The evidence is not ambiguous and it is not new. An hour of conversation, unstructured, tells you reliably how much you enjoyed the conversation.
Why it survives
Because it feels like judgement, and judgement feels like expertise. Structured assessment produces better hires and offers the assessor almost no opportunity to feel perceptive.
The organisations that changed did not become more scientific. They became more specific: same questions, same order, scored before discussion. Dull, reproducible, and considerably fairer.