科学与数学en.wikipedia.org
秘书问题
The Secretary Problem
2014 — 2022，它一次次回到首页
黑点是上首页（≥ 50 分）的那次投稿，空心点是没冲上去的投稿，点越大分数越高。
导读
维基百科关于“秘书问题”（最优停止问题）的条目：在只能顺序查看、不可回头的候选者中，用 37% 规则（1/e）最大化选中最佳者的概率。这是概率论中一个经典且反直觉的结果，常被联系到招聘、择偶、买房等现实决策场景。
为什么经久不衰：它把一个日常困境变成了优雅的数学结论，多年来反复引发关于理论假设与现实决策差距的讨论，至今仍是理解最优停止思想的最佳入门。
HN 评论精选
用 37% 规则买房的真实成功案例：先定标准再果断出手，低于市价 15% 成交。
我用这个方法买到了房子。先看了两处建立标准，然后在 Rightmove 上对大伦敦地区做搜索（以靠近 Waitrose 超市为筛选条件），再按每平方英尺单价和关键词过滤（目标 500-600 英镑/平方英尺、无需翻新、无链且能快速成交）。一出现高质量低价的房子我就直接出价（成交前看过一次）。现在我就住在那里——一套低于市价约 15% 的漂亮公寓。用秘书问题的基本解法，把绝大多数平庸的房子过滤掉后，我在第一个符合条件的选择上就停手并果断出手。
buro9查看英文原文
I found a home to purchase by implementing this. I visited two places to establish criteria, then built a simple search across Greater London on Rightmove using area searches with proximity to Waitrose. The results I then filtered again based on price per square foot and rejecting certain keywords (aiming for a price per sq ft between £500-600 without needing renovation, but looking for no chain + quick sale). As soon as a high quality and low price to square foot came up I put in an offer without having viewed it (I did view it once before completion). I live there now... It's a gorgeous apartment purchased about 15% below market rate in a nice area. Implementing a basic solution to the secretary / optimal stopping problem meant that with criteria that took out the vast majority of average places, I stopped on the first one and committed to it.
理论假设（已知总数、不可回头、分布未知）与现实场景不符，实用性存疑。
告诉别人时这是个很酷的结论，但细想之下它并不像听起来那么有用。它的几个基本假设——事先知道候选者总数、不能回头（还应加上：事先对候选者的分布一无所知）——与那些所谓的应用场景都对不上。现实中你通常能对总体分布做出估计，决策也往往不是最终性的。此外还有别的问题：零工经济者还要优化找下一份工作的时间成本，买二手车最大的问题是信息不对称而非错过就没……
t_mann查看英文原文
It's a really cool result when you tell people about it, but actually when you think about it, it doesn't seem as useful as it sounds at first. The combination of some fundamental assumptions - you know the number of candidates in advance and you can't go back to previous candidates (*edit - should have added another one here: that you don't know anything about the distribution from which candidates are drawn before you begin) - don't really seem to match any of the purported applications, nor any I could think of quickly off the top of my head. In reality, you're usually be able to gauge some estimates of the moments of the population distribution before you have to make any decision (which also aren't always final). Ie, you'll be interviewing a couple of candidates before you send out any rejections, you'll see several potential mates before you narrow in on one, you know from experience what a nice appartment looks like when you visit one,... There are also other problems in other potential applications: a gig economy worker needs to not only optimize the value of the next gig but also the time spent searching for the next gig, when buying a used car the biggest problem is the information asymmetry regarding the true value of the car, not that it's gone if you don't buy right away,...
现实中人们要的是“够好”而非“最优”，追求最优反而产生机会成本。
它在现实中真有用吗？你通常并不需要从任意样本里挑出“最好的”，只需要“超过某个门槛”就行。明明有门槛标准却非要追求样本最优，只会带来机会成本——招聘、约会、研究度假目的地都是如此。
quickthrower2查看英文原文
It is this useful in real life? You usually don’t actually need the best from an arbitrary sample, you need better than $threshold. Aiming for best from sample when you have $threshold already could just result in opportunity cost. Be it in hiring, dating, researching holiday destinations or anything.
指出该问题在内存分配器设计中的实际应用。
很多年前我在写内存分配器时遇到过这个问题：维护一个未排序的空闲块链表，但记录其长度。每次分配需要找到一个大小至少为 n 但又不要大太多的块，而链表太长不想每次都遍历完。内存分配器有很多写法，但在上述约束下，这就是同一个问题。
dreamcompiler查看英文原文
I ran into this problem many years ago when writing a memory allocator. Say you keep a list of free blocks. You don't keep them sorted, but you do keep track of how long the list is. You now need to find a block of at least size n but not much larger, and the free list is so long you'd rather not look through the whole thing for each allocation. There are a bunch of ways to write memory allocators, but with the above constraints, it's the same problem.
历次讨论
同类还值得读
如果月球只有 1 像素：一个极其精确的太阳系模型（2014）
If the moon were only 1 pixel: A tediously accurate solar system model (2014)
一个网页版太阳系模型，以“如果月亮只有 1 像素”的真实比例呈现：你必须一路横向滚动，才能体会行星之间空旷到何等程度。
科学与数学上首页 7 次 · 2014–2025joshworth.com编辑精选沉浸式线性代数（2016）
Immersive Linear Algebra (2016)
《Immersive Linear Algebra》是一本完全免费、开放在线的线性代数“超文本教科书”，其最大特色是大量可交互的图形：读者可以直接拖动、操作图中的向量、矩阵和变换…
科学与数学上首页 5 次 · 2015–2024immersivemath.com编辑精选如何将贝果切成两个相连的半环（2009）
How to Slice a Bagel into Two Linked Halves (2009)
数学家 George Hart 的经典教程，演示如何用一刀把贝果切成两个相互扣住的环，并附上切割步骤与数学原理。
科学与数学上首页 5 次 · 2009–2022georgehart.com“犹太问题”：一组用于歧视的数学题
Jewish problems
一篇 arXiv 论文，整理并讨论苏联时期犹太学生在申请大学数学系时被问到的特殊数学难题：这些题往往有简单解法却极难想到，被用作面试中拒绝犹太考生的手段。
科学与数学上首页 5 次 · 2011–2021arxiv.org你的身体并非为长久而造：人类死亡率数据的一课
Your body wasn’t built to last: a lesson from human mortality rates
文章从人类死亡率随年龄变化的统计规律出发，解释身体并非为长久使用而设计：年轻时死亡风险较低，之后随年龄近似指数上升，像复杂系统在不断累积损伤。
科学与数学上首页 5 次 · 2009–2018gravityandlevity.wordpress.com