Start with a real hook
Each question starts from a real-world news article and a specific reported claim or number.
Back of the napkin
Do I have any idea what I'm talking about?
Being wrong can feel like a punch to the gut. Being loudly wrong after making strong public claims is worse. Being full of it is no fun. One way to check yourself is to test your mental models: the little working ideas you carry around about how the world works. We use recent news items to build Fermi problems that test those models, your intuitions, and your math skills so you can see how well calibrated you are.
Matt's calibration snapshot
59.3 / 100
My average Calibration Score across 52 completed problems. Higher is better.
Explore Matt's statsWhere the points come from
Pegs10.4 / 30
Model23.4 / 30
Math6.4 / 10
Result19.1 / 30
Strongest technical area with 3+ problems: Public systems and economics, 71.1Strongest model family with 3+ problems: Unit conversion and scaling, 68.3
How to use this site.
Current events are full of hidden scale questions that help test your mental models and intuitions.
Use your pegs
A good back-of-the-napkin estimate needs memorized anchors: people, homes, gallons, joules, hours, prices, and other rough stats about the world.
Check the estimate
Then we compare the model, math, pegs, and result with authoritative references. If the reference answer looks wrong, say why.
Latest issue
Duane Arnold nuclear restart electricity
Estimate how much electricity a restarted 615 MW nuclear plant could generate and how many years of gross power value would equal a $1.9 billion loan.
Open the problemCalibration leaders
Best-calibrated scale guesses.
Higher accuracy points mean a closer scale guess: 100 is exact, 50 is a factor of 10 away. These reader points measure only the final guess; Matt's Calibration Score also grades pegs, model, and math. This prototype board only shows readers who opt in from this browser until a real backend is connected.
Top scale guesses
- No opt-in local entries yet.
Calibration Score
Earn points for getting it right.
Higher is better: earn up to 100 points across pegs, model, math, and result. The image shows the inverse: percent full of it = 100 minus the Calibration Score.
Accurate pegs: 30
Within 25% of the reference values earns 30 points, within 50% earns 20, within 100% earns 10, and more than 100% off earns 0.
Sound model: 30
A reliable calculation model earns 30 points. A model that sort of gets there earns 15. A model that cannot answer the question earns 0.
Correct math: 10
No arithmetic mistakes earns 10 points. Mistakes with no major impact earn 5. Mistakes that substantially change the answer earn 0.
Accurate result: 30
Within 10% of the reference answer earns 30 points, within one order of magnitude earns 20, within two orders earns 10, and farther off earns 0.
Weekly newsletter
Get the next gut check.
One timely problem per week, with source-backed context and reader results as they come in.