Show HN: Game to see if you're over-/under-confident in your own knowledge

Developer afc launched est.alejo.ch, a free browser-based calibration game that asks players ten "what happened first" trivia questions and tracks how well their stated confidence levels match their actual accuracy, per the Show HN post shared on August 25, 2026.

Each round presents a binary ordering question — which of two events came first — and players assign a confidence percentage to their answer. The game records success rates per confidence tier and produces a calibration graph and table summarizing whether a player consistently overestimates or underestimates their own accuracy. Results persist across sessions using the browser's local storage API, so players can accumulate aggregated statistics over many rounds, according to the HN submission.

The tool is built in pure TypeScript with no framework and runs entirely as static files, meaning there is no server-side logic and no data leaves the browser. The source code is published under the GPLv3 license on GitHub. The developer notes the project grew out of personal curiosity about calibration — the degree to which stated confidence tracks real-world accuracy — a concept used in forecasting and decision-making.

The developer reports playing 60 games totaling 480 questions and found they are relatively well-calibrated overall, though the data suggested they should avoid betting below 60% confidence on this type of binary question, per a personal stats screenshot shared in the post. More background on the project's motivation is available on the developer's blog.

The post attracted no comments at the time of publication, but the developer expressed interest in seeing calibration data from other users and invited suggestions for improvement.