Instinct, in numbers.
Real anonymous attempts. See how people perform — and how well they know themselves.
Core charts include assessed rank predictions only; empty charts are waiting for comparisons. Counts are attempts, not unique people. Only current test editions are included. Repeated play, devices and optional self-check coverage can influence distributions. Percentiles need at least 30 comparable attempts.
Explore a test.
Visual intuition · 0 completed sets
Simulated example errors, not player answers. Your measured error is shown separately. No example values enter community statistics or Core. Line Split includes both divider errors where applicable. This graph compares errors, not absolute click positions.
Every test, side by side.
| Test | Completed sets | Mean Instinct Score | Mean Instinct’s Core |
|---|---|---|---|
| Half & Half | 0 | — | — |
| Equal Pour | 0 | — | — |
| Line Split | 0 | — | — |
| Ten Seconds | 0 | — | — |
| Echo the Beat | 0 | — | — |
| Color Match | 0 | — | — |
| Color Wheel | 0 | — | — |
| Dot Flash | 0 | — | — |
| Mental Math | 0 | — | — |
| Angle Set | 0 | — | — |
| Daily benchmark | 0 | — | — |
Instinct Score measures task performance using each game’s rules. Instinct’s Core measures the accuracy of rank predictions. You can score highly on either without scoring highly on the other.
How this benchmark collects and compares results
Each locked answer contributes its normalized error, task score and optional prediction. A completed set also contributes the final prediction, its Instinct Score and its Instinct’s Core when enough comparisons exist. The server recalculates scores and rejects duplicate attempt IDs. No simulated scores, names, email addresses or IP addresses are saved in the results database.
Instinct’s Core is 1000 minus 10 times the mean absolute percentile prediction error. It uses the rank-only method. Skipped and waiting predictions are excluded, and earlier point-prediction scores are never mixed into this distribution. Each comparison requires 30 other attempts and excludes the predicting attempt. Anonymous results are shared in aggregate; full personal history stays in browser storage. No additional analytics cookies are used.