Roadmap / 0.1.2.9
All switches, measured together
v0.1.2.9, commit bb9175e. File docs/RESULTS_v0.1.2.9_all_switches.md.
Written at this version and not revised since. Later versions may change these numbers; see the roadmap.
Every module built so far, run together: all off, all on, each one removed, each one alone, and each side pulling every lever it has.
Runs: seed 20261105, 2,000 per variant for the remove-one and add-one sets; the four headline variants were rerun at 8,000 (seed 777) to cut noise. January start. Backtests pass; 18 tests pass.
Noise floor: about ±2.5 points on "US quits" at 2,000 runs. Two switches that cannot affect economic coercion (seasonal insurgency, wildlife) still show +2.1 and +2.6 there. Treat anything under 3 points as no effect.
Switch inventory
| Switch | Key | Version | Options |
|---|---|---|---|
| Weather | weather.enabled, weather.model (simple/complex) |
0.1.2, 0.1.2.6 | start_month, objective_weights, ENSO and cold-proficiency overrides |
| Seasonal insurgency | seasonal_insurgency.enabled |
0.1.2.2 | – |
| Canadian mobilization | canada_mobilization.mode (none/dmp) |
0.1.1 | territorial_core |
| US mobilization | us_mobilization.mode (standard/full) |
0.1.1 | draft |
| Potash | potash.enabled |
0.1.2.3 | embargo |
| Oil | oil.enabled |
0.1.2.4 | embargo, us_cuts_east |
| Electricity | electricity.enabled |
0.1.2.5 | canada_cut, us_cut |
| Wildlife | wildlife.enabled |
0.1.2.6.1 | north_troop_share |
| Logistics | logistics.enabled |
0.1.2.7 | us_sustainment, caf_sustainment, civilian |
| NATO and allies | nato.enabled |
0.1.2.8 | response, tripwire_troops, us_threatens_europe |
| Small arms | arms.enabled |
0.1.2.8.1 | black_market, state_suppliers, us_blockade |
"All on" means every module on with no escalatory options. "Max Canadian levers" adds the oil and potash embargoes, a Canadian power cut, 5,000 allied tripwire troops and open state arms shipments. "Max US levers" adds the cut to eastern oil, a US power cut, an arms blockade, full US mobilization and paralysed allies.
Headline (8,000 runs each)
| Scenario | Variant | US quits | US wins | Unresolved |
|---|---|---|---|---|
| Full annexation | all off | 57.8% | 41.8% | 0.4% |
| all on | 67.2% | 32.0% | 0.8% | |
| all on, max Canadian levers | 69.6% | 29.9% | 0.5% | |
| all on, max US levers | 61.7% | 37.8% | 0.6% | |
| Alberta detachment | all off | 33.3% | 63.1% | 3.6% |
| all on | 55.7% | 42.3% | 2.0% | |
| all on, max Canadian levers | 65.1% | 33.6% | 1.2% | |
| all on, max US levers | 48.2% | 49.3% | 2.6% | |
| Economic coercion | all off | 13.4% | 20.1% | 66.4% |
| all on | 42.3% | 11.2% | 46.5% | |
| all on, max Canadian levers | 50.3% | 10.9% | 38.8% | |
| all on, max US levers | 31.6% | 15.1% | 53.3% |
Costs in full annexation, all off → all on (medians): campaign 27 → 39 days; US deaths 828 → 1,230; Canadian deaths 8,087 → 10,276; Canada's peak GDP loss 20.6% → 24.6%; US peak GDP loss 1.8% → 2.1%.
What each switch is worth (change in "US quits", points)
Remove-one = all on minus the run without that switch (8,000-run baseline). Add-one = that switch alone minus all off.
| Switch | Full annexation: remove / add | Alberta detachment: remove / add | Economic coercion: remove / add |
|---|---|---|---|
| NATO and allies | +3.1 / +4.2 | +6.9 / +8.5 | +8.1 / +6.5 |
| Canadian mobilization | +4.2 / +5.6 | +6.4 / +7.5 | +0.9 / +1.0 |
| Potash | −0.4 / +0.1 | +3.6 / +4.5 | +11.9 / +9.5 |
| Oil | +0.8 / +0.9 | +2.1 / +2.6 | +9.2 / +6.6 |
| Weather (complex) | +0.2 / −1.0 | +0.7 / +3.9 | −0.3 / 0.0 |
| Seasonal insurgency | −0.6 / −0.9 | +0.1 / +0.4 | (noise) |
| Electricity | −0.7 / −1.7 | −1.1 / −0.4 | +1.9 / +0.8 |
| Wildlife | +0.5 / −2.5 | +1.7 / −0.5 | (noise) |
| Logistics | −3.6 / −1.8 | −0.6 / −2.3 | +2.2 / −1.0 |
| Small arms | +0.5 / −0.4 | −1.0 / +1.0 | (noise) |
Findings
- Together the switches move the war a lot. With everything on, the chance the US quits rises by 9 points in full annexation, 22 in the Alberta detachment and 29 in economic coercion. The US win rate in coercion is nearly halved (20% → 11%).
- Four switches carry almost all of it: allies, Canadian mobilization, potash and oil. - Allies and mobilization matter in a shooting war. - Potash, oil and allies matter when there is no shooting.
- The other six change the price, not the result. Weather, seasons, electricity, wildlife, logistics and small arms shift campaign length, deaths and GDP loss, but their effect on who wins is inside the noise. Logistics is the one that leans slightly toward the US in full annexation, because Canada's munitions run out.
- Each side's choices swing the result by 8–19 points. Canada pulling every lever against a US pulling none reaches 70% / 65% / 50%. A US pulling every lever against paralysed allies pulls it back to 62% / 48% / 32%.
- The smaller the US objective, the more these switches matter. The modules add least to full annexation, where the US already quits most of the time, and most to coercion.
- Effects are roughly additive. The big four add up to about the all-on total in each scenario, so there is no hidden interaction doing the work.
- The model is now at the limit of this way of testing it. Ten switches already need 25 variants per scenario to read cleanly, and effects under 3 points drown in noise at 2,000 runs. A full on/off grid would be 1,024 combinations. This is the test harness (0.1.2.0): common random numbers, larger runs for small effects, and screening designs instead of grids.
Still true from earlier versions
- Canadian capitulation responds only weakly to economic pain, which blunts every Canadian vulnerability (Line 5, power imports, food and drugs).
- The decisive parameters are estimates: the allied response weights, the support effects of blackouts and other narrative events, and capitulation.