After repeated no-response, end rather than keep waiting

What it is

A living-room speaker asks which playlist to play. The person has already taken a cup into the kitchen. The speaker keeps asking on a beat until someone feels the room is haunted. Termination after repeated no-response says: when several turns in a row contain no user speech, close the floor and release listening, rather than treating wait as open-ended politeness. What ends is this session’s occupancy. It is not choosing a playlist on the user’s behalf — that is a different kind of default action.

Why it happens

An open listening session occupies the acoustic channel and social space. Each unanswered chase updates the evidence that the addressee is still present, downward. Keeping the wait treats absence as a longer pause. The cost of closing too soon is that someone who still meant to answer must wake the device again. The cost of never closing is a live microphone, a device talking to an empty room, bystanders paying for inserted sound. When the count of no-response turns drives “still here” below the cost of holding the floor, the right dialogue move is a close: one ending line, listening indicator off.

“Wait a bit more” sounds gentle; it bills bystanders’ quiet for a session that may already be dead. Waiting has no natural stop. The stop has to be written as a turn cap. Hitting the cap and merely stretching the same question’s interval does not release occupancy; it only makes occupancy sparser.

Studying it

Plant “leave after the question” trials in a real room. Count at which unanswered chase returning users can still resume, and at which bystanders start reporting annoyance or unease. On session logs, treat survival: the event is “did any user speech occur after the k-th no-speech prompt.” If the curve is already on the floor after k=2 or k=3, further waiting almost never salvages the task.

Keep “the system is silent while it looks something up” out of the no-response class — that is not user absence. A lab that forbids leaving the chair will not see the empty-room case the stop rule is meant to protect.

Where it stops holding

Emergency calls and alarm acknowledgements cannot hang up on a living-room cap. People with motor or language disabilities may need more attempts; widen the cap by population rather than deleting it. When the system says “please wait” and goes to a database, it holds the floor; the user’s no-response clock should not run. A mid-question “wait a second” is an explicit hold; silence after that is not no-response and should not fire termination.

Applying it

• Write a hard cap on unanswered chases (often two, at most three). At the cap, speak a hearable close (“I’ll stop asking; call me if you need me”) and turn the listening light off.
• Do not execute the pending action as part of teardown. Ending occupancy and whether to act are separate decisions.
• On the dashboard, watch microphone-open time after last user speech. If the long tail is third and fourth chases, waiting is substituting for ending.
• Run empty-room trials in a furnished space. After the person leaves, the system must go quiet inside the cap. If it is still talking, the rule has not landed.

Related

• Same group: M1.07.1 Silence can be thinking or leaving · M1.07.2 A re-prompt must change its wording
• Nearby: M1.11 Silence, Timeouts and Re-prompts · M1.12 Opening and Closing a Dialogue · M4.01 Social Cost of Using Voice in Public
• Search terms: termination after repeated no-response · stop waiting · empty-room prompt