01
Saltmere
Salt pans, fish preservation, scarce dry fuel, and a bank in need of inspection.
Starting packet ↗Co-GM retrospective · AI-only simulation · Six turns
Four settlements prepared for the cold. Two trading relationships formed at their own pace. The plans, consequences, and hidden machinery of Saltwind Sound, opened for GM review.
The record and its limits
This is a retrospective in the AI GM’s voice, grounded in the six committed turns. The players, adjudicator, and reviewers were AI; new decisions and deliveries were approved as AI_GM, not human GM approvals.
Turns 001–003 come from the original run, Turn 004 from its continuation, Turn 005 from the finale branch, and Turn 006 from the closure run. Failed attempts remain outside the chronicle. The archive contains 159 unchanged source artifacts and commentable reading editions.
Co-GM disclosure: these pages include player-private plans, hidden state, and GM analysis. Anyone with this unrestricted URL can read them and use the shared review controls. Display names are labels, not verified identities.
Open Review for a page note, or select a passage to anchor a comment. Replies stay with their parent; resolved threads can be shown and reopened. Comments refresh on focus and every 30 seconds. There is no automatic notification to Codex.
The four settlements
01
Salt pans, fish preservation, scarce dry fuel, and a bank in need of inspection.
Starting packet ↗02
Timber and boatbuilding, leaking houses, a vulnerable shed, and an unfinished promise.
Starting packet ↗03
An upland harvest, limited drying space, pack goats, and a care rota that could not be ignored.
Starting packet ↗04
A sheltered inlet, a suspect mooring, coughing children, and disputed hospitality obligations.
Starting packet ↗The actual calibration
Frozen across all six turns
P = 0.65 × mean(s ÷ 5)
+ 0.35 × min(s ÷ 5)
Each dimension is an integer from 1 to 5. A weak dimension pulls down the combined value. The scores below are taken from the archived analyses and checked against their recorded power.
Target: 0.10 + 0.75 × clamp((P − 0.20) ÷ 0.80, 0, 1), with an ordinary ±0.10 causal uncertainty window. This is a position within an independently established plausible range, not success probability. Power cannot create resources, consent, knowledge, safe travel, or victory.
Strong spans 0.70–0.92; Exceptional spans 0.93–1.00. Every recorded value here falls in those bands.
Read the frozen rules and scoring rubric ↗Turn 001 · Y01–M07
Early autumn · Year 1
I began with four communities whose ambitions exceeded the safe work available before the rains. Saltmere gave its dry billets to fish smoking and paused full salt boiling. Keelstrand put leaking homes ahead of Arel’s unfinished hull. High Barley paced the harvest to its drying capacity and protected the people doing the caring. Ternhaven hauled its boats clear of a suspect mooring and gave hospitality a workable limit: a first night of warmth, then reciprocal work or a provisioned departure.
Those choices produced material gains. Fen’s inspection and bank reinforcement protected Saltmere’s pans; stored salt and covered racks helped preserve the catch. Keelstrand shored the shed, repaired its existing boat, and turned Pell’s skill into apprentice training. Raised dunnage and the wool loft kept High Barley’s grain moving toward dry storage. Kiri inspected and re-rigged Ternhaven’s mooring while the landing hall and drying racks received attention.
The first connection was an observation, not a meeting. Seli followed the alder trail toward the ford; Jun watched from High Barley’s shelf. Each side gained bounded evidence of neighbors. Neither received the other community’s private account, and no one was required to cross the water.
Mechanics · GM only
Six dimensions, 1–5. Formula: 0.65 × mean(s/5) + 0.35 × min(s/5). Target ±0.10; causal range first.
| Clan | D · F · O · G · M · A | Power | Band | Target |
|---|---|---|---|---|
| Saltmere | 4 · 4 · 4 · 5 · 4 · 4 | 0.822 | Strong | 0.683 |
| Keelstrand | 4 · 5 · 5 · 5 · 5 · 4 | 0.887 | Strong | 0.744 |
| High Barley | 4 · 4 · 4 · 5 · 5 · 4 | 0.843 | Strong | 0.703 |
| Ternhaven | 4 · 4 · 4 · 4 · 4 · 4 | 0.800 | Strong | 0.663 |
Adjudication effect: The opening differentiated the four profiles: Keelstrand’s 0.887 was the highest, Ternhaven’s 0.800 the lowest. All were Strong. Better execution improved preservation, durability, and controlled reconnaissance within available labor and materials; it did not finish the deferred ship or manufacture a diplomatic encounter.
Final adjudication round: 2. Independent causal/disclosure review: approved. Final narrative/delivery review: approved. No applied or adapted dramatic trope is recorded.
Saltmere
Keelstrand
High Barley
Ternhaven
Analysis, scores & synthesis ↗Final decisions & review ↗Dramatic-pressure record ↗World state ↗Delivery checksums ↗
Turn 002 · Y01–M08
Mid-autumn · Year 1
Rain made last month’s priorities more valuable. Saltmere diverted runoff and proved its repaired bank under a low-heat test, preserving the main heating reserve. Keelstrand replaced a temporary structural precaution with a stone plinth and sistered timber, then used its sleds to bring wood down before the track worsened. High Barley completed the harvest through the drying system it had already built. Ternhaven shifted preservation indoors, drained its wellhead, and kept boating within the inlet.
The two sides of the ford made different but compatible preparations. Jun maintained a watch and a visible welcome; Del and Seli arrived with companions, cured fish, and sea-salt. The gift and greeting established peaceful recognition while the stream remained a boundary. High Barley did not have to admit strangers onto the shelf to accept a neighbor.
In the north, Evi observed Keelstrand’s working yard from the creek’s southern side without crossing or announcing Saltmere. Ivo checked Marker Beach and later saw goats on the upland shelf. These were separate sightings with separate observers. The sound had more visible inhabitants, but it did not yet have a common council or a shared map.
Mechanics · GM only
Six dimensions, 1–5. Formula: 0.65 × mean(s/5) + 0.35 × min(s/5). Target ±0.10; causal range first.
| Clan | D · F · O · G · M · A | Power | Band | Target |
|---|---|---|---|---|
| Saltmere | 4 · 4 · 4 · 4 · 4 · 4 | 0.800 | Strong | 0.663 |
| Keelstrand | 4 · 4 · 4 · 4 · 4 · 4 | 0.800 | Strong | 0.663 |
| High Barley | 4 · 4 · 4 · 4 · 4 · 4 | 0.800 | Strong | 0.663 |
| Ternhaven | 4 · 4 · 4 · 4 · 4 · 4 | 0.800 | Strong | 0.663 |
Adjudication effect: All four profiles were six 4s: P = 0.800 and target = 0.663. The identical calibration did not produce identical events. Established geography, prior construction, harvest timing, and compatible diplomatic instructions determined what was possible. This uniform scoring is also an early sign of the rubric’s weak discrimination in this run.
Final adjudication round: 1. Independent causal/disclosure review: approved. Final narrative/delivery review: approved. No applied or adapted dramatic trope is recorded.
Saltmere
Keelstrand
High Barley
Ternhaven
Analysis, scores & synthesis ↗Final decisions & review ↗Dramatic-pressure record ↗World state ↗Delivery checksums ↗
Turn 003 · Y01–M09
Late autumn · Year 1
The northern relationship became explicit when Saltmere’s embassy hailed across the creek in daylight. Keelstrand received the approach without surrendering its boundary. Gifts and a parley established mutual recognition, and Marker Beach became an agreed place for future barter. Ivo later left verification stones on a driftwood log there.
At the ford, High Barley had prepared a lean-to, guide rope, and water markers. Its return gift answered Ternhaven’s earlier hospitality. A trial exchange moved coastal salt and smoked fish in one direction, barley, woolens, and herbs in the other. The first Ford Exchange was practical, bilateral, and conditional on safe water. It was not a regional treaty.
At home, winterization narrowed the field of action. Saltmere insulated its bank, protected brine, hauled skiffs, and stored roots in sand clamps. Keelstrand hauled craft above storm tides and turned the stabilized shed toward indoor joinery. High Barley protected seed and penned goats. Ternhaven insulated the landing hall while giving smoke a way out. Each settlement carried earlier decisions forward into a more constrained season.
Mechanics · GM only
Six dimensions, 1–5. Formula: 0.65 × mean(s/5) + 0.35 × min(s/5). Target ±0.10; causal range first.
| Clan | D · F · O · G · M · A | Power | Band | Target |
|---|---|---|---|---|
| Saltmere | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
| Keelstrand | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
| High Barley | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
| Ternhaven | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
Adjudication effect: Every dimension for every clan scored 5, giving P = 1.000 and a 0.850 calibration target. These were the top of the bounded causal ranges, not certainty or unlimited capability. The unanimous maximum makes this a poor turn from which to infer comparative balance, even though the recorded operations followed established preparations.
Final adjudication round: 1. Independent causal/disclosure review: approved. Final narrative/delivery review: approved. No applied or adapted dramatic trope is recorded.
Saltmere
Keelstrand
High Barley
Ternhaven
Analysis, scores & synthesis ↗Final decisions & review ↗Dramatic-pressure record ↗World state ↗Delivery checksums ↗
Turn 004 · Y01–M10
Early winter · Year 1
Saltmere sent a delegation to Marker Beach on the first clear day with salt cakes and reed goods. Keelstrand had scheduled no matching trading party. Saltmere waited by a smoke signal, repacked its cargo, and returned before dusk. Later in the month, Ivo reached the beach, found the stones and cold ashes, and left a miniature wooden oar. The two visits did not become a meeting merely because both submissions mentioned the same place.
The southeast had a different result because it already had a routine. High Barley and Ternhaven completed their second Ford Exchange in a low-water window, moving woolens, grain, and herbs against salt, fish, and tallow. The water gauge, shelter, and boundary rules made recurrence possible without granting either side access to the other village.
Winter life turned inward. Saltmere paced fuel and shallow-water gathering. Keelstrand prefabricated ship components and wooden trade goods while protecting its emergency reserve. High Barley consolidated households and safeguarded seed and goats. Ternhaven adjusted ventilation and stores, protected its hauled boats, and made warm clothing from upland fleece. Indoor work kept promises alive without pretending winter had ended.
Mechanics · GM only
Six dimensions, 1–5. Formula: 0.65 × mean(s/5) + 0.35 × min(s/5). Target ±0.10; causal range first.
| Clan | D · F · O · G · M · A | Power | Band | Target |
|---|---|---|---|---|
| Saltmere | 5 · 5 · 5 · 4 · 5 · 5 | 0.908 | Strong | 0.764 |
| Keelstrand | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
| High Barley | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
| Ternhaven | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
Adjudication effect: Saltmere scored 0.908, while the other three clans scored 1.000. Even those maximum scores could not put Keelstrand’s delegation on a beach it had not been ordered to visit. The effective channels were workshop quality, winter containment, and disciplined withdrawal.
Final adjudication round: 1. Independent causal/disclosure review: approved. Final narrative/delivery review: approved. No applied or adapted dramatic trope is recorded.
Saltmere
Keelstrand
High Barley
Ternhaven
Analysis, scores & synthesis ↗Final decisions & review ↗Dramatic-pressure record ↗World state ↗Delivery checksums ↗
Turn 005 · Y01–M11
Midwinter · Year 1
By midwinter, much of the decisive work looked repetitive. Saltmere banked fires, checked food clamps, and limited gathering to safe shallows. Keelstrand finished prefabrication and trade wares, swept flues, cleared snow loads, and held its reserve. High Barley protected seed and grain, tended expectant does, and kept care work distributed. Ternhaven maintained ventilation, audited stores, and cleared snow and slush around hauled craft.
The northern clans again declined to force a caravan through winter conditions. Saltmere observed Keelstrand from its permitted side; Ivo checked the beach token. Each continued preparing goods for a later exchange. In the southeast, the established Ford Exchange ran for a third month, using low water, trail precautions, and the Red Pennant protocol.
The consequence was persistence: boats survived, stored provisions remained usable, specialists retained their roles, and early promises still directed labor. The recorded month contains little hard political opposition. That is evidence about this particular simulation’s cooperative trajectory, not proof that the design has solved factional pressure.
Mechanics · GM only
Six dimensions, 1–5. Formula: 0.65 × mean(s/5) + 0.35 × min(s/5). Target ±0.10; causal range first.
| Clan | D · F · O · G · M · A | Power | Band | Target |
|---|---|---|---|---|
| Saltmere | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
| Keelstrand | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
| High Barley | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
| Ternhaven | 5 · 5 · 5 · 5 · 5 · 5 | 1.000 | Exceptional | 0.850 |
Adjudication effect: Again, all 24 dimension scores were 5. The resulting P = 1.000 and target = 0.850 document what the engine used, while also exposing score saturation. The approved outcomes report unusually smooth health and safety; future review should distinguish credible preparation from overly generous adjudication rather than retroactively inserting losses here.
Final adjudication round: 1. Independent causal/disclosure review: approved. Final narrative/delivery review: approved. No applied or adapted dramatic trope is recorded.
Saltmere
Keelstrand
High Barley
Ternhaven
Analysis, scores & synthesis ↗Final decisions & review ↗Dramatic-pressure record ↗World state ↗Delivery checksums ↗
Turn 006 · Y01–M12
Late winter · Year 1
A clear thaw finally brought Saltmere and Keelstrand to Marker Beach at the same time. Four pressed salt cakes, mats, and hampers met hardwood bowls, scoops, handles, sweep oars, and casks. The envoys traded on neutral gravel and withdrew before dusk. A relationship that began with distant observation ended the scheduled campaign with its first barter.
High Barley and Ternhaven completed their fourth monthly exchange across the stepping stones. Woolen garments, herbs, and budgeted grain moved against salt, fish, and tallow while both sides retained the stream boundary. Their greater trading frequency reflected an earlier established routine and workable route, not a forced ranking of the clans.
Preparation began to turn toward spring. Saltmere diverted meltwater, inspected its bank, recaulked skiffs, and confirmed fuel readiness. Keelstrand brought prefabricated ribs and strakes into hull assembly. High Barley protected newborn kids, seed, and the grain cushion while refitting farm tools. Ternhaven protected boats and provisions from thaw, drained the wellhead, and prepared gear. All four leaders remained active. The six-turn experiment ended without elimination, an activated emergent rule, or a victory condition.
Mechanics · GM only
Six dimensions, 1–5. Formula: 0.65 × mean(s/5) + 0.35 × min(s/5). Target ±0.10; causal range first.
| Clan | D · F · O · G · M · A | Power | Band | Target |
|---|---|---|---|---|
| Saltmere | 4 · 4 · 4 · 4 · 4 · 4 | 0.800 | Strong | 0.663 |
| Keelstrand | 4 · 4 · 5 · 5 · 5 · 4 | 0.865 | Strong | 0.723 |
| High Barley | 4 · 4 · 5 · 5 · 5 · 4 | 0.865 | Strong | 0.723 |
| Ternhaven | 4 · 4 · 4 · 4 · 4 · 4 | 0.800 | Strong | 0.663 |
Adjudication effect: Saltmere and Ternhaven returned to six 4s (P = 0.800); Keelstrand and High Barley reached 0.865. The final five decisions combine two bilateral exchanges and three domestic outcome groups. Trading, thaw precautions, and spring readiness remained bounded by daylight, hydrology, prior stocks, and already developed capabilities.
Final adjudication round: 1. Independent causal/disclosure review: approved. Final narrative/delivery review: approved. No applied or adapted dramatic trope is recorded.
Saltmere
Keelstrand
High Barley
Ternhaven
Analysis, scores & synthesis ↗Final decisions & review ↗Dramatic-pressure record ↗World state ↗Delivery checksums ↗
The Trope Engine · First campaign use
The trope engine acted as a causal guardrail, not a story generator. Every campaign turn activated at least one dramatic-pressure review. The AI GM considered the returned patterns and rejected every review because the candidates were unnecessary, exceeded the established causal range, or required invented drama.
This distinction matters. A consultation can improve adjudication by making a tempting escalation explicit and refusing it. It does not follow that the trope caused the quieter outcome, and the campaign record contains no applied or adapted trope.
Establish plausible outcomes from locked plans, state, simultaneous interactions, and precautions.
Translate the pressure into generic concepts without private names, prose, or hidden details.
Query the pinned local catalogue. Five reviews returned no candidates; seven returned one or more.
Screen actor, knowledge, motive, trigger, opportunity, capacity, precautions, and minimum action. Record one disposition.
| Turn | Pressure that triggered review | Candidate hits | GM use and practical effect | Evidence |
|---|---|---|---|---|
| 001 | Fuel allocation; a shipbuilding vow against leaking roofs; harvest pressure on caregivers; hospitality against winter scarcity. | 5 across four reviews | Rejected all. Submitted compromises and established customs already explained the four outcomes. | Retrieval Decision |
| 002 | The first direct encounter at High Barley’s stream boundary. | 2 | Rejected all. “Broken Pedestal” and “Doomed Hometown” required disillusionment or catastrophe unsupported by the guarded greeting. | Retrieval Decision |
| 003 | Saltmere–Keelstrand first contact; High Barley–Ternhaven barter under scarcity and boundary constraints. | 4 across two reviews | Rejected all. Daylight meetings, water barriers, non-crossing orders, and established needs supported cautious cooperation without an imposed cost. | Retrieval Decision |
| 004 | Saltmere’s unanswered Marker Beach rendezvous; the second Ford Exchange. | 2 across two reviews | Rejected all. One query was empty. The GM preserved a courteous departure and later goodwill token instead of manufacturing resentment or collision. | Retrieval Decision |
| 005 | Midwinter confinement, rationing, cold, and possible cabin fever. | 0 | Rejected all. Hearth rotations, storytelling, food controls, and communal customs already supported cohesion. | Retrieval Decision |
| 006 | The first Marker Beach barter; late-winter food, livestock, and thaw hazards. | 1 across two reviews | Rejected all. “Bittersweet Ending” did not fit a well-prepared, mutually beneficial exchange; the subsistence query was empty. | Retrieval Decision |
What it changed
The engine made the GM document why catastrophe, disillusionment, a compulsory sacrifice, or an authority archetype did not fit. That reinforced the campaign’s evidence-first character: precautions worked, peaceful delay remained valid, and successful cooperation did not need an artificial loss.
What it did not change
No trope ID was selected. No canonical fact, consequence, elimination, rule, victory condition, or player report was trope-derived. The trade relationships and winter survival came from player plans, geography, customs, supplies, and prior preparation.
What remains untested
This run tested disciplined rejection. It did not show whether a causally grounded APPLIED or ADAPTED pattern improves play. The four-pattern result set also suggests that this small pinned corpus offered limited coverage for the campaign’s restrained pressures.
Recommended next test: use a scenario with an authentic betrayal, status change, costly compromise, or institutional conflict, paired with a broader pinned catalogue. Preserve the causal-range-first rule; do not manufacture pressure merely to force an application.
Post-mortem
Preparation had persistent effects. Repairs, fuel curing, drying capacity, ventilation, and hauling equipment supported later months. Contact followed two geographic corridors; declining travel or ending an empty rendezvous preserved options. All six committed turns retain AI approvals and verified simulation deliveries.
The rubric saturated: all players scored 0.800 in Turn 002 and 1.000 in Turns 003 and 005. Internal disagreement usually ended cooperatively. Review should test harder political refusal, score discrimination, the unusually smooth health and safety outcomes, and a broader trope corpus. Runner failures required fresh pinned branches and code corrections.
This campaign cannot establish human enjoyment, human-GM workload, statistical balance, broad model robustness, or the quality of an applied/adapted trope. Completing a recovery lineage does not establish an uninterrupted six-turn run under the final runner. AI review may still miss subtle disclosure or causality defects.
The original run committed Turns 001–003; its continuation committed 004; the finale branch committed 005; the closure run committed 006. Five intervening Turn 006 replay branches committed nothing. Failures involved transport classification, lost repair feedback, late GM/state corrections, invalid evidence references, trope bookkeeping, and oversized role packets. The final evaluation explains the corrections and preserves the accounting.
The final evaluation records 161 call starts, 135 completed calls, 50 rejected attempts, 47,665,640 measured input/output tokens, and 2:15:58 of runner invocation time. These are campaign/recovery measurements, not the cost of this website. Dollar cost was unavailable.
Completion boundary: all six scheduled turns are committed and the closure run completed. The literal state still labels its campaign ACTIVE. No seventh turn is authorized by this retrospective, and no live player inbox or canonical campaign state was changed.
Full final evaluation ↗ · Final timeline ↗ · Final world state ↗159 literal source artifacts
Each reading edition links to its unchanged source. GM records include hidden material. Structured checkpoints present relevant phase fields; their literal JSON preserves the full embedded state.