How it works
What the test is
You make 25 guesses. Each guess is one of five shapes: circle, triangle, square, star or diamond. After the 25th guess, the site reveals the answer for each position and counts how many of your guesses matched. That count is your score.
If you are only guessing, each answer has a one in five chance of matching, so the average score is 5 out of 25.
Where the answers come from
The answers are not stored anywhere waiting for you, and nobody picks them. They are calculated from a value produced by a public randomness beacon (drand), which is run by an independent group of organisations. The beacon publishes a new random value every three seconds, and each value comes with a cryptographic signature.
When you finish guessing, the site chooses a beacon value that will be published a few seconds in the future. It waits for that moment, then downloads the value and checks its signature in your browser. The answers are calculated from it. Because the value does not exist until after your guesses are fixed, neither you nor the site can know the answers in advance, and no one can arrange for them to match.
Practice and study tests
There are two ways to take the test. A practice batch is for seeing how the test works, and practicing. Your results stay in this browser, are never sent anywhere, and you can delete them at any time.
Once a batch is submitted to the study, it cannot be taken back. When you make your 25th guess, your guesses are sent to us and sealed before any answer exists: independent timestamp authorities sign a fingerprint of them, so the time is vouched for by someone other than us. You get a receipt, and your browser checks the receipt itself. Only then is the beacon value chosen that the answers come from, and it is a value that had not been published when your guesses were sealed. The sealed batches are published, so anyone can download them and check the timestamps, the beacon values and the scores for themselves.
What a chance band means
Even when every guess is pure luck, scores vary. The chance band shows the range of scores that happens by luck alone. For a batch of 25 guesses with five choices each, about 9 in 10 batches land between 2 and 8 correct.
That means fewer than 1 batch in 10 lands outside the band, even with no ability at all, but it does happen. Landing outside it now and then is expected, and one batch tells you very little. The same goes for the running total chart: a line can wander outside its funnel by luck, especially when you look at it often.
What this does not show
A single good result is not evidence of anything unusual. A real finding would need many people and many batches, planned in advance and checked by others. The site will not tell you that you have a special ability, and it cannot.