Quiet-state divider

On 18 October 2017, I carried an SR Chinarro console into the rear meeting room of Liberty Wellness in Berlin. It weighed a little under nine kilograms and had two handles, although the left handle was too close to the power socket to be useful. I had been sent to replace its display and compare it with another unit operating in New Jersey.

Liberty Wellness ran a residential drug and alcohol rehab program from the Berlin building. I did not meet the clinical director. My contact was a therapist named Alina Voigt, who unlocked the equipment cupboard and stayed while I worked. There were twelve chairs in the room, a water dispenser and a low cabinet containing board games. Nothing was fixed to the floor.

The console was supposed to assist with group formation. It did not choose treatments or evaluate what members said. It registered whether a person was speaking, waiting to speak or quiescent, then suggested how the chairs might be divided between two adjoining rooms. Each resident used a plain numbered token rather than a name.

The display

The replacement screen worked normally at first. Twelve small circles appeared, one for each token. Five moved into the speaking column during the sound test. The other seven remained below a grey curve identified in the manual as a pseudo-separatrix that divides quiescent states. A value beside the curve read x < -1.

I asked what the two quiescent states meant in practice. Voigt said both meant silence. One supposedly indicated that a member was listening, while the other meant that the console had received too little information to classify the silence. She did not use those categories herself.

During the first morning, she gave me a short biography of the device’s designer and a history of the program’s formation because several members had assumed that Chinarro was the name of a former resident. According to her account, it was only a product designation. She did not know what SR stood for.

At 11:20, while the room was empty, three circles crossed the pseudo-separatrix. I checked the numbered tokens. All twelve were stacked in their charging tray, and the microphone test showed no input. The circles returned when Voigt opened the corridor door, then crossed again after she shut it.

Comparison

The New Jersey console could be reached through the maintenance connection. Its screen showed ten tokens. When I opened the Berlin door, two of those remote circles changed state at the same time as the circles on the machine beside me. I repeated this four times. The result was not identical on every attempt, but the New Jersey display always changed by at least one circle.

I disconnected the network cable in Berlin. The local console continued to react to the door for about six minutes, after which every token settled in the lower state. I cannot say whether the earlier changes came from the door sensor, the remote connection or some stored sequence. The installation sheet did not mention any link between occupancy and doors.

Voigt asked me to leave the replacement screen in place. I did so, disabled the room-division suggestion and wrote the time of each test on the service form. When I returned two weeks later, the console had been moved to the upstairs landing. It showed eleven circles although only ten tokens were present in the building. No one there knew which record the extra circle represented.