Ergodicity Explained: Why the Average Lies to You
遍历性解释：为什么平均数在骗你
EN: If a concept could change one thing about how you make decisions, it might be this one.
CN: 如果说有一个概念能改变你做决策的方式，可能就是它了。
The Core Idea in One Sentence | 一句话核心
EN: Ergodicity asks a simple question: Does what happens to the group also happen to each individual over time?
CN: 遍历性提出了一个简单的问题：发生在群体身上的事，是否也发生在每个个体身上？
If yes → the system is ergodic.
如果答案是 → 系统是遍历的。
If no → the system is non-ergodic.
如果答案是否 → 系统是非遍历的。
That's it. The implications are enormous.
The Casino Example | 赌场例子
Imagine a casino with 100 people playing a coin-flip game:
假设一个赌场里有 100 个人在玩掷硬币游戏：
The Rules:
规则：
- Start with $100 | 起始资金 $100
- Heads: win 50% (+$50) | 正面：赢 50%（+$50）
- Tails: lose 40% (−$40) | 反面：输 40%（−$40）
View 1 — The Group Average (Ensemble) | 视角一：群体平均（集合）
100 people flip once: 100 个人各翻一次：
- 50 get heads → $150 each | 50 人正面 → 各 $150
- 50 get tails → $60 each | 50 人反面 → 各 $60
- Average: $105** | **平均：$105
Looks like a 5% expected gain. Good bet, right?
看起来期望收益是 5%。好赌注，对吧？
View 2 — One Person Over Time (Time) | 视角二：一个人的时间历程
You play 100 times in a row: 你连续玩 100 次：
- Flip 1: $100 → tails → $60 | 第1次：$100 → 反面 → $60
- Flip 2: $60 → heads → $90 | 第2次：$60 → 正面 → $90
- Flip 3: $90 → tails → $54 | 第3次：$90 → 反面 → $54
- ...keep going... | ...继续...
- After 100 flips: ≈ $3.67 | 100 次后：≈ $3.67
You lose 96.3% of your money.
你损失了 96.3% 的本金。
The Paradox | 悖论
| Group Average | One Person Over Time |
|---|---|
| 群体平均 | 一个人随时间 |
| $105 (win) | $3.67 (lose) |
| $105（赢） | $3.67（输） |
How can the average win while every individual loses?
为什么平均在赢，每个个体却在输？
Because the average is not you.
因为平均数不是你。
The average includes people who got lucky and ran their money up to millions. But you can't be lucky across parallel universes. You live one timeline, and losses compound multiplicatively.
因为平均值包括了那些运气好、资金涨到几百万的人。但你不能平行宇宙里都走运。你只活一条时间线，而损失是乘法累积的。
This system is NON-ERGODIC. The ensemble average ≠ the time average.
这个系统是非遍历的。 集合平均 ≠ 时间平均。
The Math (Simple Version) | 数学（简化版）
Ensemble Average (Group) | 集合平均（群体）：
E = 0.5 × 1.5 + 0.5 × 0.6 = 1.05
Each flip, the group average grows by 5%.
每次翻转，群体平均增长 5%。
Time Average (Individual) | 时间平均（个体）：
T = √(1.5 × 0.6) = √0.9 ≈ 0.949
Each flip, your wealth multiplies by ~0.95. You lose 5% per flip.
每次翻转，你的财富乘以 ~0.95。你每次输 5%。
The geometric mean (time) is always ≤ the arithmetic mean (group). The gap is where the danger hides.
几何平均（时间）总是 ≤ 算术平均（群体）。差距就是危险所在。
Why This Matters in Real Life | 为什么这在现实生活中很重要
1. Investment Returns | 投资回报
A fund advertises "average annual return of 15%." But if the fund drops 50% in one bad year, you need a 100% gain just to get back to even. The advertised average doesn't reflect your actual experience.
某基金宣传"年均回报率 15%"。但如果基金在某一年暴跌 50%，你需要翻倍才能回本。宣传的平均数并不能反映你的真实体验。
2. Career Decisions | 职业决策
"Startup founders earn 3x more on average." But the distribution is wildly non-ergodic — a few outliers earn 100x, most earn nothing. The "average" founder is not you.
"创业者的平均收入是 3 倍。"但分布是极度非遍历的——少数人赚 100 倍，大多数人一无所获。"平均"创业者不是你。
3. Health & Survival | 健康与生存
"The average lifespan is 80 years." But this average includes people who live to 100. Your personal timeline is not an average — one fatal accident ends it completely.
"平均寿命是 80 岁。"但这个平均数包括了活到 100 岁的人。你的个人时间线不是平均数——一次致命事故就彻底结束了。
4. Russian Roulette | 俄罗斯轮盘赌
6 people play once: 5 survive, 1 dies. "Average survival rate: 83%."
6 个人各玩一次：5 人存活，1 人死亡。"平均存活率：83%。"
1 person plays 6 times: 0% survival.
1 个人玩 6 次：存活率 0%。
The most extreme example of non-ergodicity — where the absorbing barrier (death) destroys all future possibility.
最极端的非遍历性例子——吸收壁（死亡）摧毁了所有未来的可能性。
The Key Insight | 核心洞见
EN: In an ergodic system, the average tells you about your future. In a non-ergodic system, the average is a fiction — it only exists for an observer who can see all parallel outcomes simultaneously. You can't. You live one path, with no reset button.
CN: 在遍历系统中，平均数能告诉你未来的情况。在非遍历系统中，平均数是虚构的——它只存在于能同时看到所有平行结果的观察者眼中。你做不到。你只走一条路，没有重置按钮。
The practical rule: If there's a risk of "ruin" (going to zero — bankruptcy, death, irreparable damage), no positive expected value is worth the bet. Because once you hit zero, there are no more flips.
实用法则： 如果存在"归零"风险（破产、死亡、不可逆的损害），任何正的期望值都不值得下注。因为一旦归零，就没有下一次了。
How to Think in Non-Ergodic Terms | 如何用非遍历性思维思考
| ❌ Ergodic Thinking | ✅ Non-Ergodic Thinking |
|---|---|
| "The average return is positive" | "What happens to me over time?" |
| "平均回报是正的" | "我随时间会发生什么？" |
| "Most startups succeed" | "What's my personal probability?" |
| "大多数创业者成功" | "我个人的概率是多少？" |
| "I can recover from any loss" | "Is there a point of no return?" |
| "我能从任何损失中恢复" | "是否存在不可逆的临界点？" |
Practical Rules | 实用法则
Never risk ruin — No expected value matters if you can't play again
永远不要冒归零的风险——如果你无法再玩，任何期望值都没有意义Think in time averages, not ensemble averages — What happens to you repeatedly, not what happens to everyone once
用时间平均思考，不用集合平均——你重复发生什么，不是每个人一次发生什么Respect absorbing barriers — Death, bankruptcy, reputation destruction are game-overs, not temporary setbacks
尊重吸收壁——死亡、破产、声誉毁灭是游戏结束，不是暂时的挫折Prefer paths that keep you in the game — Survival > optimization
选择让你留在游戏中的路径——生存 > 优化
The Origin Story | 起源故事
Ergodicity comes from statistical mechanics (19th century physics). Physicists like Boltzmann needed to connect two ways of measuring the same system:
遍历性来自统计力学（19 世纪物理学）。玻尔兹曼等物理学家需要将测量同一系统的两种方式联系起来：
Ensemble average: Average across many identical systems at one moment
集合平均： 某一时刻许多相同系统的平均Time average: Average of one system over a long period
时间平均： 一个系统在很长时间内的平均
If they're equal → the system is ergodic. If not → non-ergodic. Most real-life systems (finance, life, career) are non-ergodic.
如果它们相等 → 系统是遍历的。如果不等 → 非遍历的。大多数现实系统（金融、人生、职业）都是非遍历的。
Nassim Taleb popularized the concept for everyday decision-making, arguing that most economics and finance models make the fatal mistake of assuming ergodicity where it doesn't exist.
纳西姆·塔勒布将这个概念推广到了日常决策中，他指出大多数经济学和金融学模型都犯了一个致命的错误——假设不存在的遍历性。
The One-Line Summary | 一句话总结
EN: The average is a statistical ghost — it describes a world you don't live in. Think in time, not in groups.
CN: 平均数是一个统计学的幽灵——它描述的世界你并不生活其中。用时间思考，不用群体思考。
Further reading: "Ergodicity Economics" by Ole Peters, Nassim Taleb's works on risk, and the Kelly Criterion for optimal betting.
延伸阅读：Ole Peters 的"遍历性经济学"、纳西姆·塔勒布的风险著作，以及凯利最优投注公式。
Comments (0)
Login to leave a comment.