Skip to main content

Probability

What's my base rate before I saw this evidence?

Prompts

Before any decision under uncertainty:

  1. "What's my base rate?" — Start with how often this happens in general
  2. "What would change my confidence by 10%?" — Identify the evidence that matters
  3. "Am I updating enough—or too much?" — Calibrate against your tendency
  4. "What would prove me wrong?" — Name the falsifying evidence

Core Principles

Explicit Priors — Start with base rates before incorporating new evidence.

  • In the Steve example: 20:1 farmer-to-librarian ratio anchors the calculation
  • In investing: 90% of crypto startups fail should anchor initial risk assessments

Iterative Updates — Use Bayes' formula to update multiplicatively.

  • If a startup CEO has a strong track record (evidence), update success probability
  • Don't reset beliefs—compound them

Process Over Outcomes — Evaluate decisions based on information available at the time.

  • A 16.7% posterior can be rational even if wrong
  • Good process sometimes produces bad outcomes—that's probability
tip

Decisions must be assessed on their process not their outcomes

The Framework

StepQuestionAction
1. PriorsWhat's the base rate?Quantify before seeing evidence
2. EvidenceWhat new information exists?Identify likelihood ratios
3. UpdateHow much should this change belief?Apply Bayes multiplicatively
4. DecisionWhat does the posterior imply?Act on updated probability

Practice Applications

ScenarioBayesian ApproachExample
InvestmentUpdate valuations using market signalsRebalance if inflation likelihood spikes
NegotiationsModel opponent priors, update with behavioral cuesAdjust offers if counterparty hesitates
Product LaunchesUse A/B testing to iteratively refineBeta test → Measure → Scale

Execution Checklist

  • Explore/Exploit — Balance experimentation (updating priors) with leveraging known strategies
  • Reversible vs Irreversible — Apply Bayesian for reversible; game theory for irreversible
  • Quantify Intuition — Assign confidence scores (e.g., "60% sure") and track calibration
  • Sensitivity Analysis — "What if the ratio is 50:1 instead of 20:1?"
  • Premortems — "Why might this NOT be true despite my evidence?"

Combating Biases

Ego Mitigation — Your prior beliefs aren't sacred. Evidence updates them.

Information Asymmetry — When one party knows more, treat gaps as signals to investigate.

Anchoring — Be aware that the first number you see biases subsequent estimates.

danger

Your ego is the enemy when it comes to making good decisions

Learn Through Games

Play games that teach probability through stakes:

  • Poker — Read opponents, size bets, manage bankroll
  • Prediction Markets — Put money where your beliefs are
  • Calibration Training — Track predictions, measure accuracy

Bets gamify commitment. They force:

  • Explore or exploit?
  • One-way or reversible?

Context