Groom shopping!
Everybody is excited about Bindu's marriage!
Prospective NRI grooms will be arriving in the city for the pre-marriage talks with Bindu. However, she has a dilemma. NRIs can stay only for a short time, because of short leave from their jobs. So if she says YES, then the marriage ceremony will be performed soon. But if she says NO, then the prospective NRI will go off and marry another lady. She cannot revisit her choice once she says no. Should she say yes to the first "A Suitable Boy" who she meets, or should she wait long enough to know her best possible choice. But then how long should she wait? Through how many people should she wait?
HR interviews may pretty much have the same decision-environment. You cannot revisit your choice once you say nay. You have to say yes to the candidate or he would leave and be taken up by another recruiter. You know the number of candidates out there. But you don't know the scores of those who have come to the interview but are waiting outside your chamber. And scores here mean, an overall assessment!
Turns out that intelligence has been applied to the "secretary problem" as it is famously known. In more technical terms, you are trying to maximise the probability of choosing the best item from the ones on offer. Rejection is once and final.
The solution is on these lines.
"The consensus among the many who have worked on this problem is that his best strategy is to let a certain fraction of the students pass and take the next one who has a score better than any of the ones seen thus far."
So if there are three students. It makes sense to let one pass and take the next candidate who has a score higher than this one candidate that has been passed. If there are lots of people, the odds are only 1/100 of the optimal choice being the first or last person. Then it makes sense to test the first 37 persons and choose the next person who is better than the previous 37.
"Finally, there is the “twelve bonk rule,” that says that if you simply want to ensure that your choice is better than 90% of the other choices available, simply ’sample’ the first 12 possibilities and pick the first person who is better after the first 12. This strategy gives you a 77% possibility of success."
Just a qualifier. Probability works in the long run, not in the short term!
And btw, Bindu eloped!
Prospective NRI grooms will be arriving in the city for the pre-marriage talks with Bindu. However, she has a dilemma. NRIs can stay only for a short time, because of short leave from their jobs. So if she says YES, then the marriage ceremony will be performed soon. But if she says NO, then the prospective NRI will go off and marry another lady. She cannot revisit her choice once she says no. Should she say yes to the first "A Suitable Boy" who she meets, or should she wait long enough to know her best possible choice. But then how long should she wait? Through how many people should she wait?
HR interviews may pretty much have the same decision-environment. You cannot revisit your choice once you say nay. You have to say yes to the candidate or he would leave and be taken up by another recruiter. You know the number of candidates out there. But you don't know the scores of those who have come to the interview but are waiting outside your chamber. And scores here mean, an overall assessment!
Turns out that intelligence has been applied to the "secretary problem" as it is famously known. In more technical terms, you are trying to maximise the probability of choosing the best item from the ones on offer. Rejection is once and final.
The solution is on these lines.
"The consensus among the many who have worked on this problem is that his best strategy is to let a certain fraction of the students pass and take the next one who has a score better than any of the ones seen thus far."
So if there are three students. It makes sense to let one pass and take the next candidate who has a score higher than this one candidate that has been passed. If there are lots of people, the odds are only 1/100 of the optimal choice being the first or last person. Then it makes sense to test the first 37 persons and choose the next person who is better than the previous 37.
"Finally, there is the “twelve bonk rule,” that says that if you simply want to ensure that your choice is better than 90% of the other choices available, simply ’sample’ the first 12 possibilities and pick the first person who is better after the first 12. This strategy gives you a 77% possibility of success."
Just a qualifier. Probability works in the long run, not in the short term!
And btw, Bindu eloped!
posted by Naveen Mandava at 11:57 PM - Permalink
<< Home