What would count as a miss of "only" at the layer the page still cares about — not a miss of the three-ingredient definition?

On one live scientific dispute, does the pre-specified copy rule (citation, failing replication, hiring, or textbook persistence) track true predictions, or is "selection on theories" a label on whatever currently scores?

Can the same inversion be produced under scarcity by arranging those three conditions, and can abundance without them fail to invert? If both, what rule separates bankruptcy from a transfer other than whether recurrence still tracks contribution?

What filter copies or shuts the Fourth Branch, and is there a measure of competence, truth, or capability that can fail while that office still scores?