| Research must justify itself as next-quarter revenue | The portfolio loses independent epistemic purposes and invites an invented application promise. | State the scientific return and resource choice separately from a qualified product use. |
| Every named branch gets its own full stack | Repeated methodology and infrastructure consume capacity without comparison. | Construct both shared provision and adequate separate provision at the specific contribution boundary. |
| Everything common goes into an office | A unit name substitutes for available contributions and authority. | Specify what is supplied, to whom, under which conditions; compare office and distributed provision only for that result. |
| One person, three available role budgets | The plan manufactures capacity by relabelling its bearer. | Reconcile all assignments against that person’s actual windows and protected commitments. |
| A shared source eliminates adaptation | Subject models, practice, access and receiving acceptance remain unpaid work. | Trace the common result through each application and include its missing operations. |
| Unknown long-run effect blocks any proposal | The inquiry demand costs more than it can change in the current choice. | Use current grounds for a bounded expert construction; obtain additional evidence only where its attainable answer matters. |
| Arithmetic supplies authority | Hours fit, but performers, clients or access owners have not agreed. | Return the exact decision and keep dependent work unavailable until its condition holds. |
| A successful cohort ends development | The provider can repeat one service but cannot address the next worthwhile class of problem. | Separate repetition, efficiency, expanded capability and development of the way development is done. |
| The teaching case becomes the method boundary | One provider’s nesting or resource split becomes a universal design. | Preserve the result/dependency/support construction and vary branches, authorities and means with their actual grounds. |