Every element you add taxes the ones already there

Addition is not free, and the arithmetic is worse than linear

A layout with three elements has three relationships to manage: A to B, A to C, B to C. Add a fourth and you have six. A fifth gives ten, a sixth fifteen. The general figure is n(n−1)/2, and it is the reason a design with eight elements is not slightly harder than one with four: it is four times harder to resolve, because every new object has to be positioned, sized and weighted against everything already present.

This is not an abstraction about beauty. It is the direct explanation for why committee flyers look the way they do. Each person adds one item that is individually reasonable (a QR code, a second phone number, an award badge, a line about parking) and the total number of relationships grows quadratically while the design time stays the same. Nobody made a bad decision. The decisions were made one at a time.

Attention is also a fixed budget

The viewer arrives with a few seconds. Whatever they spend deciding what to look at is not spent understanding anything. Two elements of equal weight cost more than the sum of their parts, because before reading either the viewer has to resolve which one comes first, and that resolution is pure overhead.

So the cost of an element is paid twice: once in the composition’s complexity, and once in the viewer’s decision time.

Ask three questions before anything goes on the page

1. What job does it do that nothing else does? If two elements do the same job (a headline and a strapline that say the same thing differently) one is redundant and should go, not be made smaller.
2. What happens if it is absent? If the answer is “nothing in particular”, that is the answer. If the answer is “somebody in the client’s office will be annoyed”, that is a political cost, and it should be named as one rather than dressed up as a design requirement.
3. What does it displace? Space is finite. The award badge does not arrive in empty space; it arrives by shrinking the headline or by eating the margin that made the headline readable.

Subtract before you rearrange

When a layout is not working, the reflex is to move things. Try removal first, because it is faster and more diagnostic. Hide one element (a layer toggle in GIMP, Krita, Inkscape, Scribus or Figma, one click, fully reversible) and look again. Go through every element in turn. You will usually find two or three that the design is better without, and at least one that you had assumed was load-bearing and is not.

This is also the most effective way to have the conversation with a client. “Here is the version with the award badge and here is the version without it” is a comparison they can judge. “The award badge weakens the hierarchy” is an opinion they have no way to evaluate.

Grouping buys you back some of the cost

You cannot always delete. The opening hours, the address and the phone number all have to be there. But three items placed close together with consistent alignment are perceived as one object, not three, because the visual system groups by proximity and similarity before you are aware of it.

That is the practical escape route. Five loose items become two groups; two groups are one relationship rather than ten. It costs you nothing except discipline about spacing: the space between groups has to be clearly larger than the space within a group, or the grouping does not read and you have five items again, now with inconsistent gaps.

The limit of the principle

“Less is more” is not always true, and applied without thought it produces work that is stylish and useless: a poster so reduced that it no longer says when the event is. The principle is not minimalism. It is that every element must earn its cost, and some elements are worth a great deal.

Dense work can be excellent. A railway timetable, a technical diagram, a product spec sheet: these carry enormous amounts of information and can be entirely calm, because the density is organised into a repeating structure the eye learns once and then applies. The thing that fails is not density. It is unstructured density: many items, each competing individually, with no system that lets the viewer stop evaluating them one by one.