Corners are cheapest but hold almost no targets

What it is

A screen has four corners, no more. Each is walled in two directions, so a flung pointer lands on that point from anywhere, cheaper than a single edge. The Windows Start button, macOS close traffic-lights toward the top-left, full-screen Exit in many players — all four slots are already a fight. Screen corner slots are not a question of cheapness. They are a question of count: once the fourth corner is assigned, there is no fifth target.

Cramming a row of icons into the lower-right does not let them share the corner’s infinite width. Only the control that actually occupies the intersection of two walls is a corner target. Its neighbours fall back to ordinary edge buttons.

Why it happens

An edge is infinite on one axis; a corner is infinite on two. ID collapses further, and a ballistic throw can aim at “that corner” from anywhere on the screen, without even choosing which edge to follow. The four corners become the scarcest pointing resource on the display.

Scarcity is an allocation problem. Each corner cleanly holds one job. Two jobs sharing a corner do not both collect the dividend — hit regions overlap, misses rise, and the dividend is spent on conflict. The historical fight is stable: close, system menu, notifications, full-screen exit. Once a platform has written a corner into muscle memory, a product that occupies it pays a conflict tax.

Studying it

A Fitts task places same-pixel targets at a corner, at an edge midpoint, and in the interior, and measures movement time. Add a condition with two adjacent targets in one corner, and test whether the second still enjoys corner time, plus the rate of hitting the first by mistake.

Independent variables: corner / edge / interior, number of targets in the corner, gap between two corner targets. Dependent variables: movement time, count of landings on the wrong corner target, whether “fling and you’re there” still holds subjectively.

Report true corners (intersection of two walls) separately from “a few pixels off the corner”; the latter is already not a slot.

Where it stops holding

Rounded screens and notches cut the geometric corner away; the true intersection may be unclickable, the slot shrinks inward, and the infinity is broken. Outer corners of a multi-monitor setup still work; inner corners (where two displays meet) leak the pointer. On touch there is no mouse inertia into a corner, so the dividend shrinks, and system gestures are easy to graze. A full-screen game that gives the four corners to aiming or skills must offer another path to system close; users can no longer be assumed to fling to a corner.

Applying it

• Assign each corner one product-level job, written as a list: close, main menu, notifications, full-screen exit. Do not stack two clickable controls in one corner.
• Sit on the actual intersection of two walls. Padding between the control and the corner wastes the slot.
• Do not take a corner the platform already owns (system notifications, gesture hot zones). Put product entries at an edge midpoint or inside.
• How to check: twenty ballistic throws from screen centre to each corner. If landings oscillate between two controls, that corner is already over capacity.

Related

• Same group: F1.05.1 Edges are unbounded in the direction of motion, so pointing cost is lowest · F1.05.3 Touch and fullscreen system gestures occupy the value of edges
• Nearby: F1.14 Semantic conventions of position · F2.12 Safe areas and irregular screens
• Search terms: screen corner · Fitts's law · infinite-width target