Discoverability of cancellation depends on pressed-state feedback
Aliases: pressed state · cancellation discoverability · touch feedback
What it is
People rarely discover or trust “slide out to cancel” without visible pressed-state feedback. Highlight, scale, colour, candidate labels, and a return to normal or cancel state on exit reveal that action is not yet committed and that movement can change the result.
Why it happens
Pressed state externalizes a hidden state machine: entry shows candidate, exit shows cancellation, and re-entry restores candidate. What actually matters here is not the coarse question of "is there feedback or not," but whether the candidate and cancelled states render as two clearly distinguishable appearances: if an implementation only builds one "highlight on press" effect, and that highlight simply stays as-is after sliding outside the target with no switch to any different look, then what a person sees is identical whether they are in the candidate or the cancelled state — there is no way to tell from the screen whether lifting now will commit. Many implementations get stuck exactly here: pressed feedback exists, but only as a single state rather than two distinguishable candidate/cancel states. That is not a matter of feedback being weak; a whole state is simply missing.
Where it stops holding
Strong animation or flicker can disrupt reading, while weak feedback may be missed under occlusion. Visual-only feedback also fails in low-vision or heavily occluded contexts, so haptic, sound, or reversible outcomes help but cannot replace clear state. Even after adding haptic or audio cues, the problem persists if the candidate and cancelled states share the exact same haptic pattern — a multichannel supplement solves "cannot see it," not "can see it but cannot tell them apart."
Applying it
- Show stable, distinct candidate and cancel feedback on press, exit, and re-entry.
- Place feedback outside finger occlusion or supplement it with haptic or audio cues.
- How to check: record a full "press—slide out—lift" sequence and step through it frame by frame, checking for any discernible visual change the instant the finger leaves the target; if the candidate and cancelled states look identical on screen, the current implementation only has one pressed-feedback state and is missing a distinct cancelled one.