Friday, April 08, 2011
I've been going through all kinds of old papers lately and having had a rant on stats recently, thought this excerpt I found was funny-- I think it was also from Dr. Witesman. This is, indeed, a very strange-sounding rule. But mathematicians have proved it works better than any other. The number thirty-seven percent is an approximation of the exact number I/e, where e is the base for natural logarithms... of course this rule can't guarantee success, but, as Churchill said of democracy, it's the worst strategy except for all the others, and it gives you a thirty-seven percent probability of making the best decision. Here is one example. Suppose over a life-time you expect to meet one hundred available candidates. If you marry the first one, the chance that you have indeed found the best of all one hundred candidates is only 1/100. Likewise, if you wait to meet all one hundred candidates, you will have rejected ninety-nine who came before, and the possibility that the last person you meet is also the best is again only 1/100. The best strategies allow you to sample for a while, in order to learn about the various candidtes; and of all such strategies, the best has you samplling thirty-seven percent of the total and then choosing the first candidate thereafter who beats all the ones who came before. Of course, there is a chance you will never find one who is better than all thirty-seven percent you've already seen. So if you are a young woman who expects to meet one hundred attractive bachelors over her dating years, you should let the first thrity-seven of them go, and marry the first one you meet thereafter who is more attractive to you than all thirty-seven young men you have already dated. Now, don't you wish your mother would give you advice like that?