No ruler, no protractor — just your eye finding the middle, the right angle, the centre. Lower error is better
How it works
A figure appears with an instruction above it. Press and drag to move the crosshair, release to lock it in. The true answer appears as a green circle straight away, with the distance between it and your mark. After 21 tasks your average error is your score.
- Seven tasks, three times each, shuffled. 21 in total.
- Move the crosshair and release. There is no undo — in eyeballing, the second guess is usually worse.
- For the right angle and the bisector the crosshair stays on the dotted circle: those tasks are about direction, not distance.
- Error is measured as a percentage of the screen width. Under 0.5% is very accurate; over 3% is a clear miss.
- The result screen breaks the error down by task, so you can see which one lets you down.
The seven tasks
| Midpoint | The middle of a line. The easiest one — above 1% here and the rest will hurt |
|---|---|
| Parallelogram | Three corners are given; mark the fourth. No guide lines |
| Centroid | Where the three medians meet. It sits closer to the long side than the visual centre |
| Circle centre | Only an arc is drawn. The shorter the arc, the harder it gets |
| Right angle | One line is given; mark the direction that turns exactly 90° at the vertex |
| Bisect angle | The direction that splits the angle into two equal halves |
| Convergence | Extend two separated segments and mark where they would cross |
Scoring
| Score | Average error over 21 tasks, as a percentage. Lower ranks higher |
|---|---|
| How error is measured | Distance between your mark and the true point, divided by the screen width. For the angle tasks, the angular difference converted to arc length |
| Device differences | None. Figures are drawn on a fixed-ratio stage and error is relative to its width, so a phone and a desktop get the same task at the same difficulty |
| Grades | Under 0.8% "draughtsman's eye", under 1.2% top tier, under 1.8% above average, under 2.6% average |
How to cut your error
- Take in the whole figure first. Look at everything before you move the crosshair; once you are staring at it you stop seeing the rest.
- For midpoints, look side to side. Fixing on one end drags your mark that way. Comparing the two gaps alternately is far more accurate.
- The centroid is not the visual centre. It leans toward the long side. Picture the line from a vertex to the middle of the opposite side and go one third of the way along it.
- Find the circle centre on a perpendicular. Imagine the chord joining the ends of the arc; the centre lies on its perpendicular bisector, which halves the problem in one step.
- Do not tilt your head for right angles. Whatever you tilt, you lose. People consistently under-open right angles that are not aligned with the screen.
- Do not dither. The first impression is the accurate one. Staring lets the centre-overestimation illusion creep in.
Frequently asked questions
Could I just measure with a ruler?
Nothing stops you, but then it is a craft project, not a test. The leaderboard belongs to people using their eyes, and only you know which one you did.
Why is there no undo?
Eyeballing is about the first judgement. If you could correct it, patience would win instead of vision.
Is a phone worse than a desktop?
No. The crosshair lands exactly where you press, and a loupe above your finger magnifies the spot it is covering. The figure size and the error scale are identical on both.
Can I score 0%?
Effectively no. Getting under 0.3% of the screen width by hand is very rare.
Related tests
- Perfect Circle — precision by drawing instead of marking.
- Hand Steadiness Test — holding a line through a narrowing path.
- Aim Trainer — speed rather than precision.
- Mental Rotation Test — turning shapes in your head.