Hacker News new | past | comments | ask | show | jobs | submit login

It is worth noting that the 1/e secretary problem solution is optimal only if your goal is to maximize the probability of choosing the single best secretary.

If your goal is to do expectimax optimization, as decision theory would dictate, you should make a decision after reviewing sqrt(n) applicants. That's assuming a uniform distribution of utility among secretary choices. If the distribution is non-uniform, another heuristic might be better.




I assure you that in dating, utility is not even close to uniformly distributed over secretaries. It is almost the least uniform naturally-occurring distribution I can think of.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: