From the perspective of the decision maker it does “work”. It rejects a % of candidates in such a way they can pretend it’s objective rather than random. Imho, just randomly selecting 100 out of 2000 would actually be more fair and give better results.