settled-reach/tests/economics/test_report_pr122.md

Test Report: PR #122 — Sprint 33 Economics Simulation

• Date: 2026-04-07
• Build: sprint-33/server → commit b19bfb32 (Layer 3)
• PR: #122 (main ← sprint-33/server)
• Tickets: #813, #805, #800, #806, #807, #808, #809
• Spec ref: D-179 (stability criteria), D-180 (event port), D-181 (signals)
• Tests run: D-179 stability suite + manual verification
• Passed: D-179 Tests 1, 2, 3 (Test 4 correctly skipped)
• Failed: 0
• Gaps: 1 (D-181 signal coverage)

---

D-179 Stability Test Results

Command: make econ-sim-stability

Loading economy data from server/data/systems.db...
  36 commodities, 21 production chains, 31 active nodes, 37 corp presences, 668 gate links
Seeding per-corporation productivity (run seed: 0)...
  37 corp×site productivity records seeded
  48 corporation behavioral archetypes loaded (inferred where not set in DB)
  301 nodes with gate connections
Seeding per-node shadow economy intensity (D-174)...
  301 nodes seeded, mean intensity 0.50

Test 1 (cold-start convergence ±5% at tick 100):     PASS  max_dev=1.05%  worst: GJ 144/medical_goods
Test 2 (long-run stability ±2% over ticks 900–999):  PASS  max_dev=0.00%  worst: GJ 144/medical_goods
Test 3 (shock response — cascade + recovery ≤200 ticks): PASS  no explosions (>20× base), no negatives across 1,116,000 records
Test 4 (cross-zone balance re-stabilizes ≤50 ticks): PASS  SKIP — no MARK_PRIMARY systems in DB

All stability checks passed.

Test 4 skip is correct. The implementation checks for MARK_PRIMARY zone data and skips gracefully when none exists (line 275-277, main.rs). Re-run after copy team delivers #820 (Compact zone assignments).

---

Build Verification

Check	Result
make econ-sim	PASS — compiled in 0.94s (release)
make econ-sim-run	PASS — 111,601 rows (100 ticks × 31 nodes × 36 commodities + header)
CSV header	PASS — node_id,commodity_id,supply,demand,price,tick,shadow_intensity,tractus_mark_rate
Negative prices	PASS — none found across 1,116,000 records

---

Model Parameter Verification

Parameter	Spec (D-178)	Actual	Result
α (price adjustment rate)	0.03	0.03 (model.rs:25)	✅
β (damping — implicit in tâtonnement)	0.4	0.4 (trade.rs)	✅
Transport cost per gate hop	5–12%	8% flat (trade.rs)	✅ (within range)
Fusion fuel demand reduction (on-grid)	~0.3×	0.3 (model.rs:39)	✅
Initial stockpile buffer	—	4× baseline demand (model.rs:32)	✅

---

Architecture Cross-Checks

corp_presence location_type: The import pipeline resolves each corp's HQ to a specific body or station and stores location_type = 'body' or 'station' per schema (import_economics.py:453-491). The sim binary queries accordingly. Consistent with schema intent.

gate_energy_connected join: Model reads gate energy via JOIN star_systems (not directly on bodies/stations). Confirmed the GATE_ENERGY_DEMAND_REDUCTION constant (0.3) is applied to fusion_fuel utility demand for on-grid nodes.

Active node count = 31: Expected. Active nodes are limited to systems with corp presence or population > 0 in the DB. The ~760 active node target (D-178) assumes a fully-authored atlas. Stability tests passing at 31 nodes is encouraging; re-run at scale when atlas authoring progresses.

---

Gaps (Non-Blocking for D-179, Required for Phase 2 Complete)

Gap 1 — D-181: Only signals 1 and 7 (partial) are produced [MEDIUM]

D-181 requires all 7 signals per active node. The TickRecord struct contains:

• Signal 1 (price_current) → price ✅
• Signal 7 proxy (shadow_intensity) → present but official_coverage_ratio (1 - shadow_intensity) is not computed ⚠️

Missing from TickRecord and CSV output:

• Signal 2: price_trend — direction + rate of change over last N ticks
• Signal 3: trade_flow_volume — freight volume through node (computed by trade.rs but not emitted)
• Signal 4: corporate_presence — which corps operate here (static, in DB, not per-tick)
• Signal 5: stockpile_weeks — stockpile IS tracked in CommodityState but not in TickRecord
• Signal 6: production_vs_baseline — not computed or tracked

D-181: "Phase 2 sim must produce all 7 signals. Phase 3 determines how the player accesses them."

Signals 4 and 7 are reasonable to defer (static data from DB + derivable from shadow_intensity). Signals 2, 3, 5, 6 require additions to TickRecord and output.rs. Signals 5 (stockpile_weeks) is the easiest — stockpile is already computed in the model; it just needs to be added to the output struct.

Recommendation: Open a follow-up task for signal completeness. Does not block D-179 tests or PR merge if the team accepts iterative delivery (D-183 allows this). Block merge only if Phase 2 is declared complete.

Gap 2 — Test 3: Warm-start proxy, not deliberate injection [LOW]

D-179 Test 3 spec: "After a single supply shock, cascade propagates realistically; recovery within 200 ticks; no price explosions or negative prices."

The implementation uses the warm-start disturbance (4× buffer initialization) as the proxy shock and verifies no explosions across 1,000 ticks. This tests the stability envelope but does not test explicit cascade propagation or recovery time measurement. The code comments acknowledge this: "Full shock-response testing will be added when D-180 event port is implemented."

Verdict: Acceptable for this sprint given D-180 port isn't implemented. Test 3 as implemented validates the core stability guarantee. The stricter cascade test follows once the event port lands. Low priority for PR block.

---

Summary

D-179 passes cleanly. The simulation is stable, builds clean, produces correct output.

Recommend PR merge with one follow-up task:

1. Add signals 2, 3, 5, 6 to TickRecord and CSV output (D-181 completeness)

Must re-run make econ-sim-stability after:

• Copy team delivers #820 (Compact MARK_PRIMARY assignments) — enables Test 4
• Atlas authoring reaches higher node counts — validates stability at scale