In all fairness, at scale. it takes one incompetent recruiter out of literally hundreds, or one badly managed case from an otherwise very competent recruiter, out of thousands a day, to get something like this to happen. Any large enough system will have component-level faults.

Sure, it is Facebook, they know how to build systems that are tolerant to faulty components. If they cared about preventing this, they could send resumes to multiple recruiters and proceed based on some sort of consensus. But that's the kicker, turns out their recruitment organization is already a fault tolerant system. The particular fault tolerance mode here is that they don't care about a certain rate of false negatives, no matter how egregious in hindsight, as long as they have few false positives. Sucks at a human level, but in some ways it probably is the optimal system based on what the objective function is for a large company.