Tasks that take more than a few seconds do not belong on the wrist
Aliases: microinteraction duration · watch task time · raised-arm budget
What it is
An active watch sequence—taps, turns, swipes, speech—should not be finished on the wrist if it needs more than a few continuous seconds. A glance is looking; this is doing: scrubbing a long playlist, filling several fields, rewriting a reply. What runs out is the raised arm and the freed hand, not the scan. A few seconds is an order of magnitude, not a magical lab cutoff. This entry only asks whether the job is short enough for the wrist. It does not ask who should receive a handoff, or how state travels.
Why it happens
Input turns a brief pose into a sustained one. The thumb keeps aiming on a patch the size of a thumbnail; the crown ticks through a long list. Each extra second raises isometric load at the shoulder and aiming error at the finger. The watch also occupies the hand’s “home”: while the wrist is up, that hand cannot take a door, a rail, or an object being passed. Socially, talking at a wrist or tapping in a burst looks more like a performance than pulling out a phone. People then abort before the task ends—wrong item, cancel, or dropping the wrist and pretending they finished. Failure is not too much information on screen. The time budget was already spent by posture and setting.
Studying it
Measure active session duration (first input to last input), not display-on time. In the wild, log crown revolutions, touch counts, and session seconds, and mark where people abandon or switch devices. In the lab, give a watch task that must be completed, add standing or walking as a secondary demand, and see which tasks stably cross a few seconds—and how errors rise after they do.
Independent variables: number of targets, input (touch / crown / speech), walking, social presence. Dependent variables: session duration, exit-before-done rate, selection errors, subjective effort in the watch-side arm.
Participants told to finish will endure durations that real settings abort; abandonment needs unincentivized logs. Do not treat time-in-app as time-on-task—the display can stay on after input has stopped.
Where it stops holding
During a timed interval in sport, or a navigation confirm that is unsafe to miss, people will hold the arm up longer; risk rewrites the cap. Lying down or resting the forearm cuts isometric load, so a longer sequence becomes tolerable—those numbers do not transfer to walking. Speech frees the hands, and the budget moves to social awkwardness and recognition round-trips; that is a different budget, not this claim failing. Experts can crush a fixed ritual (start a run, check in) under a second; that does not make a rare long flow in the same app acceptable.
Applying it
- Time each wrist flow from entry to success; if it stably exceeds a few seconds, do not offer a complete path on the watch.
- Remove long lists, multi-field forms, and multi-turn rewriting from the wrist rather than adding a “please wait” spinner.
- If it must stay, collapse it to one confirm or one rotary move, not three or more aims.
- Verify by running the real task while standing and walking. Slice first-to-last input. A median past a few seconds, or exits clustered on one flow, means delete the wrist path—not polish the animation.