Genesis Q5-NBJ pre-mortem requirement injected into prompt

Append a required output section to the genesis prompt template: 'PRE-MORTEM: In 2-3 sentences, state the most likely way this hypothesis fails the Q5 NBJ test (Necessary to the buyer / Buyable as a discrete unit / Justifiable in budget terms). Name the specific failure mode (e.g. "the buyer would not allocate budget because the workflow it replaces is unbilled"). If you cannot name a specific failure mode, output the literal string NO_PRE_MORTEM_FAILURE.' Parse the pre-mortem block out, store as candidate.pre_mortem, and the NO_PRE_MORTEM_FAILURE sentinel becomes a flag input to filter_score (raises v2_a3-equivalent describability axis by 1).

Genesis output volume drops 15-25% (self-rejection in the chain-of-thought, candidate not emitted). Among remaining candidates, the fraction tagged NO_PRE_MORTEM_FAILURE is <10%. Filter pass-through rate on candidates with substantive pre-mortems rises, because the lowest-quality shape-fixated ideas no longer reach filter.

If after 50 genesis runs the pre-mortem text becomes formulaic (n-gram similarity >0.7 across pre-mortems for distinct candidates) and NO_PRE_MORTEM_FAILURE rate stays at <2%, the prompt addition is being satisfied syntactically not semantically — revert. Also: if genesis volume doesn't drop at all, the model is not actually self-filtering and the addition is dead weight.

Rejected at the council verdict. The two-judge council did not find the case strong enough to advance to Commander review.