Given 9/10 hard-autocatalytic-core formal preconditions FAIL on the LLM-genealogical substrate (manifest §audit-substrate-translatability) and prior verdict shows RANDOM drifts with no replicator-attractor under any rule, we test the weaker Adler-SDA stability-selection predicate: selection-by-stability/persistence distinguishable from random drift. PRE-REGISTERED EXPECTATION: drift-dominant (K_DRIFT_DOMINANT) or a careful PARTIAL (K_SDA_SELECTIVE) where any low-displacement rule is the selection-proxy tautology, not a substrate SDA signature. We will NOT manufacture a positive.