阿里·阿布达尔：37%法则

嘿，朋友们。我刚读完一本精彩的书，叫《如何不孤独终老》，作者是洛根·乌里。这本书今年早些时候出版，剖析了人们在寻找爱情时容易陷入的误区，以及如何建立持久的关系。书中有几个观点让我大开眼界，今天想分享的是围绕'秘书问题'展开的那个。

Hey, friends. I've just finished reading an incredible book called, How to Not Die Alone by Logan Urie. It was published earlier this year, and it breaks down what trips people up when looking for love and how to find and build lasting relationships. There were a few bits from the book that blew my mind. One that I wanna share today is framed around the secretary problem.

假设你想招聘一名秘书。你想找到最适合这个职位的人，但只有面试后才能知道对方的能力。此时你可以选择直接聘用（永久性决定），或者继续面试其他人。唯一限制是：一旦拒绝某位候选人，就不能回头再聘用。如何最大化找到合适人选的几率？

Let's suppose you wanna hire a secretary. You wanna find the best person for the job, but you won't know how good someone is until you interview them. At that point, you can choose to hire them a permanent decision or to move on and interview someone else. The only catch is that once you've rejected someone, you're not allowed to go back and hire them. How do you maximize your chances of finding the right person?

这是最优停止理论中常用的场景，该数学分支专门研究这类问题及更复杂情况的最优解。解决方案很数学化：如果你要面试100位潜在秘书，先面试并拒绝前37位，同时记住其中最优秀者作为基准。从第38位开始，只要遇到比之前最佳更好的候选人就立即聘用。据说这样能最大化找到最佳人选的几率。

This is a scenario commonly used in optimal stopping theory, a branch of maths involved in finding optimal solutions to these sorts of problems and many other more complicated ones. The solution is pretty mathsy. Basically, if you knew you were interviewing 100 potential secretaries, you'd interview and reject the first 37, and you'd keep in mind who your best so far candidate was as a benchmark. From interview 38 onwards, you'd immediately hire the person better than your best so far candidate. This is apparently how you maximize your chances of hiring the best overall candidate.

为什么要在意这个？因为秘书问题生动展示了'探索与利用'的困境，这个困境在生活中随处可见：在选定定居城市前该考察多少地方？在决定工作前该尝试多少实习？在确定理想职业前该体验多少行业？

Why should you care? Well, because the secretary problem is a simple illustration of the explore versus exploit conundrum, which shows up in many other areas of life too. How much should you explore your surrounding areas before deciding on a base to call home? How many internships should you do before deciding on a job offer? How many careers should you try out before you figure out which your favorite is?

在决定结婚前该约会多少人？现实当然比数学问题复杂得多，但我喜欢秘书问题的原因有几个：第一，知道这个问题存在正确答案很让人安心——先探索约三分之一选项，之后选择比之前更好的。这意味着我不该在第一个居住城市就买房。

How many people should you date before deciding to settle down and get married? Obviously, real life is more complicated than a maths problem, but there are a few reasons why I like the secretary problem. One, it's quite reassuring to know that there's a correct answer to this issue. Explore the first one third ish of your options and then choose the best one beyond that point. That means that I probably shouldn't buy a home in the first city I've lived in.

我或许该多探索些再做决定。但这也意味着不该无止境寻找完美城市，而是适时安定下来。第二，令人宽慰的是，这个数学最优解的准确率也只有37%。即便用最优化公式处理最简单的'探索与利用'问题，我们仍有63%的概率找不到完美人选。

I should probably explore a bit more and then make a decision. But it also means that I shouldn't keep on sampling forever in the hope that I'll find the perfect city. At some point sooner rather than later, I should make the decision to settle down. Two, it's also reassuring that the mathematically optimal solution only gets the right answer 37% of the time. Even if we boil it down to the simplest form of explore versus exploit and we use the most optimal formula, we still fail to find perfect candidates 63% of the time.

这很棒。说明在更复杂的现实世界里，我们也不太可能找到完美的候选人、伴侣、城市或工作。所以可以放心放弃完美期待——只要找到足够好的就足够了。完美本就不存在。我大学时第一次接触37%法则，是朋友在讨论恋爱关系时提到的。

This is great. It means that in real life, which is much more complicated, we're also not very likely to find the perfect candidate, person, city, job, whatever. So we can safely give up the expectation that we should, if we find someone or something good enough, then that's fine. Perfection doesn't exist. I first came across the 37% rule at university when a friend mentioned it in the context of dating and relationships.

如果我们从数学角度思考寻找配偶这一难题，它有点像秘书问题。我们通过约会了解他人，然后在某个时刻决定安定下来，与彼此合适的人结婚。除了明显过于简化这一复杂问题外，另一个问题是我们需要提前知道要处理多少申请人，才能计算出这个数字的37%。如果我们有100个求职者，这很容易。但在约会和恋爱关系中，我们怎么可能知道一生中总共会有多少约会对象呢？

If we think about the conundrum of finding a spouse mathematically, it sort of resembles the secretary problem. We date people to find out what they're like, and then at some point, we decide to settle down and marry someone who we're mutually compatible with. The problem other than the obvious oversimplification of a complex matter is that we need to know how many applicants we're dealing with ahead of time to figure out what 37% of that number is. If we've got 100 applicants for a job, that's easy enough. But in the world of dating and relationships, how can we possibly know what our total lifetime dating pool is going to be?

我从洛根·尤里的书中得到的奇妙见解是，我们不需要用数字来思考这个问题，可以用时间来衡量。很难知道我们最终可能约会多少人，但很容易估算出我们想花多少时间约会。例如，如果我愿意在18岁到40岁之间约会，并且假设每年认识的人数没有剧烈变化，37%法则表明当我26岁时，就应该嫁给下一个最合适的人。这让我深有感触，因为当亲戚问我为什么还没结婚时，我常回答：'我只是还没遇到足够多的人'。

The magical insight that I got from Logan Urie's book is that we don't need to think about it in terms of numbers, we can think about it in terms of time. It's very hard to know how many people we could end up feasibly dating, but it's easy enough to estimate how much time we want to spend dating. If for example, I'm open to dating between age 18 to four, zero, and assuming there's no radical change in the number of people I'm getting to know each year, the 37% rule says that when I hit the age of 26, I should marry the next best person. I this really struck home for me because one of my common responses when relatives ask, why aren't you married yet? Is, well, I just haven't met enough people.

我曾以为需要约会更多人才能了解自己想要什么，然后再考虑在未来找到那个对的人。

I thought that I needed to date more people to get an idea of what I wanted and then worry about finding the one further down the line.