Skip to content

Phase 68: Consequence Enforcement

Status: Complete Block: 8 (Consequence Enforcement & Scenario Expansion) Tests: 67 new (7 test files)

Summary

Phase 68 converts the 7 highest-priority "log but don't act" patterns to actual behavioral enforcement, each gated behind an enable_* flag (default False) to prevent regressions.

What Was Built

68a: Fuel Consumption Enforcement

  • enable_fuel_consumption flag in CalibrationSchema
  • Vehicles consume fuel proportional to distance moved in _execute_movement()
  • Domain-specific rates: ground 0.0001/m, air 0.0005/m, naval 0.00005/m
  • Per-unit fuel_consumption_rate attribute override supported
  • Fuel exhaustion sets speed to 0, existing fuel gate at 0 prevents movement
  • Infantry (max_speed <= 5) exempt

68b: Ammo Depletion Gate

  • enable_ammo_gate flag in CalibrationSchema
  • _ammo_expended dict tracks rounds fired per unit_id:weapon_name
  • Weapons with magazine_capacity blocked after exhaustion
  • Weapons without magazine_capacity fire unlimited (backward compat)
  • State checkpointed in get_state()/set_state()

68c: Order Delay Enforcement

  • _pending_decisions dict tracks unit_id → execute_at_elapsed_s
  • After propagate_order(): positive delay queues decision deferral
  • Before decide(): matured decisions proceed, pending decisions skip
  • Gated by existing enable_c2_friction flag
  • State checkpointed

68d: Order Misinterpretation Enforcement

  • _misinterpreted_orders dict stores propagation results (transient, not checkpointed)
  • Four misinterpretation types enforced:
  • "position": offset movement target by misinterpretation_radius_m (default 500m)
  • "timing": re-queue with doubled delay
  • "objective": swap ATTACK/DEFEND school adjustments
  • "unit_designation": skip decide cycle entirely
  • Gated by enable_c2_friction

68e: Fire Zone Damage

  • fire_damage_per_tick field in CalibrationSchema (default 0.01)
  • Replaced log-only fire zone code with actual damage application
  • Uses existing deferred damage pattern (_apply_aggregate_casualties)
  • DUG_IN posture reduces fire damage by 50%
  • Gated by existing enable_fire_zones flag

68f: Stratagem Expiry

  • stratagem_duration_ticks field in CalibrationSchema (default 100)
  • _activation_ticks dict on StratagemEngine tracks when each stratagem activated
  • expire_stratagems(current_tick, duration) removes expired plans
  • is_active(stratagem_id) public helper
  • Battle loop calls expire_stratagems() before stratagem evaluation
  • activate_stratagem() now accepts tick parameter (backward compat via default=0)
  • State checkpointed in get_state()/set_state()

68g: Guerrilla Retreat Movement

  • retreat_distance_m field in CalibrationSchema (default 2000m)
  • Guerrilla units that disengage physically move away from nearest enemy
  • Retreat vector: opposite direction from nearest enemy, magnitude = retreat_distance_m
  • Optional ROUTING status via blend probability + PRNG
  • Gated by existing enable_unconventional_warfare flag

Files Modified

File Changes
simulation/calibration.py +6 fields: enable_fuel_consumption, enable_ammo_gate, fire_damage_per_tick, stratagem_duration_ticks, retreat_distance_m, misinterpretation_radius_m
simulation/battle.py 7 behavioral changes, +3 instance vars (_ammo_expended, _pending_decisions, _misinterpreted_orders), checkpoint support
c2/ai/stratagems.py +_activation_ticks, expire_stratagems(), is_active(), tick param on activate_stratagem(), checkpoint update
tests/unit/test_phase_60_structural.py Updated string match for renamed fire zone log
tests/validation/test_phase_67_structural.py Excluded Phase 68 flags from scenario exercise check

New Test Files

File Tests
test_phase_68a_fuel.py 8
test_phase_68b_ammo.py 11
test_phase_68c_order_delay.py 14
test_phase_68d_misinterpretation.py 11
test_phase_68e_fire_damage.py 8
test_phase_68f_stratagem_expiry.py 8
test_phase_68g_guerrilla_retreat.py 7
Total 67

Postmortem

Scope

On target. All 7 planned substeps delivered. 67 tests vs ~54 planned (+24%).

Integration

Fully wired — 1 minor finding: - is_active() on StratagemEngine defined but not called from battle.py. Available as public helper for tests and future use. Not a gap — expiry removes plans from _active_plans, which prevents bonus application naturally. - _misinterpreted_orders correctly NOT checkpointed — transient intra-tick buffer consumed in same OODA cycle.

Test Quality

Unit-level coverage strong, integration-level coverage moderate: - 68a (fuel) and 68f (stratagem expiry) have strong integration tests calling actual engine methods - 68b (ammo), 68c (delay), 68d (misinterpretation), 68g (guerrilla retreat) test logic in isolation rather than through full battle loop - Acceptable for Phase 68 — the behavioral code is structurally wired and gated behind flags that default to False

Deficits (accepted limitations)

  1. Hardcoded domain fuel rates — ground/air/naval rates baked into battle.py (0.0001/0.0005/0.00005). Per-unit fuel_consumption_rate attribute provides override path. Schema-level domain rates deferred.
  2. Hardcoded fire posture multiplier — DUG_IN takes 0.5x fire damage. Should be in CalibrationSchema if tuning needed.
  3. Hardcoded vehicle speed threshold — 5.0 m/s determines fuel eligibility. Consistent with Phase 58e gate.
  4. Phase 68 flags not in scenarios — enable_fuel_consumption and enable_ammo_gate not enabled in any scenario YAML yet. Will be enabled in a future integration phase (like Phase 67 did for Block 7 flags).

Performance

67 new tests run in 0.62s. Full suite ~21 min, consistent with prior phases. No performance regression.