The delete target on a selected chip must be large enough
Aliases: chip dismiss · close affordance · tap slop
What it is
The dismiss mark on a selected chip must be a real, large enough hit region, not a few decorative pixels in the corner. An adequate dismiss target lets someone take one chip away without grazing its neighbour and without retriggering the chip body as “select again.” When the entry is too small, delete becomes a mis-delete of the next chip, or a tap on the body whose result is illegible. Add/remove symmetry lands on the finger here: joining often taps the whole chip, leaving must be more precise, so the leave target cannot be smaller.
Why it happens
The close mark is often drawn at eight to twelve pixels, far inside a finger’s contact ellipse. If the centroid sits between the mark and the text, the system has to arbitrate “delete this” versus “toggle this.” Unstable arbitration feels random: sometimes the chip vanishes, sometimes it stays and looks lit. Two neighbouring marks packed tight let one ellipse cover two targets; Fitts overshoot deletes the wrong object.
Making the whole chip delete would enlarge the target, but it fights “tap to edit / select again.” Delete therefore needs an exclusive region inside the chip, and that region’s outer edge must sit far enough from the text and from the next chip that an ellipse can belong to only one action.
Studying it
Run a pointing task on touch: given one chip, delete only that chip; neighbour deletes and body toggles both count as errors.
Independent variables: visible size of the mark, invisible hit expansion, chip spacing, whether the whole chip deletes. Dependent variables: correct-delete rate, neighbour mis-deletes, body misfires, time to finish a set of deletes.
Log the contact ellipse, not the click coordinate. When the ellipse covers both mark and text, the arbitration rule is itself an independent variable.
Where it stops holding
A mouse can live with a smaller mark because the hotspot is tighter than a finger; the same component still needs expansion on touch. Keyboard users delete with backspace or a per-chip delete key and do not depend on the pixels, but the mark must remain for pointers. Read-only chips have no dismiss; do not draw a dead mark. Destructive removal (dropping the last admin) must not complete on a large mark alone — size solves hitting, not confirmation. If the chip itself is below minimum touch size, enlarge the chip before talking about the mark; a huge internal mark still lives inside an illegal small target.
Applying it
- Make the delete hit region at least the minimum touch edge; the visible mark may be smaller if the hot zone expands.
- Keep that hot zone off the neighbour and off the text; leave a gap two ellipses cannot cover as two marks at once.
- Partition delete from “tap body to edit”: the mark owns one end, the text the rest.
- How to check: delete five chips with a finger, not a mouse. A neighbour vanishing, or a chip lighting without leaving, means the entry is too small or the partition is unclear.