CMH MATH PATH — LESSON 1: WHY CROWDS ARE PREDICTABLE

The Law of Large Numbers & Central Limit Theorem

PART 1: THE LAW OF LARGE NUMBERS — WHY MILLIONS ACT LIKE ONE

The Cookie Jar Experiment

Imagine you have a giant jar with 1 million marbles. Half are red (angry), half are blue (calm).

Experiment 1: Pull out ONE marble.

  • Can you predict the color? No. It could be red or blue.
  • This is like predicting one person’s behavior — nearly impossible.

Experiment 2: Pull out 10 marbles.

  • Maybe you get 6 red, 4 blue. That’s 60% angry.
  • Still unreliable — could easily be 7-3 or 5-5.

Experiment 3: Pull out 10,000 marbles.

  • Now you’ll get very close to 5,000 red, 5,000 blue.
  • The percentage will be almost exactly 50% angry.

Experiment 4: Pull out 100,000 marbles.

  • The count will be even closer to 50/50.
  • You can now say with confidence: “About 50% of marbles are red.”

What Just Happened?

This is the Law of Large Numbers:

“When you look at more and more things, the average becomes more and more stable and predictable.”

Real CMH Example:

  • We can’t predict if Sarah will vote in the election.
  • But with 10 million voters, we can predict voter turnout within 1%.
  • Individual randomness cancels out in large groups.

The Math Behind It:

Individual Chaos = High
Group Chaos = Individual Chaos ÷ √(Number of People)

For 100 people: Group Chaos = High ÷ 10 = Much Lower
For 1,000,000 people: Group Chaos = High ÷ 1000 = Almost Zero

The chaos shrinks as the group grows!

PART 2: THE CENTRAL LIMIT THEOREM — THE MAGIC BELL CURVE

The Birth of a Bell Curve

Let’s play a dice game with society’s “happiness score” (1-6, like dice):

Step 1: Ask ONE person: “How happy are you today, 1-6?”

  • They roll their personal “mood dice.”
  • You might get a 4.

Step 2: Ask TWO people, average their scores:

  • Person 1 rolls 2, Person 2 rolls 6 → Average = (2+6)/2 = 4
  • This average could be 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 5.5, or 6

Step 3: Ask FIVE people, average their scores:

  • Possible averages: 1.0, 1.2, 1.4, …, 5.8, 6.0
  • Notice: Extreme averages (1.0 or 6.0) become RARE.
  • Why? To get average 1.0, all 5 people must roll 1! Unlikely.

Step 4: Ask 100 people, average their scores:

  • Now the averages cluster around 3.5 (the true average of 1-6)
  • They form a beautiful bell curve!

Why This Is Magic

The Central Limit Theorem says:

“No matter how weird or messy your starting data is, if you take enough averages of groups, those averages will always form a bell curve shape.”

Even if people’s moods are all over the place…

  • Some are super happy (6)
  • Some are miserable (1)
  • Some are medium (3, 4)

The AVERAGE of large groups always settles into this pattern:

Most averages are in the middle, fewer are at the edges.

The Three Rules of the Bell Curve

  1. The Peak = Most likely outcome (the true average)
  2. The Width = How much variation is normal
  3. The Tails = Rare but possible extremes

CMH Example: Average Income

  • Survey 1,000 people, calculate average income
  • Do this 100 times with different groups of 1,000
  • Plot the 100 averages → Bell curve!
  • Center = True average income of whole population
  • Width = How much samples might vary
  • Far right tail = Rare case where you accidentally survey only billionaires

PART 3: WHY THESE IDEAS ARE CMH SUPER POWERS

Super Power 1: We Can Measure What We Can’t See

We don’t need to ask every single person in a country:

  • “Are you angry at the government?”
  • “Will you protest next month?”

Instead, we ask 1,000 carefully chosen people and get a reliable average.

Why it works:

  • If the true protest willingness is 30%…
  • Our 1,000-person survey will show 28-32% (close enough!)
  • The bell curve tells us exactly how close “close enough” is

Super Power 2: We Can Spot Real Changes

Suppose last month, 25% were willing to protest.
This month, our survey says 31%.

Is this a real change or just random noise?

The bell curve tells us:

  • If the width of our bell curve is ±3%…
  • Then 25% → 31% is 2 standard deviations apart
  • This means 95% chance it’s a real change, not luck!

Super Power 3: We Can Predict Ranges, Not Just Guesses

Instead of saying:

“Unemployment next year will be 5.2%”

We say:

“Unemployment will likely be between 4.8% and 5.6% (95% confidence)”

This “range prediction” comes directly from the width of the bell curve.

PART 4: HANDS-ON BELL CURVE LAB

Exercise: The Mood of a Virtual Town

Let’s build this in a spreadsheet:

Step 1 — Make “People” with Random Moods:

Column A (Person 1-500): =RANDBETWEEN(1, 6)

Each cell is one person’s mood today (1=terrible, 6=ecstatic).

Step 2 — Create Survey Teams:

Column B (Team 1 Average): =AVERAGE(A1:A20)
Column C (Team 2 Average): =AVERAGE(A21:A40)
... Make 25 teams total

Each “survey team” asks 20 people and calculates average mood.

Step 3 — Watch the Bell Curve Form:

  1. Press F9 to generate new random moods
  2. Watch the 25 team averages change
  3. Notice they always cluster in the middle (around 3.5)
  4. Extremes (below 2.5 or above 4.5) are rare!

Step 4 — Calculate the Magic Numbers:

Mean of all team averages: =AVERAGE(B1:B25)
Standard Deviation (width): =STDEV(B1:B25)

The Magic Rule:

  • 68% of team averages will be within ±1 standard deviation
  • 95% will be within ±2 standard deviations
  • 99.7% will be within ±3 standard deviations

Try it! Count how many team averages fall within each range.

PART 5: COMMON MISTAKES & HOW TO AVOID THEM

Mistake 1: The “Small Sample” Error

  • Wrong: “I asked 3 people, 2 were angry → 67% of country is angry!”
  • Right: “With 3 people, uncertainty is huge. Could easily be 30-90%.”

Rule: Sample size matters more than percentage!

Mistake 2: The “Weird Group” Error

  • Wrong: Survey only people at a protest rally about protest attitudes
  • Right: Survey random people from the whole population

Rule: Your sample must represent the whole group!

Mistake 3: The “One Time” Error

  • Wrong: One survey tells the whole story
  • Right: Track changes over time — is today different from yesterday?

Rule: One measurement = snapshot. Many measurements = movie.

PART 6: YOUR CMH FIELD GUIDE TO LARGE NUMBERS

Quick Reference Table

If Your Group Size Is:You Can Expect:Good For:
< 30 peopleWild fluctuations, unreliable averagesCase studies, stories, not predictions
30-100 peopleSome stability, but still noisyPreliminary surveys, early warnings
100-1,000 peopleReasonable estimates (±5-10%)City-level predictions, policy testing
1,000-10,000 peopleGood precision (±1-5%)National trends, election forecasts
> 10,000 peopleHigh precision (±0.1-1%)Scientific research, CMH models

The Golden Rules:

  1. N > 1,000 for national predictions
  2. Track changes, not single numbers
  3. Always report the range, not just the average
  4. Check your sample represents the population

YOUR MATH HOMEWORK:

Exercise 1: The Coin Flip Nation

  1. Flip a coin 10 times. Record % heads.
  2. Flip 100 times (or simulate in spreadsheet). Record % heads.
  3. Which was closer to 50%? Why?

Exercise 2: Find the Bell Curves

Look at these real-world examples. Which would form bell curves if we averaged enough samples?

  • Height of adults in your country ✓ (Yes, forms bell curve)
  • Number of social media posts per person per day
  • Distance people live from their workplace
  • Number of siblings people have

Exercise 3: Design a CMH Survey

You want to know: “What % of young people in your city would join a climate protest?”

  1. How many people would you survey? Why?
  2. How would you choose them randomly?
  3. If your survey says 40%, what range might the true answer be?

Exercise 4: Spot the Error

Find a news article that says something like:

  • “70% of Americans believe X” (from survey of 500 people)
  • Calculate the margin of error: ±1/√N = ±1/√500 = ±4.5%
  • So really it’s 65.5% to 74.5%
  • Does the article mention this range?

REAL CMH APPLICATIONS RIGHT NOW:

  1. Election Polling: Surveys of 1,000-2,000 voters predict national elections within ±3%
  2. Economic Indicators: Unemployment rate from 60,000 household surveys
  3. Disease Tracking: COVID rates from random testing samples
  4. Social Unrest Forecasting: Protest likelihood from social media sentiment of millions

KEY TAKEAWAY:

Individuals are unpredictable. Small groups are noisy. But large populations obey mathematical laws. The chaos of human freedom, at sufficient scale, becomes the order of statistical law. This is the foundation upon which all Computational Macrohistory is built.

When you understand this, you stop seeing society as 8 billion separate stories, and start seeing it as a mathematical system with predictable properties.

Next Lesson: Probability Distributions — Beyond the Bell Curve
(Where we’ll learn to handle revolutions, pandemics, and other “rare but earth-shaking” events that live in the tails of the distribution.)

Thought to Ponder: If individuals are free but crowds are predictable, what does that say about human freedom? Can we be both free and predictable at the population level?

Similar Posts