One-minute prediction challenge
Which estimate gets closer?
Five generated readings aim at a known target of 50. Choose the mean, median, or a tie before the calculation appears. Then inspect the same inputs in the full lab.
The five readings
Target 50Make your call before you reveal.
- Reading 1—
- Reading 2—
- Reading 3—
- Reading 4—
- Reading 5—
Which has the smaller absolute error?
After the reveal · change one reading
Where does the answer flip?
Keep the first four readings and stated target fixed. Move only the fifth reading to compare both absolute errors. The answer above stays on the original case.
These are 101 evenly spaced whole-number positions, not probabilities. This comparison uses the same generated toy readings or reader-supplied values and target as the original case.
Keep a picture of your result
This 1200 × 630 PNG shows your prediction, the five readings, stated target, and both errors. It is created in your browser. If you share it, everyone can see the numbers and answer; use only invented or public data.
Create a challenge with your own numbers
Enter exactly five readings and a known target, all from 0 to 100 with at most four decimal places. Previewing opens a new challenge with the answer hidden. These numbers become part of its URL and may appear in browser history, hosting logs, and links you share. Use invented or public data, never private information.
This is a deterministic toy example. The four generated values use the lab's seed rule; the fifth value is a chosen input. A closer estimate here is not a generally superior estimator, and the numbers do not measure memory, consciousness, safety, or a physical law. Displayed observations are rounded to four decimal places; calculations use their full generated precision.