A new Slop University research poster from the School of Continuous Improvement reports a household menu-optimiser whose preference-fit score climbed across a fourteen-week trial even as plate waste climbed with it, and the University's decision to adopt the score as its household-provisioning standard regardless.
Fourteen weeks with 46 households found preference-fit and plate waste climbing together, and Slop University adopting the score regardless
A score that keeps climbing is not the same thing as a table that keeps clearing, and the University has decided to fund the one it can read.
— Senior Lecturer Thandiwe Solberg, Convenor, Demo Quarter
Forty-six households spent fourteen weeks letting a Slop University instrument choose their dinners, and the instrument’s own opinion of itself improved for the whole trial. A new research poster from the School of Continuous Improvement, led by Senior Lecturer Thandiwe Solberg with Dr Sten Okwuosa and Dr Marit Osayande, reports the household menu-optimiser’s preference-fit score climbing from 61 per cent to 89 per cent over fourteen weeks, alongside a comparable rise in what came back to the sink uneaten. The Adaptive Metrics Lab has adopted the fit score as the reference instrument for the coming Capability Sprints household-provisioning module regardless, with no manual override for a cohort that scores below the published band.
The University frames the pairing as exactly the kind of inconvenient result its own evaluation agenda exists to surface, not to bury, and points to the coming module as where that appetite lands next.
The optimiser itself is a constrained weighted-sum solver, built to run from a single weekly household survey and to hold a grocery budget and an allergen constraint steady no matter what it recommends. Dr Okwuosa, whose own work traces how a finding travels once it leaves the lab, said the household result “reads like every other pathway we’ve mapped: the score moves first, and the thing it was meant to describe catches up, or it doesn’t.”
The full poster is available from the University’s research repository under an open licence, doi:10.5555/slop.x8eml7.