LESSON 4: PROBABILITY — WHY WE SAY “MAYBE” INSTEAD OF “YES/NO”

In Lessons 2 and 3, we learned to measure a society’s vital signs and model how they change.
But if you’ve ever tried to predict the weather, you know forecasts say “60% chance of rain” — not “It will rain.”

That’s because complex systems are full of uncertainty.
In CMH, we don’t predict the future — we predict possible futures and assign each a probability.

1. WHY CERTAINTY IS IMPOSSIBLE

Let’s go back to our simple “Unrest Engine” from Lesson 3:

U(t+1) = U(t) + 0.2 × Y(t)

This assumes nothing unexpected happens.
But in reality, many unexpected things do happen:

  • A natural disaster strikes
  • A charismatic leader emerges
  • A new technology goes viral
  • A foreign power intervenes

These are called exogenous shocks — surprises from outside the model.

So, our equation should really be:

U(t+1) = U(t) + 0.2 × Y(t) + ε

Where ε (epsilon) is a random shock.
Some years ε is small (≈ 0). Some years it’s huge (± 5).

2. THE “PROBABILITY CLOUD”

Because of ε, we can’t draw one clean line into the future.
Instead, we draw a cloud of possibilities.

Unrest Next Year (U(t+1))
    ↑
    |       • • • • •
    |     • • • • • • •
    |   • • • • • • • • •
    | • • • • • • • • • • •
    +---------------------→ Unemployment Now (Y(t))

Each dot is one possible future.
The dense middle is the most likely outcome.
The spread-out edges are less likely, but possible.

This cloud is called a probability distribution.

3. READING THE PROBABILITY CLOUD: THE “CRISIS ZONE”

Let’s say we define a crisis as U > 7.
We can’t say “Crisis will happen.”
We can say: “There is a 40% chance that U will be > 7 next year.”

How do we get that number?

  1. We run our model 10,000 times, each time adding a different random shock ε.
  2. We count how many runs ended with U > 7.
  3. If 4,000 runs had U > 7, then:
P(Crisis) = 4,000 / 10,000 = 0.4 = 40%

This method is called Monte Carlo simulation — named after the famous casino, because it uses randomness like a game of chance.

4. A SIMPLE PROBABILITY TOOL: THE “SIGMOID” FUNCTION

Sometimes we want to turn a score (like our Unrest Index) directly into a probability.
We use a special function called the sigmoid.

It looks like this:

Probability of Crisis
    ↑
1.0 |                ┌───┐
    |               /
0.5 |             ──
    |             /
0.0 └------------┘
    ←───────────────→ Unrest Index (U)
          5           10

What it does:

  • If U is low (≈ 2), probability is near 0%.
  • If U is near the threshold (≈ 5), probability is 50%.
  • If U is high (≈ 8), probability is near 100%.

The formula is:

P(Crisis) = 1 / (1 + e^(-(U - 5)))

Don’t worry about the math — just see the curve.
It’s a smooth, sensible way to turn a number into a “maybe.”

5. PUTTING IT ALL TOGETHER: A CMH PREDICTION REPORT

A real CMH forecast doesn’t say “Revolution in 2026.”
It says something like this:

COUNTRY X — 5-YEAR OUTLOOK

  • Baseline Unrest Index (U) today: 4.2
  • Projected U in 5 years: 6.8 ± 1.5 (range: 5.3 to 8.3)
  • Probability of entering crisis zone (U > 7): 35%
  • Most likely trigger: Rising youth unemployment
  • Confidence in forecast: Medium (due to climate volatility)

This is honest, useful, and actionable.
Policymakers can’t prevent a “certain” future — but they can lower the probability of a bad one.

6. THE GOLDEN RULE OF CMH

All predictions are probabilities. All probabilities have error bars. All error bars grow with time.

The further we look ahead, the wider the “cloud” becomes, until the forecast is no better than a guess.

YOUR CMH HOMEWORK:

  1. Estimate Your Own Probability: On a scale of 0–10, how stressed is your country today? Use the “sigmoid” idea: what’s your gut-feel percentage chance of a major protest or crisis in the next 2 years?
  2. Find a Random Shock: Look at the news. Identify one event this week that would be a “random shock” ε in a CMH model (e.g., a sudden election, a natural disaster, a viral social movement).
  3. Think About Error Bars: If CMH says “40% chance of crisis,” what should a leader do? Prepare? Ignore it? What’s the right way to use a probabilistic forecast?

When you’re ready, we’ll move to Lesson 5: Testing the Model — How to Know If We’re Wrong (And Why That’s Good).

This is where science happens — not in being right, but in learning from mistakes.

Similar Posts