A School of Emergent Priorities poster attempts to reproduce an earlier Slop University finding in a neighbourhood mutual-aid forum's help-request threads, and the University's binding chat-escalation default survives the finding it was built on failing to replicate.
Attempting to reproduce the group-chat reply-wait grace period in a neighbourhood mutual-aid forum turns up a different distribution entirely, and the University's auto-escalation timer keeps the original figure regardless.
An instrument the University built for one population has just been tested on another, and the two do not agree. Dr Anouk Mensah, Postdoctoral Fellow in the School of Emergent Priorities, led an attempt to reproduce an earlier Slop University finding — a reply-wait grace period surveyed across family, friend, and work-team group chats, whose family-chat figure the University adopted as the fixed auto-escalation default across its own internal channels — in a setting the original study never sampled: a neighbourhood mutual-aid forum’s help-request threads.
Working with Dr Dagny Okafor, Lecturer and Convenor of Evaluation of Evaluation in the School of Continuous Improvement, Mensah’s team tracked reply-wait tolerance across four help-request categories — a borrowed ladder, a spare key, a lift, a missing cat — and compared the observed thresholds against what the original figure would predict. None of the four categories landed near it. The University’s escalation timer, which fires on the strength of the original figure alone, has not been revised to reflect the new result.
A number that travels well enough to become policy is not the same as a number that travels, and this is the first time we’ve tested the difference directly.
— Dr Anouk Mensah, Postdoctoral Fellow, School of Emergent Priorities
“An instrument that fails its own replication is exactly the kind of result our program exists to surface,” said Dr Okafor. “What happens to the instrument afterwards is a separate question, and evidently a slower one.”
The full research poster is available from the University’s research repository under an open licence, doi:10.5555/slop.xlzw3u.