Published on 2026-07-25
Why does a bigger icon lie about a number nobody changed?
Draw a coin twice as tall to show double the money, and your eyes see roughly four times as much. Depth and size play tricks that plain bars never do.
Draw a coin twice as tall as another one, to show that a budget doubled. Simple enough, except your eyes don't measure height. They measure area, the whole shape at once, and a coin twice as tall is also twice as wide, which makes it four times the area. Nobody changed the number. The picture just told a bigger story than the number did.
A shape grows faster than a number
This is basic geometry doing damage. Scale a flat shape up by a factor of two in every direction and its area grows by four, because area is a square of length. Scale a solid object, a stack of coins, a rendered 3D bar, and its volume grows by eight, because volume is a cube of length. A researcher writing in the peer-reviewed literature on data visualization pitfalls gives a real case of exactly this: a diagram using a circle's diameter to represent two economies, one worth 14.5 trillion and the other 5.7 trillion. The true ratio between them is 2.56. Because the value was mapped onto the diameter rather than the area, the bigger circle ends up looking about 6.5 times larger than the smaller one. The reader never sees 2.56. They see something closer to six and a half.
Depth plays the same trick a different way
Three-dimensional charts add a second version of the same problem, this time through perspective rather than area. Claus Wilke, who wrote the widely used textbook 'Fundamentals of Data Visualization', shows what happens when you tilt a chart of the Titanic's passenger classes into 3D. The real numbers are plain: 322 people in first class, 279 in second, 711 in third. Once the bars are rendered at an angle, with depth and a vanishing point, the second-class bar that should represent 279 people visually reads as something closer to 210 or 220. The bars further back get quietly shrunk by the same camera trick that makes a road look narrower in the distance. Wilke shows the same problem in pie charts: tilt a pie with a 25 percent slice toward the viewer, and that slice can look considerably larger than a quarter of the circle, simply because it sits closer to you than the rest.
A 3D chart doesn't add a dimension of information. It adds a dimension of distortion, and hides the number behind it.
Nothing here required a fake number
What makes this family of tricks so persistent is that no one has to falsify anything. The designer can start from a completely honest spreadsheet, use a completely honest piece of software, and still hand you a picture that misleads, because the distortion lives in the geometry, not in the data entry. A pie chart tilted for visual flair, an icon scaled up to look impressive, a bar chart rendered with depth because it looks more 'modern' on a slide: none of these choices touch a single figure, and all of them change what you walk away believing.
The fix is almost embarrassingly simple once you know to look for it. Compare the numbers printed next to a chart, not the shapes drawn to represent them. If a chart uses size, area, or 3D depth to show a difference, treat the visual impression as decoration and go straight to the labels. A flat bar chart, ordinary and a little boring, resists nearly all of this: a bar twice as long really does represent twice the value, because length alone scales the way our eyes expect it to. The moment a chart adds a second or third dimension to look more exciting, that's exactly the moment its shape stops being trustworthy.
It helps to remember that a designer rarely sits down planning to deceive anyone. Depth and scaled icons usually get added because they look sharper on a slide, more finished, more like something worth your attention. The deception, when it happens, is a side effect of style rather than the point of the exercise. That doesn't make the distortion any smaller once it reaches you. A reader who never checks the printed figures ends up trusting the designer's sense of proportion instead of the actual data, and those two things only agree by coincidence.
Boring charts earn your trust
None of this is an argument against pictures. A good chart still says in one second what a table says in one minute. It just means the charts worth trusting are usually the plainer ones: flat, one dimension of length doing the work, nothing tilted toward you for effect. Next time a chart looks unusually striking, that's worth a second glance, not because striking charts are always wrong, but because striking is often exactly what a distorted shape is built to look like.