Review decisions in three steps: re-read reasoning blind, predict outcome, then compare — this sequence defeats hindsight bias

Why This Is a Rule

Hindsight bias — "I knew it all along" — is one of the most robust and damaging cognitive biases. Once you know an outcome, your brain retroactively adjusts your memory of what you knew and predicted before the outcome, making the result seem more predictable than it was. This prevents learning from decisions because every outcome feels like it was foreseeable, eliminating the feedback signal that would improve future predictions.

The three-step review sequence is specifically designed to defeat hindsight bias by controlling information exposure. Step 1 (re-read reasoning with outcome hidden) forces you to re-enter your pre-decision mental state — the actual uncertainty you faced, the factors you weighted, the logic you followed. Step 2 (predict outcome from reasoning alone) generates a genuine prediction that reflects your current assessment of the reasoning quality — before outcome knowledge can contaminate it. Step 3 (reveal outcome and compare) creates a three-way comparison: original prediction, current prediction, and actual outcome.

The diagnostic value is in the gaps: if your current prediction matches the original → your reasoning quality is stable. If it differs → your current self has information your past self lacked (genuine learning) or is being influenced by something your reasoning doesn't capture. If the outcome differs from both predictions → the situation contained genuine uncertainty that no amount of reasoning would have resolved.

When This Fires

• When reviewing any decision journal entry (Decision journal entries need six fields captured before outcomes: date, decision, reasoning, prediction, confidence %, and your current state) after the outcome window has passed
• During monthly or quarterly decision journal review sessions
• When someone asks "would you make that same decision again?" — use this protocol to answer honestly
• When trying to distinguish good process from good luck

Common Failure Mode

Reading the outcome first: "Let me see how that decision turned out... ah, it went badly. Now let me re-read my reasoning to see where I went wrong." The reasoning review is now contaminated — you'll zero in on the factors that predicted the bad outcome and ignore factors that supported the decision. The actual error in your reasoning may not be the one that matches the outcome.

The Protocol

(1) Open a past decision journal entry. Cover or hide the expected outcome, confidence, and any post-decision notes. (2) Read only the decision, reasoning chain, and state. Re-enter the pre-decision mindset. (3) Based only on the reasoning you just re-read, predict: "Given this reasoning, what outcome would I expect?" Write your prediction. (4) Now reveal the original expected outcome and confidence percentage. Compare your current prediction to your original prediction. (5) Now reveal the actual outcome. Compare all three: original prediction, current prediction, actual outcome. (6) Extract the lesson: was the reasoning sound regardless of outcome? Was the confidence well-calibrated? Did the state (tired, stressed) correlate with worse reasoning? Record the meta-lesson.