Why a queue that moves is tolerated longer than a shorter one that stalls

Two counters. One line shuffles forward every twenty seconds for eleven minutes; the other stands still for four. People leave the second one, complain about it afterwards, and estimate it at three times its length.

What everyone sees

Waiting time is assumed to be experienced roughly as measured, so reducing minutes should reduce dissatisfaction proportionally. It is not experienced as measured. In the absence of information about the endpoint, movement is the only evidence available that the wait is finite, and an unoccupied stall provides none.

What is actually happening

Maister’s account of waiting-line psychology set out the propositions that operations research had ignored: unoccupied time feels longer than occupied time, unexplained waits feel longer than explained ones, and uncertain waits feel longer than known finite ones. Hui and Tse tested the informational half experimentally, finding that giving people queue-position information improved their evaluation of medium and long waits by changing the perceived duration rather than the actual one, while for short waits the information mattered little. Together they explain why the shorter stalled line loses. It withholds both the progress signal and the endpoint estimate, so the only inference available is that this could continue indefinitely.

Why it stays hidden

The mismatch hides because the operations data is accurate. Average wait time falls, throughput improves, and complaints rise, so the numbers appear to contradict the customers. What the numbers do not contain is variance within the wait, and variance is what the customer was reporting on.