Community Engagement & Moderation Simulator

Community Levers

Content quality 65%

Posting cadence 3 / day

Moderation strength 50%

Moderator response speed medium

Live Metrics

Members

120

Avg. sentiment

0.00

Engagement rate

Daily churn

Simulation started…

How it works

Each of the N dots is one community member with a hidden sentiment s ∈ [−1, 1] and an engagement propensity e ∈ [0, 1]. Every simulated "day" the model runs three coupled updates:

Post arrival:  posts/day = cadence
Reaction:      Δs_i = k1·(quality − 0.5) − k2·toxicity_i + noise
Toxicity:      toxicity_i ~ Bernoulli(p_troll), scaled by (1 − moderation)
Moderation:    removal_prob = modStrength × modSpeed  (per toxic post, per tick)
Churn:         P(leave) = σ(−s_i · c1 − e_i · c2 + c3)     [logistic]
Growth:        new members/day ∝ max(0, avg sentiment) × retention appeal

High content quality raises baseline sentiment drift for every member each tick. High posting cadence multiplies how many reaction events fire per tick — great for engagement, but it also multiplies how often toxic posts get a chance to appear, so cadence and moderation trade off. Moderation strength sets what fraction of toxic content is actioned; response speed sets how quickly (fewer ticks of exposure before removal). Members whose sentiment stays negative for several ticks in a row churn out (turn red, shrink, and drift outward before being removed); members with rising sentiment and engagement turn green and pull toward the center, the "core" of the community.

This mirrors real community-management tradeoffs: a Discord/Reddit/forum team that posts too rarely loses engagement, posts too often without quality dilutes sentiment, and under-moderates lets toxic actors drive out everyone else — the same dynamics used informally by trust-and-safety teams to reason about moderation staffing.

Engaged / positive Neutral / lurking Toxic / churning

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.