Answer a question to begin.
NOW OPEN · ADMISSION: AN OPEN MIND
HOW SURE
ARE YOU,
REALLY?
Welcome to the only arcade where “I don’t know” is a perfectly good move. Test 12 facts. Place your confidence on the dial. Learn the difference between being right and being well-calibrated.
✓ No ranking of people✓ No account✓ No tracking
EST. IN THE SPACE BETWEEN
“I THINK” & “I KNOW”
01 / YOUR TICKET TO REALITY
The rules are simple.
01
Pick an answer
Three plausible options. You can also pass. There is no prize for pretending.
02
Price your certainty
Set 0–100% for the chance your selected answer is correct. This is a forecast, not a personality trait.
03
Check your receipt
See the evidence, your prediction's Brier error and how confidence differs from accuracy.
02 / THE MAIN ATTRACTION
THE QUESTION BOOTH.
00/ 12 TICKETS
STAMPED
STAMPED
TICKET 01 / 12SCIENCE
Choose one statement. Your confidence comes next.
THE BOOTH IS CLOSED · THE LEARNING IS NOT
You made room
for maybe.
That is the whole point. The receipts below are about this small set of predictions, not a verdict on your abilities.
CURIOUS / NOT COMPETITIVENO. 0001
03 / YOUR PRIVATE READOUT
THE RECEIPTS.
Session-only. Nothing is uploaded, saved to an account, or compared with anyone else.
Lower is better for these binary forecast tickets.
Your predicted chance of being correct.
Expectation meets reality
EXPERIMENTAL / SMALL SAMPLEEach row groups answers by stated confidence. Compare the probability you assigned with the fraction actually correct.
Mean confidence Actual accuracy
Your chart appears after your first answered ticket. Passes are excluded.
No forecasts yet.
✳
What should you conclude?
Twelve questions can offer a useful reflection, not a stable estimate of anyone’s judgment. Play, look for surprises, and resist the urge to turn a small sample into an identity.
04 / THE MATH, MINUS THE MATH ANXIETY
THE BRIER
FUNHOUSE.
A confidence forecast is a small bet in the mathematical sense, not a wager. The Brier score asks: How far was your predicted probability from the observed outcome?
For one selected answer: (probability − outcome)². Outcome is 1 if your answer was correct, 0 if not. A lower number means a smaller squared prediction error. This is a binary-event version of the score, not a full three-choice probability model.
Worked example: You say 80% (0.80) that your answer is right. If right, (0.80 − 1)² = 0.040; if wrong, (0.80 − 0)² = 0.640. It is the size of the prediction error, not a score for a person.
A Brier score is not the same thing as calibration; examine the chart too. One lucky result is not proof you forecast well. Read the scoring nuance ↗
(0.80 − 1)²0.040
Confident and correct: a small error.
05 / SIX MINIATURE THOUGHT TRAPS
CONFIDENCE CAN LIE.
IN BOTH DIRECTIONS.
These are illustrative scenarios, not diagnoses or measured traits. Open a card to discover a more useful question than “Am I good at this?”
OVERCONFIDENCE / 01“I remember the meeting perfectly.”
Memory is reconstructive; being vivid does not guarantee being complete. Try: Which detail can minutes, notes or another record verify?
On memory systems ↗UNDERCONFIDENCE / 01“I checked it twice, but I’ll say 50%.”
Hedging can be inaccurate too. If evidence is strong, a higher probability may be the honest forecast. Ask: What would I tell a stranger with the same evidence?
OVERCONFIDENCE / 02“One viral anecdote proves the trend.”
A moving story can be real without being representative. Ask: What is the denominator? Which cases are missing?
Explore population data ↗UNDERCONFIDENCE / 02“Unless it’s certain, I know nothing.”
70% is not ignorance. Many decisions require acting with imperfect information. Ask: Which action is sensible across the plausible outcomes?
OVERCONFIDENCE / 03“I saw everything in front of me.”
Attention has limits; unexpected objects may go unnoticed during a demanding task. Ask: What was I focusing on that might have hidden something else?
See attention research ↗UNDERCONFIDENCE / 03“One wrong answer invalidates all my learning.”
A miss is evidence about one claim. Revisit the source, update the claim, and keep what remains justified. Ask: What exactly changed?
KEEP THE CONVERSATION OPEN
Have a thought, challenge,
or idea? Write to Sahil.
A source worth adding, a question worth asking, or a disagreement worth examining — the booth is listening.