Base rates: the number you don't want to look at

Some uncomfortable numbers:

• Five-year startup survival rate: ~50%
• First-year restaurant failure rate: ~60%
• Job changes that yield more than a 20% raise: ~30%
• Sustained 10kg weight loss for over a year: under 10%

These are base rates — historical probabilities of an event class in large samples. Kahneman keeps reminding us: humans systematically ignore base rates when forecasting.

Why they get ignored

Because base rates are "outside numbers," but when you decide, you're thinking your story.

"Startup 50% failure rate? Sure, but I have specific advantages — my team, my idea, my execution."
"Restaurant 60% failure? Sure, but I'll do it differently — my menu, my location, my customer focus."

These additions might all be true, but they are stories, not numbers. Stories are persuasive; numbers are accurate — and you're betting on accuracy.

The right order

Don't start with "my specialness." Instead:

1. Find the base rate — historical success rate for similar situations
2. List your differences — advantages or disadvantages absent from the statistical sample
3. Quantify the difference — how far do these pull you from the base rate? 10%? 20%? Double?
4. Honest check — if a friend used the same argument to convince you, would you accept it?

After these four steps, your estimate is usually one notch lower than first instinct. That's the real probability.

When to use it

• Predicting your or others' success rates
• Evaluating "can I really pull this off"
• Estimating timelines, costs, progress

When not to use it

When your situation truly diverges from the reference class — you possess advantages absent from the sample. But "I'm special" itself needs to be proven first.

Full questioning lives on Hone's framework page.