A threshold that is too small misclassifies natural hand tremor as drag intent

What it is

A drag threshold is the minimum post-press movement before an interaction becomes a drag rather than a click. When too small, it mistakes natural tremor, mouse micro-motion, or trackpad noise for deliberate dragging, turning intended selection or activation into movement, text selection, or another side effect.

Why it happens

The system records a press origin and calculates displacement until it crosses a threshold. Even a hand attempting stillness produces small movement, so a tiny value cannot separate press error from deliberate travel. This gets amplified further by control-display gain: the higher the gain, the larger the on-screen pixel displacement produced by the same amount of physical hand tremor, and the more easily it crosses a threshold defined in screen pixels. "How small is too small" is therefore never an isolated parameter — it is tied to whatever gain is currently in effect, and the same threshold value filters noise quite differently on a low-gain mouse than on a high-gain trackpad.

Studying it

Test single click, double click, short drag, and long drag across values, recording false drags, click success, start time, and undo. Include high-tremor conditions, different gains, and small targets instead of choosing from a developer's stable pointer trace. A more direct approach measures the noise floor on its own first: have participants try to "hold perfectly still" for several seconds, record the distribution of displacement the device actually reports during that time, and take a high percentile of it (covering, say, over 95% of stillness attempts) as the reference for the threshold's lower bound, rather than guessing a pixel count from experience.

Where it stops holding

Increasing threshold reduces false drags but harms real short drags. Press feedback, reversible actions, and larger targets also reduce consequences. Device noise and precision differ, so one number is not universal.

Applying it

• Set a start range that tolerates natural tremor using real false-drag data combined with whatever control-display gain is currently in effect, rather than picking an isolated pixel count.
• Preserve click-candidate state before crossing; do not move objects or show drag side effects.
• How to check: collect real "hold still" displacement data from a sample of users and calibrate the threshold's lower bound from a high percentile of that distribution, rather than judging "small enough" by feel.

Related

• Same group: C1.22.2 A threshold that is too large discards real short drags as clicks · C1.22.3 Drag thresholds must scale with input precision and display density, not fixed pixels · C1.22.4 While held below threshold, the cursor should show no drag feedback · C1.22.5 A directional threshold can distinguish horizontal dragging from vertical scrolling at gesture start
• Nearby: C1.20 Hover delay and false activation · C1.23 Two-finger and horizontal scrolling
• Search terms: drag threshold · tremor · intent