Greywater Reach — GM Operating Evidence
Evidence horizon: Campaign operations through Turn 006 closure; player debrief responses pending
Recorded elapsed workflow intervals
These are wall-clock intervals between recorded gates, not claims of continuous active labor. Long spans primarily represent human-GM review or an overnight pause.
| Turn | Intake lock → analysis | Analysis → adjudication | Adjudication → drafts | Drafts → delivery approval | Approval → verified delivery | Delivery → closure |
|---|---|---|---|---|---|---|
| 1 | 8m 10s | 4m 42s | 5m 11s | 3m 19s | 45s | 6m 47s |
| 2 | 6m 16s | 33m 53s | 2m 23s | 20h 58m | 1m 10s | 36s |
| 3 | 8m 17s | 31m 19s | 6m 48s | 50m 02s | 44s | 17m 08s |
| 4 | 8m 07s | 10h 51m | 5m 06s | 3m 39s | 45s | 2m 28s |
| 5 | 5m 12s | 1h 43m | 4m 30s | 3m 15s | 24s | 7m 13s |
| 6 | 6m 41s | 7m 16s | 2m 36s | 3h 24m | 1m | 8m 36s |
Decision burden
| Turn | Docket decisions | Character of review |
|---|---|---|
| 1 | 4 | Ordinary local adjudication |
| 2 | 4 | Ordinary local adjudication |
| 3 | 5 | Local adjudication plus midpoint attractor review |
| 4 | 4 | Ordinary local adjudication |
| 5 | 6 | Four local outcomes, emergent-rule gate, attractor review |
| 6 | 6 | Four local outcomes, first-contact gate, final attractor review |
Four ordinary turns stayed within the three-to-five target. Turns 5 and 6 exceeded it by one because each included separately controlled experiment-level questions in addition to local adjudication. Routine provenance, delivery verification, and canonical generation were not counted as substantive decisions.
Character and conflict evidence
- Named actors repeatedly changed or constrained plans: Oren and Pell at Ridgeford; Nessa, Ilya, Beren, and Ava around Hazelstead; Rusk and the injured households at Claybank; Toren and Tovan around Merehaven.
- Strong plans generally contained conflict through stops, records, hearings, custody rules, and review dates without erasing material scarcity, rival claims, injury, or distrust.
- Consequences remained causally traceable: crossing maintenance, visible harvest loss, enduring disability, impaired reserve hull, deferred production, and unresolved land and inlet claims.
- Quiet conflicts were most effective when tied to named authority and a future decision. They risked feeling overly orderly when every disagreement resolved into a bounded procedure.
Operational defects and judgment
- No privacy leak, wrong-player result delivery, checksum mismatch, unauthorized elimination, or unauthorized victory promotion occurred.
- A validator defect counted historical decisions against the current-turn ordinary-decision target. The narrow correction scopes the count to the current turn and is covered by automated tests.
- Structured-state generation reduced stale-view drift but did not replace human judgment for prose disclosure, causal fairness, character motive, cross-player observability, or whether an attractor deserved promotion.
- No player absence occurred, so the approved absence procedure and hard third-miss boundary were not exercised in live play.
- Direct player contact remained optional and occurred only after a player deliberately funded and gated it in the finale.
Victory-attractor finding
Witnessed bounded authority became a recognizable design direction across all four communities. It did not mature into a credible victory structure: there was insufficient sustained cross-player comparison, shared dependency or conflict, and no common competitive axis. The final canonical status REVIEWED_NO_PROMOTION is supported.