HSpector1 · HSpector1 · Jul 29, 2026 · Jul 29, 2026 · Jul 29, 2026 · Jul 29, 2026
diff --git a/docs/DECISION-LOG.md b/docs/DECISION-LOG.md
@@ -5,6 +5,43 @@ consequences, owner, and conditions to revisit. Newest first.
 
 ---
 
+### 2026-07-30 · D-029 · Menu-responsive demand rewrite: distinct regimes achieved; universal-generalist gate borderline (economic root cause)
+- **Status:** provisional · **Owner:** Builder → owners · **Follows** the PR #5 adjudication (D-028's Fail/Rewrite,
+  refined: the fix is a headless demand-model rewrite, not M1). Branch `rewrite/m0-menu-responsive-demand`.
+- **Decision/finding:** Replaced menu-independent arrival composition with a **menu-responsive consideration
+  model** (`DemandModel.Capture`): per-segment depth + positioning focus + occasion variety + a representative
+  price-position gate set BOTH arrival volume and composition; purchase (`PickBest`) stays separate. This makes
+  the per-market frontiers **distinct regimes** — value/throughput wins the value lunch, premium wins the
+  enthusiast evening, mixed wins social — which the old fixed-composition model could not. Manipulation-resistant
+  (filler/never-ordered/anchor bounded), legible, deterministic; forecast reuses the same `Capture`.
+- **Residual (honest):** the strict "no plan within 10% of the frontier in every market" gate is **borderline-
+  missed on held-out seeds** — a premium-anchored generalist's worst-market regret is ~6–13% (median ~9%, sweep
+  over 4 bases), dominator on 2/3 fresh bases. **Decisive cause (economic, not demand):** consideration is
+  market-name-agnostic, so a premium menu serves the same fraction of the premium slice in every market, and
+  premium multi-course checks (~5× a value cover) make that profitable everywhere. Regret is fixed by (pool mix)
+  × (per-cover economics), both outside the demand model → no composition candidate (C1–C4) can close it. This is
+  the concrete answer to §22's "does the product need an economic rebalance / broader redesign": **yes, the
+  residual needs an economic rebalance or the excluded systems (reputation/repeat-visits), NOT abandoning the
+  demand model** (which is correct and kept).
+- **One fixture change:** lunch pool 65→82% value + 120→150 arrivals, to match its stated "value-heavy,
+  high-volume" identity (achieves 3 of 4 gates). Pushing further (~95% value) to force the 4th gate is
+  gate-chasing and was declined. Flagged for owner scrutiny.
+- **Consequence / gate:** `Verdict: Conditional-Pass / Action: Continue` (open PR for independent review + owner
+  decision). Supersedes D-028's "needs M1" framing: the demand rewrite (headless, in scope) delivered most of the
+  property; the remaining residual is economic. Goldens re-baselined (see D-030). Do not merge / no human tests /
+  no M1. Full report: `reports/m0/M0-DEMAND-REWRITE-REPORT.md`; iteration evidence:
+  `reports/m0/demand-rewrite/candidate-1.md`.
+
+### 2026-07-30 · D-030 · Goldens re-baselined for the demand rewrite; Focused Value strategy rebuilt
+- **Status:** locked · **Owner:** Builder
+- **Decision:** The menu-responsive demand changes (arrival volume/composition, MakeParty sampling the captured
+  mix, the value-heavy lunch pool) and the rebuilt **Focused Value** strategy (a full-throughput value operation
+  so a value plan wins the value-heavy lunch) intentionally change the simulated order stream, so the three
+  golden checksums were re-baselined: Focused Value/lunch `0xFB04479B19656E6B`, Premium Craft/enthusiast
+  `0x198E44A139793975`, Balanced Competent/social `0x6CBD8F76B46A51E1`. Prior (D-027): `0xC814AFAA4D8752DE` /
+  `0x4D834261A7A6813E` / `0x2D2C5DCA4431A5EF`. Also recalibrated the causal report's weak-demand attraction floor
+  (70%→35% of pool) for the new demand regime (does not affect the checksum; cause strings are not hashed).
+
 ### 2026-07-29 · D-028 · CORRECTION to D-025: High B is NOT resolved — a mixed dominator remains (Fail/Rewrite)
 - **Status:** locked · **Owner:** Builder · **Supersedes the "no dominator" result in [[D-025]].**
 - **Decision/finding:** After the affordability fix, a **strengthened** harness search (multi-start hill-climb,

diff --git a/docs/design/M0-DEMAND-REWRITE-CONTRACT.md b/docs/design/M0-DEMAND-REWRITE-CONTRACT.md
@@ -0,0 +1,116 @@
+# M0 Demand-Model Rewrite Contract (design lock)
+
+**Status:** Design-locked, implementation NOT complete. Branch `rewrite/m0-menu-responsive-demand` off PR #5 head
+`7bd8060`. This replaces one failed load-bearing M0 assumption. It is not M0.5, M1, visual, reputation,
+repeat-visit, or a scope expansion.
+
+## 1. Product question
+Can menu, pricing, staffing, and capacity decisions create multiple understandable and viable restaurant
+strategies **by influencing both customer attraction and service outcomes**, with visible consequences that
+make players want to revise their plan and run another service?
+
+## 2. Failed prior assumption (preserved, not euphemized)
+> Customer arrival composition can remain fixed independently of the restaurant's menu and prices while still
+> producing meaningful market-specific strategies.
+
+Independently adjudicated **Fail/Rewrite** (PR #5): a fixed mixed value-chassis + premium-anchor plan is
+near-optimal in all three markets because `MakeParty` samples each party's segment from the fixed
+`SegmentMixBp` and the per-segment `ConversionBp` signal (`segConv`) is computed then discarded. The premium
+anchor skims the always-present high-budget minority at no attraction cost. See `M0-CORRECTION-3-REPORT.md`
+and the PR #5 adjudication comment.
+
+## 3. New assumption under test (must be able to fail)
+A deterministic, single-service, headless model of **restaurant consideration** can create credible
+positioning tradeoffs **without** repeat visits, reputation, marketing, or a living city.
+
+## 4-8. Model sequence (conceptually distinct stages)
+```
+Potential market pool (SegmentMixBp = pool, NOT arrivals)
+ → RestaurantOffer derived from menu/prices/expected operation
+ → per-segment Consideration (offer vs segment budget/occasion/quality/pace)
+ → Captured demand (pool × consideration) — sets arrival volume AND composition
+ → seeded Arrival realization
+ → in-restaurant purchase (order/substitute/decline — UNCHANGED: WTP + affordability)
+ → service → outcome
+```
+`RestaurantOffer` is a small explicit vector (e.g. ExpectedMealCost, ValueAvailability, PremiumAvailability,
+ExpectedQuality, ExpectedPace, MenuBreadth) derived from actual choices — never a player-picked label, never
+inferred from max price alone (must resist filler-dish manipulation; weight by likely orders). Purchase stays
+separate: a customer may choose the restaurant and still reject an overpriced dish.
+
+## 9-10. Forecast & causal relationships
+Forecast exposes the funnel: potential → captured → arrivals → completed covers → revenue → contribution, and
+names attraction risks (weak value appeal, premium demand too small, high expected meal cost, poor occasion
+fit, capacity shortfalls) in plain language (no raw utility formulas). Causal report distinguishes
+**attraction** failures (never considered / wrong segment / price-position deterrence) from **conversion**
+failures (rejected prices, no suitable dish) from **service** failures (seating/kitchen/FOH/execution). It
+must not blame operations for customers who never intended to visit.
+
+## 11-13. Numerics / RNG / scope
+Integer & fixed-point only, `FixedMath.MulDivRound`; named seeded RNG streams (no wall-clock, no
+`Math.random`); smooth consideration (no cliffs), permits low and high consideration, segment-varying,
+name-agnostic (no market-name / strategy-fixture branch). Preserve all existing invariants (accounting,
+determinism, immutable forecast, affordability, WTP, capacity, patience, execution, attribution, FIFO, labor,
+guards). Replace only: fixed arrival composition, demand generation that ignores the offer, forecast demand
+assumptions, exogenous-demand reports, search evaluation that conflates attraction with conversion.
+
+## 14-15. Non-goals (hard)
+Repeat visits, reputation, reviews, critics, marketing, loyalty, word-of-mouth, CLV, neighborhood/city,
+competitors, travel, reservations, delivery, inventory/suppliers/spoilage, new recipes/markets/segments,
+hiring, graphics/Godot/pathfinding/save/multi-restaurant/managers/campaign/audio/assets/multiplayer/mod.
+No content-volume increase.
+
+## 16-17. Strategy-integrity & positioning gates (pass criteria)
+- **No fixed complete plan** within **10%** of the best discovered held-out result in all three markets
+  (report 5/10/15/20% sensitivity). Value/throughput plan within 5% of best lunch; premium plan within 5% of
+  best enthusiast; lunch vs enthusiast optima differ on ≥2 of {menu, price regime, staffing, capacity,
+  captured mix, bottleneck}. **Mixed menus remain viable** (esp. social) — the deterrent must be occasion-fit
+  opportunity cost, NOT a mixed-menu penalty, breadth tax, cuisine-purity score, name branch, or hard ban.
+- Demand gates: menu/prices materially shift arrival composition; a premium anchor changes attraction not
+  only order mix; a broad menu does not capture every segment at full strength; filler/never-ordered dishes
+  have bounded influence.
+- Forecast gates: ≥85% overall / ≥75% per-market & per-family directional agreement (price, positioning, menu,
+  seat, kitchen, FOH); audited price & seat directions correct; positive ranking correlation each market.
+
+## 18. Required owner decision
+After the build: does the menu-responsive demand model establish the strategy-integrity property (Continue to
+independent technical re-review), or does it hit a §7 fail condition (Rewrite/Abandon the demand approach)?
+
+---
+
+## Implementation status & the load-bearing risk (honest)
+**IMPLEMENTED (Candidate 1, "relevant-option depth"). Verdict: Conditional-Pass / Continue.** See
+`reports/m0/M0-DEMAND-REWRITE-REPORT.md` (§A–H) and `reports/m0/demand-rewrite/candidate-1.md` (iteration log).
+`DemandModel.Capture` now sets arrival volume AND composition from per-segment depth + positioning focus +
+occasion variety + a representative price-position gate; purchase stays separate. The per-market frontiers are
+now **distinct regimes** (value wins lunch, premium wins enthusiast, mixed wins social); manipulation-resistant,
+legible, deterministic; 139 tests green. **The strict universal-generalist gate (§16 below) is borderline-missed
+on held-out seeds** (premium-anchored generalist worst-market regret ~6–13%, median ~9%; dominator on 2/3 fresh
+bases). Decisive cause is ECONOMIC (market-agnostic consideration ⟹ a premium menu serves the same fraction of
+every market's premium slice, and premium multi-course checks earn ~5× a value cover), so no composition
+candidate can close it; it needs an economic rebalance or the excluded systems. One principled fixture change was
+made (lunch pool 65→82% value, matching its "value-heavy" identity); pushing to ~95% to force the gate was
+declined as gate-chasing. **Owner decision (§18) required.**
+
+The failed v1 prototype (below) is preserved as evidence:
+**v1 (reverted):** `RestaurantOffer` = average main price; consideration = price-level fit × identity fit;
+captured mix + a mix-weighted `drawBp` volume scalar. Measured on held-out seeds it **FAILED both ways**: the
+mixed dominator still won all three markets (0% regret), and focused value/premium plans got *weaker* (the
+mix-weighted `drawBp` penalized focused menus, and price-level fit alone does not deter the minority skim). The
+surviving Candidate 1 fixes this by (a) driving composition from undistorted per-segment ratios, not a
+mix-weighted scalar, and (b) using a mean-price position gate + occasion-specific depth rather than price alone.
+
+**Lesson / critical uncertainty (per §33):** consideration based on price-fit + best-dish-fit is *necessary
+but insufficient* — a mixed menu genuinely fits everyone somewhat, so a naive model lets it attract everyone
+(reproducing the exploit) OR, if volume is mix-weighted, it wrongly punishes focused menus. The load-bearing
+design is the **coupled second piece**: an occasion/positioning-coherence term where a market's *occasion*
+(derived from the pool's own parameters — budget distribution, patience, not the market name) rewards a
+restaurant that is a *clear fit* for that occasion and gives a muddled cheap+premium offer only a partial draw
+from each pole, so a **focused** restaurant captures more of its pool than the mixed one — while a mixed
+restaurant still wins the market whose occasion *is* "moderate/varied" (social). Calibrating this to (a) drop
+the mixed generalist below the 10% band, (b) keep mixed viable in social, (c) not punish mixed arbitrarily, is
+the crux and the likely stop point if the model cannot be made to work.
+
+**Estimate:** ~24-40 focused hours across the six workstreams (offer, consideration/arrival, forecast funnel,
+causal attraction layer, search, docs). Critical path = consideration/arrival model + calibration (highest
+uncertainty). This exceeds a single builder session; the branch, contract, and v1 evidence are the handoff.
diff --git a/reports/balance/distribution.md b/reports/balance/distribution.md
@@ -1,24 +1,24 @@
 # M0 Balance / Distribution Report
 
-Generated by `RestaurantSim.Harness`. Seeds per cell: **200** (base seed 700000). 
+Generated by `RestaurantSim.Harness`. Seeds per cell: **80** (base seed 700000). 
 Contribution is per single service in integer cents rendered as dollars. "win%" = share of seeds 
 where this strategy had the highest contribution in that scenario (same seed compared across strategies).
 
 ## Neighborhood Lunch Rush  (`lunch-rush`)
 
-_Value-heavy, high-volume, price-sensitive, small parties, early peak. Rewards fast throughput and lean value menus._
+_Value-heavy, high-volume, price-sensitive, small parties, early peak. Rewards fast throughput and lean value menus. Pool is strongly value-dominated to match its stated identity: under menu-responsive demand a market's winner is set by its customer pool, so a genuine value-lunch occasion must be mostly value diners (few premium diners to skim)._
 
 | Strategy | median contrib | p10 | p90 | loss% | avg covers | avg sat | avg ticket | win% |
 |---|--:|--:|--:|--:|--:|--:|--:|--:|
-| Focused Value | $838.65 | $404.50 | $1127.80 | 0% | 93 | 521 | 16m | 68% |
-| Premium Craft | $421.64 | -$76.56 | $865.04 | 14% | 26 | 748 | 15m | 16% |
-| Broad Menu | $219.00 | -$421.00 | $931.50 | 38% | 83 | 543 | 19m | 15% |
-| Overcapacity | -$635.50 | -$843.50 | -$476.50 | 100% | 22 | 524 | 20m | 0% |
-| Understaffed | -$476.00 | -$558.00 | -$376.50 | 100% | 12 | 497 | 22m | 0% |
-| Balanced Competent | -$36.00 | -$445.50 | $483.50 | 53% | 64 | 554 | 20m | 0% |
-| Overpriced Weak Execution | -$78.60 | -$377.00 | $323.90 | 62% | 10 | 435 | 25m | 0% |
-| Station Bottleneck | -$495.50 | -$634.50 | -$371.00 | 100% | 15 | 578 | 18m | 0% |
-| Intentionally Bad | -$140.00 | -$284.50 | -$7.50 | 91% | 7 | 568 | 19m | 0% |
+| Focused Value | $1682.50 | $1480.00 | $1892.00 | 0% | 170 | 604 | 11m | 93% |
+| Premium Craft | $1064.12 | $635.46 | $1549.56 | 1% | 42 | 719 | 17m | 6% |
+| Broad Menu | $393.00 | -$121.50 | $933.50 | 15% | 79 | 580 | 19m | 0% |
+| Overcapacity | -$572.00 | -$756.00 | -$424.50 | 100% | 23 | 538 | 19m | 0% |
+| Understaffed | -$402.00 | -$529.00 | -$261.50 | 100% | 14 | 512 | 21m | 0% |
+| Balanced Competent | -$20.50 | -$429.50 | $388.00 | 53% | 64 | 576 | 19m | 0% |
+| Overpriced Weak Execution | -$245.90 | -$522.50 | $219.20 | 76% | 10 | 404 | 26m | 0% |
+| Station Bottleneck | -$316.00 | -$473.50 | -$50.50 | 93% | 19 | 603 | 19m | 0% |
+| Intentionally Bad | -$118.00 | -$237.50 | $32.00 | 80% | 9 | 548 | 21m | 0% |
 
 **Winner:** Focused Value  ·  **profitable strategies (median >= +$150):** 3
 
@@ -28,40 +28,40 @@ _Mid-budget social diners in groups, multi-course, moderate patience. Rewards a
 
 | Strategy | median contrib | p10 | p90 | loss% | avg covers | avg sat | avg ticket | win% |
 |---|--:|--:|--:|--:|--:|--:|--:|--:|
-| Focused Value | $1118.20 | $615.75 | $1501.05 | 0% | 102 | 579 | 17m | 14% |
-| Premium Craft | $1963.26 | $608.50 | $3024.90 | 2% | 75 | 612 | 24m | 72% |
-| Broad Menu | $786.50 | -$189.00 | $1620.50 | 13% | 94 | 560 | 24m | 8% |
-| Overcapacity | -$289.00 | -$683.00 | $84.00 | 84% | 36 | 568 | 21m | 0% |
-| Understaffed | -$393.50 | -$556.00 | -$198.00 | 99% | 17 | 510 | 25m | 0% |
-| Balanced Competent | $688.00 | -$76.00 | $1568.50 | 11% | 86 | 598 | 22m | 5% |
-| Overpriced Weak Execution | -$646.50 | -$900.10 | -$298.20 | 98% | 9 | 397 | 28m | 0% |
-| Station Bottleneck | -$471.00 | -$679.00 | -$175.00 | 97% | 20 | 611 | 20m | 0% |
-| Intentionally Bad | -$204.50 | -$393.00 | $65.50 | 84% | 11 | 530 | 24m | 0% |
-
-**Winner:** Premium Craft  ·  **profitable strategies (median >= +$150):** 4
+| Focused Value | $1816.50 | $1461.00 | $2195.00 | 0% | 153 | 722 | 10m | 60% |
+| Premium Craft | $1412.94 | $274.46 | $2633.06 | 6% | 70 | 590 | 24m | 36% |
+| Broad Menu | $241.50 | -$449.00 | $1153.00 | 30% | 89 | 529 | 25m | 2% |
+| Overcapacity | -$578.50 | -$882.50 | -$211.00 | 98% | 33 | 573 | 21m | 0% |
+| Understaffed | -$421.00 | -$553.00 | -$199.00 | 100% | 17 | 517 | 25m | 0% |
+| Balanced Competent | $135.00 | -$419.50 | $943.00 | 36% | 75 | 597 | 22m | 1% |
+| Overpriced Weak Execution | -$719.40 | -$986.50 | -$438.50 | 98% | 9 | 402 | 28m | 0% |
+| Station Bottleneck | -$521.00 | -$740.00 | -$209.50 | 100% | 20 | 617 | 21m | 0% |
+| Intentionally Bad | -$237.50 | -$381.00 | $10.50 | 87% | 11 | 519 | 24m | 0% |
+
+**Winner:** Focused Value  ·  **profitable strategies (median >= +$150):** 3
 
 ## Destination Enthusiast Evening  (`enthusiast-evening`)
 
 _High-budget enthusiasts, lower volume, high quality expectations, patient. Rewards premium craft and skilled execution._
 
 | Strategy | median contrib | p10 | p90 | loss% | avg covers | avg sat | avg ticket | win% |
 |---|--:|--:|--:|--:|--:|--:|--:|--:|
-| Focused Value | $667.70 | $400.00 | $927.50 | 0% | 63 | 631 | 13m | 0% |
-| Premium Craft | $2471.16 | $1379.50 | $3127.42 | 0% | 75 | 595 | 24m | 83% |
-| Broad Menu | $1448.50 | $611.50 | $2061.50 | 1% | 89 | 565 | 24m | 9% |
-| Overcapacity | $399.00 | -$129.00 | $1102.50 | 15% | 51 | 556 | 22m | 0% |
-| Understaffed | -$208.00 | -$407.00 | $27.00 | 85% | 22 | 464 | 27m | 0% |
-| Balanced Competent | $1256.00 | $800.00 | $1655.00 | 0% | 79 | 634 | 18m | 7% |
-| Overpriced Weak Execution | -$585.80 | -$878.30 | -$196.30 | 97% | 10 | 335 | 29m | 0% |
-| Station Bottleneck | -$154.00 | -$499.50 | $207.00 | 70% | 26 | 575 | 22m | 0% |
-| Intentionally Bad | -$170.50 | -$345.50 | $108.50 | 79% | 12 | 469 | 25m | 0% |
+| Focused Value | $394.50 | $95.50 | $691.50 | 2% | 71 | 716 | 10m | 0% |
+| Premium Craft | $2313.06 | $845.32 | $3445.88 | 0% | 76 | 583 | 25m | 76% |
+| Broad Menu | $1183.00 | -$21.00 | $2072.00 | 10% | 91 | 549 | 25m | 6% |
+| Overcapacity | $348.00 | -$106.00 | $998.50 | 13% | 50 | 575 | 21m | 0% |
+| Understaffed | -$188.50 | -$405.50 | $60.00 | 86% | 24 | 480 | 27m | 0% |
+| Balanced Competent | $1359.00 | $866.50 | $1695.00 | 0% | 83 | 654 | 18m | 17% |
+| Overpriced Weak Execution | -$613.50 | -$903.80 | -$248.80 | 93% | 10 | 344 | 28m | 0% |
+| Station Bottleneck | -$153.00 | -$518.00 | $231.00 | 68% | 26 | 588 | 22m | 0% |
+| Intentionally Bad | -$166.00 | -$334.00 | $151.50 | 73% | 13 | 468 | 25m | 0% |
 
 **Winner:** Premium Craft  ·  **profitable strategies (median >= +$150):** 5
 
 ## Dominance check
 
 - **lunch-rush** best strategy: `Focused Value`
-- **social-dinner** best strategy: `Premium Craft`
+- **social-dinner** best strategy: `Focused Value`
 - **enthusiast-evening** best strategy: `Premium Craft`
 
 Distinct winning strategies across scenarios: **2** of 3. No single strategy wins across all markets: context changes the best plan. ✅