高性价比人生指南:498 条循证建议,每条写明成本、收益和证据等级,优化四种资源

HowToLiveBetter: 498 Evidence-Based Life Tips — Cost, Benefit, and Evidence Grade for Each, Optimizing Four Resources

Research #open-source#evidence-based#life-guide#Chinese#health#finance#legal#Unlicense#productivity
更新于
🇨🇳 中文

📌 开源仓库:https://github.com/eternity4719/HowToLiveBetter 在线检索页:https://eternity4719.github.io/HowToLiveBetter/ License:Unlicense | Stars:2,366 | Forks:195


大多数生活建议有两个问题:要么没有来源,要么有来源但不告诉你”值不值”。

HowToLiveBetter 把这两件事同时做了:498 条建议,每条注明花掉什么、换回什么、证据多硬、来源在哪,按性价比从高到低排列——不是按类别排,而是成本低、收益大的条目排最前面。

2026 年 9 月 7 日上线,三周内 2366 stars,195 人 fork。在线检索页已经开放,可以按关键词、章节、证据等级、成本三个维度(花钱/花时间/要毅力)组合筛选。


每条建议长什么样

以”把家里的食盐换成低钠盐”为例:

  • 成本:每袋贵几元
  • 说人话:两万人的随机试验里,换成低钠盐的人五年内死亡概率低约 12%,中风低约 14%
  • 收益:脑卒中降 14%,心血管事件降 13%,总死亡率降 12%
  • 证据等级:A(来自 RCT,有具体数字)
  • 来源:Neal B, et al. (2021). NEJM. DOI 链接
  • 备注:争议——附上了反方研究(生态学研究,证据等级低)和限制条件(肾功能不全者不适用)

这是全书的标准格式:每条都有”说人话”翻译一行,把 HR/RR/置信区间这些统计数字还原成”低约两成”这种判断,供直接使用;原始数字和置信区间保留,供你自己核实。


四种资源,不跨口径比较

全书优化四类东西,且明确不跨口径换算:

资源衡量方式
寿命总死亡率(全因死亡率)变化、特定死因下降幅度
时间与精力每天/每周节省的小时数、注意力消耗降低
金钱金额:万元级/数百至数千/几十元
人身自由避免刑事责任/拘留/民事纠纷

“总死亡率降 12%“和”每年省 500 元”不在一把尺子上,不做换算。这避免了很多伪量化。


31 章覆盖范围

健康与生命

  • 不要早死(外因、燃气中毒、疫苗、筛查、家庭应急)
  • 不要慢慢死(烟酒、运动、睡眠、饮食、久坐)
  • 急救先做什么(心脏骤停 CPR、大出血、骨折固定、一氧化碳、火灾溺水迷路蛇咬)
  • 得了慢性病之后(门诊慢特病跨省、服药依从、并发症筛查)

钱与法律

  • 不要浪费钱(订阅/彩票/费率/保险/直播带货)
  • 没钱时怎么活(救助、补贴、欠薪维权)
  • 别把自己搭进去(AI 换脸诈骗、被诬告后的救济、彩礼婚前财产)
  • 普通人容易踩的法律红线(谣言、传播色情、兼职洗钱、高空抛物、仿真枪、无人机)
  • 程序员容易踩的红线(外挂、爬虫、抢票脚本、删库、带走源码、开源许可证、ICP 备案)
  • 创业与做生意(主体选择、涉税诈骗、量产知识产权、退场)
  • 租房与买房(押金被扣、暴力腾退、资金监管)

人生大事

  • 恋爱和结婚划不划算(择偶策略、异地恋、健康账/时间账/钱账、退出成本)
  • 养孩子划不划算(育儿补贴、产假、三期保护、时间账)
  • 怀孕和生产(叶酸、产检、生育保险、立刻去医院的信号)
  • 刚出生的孩子(安全睡眠、疫苗时序、维生素 K)

信息与工作

  • 账号与信息安全(二次验证、SIM 卡、银行卡盗刷、人脸识别)
  • 在职、离职和工伤(加班费算法、N/2N 补偿、工伤认定、职业病体检)
  • 学什么技能划算(教育回报率、山寨证书、抗自动化能力、紧缺职业查询)
  • 做一个网站或平台(支付红线、ICP/直播资质、内容治理、数据出境)
  • 十八岁之后有哪几条路(当兵、基层服务、自考、灵活就业社保)

证据分级与性价比档

证据等级(回答”数字可不可信”):

等级含义数量
A荟萃分析、大型队列或 RCT,有具体数字323 条
B有研究支撑但难以量化,或小样本/单一研究126 条
C作者经验或普遍共识,无直接文献49 条

另有 45 条标注”争议”(A/B 级但附有反方证据),39 处标注”TODO 待核实”。

性价比档(回答”值不值得做”,与证据等级正交):

档位定义数量
极高收益大且三项成本(钱/时间/毅力)全为零88 条(18%)
收益大但成本较低,或收益中但成本为零248 条(50%)
一般其余——需要自己权衡那笔花销162 条(33%)

作者说得很直白:“一般”不等于不该做,全书条目都是建议做的,只是这一档需要你自己权衡。


一些值得关注的具体条目

极高性价比、证据 A 级的交集

  • 把家里的盐换成低钠盐(总死亡率降 12%,多花几元/袋)
  • 学会 CPR(中国院外心脏骤停 79.2% 发生在家里,首要受益人是家人)
  • 戒烟(若抽烟)
  • 家里装烟雾报警器和一氧化碳报警器

程序员看了可能会心一跳的条目:爬虫在什么情况下会被定罪、开源许可证 GPL 的传染性边界在哪、带走前公司源码的法律后果、ICP 备案漏办的处罚档次——这些在”程序员和技术人容易踩的红线”一章里,全部注明了证据等级和来源文件。

一条测试作者认真程度的条目:书里对”运动能延寿多少”这个问题,不只给了个结论,还列出了 HR 和 95% 置信区间,并注明”观察性研究,不能完全排除反向因果”——生病的人运动少,所以不能简单认为运动少导致死亡。


在线检索页的用法

打开 https://eternity4719.github.io/HowToLiveBetter/ 之后:

  • 直接搜关键词(如”低钠盐”、“工伤”、“ICP”)
  • 勾选证据等级 A,只看 323 条最硬的
  • 勾选性价比”极高”,得到 88 条不花钱不花时间还有大收益的
  • 三个成本维度(花钱/花时间/要毅力)可以任意组合
  • 把鼠标放到带虚线的统计术语上会弹出解释(HR、RR、荟萃分析等)

值得单独看的长文

四篇长文附在 docs/ 目录下,独立成文:

  • 结婚划不划算:完整的恋爱婚育账本,时间/钱/法律后果分开算
  • 家庭应急装备清单:按优先级排的清单,每样写了用途和不买的理由
  • 遇到陌生人出事该不该停:把免责条款、被讹风险、救人风险分开讨论
  • 做平台要办哪些证:从 ICP 到直播资质到数据出境,按证件类型列清单

项目的一个特点

作者在设计上做了一件不常见的事:陌生人受益排在你自己受益后面,两种不合并计算

全书条目的受益对象从高到低分四档:① 你自己、② 配偶和直系亲属、③ 朋友/同事/其他亲属、④ 陌生人。急救那一章明确写着:中国院外心脏骤停 79.2% 发生在家里,学 CPR 首先是为了用在家人身上;“看到陌生人溺水自己不下水”本身就是自保规则,不是冷漠,是为了防止旁观者变成第二个伤者。

帮陌生人要不要出手,书里的条目会把免责条款、自保动作和风险都写出来,但不替你做决定。


开源代码与模型仅供学习、勿直接用于工作。

© 2026 Author: Mycelium Protocol. 本文采用 CC BY 4.0 授权——欢迎转载和引用,须注明作者姓名及原文链接,不得去除署名后以原创发布。

🇬🇧 English

📌 Repository: https://github.com/eternity4719/HowToLiveBetter Live search: https://eternity4719.github.io/HowToLiveBetter/ License: Unlicense | Stars: 2,366 | Forks: 195


Most life advice has two problems: either it has no sources, or it has sources but doesn’t tell you whether it’s worth doing.

HowToLiveBetter does both at once: 498 tips, each stating what it costs, what it returns, how strong the evidence is, and where the source is — ranked by cost-effectiveness, not by category. Low-cost, high-benefit entries come first.

Launched September 7, 2026. Hit 2,366 stars and 195 forks within three weeks. A live filterable search page is already open.


What each entry looks like

Using “swap table salt for low-sodium salt” as an example:

  • Cost: A few extra yuan per bag
  • Plain language: In a 20,000-person randomized trial, households that switched to low-sodium salt had ~12% lower mortality and ~14% lower stroke risk over five years
  • Benefit: Stroke −14%, cardiovascular events −13%, all-cause mortality −12%
  • Evidence grade: A (RCT with concrete numbers)
  • Source: Neal B, et al. (2021). NEJM. DOI link
  • Note: Contested — includes the counterargument (ecological study, lower evidence grade) and contraindications (renal impairment)

Every entry has a “plain language” line that translates HR/RR/confidence intervals into “roughly 20% lower” — for direct use without needing stats knowledge. The full numbers are kept for those who want to verify.


Four resources, no cross-metric comparison

The guide optimizes four things, and explicitly refuses to convert between them:

ResourceMeasured as
LifespanAll-cause mortality change, specific cause reduction
Time and energyHours saved per day/week, attention drain reduced
MoneyAmount: tens of thousands / hundreds to thousands / tens of yuan
Personal freedomAvoiding criminal liability / detention / civil disputes

“12% lower mortality” and “save 500 yuan/year” are on different scales — no conversion is made. This avoids a lot of pseudo-quantification.


31 chapters

Health and life: don’t die early (external causes, gas poisoning, vaccines, screening, home emergency kit), don’t die slowly (smoking/alcohol/exercise/sleep/diet/sitting), what to do in emergencies (CPR, heavy bleeding, fracture stabilization, CO, fire, drowning, snake bites), living with chronic illness.

Money and law: don’t waste money (subscriptions, lottery, fees, insurance, live-commerce traps), surviving without money (assistance programs, unpaid wage recovery), self-protection (deepfake scams, being falsely accused, engagement gifts, premarital property), common legal landmines (rumors, obscenity distribution, unwitting money laundering, high-rise objects, toy guns, drones), developer-specific legal risks (scrapers, game bots, ticket bots, deleting databases, taking source code when leaving, GPL license infection, ICP filing).

Life events: relationships (partner selection strategy, long-distance, financial/time/health accounting), pregnancy and birth (folate, prenatal care, maternity insurance, emergency signals), caring for newborns (safe sleep, vaccine timing, vitamin K), aging parents (guardianship, wills, elder care traps).

Work and skills: job security and termination (overtime calculation, N and 2N severance, work injury recognition, occupational disease exams), what skills are worth learning (ROI by credential type, automation resistance, how to find shortage occupations), building a website or platform (payment red lines, ICP/livestream licenses, data export rules), paths at 18 (military service, community programs, adult education, gig economy social insurance).


Evidence grades and value tiers

Evidence grade (answers “can I trust this number?”):

GradeMeaningCount
AMeta-analysis, large cohort, or RCT; concrete numbers323
BResearch-backed but hard to quantify, or small/single study126
CAuthor experience or common consensus; no direct literature49

45 entries marked “contested” (Grade A/B with counterevidence included), 39 marked “TODO: needs verification.”

Value tier (answers “is it worth doing?”, orthogonal to evidence grade):

TierDefinitionCount
TopHigh benefit AND all three costs (money/time/willpower) = zero88 (18%)
HighHigh benefit at low cost, or medium benefit at zero cost248 (50%)
StandardEverything else — personal tradeoff required162 (33%)

The author is clear: “Standard” doesn’t mean skip it — all entries are recommendations. It just means this tier requires you to weigh the costs yourself.


The search page

Open https://eternity4719.github.io/HowToLiveBetter/:

  • Search by keyword (“low-sodium salt,” “work injury,” “ICP”)
  • Filter to Grade A only (323 entries with the hardest evidence)
  • Filter to “top value” (88 entries with large benefit and zero cost)
  • Combine three cost dimensions (money / time / willpower) in any combination
  • Hover over statistical terms for inline definitions (HR, RR, meta-analysis, etc.)

A design choice worth noting

The guide explicitly ranks whose benefit counts, from high to low: ① yourself, ② spouse and immediate family, ③ friends/colleagues/extended family, ④ strangers. Benefits to different tiers are not added together.

The first-aid chapter states plainly: 79.2% of out-of-hospital cardiac arrests in China happen at home — learning CPR is primarily so you can use it on your own family. “Don’t jump in to save a drowning stranger yourself” is a self-protection rule, not indifference — it’s to prevent the bystander from becoming the second victim.

For entries involving strangers, the guide spells out the liability protections, the self-protection moves, and the risks, but doesn’t make the decision for you.


Open-source code is for learning purposes only — do not use directly in production work.

© 2026 Author: Mycelium Protocol. Licensed under CC BY 4.0 — free to share and adapt with attribution.

💬 评论与讨论

使用 GitHub 账号登录后发表评论

关于本站 · 免责声明

🍄 Mushroom Research Blog 是非营利、免费公开的个人科技观察博客与公众号 XStack18,不接受商业合作、不代表任何企业或机构立场,也不谋求商业利益。我们以个人视角客观中立地记录和分析 AI、Web3 等领域的最新模型发布与技术动态——不止转述新闻标题或二手信息,而是给出有独立思考的深入分析,希望帮更多人获得有价值的一手科技认知。

⚠️ 文中介绍的开源代码与模型,仅供学习交流与技术借鉴。它们大多仍处于早期阶段,有待进一步研究和验证,请勿直接用于工作或生产环境;如需采用,请先自行充分测试,并核实其许可证与安全性。
Open-source code and models featured here are shared for learning and reference only. Most are early-stage and still need further study and verification — please don't use them directly in your work or in production. Test them thoroughly and check their licenses and security first.

  1. 本站文章均为作者基于公开信息的个人研究与观点整理,不代表文中提及的任何公司、产品、模型的官方立场,未与其构成商业关联或合作关系。
  2. 科技行业信息更新极快,我们尽力保证内容准确、及时,但不对完整性、实时性做绝对保证,具体请以相关企业/项目官方公告为准。
  3. 文中引用的第三方商标、产品名称、图片、数据等版权归原权利人所有,我们会尽量注明来源;如你认为存在版权疑问或侵权,请通过下方邮箱联系我们,收到通知后会尽快核实处理(更正、加注来源或删除)。
  4. 文章内容仅为技术科普与个人观点,不构成投资、法律或其他专业建议,据此进行任何决策的后果需自行判断和承担。

📮 侵权 / 勘误 / 合作咨询:[email protected]