[deadp]

DRAFT, game-value change, measure before merge.
Diagnosis (from code, no sims run): the chapter 2 goal of 5 item kinds gates chapter 3 (woolshed opens at season 7; the sim farmer repeats a few kinds), and the rare screen and fair open with chapter 3.
Change: ITEM_TYPES_FOR_CH2 5 to 3 (and ORDERS_FOR_CH2) moved into config.ts; the chapter 2 test and the DESIGN-v3 §10.1 table updated.
Tests: npm test (381) and typecheck pass.
Measure: run packages/game/scripts/blind.ts in chunks of at most 8 seeds (1-8, 9-16, 17-24, 25-30); compare actSeason[3] median and rare-by-27 against 15 and 2/30. I could not confirm the exact invocation from the docs.

Queued from the dashboard (item 544). The request:

Goal (proposal 8, Blue Sheep chapter ladder): Joshua chose "fix chapter 3 timing". In the 30-seed blind sim the median arrival at chapter 3 (true blue + a fair ribbon) is season 16-17; the DESIGN-v3 target is about season 11. Chapter 3 opens the rare-gene screen and southern rams, so late arrival is why rare fleece comes late (2/30 by season 27, target 15). PR 13 did not help; do not revert it.
Do: (1) Read docs/BACKLOG.md, docs/DESIGN-v3.md section 10.1, core/acts.ts, core/config.ts, core/pacing.ts, scripts/policy.ts. Find what holds chapter 3 back (true blue odds, ribbon/fair timing, chapter 2 goals, or the sim farmer's behaviour). Use the existing sim results in the repo docs; do NOT run long sims. (2) Make the smallest change that moves the chapter 3 goal earlier in a way a real player also feels (goal thresholds, fair timing, or an earlier route to a ribbon). Keep numbers in config.ts only. (3) Update or add unit tests for the changed goal; npm test and typecheck must pass. (4) Put in the PR body: what held chapter 3 back, what changed, expected effect, and the exact blind.ts command to measure it in chunks of at most 8 seeds.
Done when: PR open as a DRAFT (game-value change; it must not auto-merge before measurement), tests and typecheck pass, diagnosis written in the PR body.
Do not touch: world3d/, wool look, rare allele rates (mu), save format or save key, the forecast, CI files. No sim runs, no tuning of other values. One outcome. 30 minutes.

Opened by the queue runner on LXC 113. Merged once CI checks pass and the reviewer's verdict is recorded, unless it touches a hard line (qrunner/hardlines.toml): then it waits for Joshua's yes on the dashboard.

🤖 Generated with Claude Code