A wait with feedback is perceived as shorter

What it is

The same clock wait, given progress, a readable structure, or “which step is running”, is usually judged shorter than empty waiting. That is an occupied wait: attention is taken by information that belongs to the wait, and the gate is pulled off pure timing. Feedback is not comfort copy. It is task-related change in the interval that can be processed. An unfeedbacked wait leaves only a clock.

Why it happens

Maister wrote the queue rule as “occupied time feels shorter than unoccupied time”: idleness forces attention onto monitoring time. Feedback in an interface turns idleness into light work — read a title that already arrived, watch a bar rise, watch a stage name change. The work need not be interesting; it only has to be legitimate information about this wait, and the gate will spare it a share. One of Myers’s reasons for percent-done was to let people decide whether to keep waiting; the decision itself occupies. Harrison and colleagues later showed that progress-bar motion can push felt speed with clock time fixed: the shape of feedback modulates attention, not merely “is there a bar”.

Feedback has to be real change. The same “please wait” refreshed ten thousand times occupies nothing; after habituation it is noisier than a still spinner. Fake progress (bar moving, work not) can occupy for a short stretch; once it is seen through, perceived duration is billed back in hindsight, with broken trust on top.

Studying it

Hold objective delay fixed and compare empty wait, indeterminate motion, determinate progress, and a wait in which people can keep working. Collect judged duration and stay-or-go.

Independent variables: kind of feedback (none / cyclic motion / accumulating progress / stage names / already-arrived readable content), whether feedback is coupled to real work. Dependent variables: duration estimates, cancel and switch-away, post-hoc “I had something to do”.

Occupation and filled duration must be manipulated separately: meaningless flashes are fill, not necessarily occupation of the task; task-related progress is occupation. Comparing only “animation / no animation” smears the two. When cancel is allowed, the feedback group often cancels later and more accurately — decision quality, not only “felt short”.

Where it stops holding

Feedback works best inside the attention-holding window; past that ceiling people leave, and a slightly shorter feeling will not rescue someone already gone. If feedback blocks work already possible (a full-screen spinner over readable body), occupation becomes interruption and the wait is judged longer. On time-critical, must-hit-a-mark tasks (submit-by countdown) people watch a clock on purpose; feedback cannot pull the gate off. Assistive channels that announce “loading” once do not share the visual occupation; supply textual stages. Once fake occupation is spotted, later true feedback is treated as noise too.

Applying it

• When a wait crosses the continuity-of-thought window, give work-related feedback: progress, stages, already-arrived parts that can still be read — not cyclic motion alone.
• Let people do one legitimate thing during the wait (read arrived body, tidy a selection). That is more stable than a livelier bar.
• Do not pass off task-irrelevant animation as occupation.
• How to check: same delay, empty spin versus “bar rising or body already readable”. Ask “how long was that” and whether they wanted to leave. The feedback group should estimate shorter and leave less without cause; if the bar moves and the estimate does not, feedback did not occupy timing attention.

Related

• Same group: I2.07.1 Perceived duration can diverge from clock duration · I2.07.3 Optimising perception does not replace optimising the clock
• Nearby: I2.01 Choosing skeleton versus spinner · I2.02 Predictable progress · E6.09 Determinate and indeterminate progress
• Search terms: occupied wait · progress feedback · perceived wait time