Suppose the case of interview for a job where half the applicants are unqualified. This algorithm has an ~20% chance of hiring someone unqualified. This algor

Suppose the case of interview for a job where half the applicants are unqualified. This algorithm has an ~20% chance of hiring someone unqualified. This algorithm seems optimized towards getting near-optimum rather then limiting risk. I think for real world problems it'd be much better to lower your standards as you near the end of the pool.