WARM-UP
Before you decide
State your estimate in percentages. “Likely” becomes something you can later learn from.
YOUR DECISION PRACTICE
Dashboard
THE CORE DISTINCTION
Judge the process, not only the result.
A good decision can lose. A bad decision can win. Review both separately.
AT A GLANCE
Practice dashboard
NEXT ACTION
Review queue
RECENT THINKING
Latest decisions
DECISION JOURNAL
Make your thinking auditable
Capture what you knew, what you believed, and what could prove you wrong—before hindsight edits the story.
YOUR RECORD
Decision journal
A record of beliefs at the time they were made.
LEARNING LOOP
Close the feedback loop
Record what happened without allowing the result to erase whether the decision process was sound.
Select a decision to review
Choose an open or due decision from the queue. Your original forecast will remain visible while you review it.
PATTERN DETECTION
Insights, not verdicts
Small samples are noisy. Use these signals to generate better questions—not to declare yourself permanently accurate or inaccurate.
CALIBRATION
Forecast vs. outcome
For resolved binary forecasts only. Perfect calibration means an 80% bucket succeeds about 80% of the time.
PROCESS
Decision quality
BLIND SPOTS
Common pattern prompts
PRACTICE MANUAL
Field guide
Use the prompts to build a habit of explicit probabilities, dissent, and process reviews.
Wanna bet?
Translate claims such as “probably,” “definitely,” and “unlikely” into a probability. The purpose is not to be right on demand; it is to learn how your confidence maps to reality.
Prompt: “What probability would I put on this, and what would I stake to show I mean it?”
Separate result from process
An outcome is evidence, but it is not a complete grade. Ask whether your information, reasoning, alternatives, and execution were good before considering whether luck helped or hurt.
Prompt: “Would I endorse this decision again with the information I had then?”
Use the outside view
Start with a reference class: comparable decisions, projects, people, or historical cases. Then adjust cautiously for what is genuinely distinctive about your case.
Prompt: “What usually happens in situations like this?”
Invite dissent
Seek reasons you may be wrong before you commit. Treat disagreement as information rather than a threat to your identity or status.
Prompt: “What would the strongest case against this forecast be?”
Run a pre-mortem
Assume the decision failed and work backward. Concrete failure paths are easier to prevent or monitor than generic anxiety.
Prompt: “It is a year later and this failed. What most likely happened?”
Update, do not defend
When the evidence changes, your estimate should change too. Updating is not inconsistency; refusing to update is often a way of protecting a story.
Prompt: “What evidence would move me by 10 percentage points?”