The split test asks for one thing: drag a straight line across a shape so that the two pieces have equal area. Most people expect to be close. Most people are not, and they miss in the same direction for the same reason. This page is about why, and about the small piece of mathematics that means a perfect answer always exists even when you cannot see it.
Play today's split test Five shapes, one attempt, a new set every day.
Your eye reads outline, not area
Area is not something the visual system measures directly. What it measures well is length, width and the contour of an edge, and it estimates area from those. That estimate has a consistent flaw: regions that are long and thin are over-weighted, because they contribute a lot of edge, and regions that are short and wide are under-weighted, because they contribute very little. A shape with a narrow tail and a fat lobe therefore looks balanced well before it is, and the line goes down too far toward the tail.
The same bias shows up in every task that asks people to compare areas. Pie charts are read badly for exactly this reason; so are maps with irregular regions. It is not carelessness and it does not improve much with concentration. What improves it is changing what you attend to.
A perfect cut always exists
Pick any direction. Now imagine sliding a line in that direction across the shape from one edge to the other. At the start, all of the area is on one side; at the end, all of it is on the other. Because the shape is a single bounded region, the fraction on one side changes continuously as the line moves — it cannot jump. So somewhere in between it passes through exactly one half.
That is the intermediate value theorem, and it means the angle of your line is completely free. There is no wrong angle, only a wrong position along it. The practical consequence is that you should pick the angle that makes the comparison easiest — usually one that cuts across the bulk of the shape rather than along it — and spend all your attention on where the line sits.
How the score is worked out
After your cut, the shape is clipped along the line into two polygons and the area of each is calculated exactly. If the split is 48 to 52, you are two points from even, and the score is 100 minus twice that distance: 96. A 45–55 split scores 90, a 40–60 split scores 80, and a cut that leaves everything on one side scores zero. Five shapes make a run and the result is the average of the five.
| Your split | Score | Verdict |
|---|---|---|
| 50 / 50 | 100 | Perfect |
| 48 / 52 | 96 | Excellent |
| 45 / 55 | 90 | Good |
| 40 / 60 | 80 | Fair |
| 30 / 70 | 60 | Poor |
The percentile alongside it places your five-cut average against typical results. Around 85 is average; consistently above 93 is strong; above 96 is rare and usually means someone has learned to look for the centre of mass rather than the middle of the outline.
What actually improves the number
Look for weight, not shape. Ask which side would be heavier if the silhouette were cut from steel plate. That question routes around the outline bias, because heaviness is judged by bulk rather than by edge.
Distrust near-symmetry. A shape that is almost symmetrical is the one people miss most, because the eye finds the axis and stops checking. The small asymmetry is precisely what moves the true line.
Blur it. Squinting removes the contour detail your eye is over-reading and leaves the distribution of mass. The correct line is often obvious with your eyes half closed and invisible with them open.
Use practice mode for the shapes that catch you. The daily test gives one attempt on five shapes. Practice mode gives unlimited shapes with no effect on your daily result, and it is where the centre-of-mass habit is actually built.
Common questions
Is it really possible to cut any shape exactly in half?
Yes, at any angle you choose. For a bounded region, sliding a line across it moves the fraction on one side continuously from zero to one, so it must pass through a half. The angle is free; only the position matters.
Why do my cuts always land on the same side?
Because the bias is systematic. Long thin parts of a shape are over-weighted by the eye and short wide parts are under-weighted, so the line drifts toward the thin part. Once you know which way you miss, you can correct for it deliberately.
What is a good split test score?
An average around 85 across five shapes is typical. Above 93 is strong and above 96 is rare. The score is 100 minus twice the distance from an even split, so a 48–52 cut scores 96 and a 45–55 cut scores 90.
Does the split test measure anything real?
It measures a real and well-documented perceptual bias in area estimation, the same one that makes pie charts hard to read. It is not a test of intelligence, and it improves quickly with the right strategy, which intelligence measures are designed not to do.