Welfare Ladder
Per-user sentiment tracking over time with multi-rung escalation, crisis-level human intervention, and a full welfare audit trail.
What is it?
The Welfare Ladder (COS-005) is Warburton's system for tracking individual member wellbeing over time and escalating appropriately when someone needs help. It uses longitudinal sentiment patterns — not single-message analysis — to identify members who may be struggling, and provides a structured escalation path that always ends with a human when it matters.
Why does it exist?
Most people don't need rescuing. They need pointing towards another human.
When someone in a community shows signs of distress, the worst response is a clinical disclaimer. The second worst is ignoring it entirely. The Welfare Ladder exists because online communities have a duty of care to their members, and that duty requires a system — not good intentions alone.
People in distress rarely announce it clearly. The signals are in the patterns: changes in posting frequency, shifts in tone over weeks, withdrawal from conversations they used to engage with. A single message might mean nothing. A pattern over time might mean everything.
Why is it different?
Most community tools either ignore welfare entirely or provide a keyword-triggered response — if someone says certain words, they get a canned message with a helpline number. This is performative care, not actual care.
The Welfare Ladder tracks sentiment per user over time, using longitudinal patterns rather than single-message triggers. It escalates through multiple defined rungs, each with specific criteria and specific responses. And it never makes the final call on welfare — crisis-level intervention always routes to a human operator.
When Warburton detects signs of distress, it doesn't issue a clinical disclaimer. It uses its character voice to gently suggest the person speak to a real human. The referral is discreet and does not alert other members that the system noticed.
How does it work?
- Per-user sentiment tracking: The system maintains a sentiment profile for each community member, tracking changes over time rather than evaluating individual messages in isolation. This longitudinal approach catches gradual declines that single-message analysis would miss.
- Multi-rung escalation: The ladder has multiple defined rungs, each with specific entry criteria and defined responses. Lower rungs involve passive monitoring. Middle rungs involve gentle character-voiced engagement. Higher rungs involve operator notification. The criteria at each level are explicit — no ambiguity about when escalation occurs.
- Crisis-level human intervention: At the highest escalation levels, the system always routes to a human operator. No AI makes the final call on welfare. The system provides the operator with the full context — the member's sentiment history, the triggers that caused escalation, and the timeline of changes.
- Full welfare audit trail: Every escalation is logged with timestamp, trigger, rung level, action taken, and outcome. This creates a complete record that can be reviewed for compliance, safeguarding audits, or post-incident analysis.
Example
Why should anyone care?
If you run a community, someone in it is having a hard time right now. You probably don't know who. The Welfare Ladder doesn't replace human judgement — it extends it. It watches when humans can't, tracks patterns humans wouldn't notice, and ensures that when someone needs help, the path to that help is structured, discreet, and documented.