A regression-discontinuity study finds a messaging platform's three-report auto-hide rule predicts almost nothing about whether a flagged message is an actual rule violation, and that authors post less after being auto-hidden regardless.
A School of Emergent Priorities regression-discontinuity study of 214 community group chats finds a platform's own auto-hide threshold barely tracks what a moderator's manual read confirms
A parenting group’s flagged message disappears the moment three members report it, whatever it says — and, a paper published today by the School of Emergent Priorities finds, whatever the sender does next tends to change too. The rule sits inside a messaging platform used by parent groups, street groups, and sporting clubs alike, applied identically regardless of what a group discusses. Pairing 214 of these chats with the platform’s own report-count rule, the study compares what happens either side of the three-report line: a message that clears it is hidden within moments, one sitting a report short stays visible however many members flag it. The report count predicts hiding almost perfectly, and predicts almost nothing about what a moderator’s later, unhurried read confirms.
A rule that sorts perfectly on one thing and not at all on another is not a failure of the rule. It describes what the rule was built to do, and it’s the people running the roster who are left holding what that means.
— Associate Professor Kwame Lindqvist, Director, Master of Priority Studies, School of Emergent Priorities
Co-author Dr Ronja Oyelaran called it an unusually clean natural experiment, a threshold applied the same way whatever a group discussed. The sharper finding sits downstream: authors whose message crossed into hidden territory posted noticeably less the following week than authors whose materially identical message stayed visible, across all three group types.
For the School, the result lands inside a question several recent projects have circled: what a rule like this actually buys the person left holding it. Associate Professor Lindqvist’s Master of Priority Studies practicum has set its current cohort to design a duty-moderator rotation policy for partner community groups.
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.nko2q7.