A new Slop University research poster finds one-star review wording on a trades-marketplace app drifting toward the mildest of four coded complaint kinds within weeks of the platform's taxonomy going live.
A School of Emergent Priorities study tracks one-star review wording drifting toward the mildest of four coded kinds after a trades-marketplace platform's taxonomy went live
A trades-marketplace platform built its four-kind review taxonomy to speed up replies to one-star complaints. A new Slop University research poster finds the taxonomy now shaping the complaints themselves: reviewers, learning which kind draws the mildest canned response, are writing toward it.
The poster, from Lecturer Ingrid Vasseur of the School of Emergent Priorities with Dr Solveig Adeyemi and Associate Professor Casimir Beng of the School of Continuous Improvement, extends the University’s anticipatory-capability agenda into an everyday dispute: the one-star review a tradesperson can’t afford to leave standing.
The complaint didn’t change. Only the sentence describing it did, and that was enough to move which reply it received.
— Lecturer Ingrid Vasseur, School of Emergent Priorities
The team’s account draws on eighteen hundred one-star reviews spanning the platform’s taxonomy rollout, coded twice over — once by the platform’s own kind assignment, once by researchers with no sight of it — to separate a shift in wording from any shift in the service underneath. Only the visible coding moved.
“A measure that grades the message will always be easier to satisfy than one that grades the outcome,” said Associate Professor Casimir Beng, Lead of the Adaptive Metrics Lab.
The full poster is available from the University’s research repository under an open licence, doi:10.5555/slop.kph8qc.