Secretary — the 37% rule: reject the first 37% of candidates to calibrate, then hire the first who beats them. The optimal-stopping problem, playable, with a live success-rate curve peaking at 1/e.

Personnel Office · Optimal Stopping
One post, one queue of applicants, no going back. Look, then leap — and let 1/e tell you when.
The hiring desk · appoint one of 2401 / 24
37%
◂ look & calibrateleap ▸
Hired
Applicant No. 01Calibrating

Assessment
★ best yet
benchmark —
The rule is watching…

Rounds played 0 Best hired 0 The rule wins ≈ 37% of the time
On the mathematics of stopping
I · The rule

Reject the first thirty-seven.

Interview a third of the queue and hire no one. Note only the best of them — your benchmark. Then appoint the very next applicant who clears it. That single instruction is the whole strategy, and it is provably the best you can do.

II · Why 1/e

The threshold hesitation earns.

Look too little and you leap before the strong candidates arrive. Look too long and the best has already walked out the door. The two failures balance at exactly one part in e — 0.3679 — where the win rate crests near thirty-seven per cent.

It holds for four applicants or four thousand; the fraction never moves.

III · Elsewhere

The same shape, everywhere.

Flats viewed once and gone. A single parking bay on a filling street. The apartment, the offer, the person — any queue you meet in order, judge on the spot, and can never revisit obeys the same arithmetic of patience.

Calibrate, then commit.