Muster v2 — Concept Prototype

Status: Concluded 2026-09-15 — verdict PIVOT (narrow) Independent playtest complete — see PLAYTEST-independent.md. The condition left open by REPORT.md is now closed, and the independent run changed the verdict. Predecessor: ../muster-mechanic-concept/ (verdict PIVOT)

Hypothesis

If military readiness has no computable safe threshold (raid strength scales, gear decays, forging takes 3 turns) and the economic alternative has a real failure state (ploughs feed a population that can starve), then the iron allocation stays a live decision across a whole season — and the player can articulate, unprompted, what their allocation cost them in both directions.

How to run

Paper, 3 seasons of 8 turns. Read rules.md. Roll each season's raid turn (1d5+3) and strength in advance and keep them hidden. Record allocation per turn and both failure causes if they occur.

The adversarial pass — the step v1 skipped

adversarial-pass.py brute-forces the optimal policy before any human plays. Run it: python3 adversarial-pass.py

It answers the question that killed v1: *what is the optimal play, and does it make the decision disappear?* For this ruleset it does not — adaptive play beats the best constant ratio by 8 points, and both extremes (all-ploughs, all-spears) die 100% of the time, which confirms the instrument can detect failure.

Three instrument bugs were found and fixed while building it:

• Clamping losses at the floor made every policy survive at 100% — the meter read a constant.
• Resetting harvest each season made ploughs near-worthless late.
• No control run. Without verifying the model can report a death, a survival number means nothing.

Findings

Played 2026-09-15. 6 of 12 settlers survived 3 seasons. The allocation stayed live for all 24 turns; three seasons produced three outcomes from three causes (correct planning, a lucky roll, a lead-time failure).

Season 3 is the mechanic's thesis: expected strength 9.8, raid landed turn 4 with only 4 of 10 spears ready. Six settlers died. The remaining six spears arrived on turns 5 and 6 — after the funeral. The iron was committed; it was three turns too late.

Caveat: the solver showed this only holds in a narrow difficulty band. A friendlier tuning pass silently destroys the mechanic. See REPORT.md.

Files

• rules.md — the ruleset, with solver findings and what to watch
• adversarial-pass.py — the policy solver (run before play)
• ledger.py — rule-enforcing state tracker used during the session
• REPORT.md — the author's own run (PROCEED WITH CONDITIONS — superseded)
• PLAYTEST-independent.md — the authoritative result: independent tester, PIVOT
• live-ledger.py — the rule-enforcing ledger used during the independent session