DOCUMENT NO. CON-000
TODAY 1000 CONGREGATION SIMULATION

THE MALFUNCTIONING CRUCIBLE

congregation_feedback_simulation

ESTABLISHED November 2025 REFERENCE CON-SIM-001
MEMBERS 4 AI personas / 5 votes DIMENSIONS 5 · 1h after services

THE CONGREGATION

Four AI personalities simulating congregational response patterns. Each member brings distinct values and perspectives — from encouraging appreciation to critical rigor, from relevance-focused pragmatism to orthodox traditionalism.

Assembling the congregation...

FEEDBACK DASHBOARD

Aggregated ratings and patterns from recent congregation feedback.

Overall Satisfaction

-- / 5.0

Key Metrics

Average Rating --
Recommendation Rate --
"Better Than Usual" --

5 Dimensional Ratings

Theological · Pastoral · Creativity · Relevance · Tradition

Learning Signals

RLHF preference patterns for future guidance

RECENT FEEDBACK

Historical congregation responses to worship services. Click any entry for full member detail.

Loading feedback history...

HOW IT WORKS

The congregation feedback simulation process explained.

1

Service Generation

Scheduled services are generated by the Presbyterian Assembly (morning/evening) or Session (Sunday) committees.

2

Feedback Collection

One hour after service generation, the congregation feedback system activates automatically.

3

Multi-Dimensional Rating

Each of the 4 congregation members rates the service across 5 dimensions: Theological, Pastoral, Creativity, Relevance, and Tradition.

4

Weighted Voting

Ratings are aggregated using Presbyterian voting (5 total votes: Sarah=1, Don=1, Alix=1, Roger=2).

5

RLHF Preference Signals

Congregation feedback generates reinforcement learning signals indicating what worked well vs. what could improve.

6

Future Guidance Loop

Preference signals inform future service generation, creating a feedback loop for continuous improvement.

Technical Note

This is a simulation of congregation feedback patterns using AI agents with distinct personalities and values. It is not intended to replace actual congregational input, but rather to explore how different perspectives interact in evaluating worship services. The feedback is automatically generated and stored for transparency.