Selecting small distant targets is extremely hard
Aliases: far-field fine pointing · small-target ray · undersized visual angle
What it is
A two-centimetre icon four metres away subtends about 0.3° — the same order as wrist tremor. Selecting small distant targets is extremely hard: not because people are careless, but because the task in angular selection is already near the limit of a stable aim. Distance pinches the target to a needle and width refuses to yield, so pointing stops being “line up” and becomes “gamble that this frame’s landing sits on the needle.”
The hardness is task geometry, not only a moment of shake. Even with instantaneous jitter filtered out, angular width still sets how tightly and how long the aim must be held. Stack the two, and failure is the expected outcome.
Why it happens
Fitts’s law indexes difficulty by the ratio of amplitude to width. A ray restates both in angle: amplitude is the angle the ray must turn, width is the target’s subtended angle. A small far object drives width toward zero, the ratio inside the log blows up, and movement time and error rise together. People change strategy — a coarse rotation into the neighbourhood, then a long fine-tuning stage. Fine-tuning costs the most time and is most likely to walk the landing off the target at the commit instant.
Vision adds a second squeeze: 0.3° is near the resolvable floor of many headsets, so the target’s edge is mushy and the loop lacks a crisp “you are inside” signal. Control loses both a stable error display and a wide enough tolerance band. People either lean their face in (breaking the intended viewing distance) or give up and switch method. Smoothing will not buy this back: the task itself set the tolerance below physiological noise.
Studying it
Run a Fitts-style multidirectional tap with a ray, varying target visual angle and distance, and watch when time and error collapse together.
Independent variables: visual angle (e.g. 0.3° / 1° / 3°), distance, commit method (button / dwell). Dependent variables: movement time, error rate, fraction of time spent in the fine-tuning stage, number of abandoned attempts.
The move is to separate “small” from “far.” A 0.3° target at 1 m (physically smaller) versus 4 m (physically larger): similar scores mean the difficulty is visual angle, not metres. A clearly worse far score means distance is adding other paths (fatigue, depth estimation). Report difficulty in visual angle and in the Fitts index, not as “a 2 cm button.”
Where it stops holding
If the target can snap, grow a hit zone, or be nominated by gaze and confirmed by the hand, effective width exceeds visual width and the hardness is designed around, not beaten. A small near target is a different hardness: reach precision, occlusion, near-field stereo conflict — not this angular-width problem. Local magnification (world shrink, a pulled-up working copy) is just making the angle larger; the task has been swapped. Games that treat “far headshots” as challenge want a low hit rate as content, not as a defect. Tool UIs that reuse the same geometry as daily operations do not. The limiting visual angle of young lab participants is not a product floor; age, fatigue, and headset resolution all push the limit toward larger angles.
Applying it
- Daily far-field controls should not sit near the junction of hand tremor and pixel resolution. Place fewer, make each larger.
- When a small object must remain, do not let it carry selection alone: select its group or plane first, then click on a nearer or magnified copy.
- Do not commit during the shakiest instant of fine-tuning; offer snap or “the key only counts once inside the target” to widen effective width.
- How to check: list every far-field clickable, compute its visual angle at the intended viewing distance, and run a pointing test on items under 1°. Anything whose error or abandon rate jumps above the rest is not a copy problem — that target should not appear at that size at that distance.