Skip to content

Phase 64: C2 Friction & Command Delay

Status: Complete Tests: 60 new (6 test files) Source files modified: 4 (zero new source files)

Overview

Wires four dormant C2 engines into the simulation loop: OrderPropagationEngine, PlanningProcessEngine, ATOPlanningEngine, and StratagemEngine. All behavioral changes are gated by enable_c2_friction=False (default from Phase 63). This is Block 7, Phase 7 of 10.

Implementation

Step 0: CalibrationSchema Infrastructure (10 tests)

Added 5 new fields to CalibrationSchema: - planning_available_time_s: float = 7200.0 — planning budget (2hr default) - stratagem_concentration_bonus: float = 0.08 — +8% ATTACK decision score - stratagem_deception_bonus: float = 0.10 — +10% ATTACK decision score - order_propagation_delay_sigma: float = 0.4 — log-normal sigma for delay variation - order_misinterpretation_base: float = 0.05 — base misinterpretation probability

All values chosen for realistic defaults; zero behavioral change when enable_c2_friction=False.

64a: Order Propagation Wiring (11 tests)

None guards in propagation.py — 3 guards added for self._command: - Authority check skipped when command is None - Staff effectiveness computation skipped when command is None - Sender degradation status check skipped when command is None

battle.py DECIDE handler — when enable_c2_friction=True: - Creates FRAGO Order, calls propagate_order(), logs delay/misinterpretation - Calibration-to-config forwarding: order_propagation_delay_sigma and order_misinterpretation_base from CalibrationSchema to PropagationConfig - Propagation failure causes DECIDE skip (continue)

64b: Planning Process Wiring (12 tests)

engine.py — planning update() per strategic tick: - Auto-advance via advance_phase() - complete_planning() called when ISSUING_ORDERS phase reached

battle.py — DECIDE gate: - If planning status is not IDLE or COMPLETE, skip DECIDE - If IDLE, initiate planning with configurable planning_available_time_s - Method selection: echelon 5 (Company) with 7200s selects RAPID planning

64c: ATO Management Wiring (9 tests)

engine.py — lazy aircraft registration every strategic tick: - AIR domain units registered automatically - Sortie availability logging when enable_c2_friction active

battle.py — ATO sortie gate in _route_air_engagement(): - get_available_sorties() <= 0 causes air engagement skip - Returns (True, None) — engagement handled but no result

64d: Stratagem Activation Wiring (10 tests)

Ordering fix — moved stratagem evaluation BEFORE decide() so bonuses flow into school_adjustments. Phase 53c placement had stratagems evaluated after decide(), rendering bonuses ineffective.

Three stratagem types wired: - Concentration: compute enemy center of mass, plan + activate, +8% ATTACK bonus via stratagem_concentration_bonus - Deception: feint/main split, plan + activate, +10% ATTACK bonus via stratagem_deception_bonus - Economy of force: last 2 units when 5+ available

Step S: Structural Verification (8 tests)

Source-level string assertions for all Phase 64 wiring points.

Files Modified

File Changes
stochastic_warfare/simulation/calibration.py 5 new fields (delay sigma, misinterp base, planning time, stratagem bonuses)
stochastic_warfare/c2/orders/propagation.py 3 None guards for _command in propagate_order()
stochastic_warfare/simulation/battle.py _get_unit_position() helper, order propagation call, planning gate + initiation, stratagem plan+activate with decision score boost (reordered before decide()), ATO sortie gate in air routing
stochastic_warfare/simulation/engine.py Planning update() + auto-advance per tick, ATO lazy aircraft registration + sortie logging

Test Files

File Tests
tests/unit/test_phase_64_infra.py 10
tests/unit/test_phase_64a_order_propagation.py 11
tests/unit/test_phase_64b_planning_process.py 12
tests/unit/test_phase_64c_ato_management.py 9
tests/unit/test_phase_64d_stratagem_activation.py 10
tests/unit/test_phase_64_structural.py 8

Design Decisions

  1. Reused enable_c2_friction flag from Phase 63 as master gate for all Phase 64 changes. One flag controls all C2 friction behavior.
  2. command_engine=None guard rather than full CommandEngine instantiation — avoids heavy HierarchyTree + TaskOrgManager dependencies.
  3. Planning auto-advance without sub-module result injection — timer delay is the key friction; content fidelity deferred.
  4. Stratagem before decide() — moved stratagem evaluation from after decide() (Phase 53c placement) to before, so bonuses flow into school_adjustments.
  5. Order delay logged, not queued — no deferred delivery infrastructure; delay value computed but execution proceeds immediately.

Deviations from Plan

  • Plan specified ato_registration_done as an internal flag on CalibrationSchema — correctly omitted. Kept as instance-level tracking via engine re-registration each strategic tick instead.
  • Postmortem discovered and fixed: stratagem bonus applied after decide() (ordering bug), None position handling in enemy centroid calculation.

Deferrals

  1. CommandEngine full hierarchy wiring — authority check skipped when command_engine=None (Medium)
  2. Order delay enforcement queue — delay computed and logged but not enforced; orders execute immediately (High)
  3. Misinterpretation parameter modification — was_misinterpreted logged but order params unchanged (Medium)
  4. Planning result injection — MDMP auto-advances without COA development or wargaming results (Medium)
  5. ATO entry consumption — sorties_today never incremented after air engagement; sortie gate never triggers (Medium)
  6. Stratagem duration and expiry — active stratagems never expire (Medium)
  7. Deception effect on enemy AI — no false force disposition injected into FOW (Medium)
  8. echelon_level=5 hardcoded for all units; should look up actual echelon (Low)
  9. mission_type=0 (ATTACK) hardcoded for all orders (Low)
  10. Economy-of-force and feint unit selection by list position, not tactical criteria (Low)
  11. _prop_cfg private attribute mutation from battle.py for calibration forwarding (Low)

Postmortem

Scope: On target (60 tests vs planned ~60)

All 4 planned subsections delivered plus infrastructure and structural verification. No scope creep, no unplanned reductions. One ordering bug (stratagem before decide) discovered and fixed during postmortem.

Quality: High

  • All 60 Phase 64 tests pass
  • Calibration coverage test caught order_propagation_delay_sigma and order_misinterpretation_base as unconsumed — fixed immediately with consumer wiring
  • Postmortem caught stratagem ordering bug (bonuses applied after decide consumed school_adjustments)

Integration: Fully wired

  • All 4 engines connected to simulation loop (order propagation, planning, ATO, stratagem)
  • All 5 new CalibrationSchema fields consumed
  • Master enable_c2_friction flag gates all behavioral changes
  • Zero new source files — all changes in existing modules

Deficits: 11 deferrals (see above)

D2 (order delay queue) and D5 (ATO sortie consumption) are the most impactful deferrals. Both represent infrastructure that is structurally wired but not functionally complete — delays are computed but not enforced, sorties are tracked but not consumed.

Lessons Learned

  • Postmortem catches ordering bugs: stratagem bonus applied after decide() consumed school_adjustments — zero gameplay effect until fixed. Always verify data flow ordering, not just data flow existence.
  • None position guard pattern: getattr(e, "position", Position(0,0,0)) returns None when attribute exists but is None. Use (getattr(e, "position", None) or Position(0,0,0)) instead.
  • Calibration coverage test is invaluable: immediately caught order_propagation_delay_sigma and order_misinterpretation_base as unconsumed fields, forcing consumer wiring before commit.
  • Events as observability, not control flow: All 4 C2 events (PlanningStarted/Completed, ATOGenerated, StratagemActivated) are published but have no production subscribers. This is the project's event architecture pattern — events feed the recorder/narrative, not the control loop.