diumTrust Score: 11 Events That Should Move It, and the One That Shouldn't
Reputation gamified badly is karma farming. Reputation gamified well is your moderation team multiplied tenfold.
Trust and health
Moderation, reputation, slow mode, health metrics and search visibility: how to keep a community useful as it grows.
Reputation Gaming Is Real: the 4 Design Choices That Defang Goodhart's Law
Multi-factor scores, peer verification, time-decay, and anti-gaming heuristics. Together they make ring-voting structurally unprofitable.
The 6-Level Trust Ladder You Can Copy from RentFox and Ship by Friday
Tactical post: a level-not-number display that maps to specific permission unlocks at each tier.
Anonymous Posting That Doesn't Poison the Well: the 3-Rule Design
Per-thread opt-in, hidden in display only, mod-auditable. Three rules that make anonymous posting useful without enabling abuse.
Moderation Queue That Actually Gets Cleared: the 'Top of Inbox' Redesign
Sort by reporter trust × content severity × time-since-report, not FIFO. Mod queues clear 3x faster.
The Community Health Scorecard: 12 Metrics, With Formulas
One composite, six diagnostic sub-scores, twelve underlying metrics, and an operating cadence.
Auto-Raid Detection Without False Positives: the 4 Signals That Catch 90%
Join-velocity spike + low-trust authors + repeated-content fingerprint + cross-thread pattern within 5 minutes.
Answer Engine Optimization for Community Platforms: the 9 Schema Patterns ChatGPT Actually Cites
Definition Box, Comparison Matrix, HowTo Steps, FAQ, Spec Sheet, Glossary Anchor, Pricing Box, Schema Snippet, Named Formula.
Why Your Community Needs Three Suspension States, Not One
Mute (read-only) / suspend (time-boxed) / ban (permanent with appeal): three discipline tiers reduce repeat offenses by 41%.
Why Community SaaS Pages Are Invisible to ChatGPT (and the Fix Is a Glossary, Not Backlinks)
LLMs cite glossary pages 4x more often than blog posts because glossaries hit the entity-disambiguation step in retrieval.
The 'Apology Economy': How Appeals and Ban-Expiry Rebuild Trust Faster Than Perma-Ban
Time-bounded suspensions with documented apology pathways recover 28% of suspended members. Perma-ban recovers 0%.
The 'Definition Box' Page Template That Wins AI Overviews for Community Queries
First 60 tokens directly answer 'X is...'; entity reinforced in URL, H1, and meta; FAQPage schema in the head.
Profanity Filters Are the Wrong Layer: Where Toxicity Actually Lives
Profanity filters catch 12% of harm. Intent-classified toxicity at the embedding layer catches 78% with fewer false positives.
The 'Comparison Page' That Cite-Bait Works On
Comparison pages cited by ChatGPT use a labeled-cell matrix in the first 800 tokens, named entities in column headers, and verifiable numbers per cell.
The Audit Log Every Community Needs (and the 4 Events Nobody Logs)
Role changes, content edits, moderation actions: most platforms log these. The four nobody logs: session impersonation, P2P transfer metadata, retention deletes, key rotations.
GEO for Forums: How to Get Cited When Someone Asks 'What's the Best Community Platform for X'
Vertical AEO. Per-vertical comparison pages with named-criterion grading, not generic top-10 lists.
The 'Lurker Problem' Is Fake: Your Platform Is Just Unaddressable
The 90-9-1 rule is a UI symptom, not a behavior law. Addressable threads with single-question prompts move lurkers to repliers at ~22%.
Why Every Community Should Ship a 'Leaving' Workflow
Communities with structured offboarding forms recover 11% of departing members within 6 months.
The 'Show Me What You're Working On' Weekly Thread Is the Best Growth Hack You're Not Running
Weekly recurring Execution thread averages 38% reply rate and seeds 60% of next month's office-hours topics.
Why Your AI Summarizer Should Write the Next *Question*, Not the Recap
Summarize-then-ask outperforms summarize-only by 41% on next-day reply rate. The AI's job is to set up the next contribution.
Voice-to-Thread: Turning a 60-Min Zoom Into 1 Broadcast + 5 Inquiries + 1 Execution Thread
Voice-to-thread parses meeting transcripts into a Broadcast (decisions), Inquiries (open questions), and an Execution thread (action items).
The 5 AI Features Your Community Should NOT Ship (and the 1 That 10x's Value)
Skip auto-replies, AI hosts that talk, AI moderation hard-actions, AI personalized feeds without explainability, AI engagement bots. Ship: AI duplicate-question detection.
Open Dataset: 100 Anonymized Wave Threads (CSV) for Researchers
Backlink magnet: 100 anonymized waves with type, lifecycle, reply count, trust scores. Reproducible community research.