Skip to main content
A five-kind review taxonomy now decides which drivers face retraining

A five-kind review taxonomy now decides which drivers face retraining

A new Slop University paper sorts low-star gig-delivery driver reviews into five kinds and reports what changed once the assigned kind began deciding, without appeal, who enters a mandatory retraining register.

The kind assigned to a gig-delivery driver's low-star review is the sole trigger for enrolment in a published retraining register, with no appeal.

A driver told to slow down and a driver accused of rudeness can land in the same one- and two-star review, unless something more precise is doing the sorting. A new paper from the School of Emergent Priorities and the School of Continuous Improvement reports on a five-kind scheme built to make that distinction, and on what happened once a partner gig-delivery platform started acting on it.

The scheme — route or navigation deviation, communication breakdown, vehicle condition, conduct, and unattributable — was checked against blinded human coding before it went live, and each kind now carries straight into a driver’s record. Since a July rollout, any driver whose reviews sort into the conduct kind three times within ninety days is entered into a published retraining register, with no process to contest the kind a review was given. The release extends the two Schools’ shared interest in the point at which a description of something starts to decide it instead.

The five kinds were built to describe what a review is about. What they have started to decide is a separate question, and it is the one we report on here.

— Associate Professor Kwame Lindqvist, Director, Master of Priority Studies, School of Emergent Priorities

The paper’s authors caution that a coding scheme’s reliability and its consequences, once automated, are not the same thing to measure, and report both separately.

The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.0i7mog.