Skip to main content
A station piano's repertoire holds to three songs for six months

A station piano's repertoire holds to three songs for six months

A School of Continuous Improvement paper codes six months of a rail-station concourse piano's played repertoire and finds three phrases account for three-fifths of every session, unmoved by an operator working group's request for variety.

A School of Continuous Improvement study logs a rail-station piano's played repertoire for 26 weeks, and finds a working group's sign changed nothing about it

A tune everyone already knows is the easiest one to keep playing, whether the audience notices or not.

— Associate Professor Casimir Beng, Lead of the Adaptive Metrics Lab

For six months, a contact microphone installed inside a rail-station concourse piano to monitor its tuning has instead been logging what commuters actually choose to play on it. Adaptive Metrics Lab researchers coded 1,142 playing sessions into a fixed repertoire scheme and found three recognisable phrases accounting for three-fifths of everything logged, a concentration that barely moved across 26 weeks of ordinary service disruption, weather, and passing foot traffic.

The pattern held tightest during peak commuter hours, when sessions ran shortest, and loosened through the middle of the day and into the evening as sessions lengthened. When the piano’s own operator convened a working group in response to an early reading of the results and posted a sign asking players to vary the programme, the researchers found the concentration unchanged over the following fourteen weeks.

“A shared instrument that nobody is told what to play on will still find its own settled answer,” said Dr Ingrid Vasseur, Lecturer and Convenor of Futures in Committee. “What we hadn’t expected was how little a working group’s own intervention moved it once that answer had already set.”

The finding sits within the School’s continuing appetite for measuring what an instrument leaves unchanged, alongside its wider programme tracking how measurement systems shape, and fail to shape, the behaviours they observe. The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.wd8d3v.