| How should alternatives be connected to consequences? | Use explicit outcome models with the conditions and dependencies needed by the decision. | Rank isolated attributes or tool outputs without a consequence path. | Adapt: :4.1–:4.4 reconstruct only decision-changing consequence paths and shared burdens. The added effort is accepted when an omitted interaction could reverse the comparison; a qualified existing account is reused. | Borgonovo et al.’s 2026 decision-analysis review supplies the current graphical-model and uncertainty line. It does not validate a local mechanism or establish causal effects from arrows. C.11.CRC supplies a missing finite configuration-relative contribution, not field calculations. | Reopen when a changed configuration, mechanism, affected subject, or horizon invalidates a consequence path. |
| What comparison is justified by conflicting or incomplete values? | Use a declared preference model and preserve its partial or model-dependent relations. | Impose one total weighted rank. | Adapt: :4.5–:4.6 retain ties, incomparability, and missing-input distinctions. The deliberately accepted trade-off is less rhetorical decisiveness for a truthful relation. | Greco, Słowiński, and Wallenius’s 2025 MCDA review is a comparison candidate for alternative preference and recommendation forms. Its finite-action methods do not supply local values, probabilities, or decision authority. | Reopen if elicitation or evidence changes the admitted model set or resolves a previously missing comparison. |
| How can early qualitative consequences be constructed and questioned before a complete domain model exists? | Make the intermediate relationships and their grounds explicit, carry each alternative through them, and use attributed domain judgement where evidence does not determine the result. | Draw influence arrows, count favorable signs, or declare a complete model necessary before any useful comparison. | Adapt: :4.2.1 and :5.4 supply construction, questioned propagation, conditional backward exploration and local revision; no new notation or solver is required. | Horkoff and Yu’s PoEM 2009 procedure and ER 2010 backward analysis are historical method anchors, not current empirical validation or iStar 2.0 semantics. Qualitative support does not establish causal magnitude, participant commitment or realized success. | Reopen when a changed relation, stakeholder judgement or domain model changes a material consequence; use a stronger model when quantitative interaction or feedback decides the answer. |
| Does generic candidate stewardship already compare the consequences? | Keep profile comparison, finite contribution, and archive/front stewardship distinct. | Treat C.18 as a general consequence-ranking Method. | Adopt: :4.2 and :4.5 use the exact C.11.CRC and A.19.CPM contributions when needed; the PSD consequence account retains its field question. No archive machinery is required for a small comparison. | Current C.18 governs generation, archive, and front objects; A.19.CPM governs its admitted set-valued comparison; direct domain Methods supply consequence models. Their different result boundaries defeat the broader attribution. | Reopen when the current question actually becomes archive stewardship or a direct comparison contribution changes. |