Roadmap / 0.1.3.4
Where the US puts its troops, and provinces pacified one at a time
v0.1.3.4, commit aa377e2. File docs/RESULTS_v0.1.3.4_allocation_pacification.md.
Written at this version and not revised since. Later versions may change these numbers; see the roadmap.
Switch: spatial.enabled.
Options: allocation (need / population / resources), pacification (national / provincial), pacification_calibration (median / window).
Defaults: allocation by need (unchanged), pacification provincial with the median calibration (new).
Runs: 3,000 paired runs per variant, every other module on. 29 tests pass; all harness checks pass, including the check that the uniform split still equals the single-area model.
Scope: provinces as aggregates. No sites, routes or tactics.
Headline
- The provincial split now moves outcomes, and the direction depends on one number nobody has: how fast a single province goes quiet. With the slower of two defensible calibrations, pacifying province by province helps Canada by 1 to 3 points. With the faster one it helps the US by 5 to 10.
- Concentrating troops cuts US deaths by roughly a quarter under either calibration. Whether it also wins depends on the same number.
- Putting resource provinces first cuts the US gasoline shock by a third (30 cents a gallon to 20). Putting population centres first makes it half as bad again (45 cents), because the oil is left unheld.
What was added
Three allocation rules
- Need (the rule until now): troops are spread in proportion to what each province needs, so a shortfall is shared equally.
- Population first: provinces are filled to their full requirement in order of population (Ontario, Quebec, British Columbia, Alberta, …) until the troops run out. The rest get nothing.
- Resources first: the same, in order of resource value to the US (Alberta, Saskatchewan, Newfoundland and Labrador, British Columbia, Manitoba, Ontario, …).
Holding only a chosen set of provinces as a war aim ("take the West") is not in this version. It needs its own scenario: a different opening campaign, a Canada that goes on existing, and different politics on both sides. It is the next version.
Province-by-province pacification
- Until now the insurgency ended everywhere at once, as one national event.
- Now each province can go quiet on its own. The monthly chance uses the same formula as the national one (higher with more control, lower with more resistance and more allied support), times a scale.
- A pacified province keeps 5 to 30% of its resistance and needs only a holding garrison. The troops it frees go to the others under the allocation rule.
- If the garrison is neglected and resistance climbs back past half its old level, the province is contested again.
- The occupation is consolidated when 90% of the occupied population lives in pacified provinces.
- Provinces share part of each month's luck (30 to 70% is local).
The calibration problem
The national rate comes from RAND's finding that insurgencies last a median of about ten years. Nothing comparable exists for a province. Two ways to set the scale, both defensible:
| Calibration | Rule | Scale | A province held at full strength goes quiet in a median of |
|---|---|---|---|
| Median (default) | The time for the whole country to be pacified keeps the ten-year median | about 2.5 | about 5 years |
| Window | The chance of the whole country being pacified within two years stays what the national model gave | about 15 | about 9 months |
I made the median calibration the default because it rests on the same published figure as the rest of the occupation model. The window calibration is the fast bound.
Results: full annexation (change vs v0.1.3.3, 95% interval)
Baseline (need, national pacification): US quits 70.5%, 1,649 US deaths, gasoline up 30 cents.
| Allocation | Pacification | Chance the US quits | US deaths | Occupation consolidated | Gasoline rise (cents a gallon) |
|---|---|---|---|---|---|
| Need | National (v0.1.3.3) | 70.5% | 1,649 | 1.5% of runs | 30 |
| Population first | National | +0.7 (0.2 to 1.2) | 1,203 (−27%) | 2.3% | 45 |
| Resources first | National | +0.3 (−0.1 to 0.6) | 1,289 (−22%) | 2.0% | 20 |
| Need | Provincial, median (new default) | +1.1 (0.7 to 1.5) | 1,632 | 0.3% | 30 |
| Population first | Provincial, median | +2.6 (2.0 to 3.3) | 1,189 | 0.7% | 45 |
| Resources first | Provincial, median | +1.8 (1.3 to 2.4) | 1,274 | 0.7% | 20 |
| Need | Provincial, window | −4.9 (−5.9 to −4.0) | 1,396 | 10.0% | 30 |
| Population first | Provincial, window | −9.4 (−10.7 to −8.2) | 963 | 18.4% | 45 |
| Resources first | Provincial, window | −9.6 (−10.8 to −8.3) | 1,024 | 17.8% | 19 |
Alberta detachment: exactly unchanged (one province, so the provincial event is the national one). Economic coercion: unchanged (no occupation).
New baselines with every module on: full annexation 71.6%, Alberta 61.7%, economic coercion 34.4%.
What it means
1. Slow pacification helps Canada; fast pacification helps the US. Under the median calibration the whole country has to go quiet province by province, and that almost never happens inside a two-year war: consolidation falls from 1.5% of runs to 0.3%. Under the window calibration each province has about a one-in-four chance of going quiet during the war, each one frees troops for the next, and consolidation rises to 10 to 18% of runs. The cascade is the mechanism, and its speed is the unknown.
2. Concentration saves US lives either way. Troops massed at full strength take fewer casualties than troops spread thin, and provinces with no US troops produce no US casualties. Population first cuts US deaths by 27% and resources first by 22%, before any pacification effect.
3. But concentration only wins if provinces go quiet quickly. With slow pacification, concentrating is worse for the US than spreading out (72 to 73% against 71.6%): the provinces left alone harden, and nothing is finished in time. With fast pacification, concentrating is much better (61% against 65.5%): take a province, quiet it, move on.
4. Which provinces come first decides the economic bill.
| Rule | Peak US oil lost (million barrels a day) | Gasoline rise | Peak potash shortfall |
|---|---|---|---|
| Need | 2.28 | 30 cents | 38% |
| Population first | 2.97 | 45 cents | 55% |
| Resources first | 1.65 | 20 cents | 26% |
Resources first holds Alberta, Saskatchewan, British Columbia, Manitoba and Newfoundland and Labrador at about 91% control and leaves Quebec at 10% and the Maritimes and the North near zero. Population first holds Ontario at 92% and Quebec at 71% and leaves Alberta at 15% and Saskatchewan at 6%.
5. Who is pacified first (share of runs in which the province is ever pacified):
| Province | Need, median | Resources first, median | Population first, median | Need, window | Resources first, window | Population first, window |
|---|---|---|---|---|---|---|
| Alberta | 6% | 16% | 4% | 28% | 56% | 28% |
| Saskatchewan | 5% | 17% | 2% | 27% | 53% | 19% |
| Manitoba | 6% | 17% | 3% | 27% | 54% | 23% |
| British Columbia | 6% | 16% | 6% | 27% | 55% | 37% |
| Newfoundland and Labrador | 5% | 16% | 1% | 26% | 52% | 12% |
| Ontario | 5% | 8% | 17% | 26% | 41% | 57% |
| Quebec | 5% | 2% | 12% | 28% | 23% | 48% |
| New Brunswick, Nova Scotia, PEI | 5% | 1% | 1 to 2% | 26% | 10 to 12% | 13 to 16% |
| The North | 5% | 1% | 1% | 25% | 8% | 11% |
Limits
- The pacification speed is unanchored, and it decides the sign of the result. This is now the most important unknown in the provincial model. Evidence that would help: how long regions inside occupied or insurgent countries took to go quiet (for example provinces in Iraq and Afghanistan, or regions under the German occupation of western Europe).
- The residual resistance in a pacified province and the relapse rule are my estimates.
- A single-province theatre uses the national rate unscaled. So Alberta alone (v0.1.2 scenario) and Alberta inside a full annexation pacify at different rates. That is a consequence of calibrating on whole conflicts; it is not resolved.
- The priority rules are all-or-nothing: a province gets its full requirement or what is left. Real allocation would be smoother.
- The US does not choose its rule in response to events, and Canada does not respond to the US's choice (for example by concentrating resistance where the US is absent).
- Unoccupied provinces under a priority rule still count as part of the occupation for Canada's government and the economy. A proper partial annexation is the next version.
Next
v0.1.3.5: partial annexation as a war aim. The US takes and holds a chosen set of provinces (Alberta and Saskatchewan are 15% of the people and 90% of the resource value) and leaves the rest of Canada standing. This needs its own scenario, with a smaller opening campaign and a Canadian government that keeps functioning.