| Local structure win hides other-scope loss. A module split, control placement, evidence scope, or direct responsibility-relation change helps one concern while worsening another architecture characteristic. | Rebuild the selected-structure contribution rows and record the gained and lost characteristics before comparison; do not infer responsibility from a team label or assignment. |
| Function and architecture characteristic collapse. The candidate is argued from user-visible function while evolvability, coupling, cohesion, latency, evidence burden, or another architecture characteristic remains unnamed. | Recover the function through A.6.F or the structural-view pattern, then name the architecture characteristic separately. |
| Function without feasible bearer. A functional architecture, workflow, Method step, or searched graph asks for a function but A.6.F identifies no bearer that satisfies the functional predicate under the relevant module, resource, placement, control, evidence, local-kind, classification, or assignment constraints. | Repair the bearer claim before admitting the candidate. |
| No real trade-off. Only one configuration is visible, or alternatives differ only by description. | Generate structurally different candidates, or state why the project work is not architecture synthesis and use the subject pattern. |
| Description artifact stands in for candidate content. A diagram, ADR, view, dashboard, benchmark output, or digital-twin view is the visible work product, but the selected structures and architecture-characteristic trade-off are still missing. | Keep the visible work product with its description-use, C.29 mathematical-lens, benchmark, publication, or source-use pattern and recover candidate content before C.32 use. |
| Front member treated as durable optimum. A front member, local winner, or benchmark leader is used as if the evolution window will stay fixed. | Record evolution window, source-return condition, and retained alternatives under the exact C.18 or C.19 predicates; use G.5 only to declare a selected-set result from those alternatives. If the result is published, use E.17 for a source-backed face and source return and E.24.PUB for the publication occurrence and audience availability. |
| Software-source overfit. A software architecture source supplies a useful architecture-change idea, but the described holon is not a software system. | Translate only the change over selected structures and characteristics; do not import the software ontology. |
| Architecture-influence source omitted. The candidate architecture for a changed referent cannot be built, tested, deployed, certified, or evolved under the current architecture, Work, communication, method, tool, deployment, evidence, selected-structure, or other source, but that source’s exact kind and influence status are hidden. | Open C.32.CONWAY; recover the source kind and either its exact obtaining direct relation or the precise provisional disposition, keep the acting System, any local system-role kind or assignment, Work, changed referent, and any actual transformation distinct, and prepare influence-source-side change, transformed-side change, joint change, and bounded mismatch as candidate alternatives or comparison inputs. |
| Method-defined dimensions lose their semantics. A BIM, digital-twin, or view-method dimension already carries method-defined structure, constraint, cost, schedule, use-phase, or maintenance semantics, but the synthesis text keeps only the dimension name or dimension count. | Preserve the method semantics and map them to selected structures, constraints, characteristics, and source-return conditions. |
| Ideality shortcut. Fewer bearers, fewer modules, or one universal module is only a candidate direction until functions, architecture characteristics, scale window, safety, admissibility, and losses are named. | Keep it as one candidate and expose those missing tests before comparison. |