I think companies could test their interview process this way. Fill some positions by choosing from the candidate pool at random (there could be a baseline filter to ensure a minimum level of competence) and compare the random hires with the regular hires to see how much of a difference the interview process really makes.