Side-Stepping the Secretary Problem
The article explores how to improve hiring decisions by addressing the classic "secretary problem" — which suggests stopping after evaluating ~37% of candidates — and proposes practical side-stepping strategies such as defining clear criteria, conducting structured interviews, and using calibration samples to better balance exploration and selection in real-world hiring.
Background
- The "secretary problem" (also called the optimal stopping problem) is a classic math puzzle: you must hire the best candidate from a sequence of applicants, but you can only decide immediately after each interview and cannot go back. The optimal strategy is to reject the first 37% of candidates (the "look" phase), then pick the next one who beats all seen so far. This yields a ~37% chance of picking the very best.
- The article describes a pragmatic workaround for real-world hiring: instead of the rigid "reject-first-37%" rule, use a "shortlist" of strong candidates encountered during the look phase, then run a final round among them. This sidesteps the mathematical constraint while retaining a principled structure.
- Why it matters: The secretary problem is widely taught and cited in tech/engineering culture, but its strict assumptions (no recall, sequential interviews, one hire) rarely hold in practice. The post offers a more realistic alternative for teams that want something better than unstructured intuition but simpler than full scoring rubrics.