A new Slop University paper instruments a share-house energy-budget power-board across 31 households and finds its published fairness rotation predicts almost nothing about which resident's appliance actually gets cut.
Instrumenting 31 share houses over sixteen weeks finds live appliance load, not the published rotation, decides whose power a shared energy budget switches off
The School of Continuous Improvement keeps asking the same question of one measurement instrument after another: once a rule is checked at every decision, is it the rule doing the deciding? Its newest paper puts the question to a share-house power-board that enforces a household energy budget by switching off whichever appliance is drawing power once the day’s allocation runs out.
The device publishes a fortnightly rotation naming one resident per day whose circuits are shielded from cutoff. Researchers instrumented 31 share houses running one for sixteen weeks, logging every cutoff against the day’s rotation and the live draw of every connected appliance. Which resident’s appliance was actually switched off could be predicted with 89.7% accuracy from live load alone; folding in the rotation’s own protected-day status moved that figure by well under a point. A rotation that shields one resident at a time, the team notes, leaves everyone else exposed on every day it runs, regardless of whose turn it formally is.
A rule that only ever gets to protect one person a day was never going to look like the mechanism doing the choosing; we just hadn’t measured how little of the choosing it was doing.
— Lecturer Marit Osayande, Convenor of the Living Dashboard, School of Continuous Improvement
“Residents already suspected the rotation was more of a courtesy than a decision,” said Senior Lecturer Ronja Oyelaran of the School of Emergent Priorities. “What the load data adds is a number for the suspicion.”
The team worked with a share-housing management platform to recruit properties already running the device, tagging every cutoff with the day’s rotation and a reading of which appliance was drawing the most power at the moment the allocation ran out.
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.ws53l7.