentionally directed: compression and interpretation need not be invertible” (§6), and the normative choice of observables (§11). Neither says that a rendering may rightly fail to be faithful. A39 §5.4 made a neighbouring distinction, also dropped: “fidelity as an explanatory criterion” versus “fidelity as an optimized translation objective”.
Note of 2026-10-29: what holds it back
Its formal half has no measure to state: the record holds no one-way (simulation) distance for observational profiles, only the symmetric bisimulation metric (THEORY-247), so the “relevant possibilities” an adaptive rendering must reproduce are not defined. Its defeat condition, how readers judge deliberately adaptive renderings, has not been observed. The legal readings (NOTE-1019, NOTE-1021, NOTE-1024) sort renderings by these goals but, as NOTE-1019 says, ask who may make a rendering, not how to judge one.