Skip to content
How to ThinkIn the Age of AI

Question

Why does warranted confidence fail?

beginner

Quick Answer

Treating the feeling of confidence as evidence of correctness. You finish testing a schema, find three supporting cases, and feel certain — but you never checked for disconfirming evidence, never tested the boundary conditions, never asked whether the supporting cases were independent. High confidence without a validation trail is just comfort wearing the costume of knowledge.

The antidote is a simple question: 'What would change my mind?' If you cannot answer it, your confidence is not warranted — it is just familiar.

The fix: Pick one belief you hold with high confidence — something you'd rate at 90% or above. Write it down as a testable claim. Now list the actual evidence you have for it: not impressions, not 'everyone knows this,' but specific observations, experiences, or data points. Count them. Then list any counter-evidence you've encountered and dismissed or forgotten. Assign a new confidence level based solely on the evidence inventory. Most people find a gap of 20-40 points between their felt confidence and their evidence-warranted confidence. That gap is the lesson.

The underlying principle is straightforward: Confidence based on tested schemas is categorically different from confidence based on untested assumptions.

Learn more in these lessons