Scenarios
The model does not have one answer. It has five, and the difference between them shows how much the result depends on the assumptions — which tells you more than any single figure.
Each scenario is stored as data, not hidden in the software. Every published number names the scenario that produced it, and every setting that could change a number is stored too, so that a calculation can be repeated exactly and checked by someone who disagrees with it.
What scenarios may not do
No scenario deletes harm, and none stretches the remaining shares to fill a gap. Every scenario keeps an "Other" share, and with it included the shares always add up to 100%. When a scenario declines to pass a share on, the share moves to Other, where it stays counted and stays visible.
This matters more than it sounds. Suppose only 40% of a war’s link to the economy has evidence behind it. If the rest were thrown away and the 40% stretched back to 100%, an oil share of 30% would become 75% — the model would make its strongest claim about oil exactly where its evidence is weakest. Keeping the rest in "Other" leaves every supported share at its true size and turns the gap into a number you can see.
renormalise: false — Every scenario stores this as "renormalise: false", and both the setup scripts and the automatic tests check it — so it is a fact about the data, not a promise in words.
Economic-embedded S-MAX
The main scenario. It is the idea the project was founded to test, and the site leads with it rather than hedging about it.
Passes all harm that can be linked to the economy through it, including the part that no particular industry has evidence for. That part is shared out according to what the country where each death was counted produces — an assumption graded LOW. It measures the economic activity tied to a war, not the reasons anyone fights.
| Passes the "Other" share on to products | yes |
|---|---|
| Needs documented evidence before a share reaches a named industry | no |
| Lowest confidence a link may have and still be passed on | LOW |
| Industries that may receive a share | all |
| Links harm from poverty and pollution to products | yes |
| Allows for work injuries that go unreported | 19.5–24.1 (×22.7) |
| Deaths a war causes indirectly, as a multiple of deaths in fighting | 3–15 (×8) * |
| A war nobody scored follows the production basket of each country where it was counted — what that country produces | yes |
| Any death left unassigned follows the production basket of the country where it was counted — what that country produces | yes |
| Share for unplaced harm | OTHER |
| Stretches the other shares after holding one back | no |
Evidence-weighted S-EVID
Only shares with documented evidence reach a named industry. The rest is kept in Other and is not passed on to products — it is still counted, still shown, and never stretched over the other shares.
| Passes the "Other" share on to products | no |
|---|---|
| Needs documented evidence before a share reaches a named industry | yes |
| Lowest confidence a link may have and still be passed on | LOW |
| Industries that may receive a share | all |
| Links harm from poverty and pollution to products | yes |
| Allows for work injuries that go unreported | 19.5–24.1 (×22.7) |
| Deaths a war causes indirectly, as a multiple of deaths in fighting | 3–12 (×5) * |
| A war nobody scored follows the production basket of each country where it was counted — what that country produces | no |
| Any death left unassigned follows the production basket of the country where it was counted — what that country produces | no |
| Share for unplaced harm | OTHER |
| Stretches the other shares after holding one back | no |
Resource-only S-RES
Only energy, minerals, farming, land, water and forestry are passed on to products. Shares belonging to any other industry move to Other rather than being deleted.
| Passes the "Other" share on to products | no |
|---|---|
| Needs documented evidence before a share reaches a named industry | yes |
| Lowest confidence a link may have and still be passed on | LOW |
| Industries that may receive a share | ENERGY, MINING, AGRICULTURE, LAND, WATER, FORESTRY |
| Links harm from poverty and pollution to products | yes |
| Allows for work injuries that go unreported | 19.5–24.1 (×22.7) |
| Deaths a war causes indirectly, as a multiple of deaths in fighting | 3–12 (×5) * |
| A war nobody scored follows the production basket of each country where it was counted — what that country produces | no |
| Any death left unassigned follows the production basket of the country where it was counted — what that country produces | no |
| Share for unplaced harm | OTHER |
| Stretches the other shares after holding one back | no |
Arms and resources S-ARMS
Natural resources plus the arms industry. It tests a founding point of the project: that weapons are a product made by the economy, not something outside it.
| Passes the "Other" share on to products | no |
|---|---|
| Needs documented evidence before a share reaches a named industry | yes |
| Lowest confidence a link may have and still be passed on | LOW |
| Industries that may receive a share | ENERGY, MINING, AGRICULTURE, LAND, WATER, FORESTRY, ARMS, MILITARY_PROCUREMENT |
| Links harm from poverty and pollution to products | yes |
| Allows for work injuries that go unreported | 19.5–24.1 (×22.7) |
| Deaths a war causes indirectly, as a multiple of deaths in fighting | 3–12 (×5) * |
| A war nobody scored follows the production basket of each country where it was counted — what that country produces | no |
| Any death left unassigned follows the production basket of the country where it was counted — what that country produces | no |
| Share for unplaced harm | OTHER |
| Stretches the other shares after holding one back | no |
Conservative S-CONS
Only the best-supported links are passed on; harm from poverty and pollution is not linked to products at all; and the work injuries countries report are used as reported, with no allowance for under-reporting. It exists so that a reader who rejects the model's more debatable steps still has a number.
| Passes the "Other" share on to products | no |
|---|---|
| Needs documented evidence before a share reaches a named industry | yes |
| Lowest confidence a link may have and still be passed on | HIGH |
| Industries that may receive a share | all |
| Links harm from poverty and pollution to products | no |
| Allows for work injuries that go unreported | no |
| Deaths a war causes indirectly, as a multiple of deaths in fighting | 1–3 (×1) * |
| A war nobody scored follows the production basket of each country where it was counted — what that country produces | no |
| Any death left unassigned follows the production basket of the country where it was counted — what that country produces | no |
| Share for unplaced harm | OTHER |
| Stretches the other shares after holding one back | no |
Settings not yet backed by evidence *
Some settings do not yet have evidence behind them, and the scenario marks them. The biggest is how many deaths a war causes indirectly: published estimates range from three to fifteen times the deaths in fighting, and every point in that range has serious support. It is kept as a setting with a wide range rather than a fixed number, and it changes the headline more than anything else in the model.
Questions
Why is Economic-embedded the main scenario?
Because the project exists to measure an idea, not to hedge about it: that organised violence is tied up with economic activity. The model does not claim to know anyone’s motives — it measures the economic activity linked to a war. A war that began in religious hatred still buys rifles, burns fuel, takes land and sells what it digs up.
What if I reject that idea?
Then read the Conservative scenario — that is why it exists. It passes on only the best-supported links, does not link harm from poverty or pollution to products, and uses reported work injuries without adjustment. It gives a much smaller number, and the difference between the two is the most honest thing the model can show you.
Can I compare figures from two scenarios?
Compare them with each other, yes — that is the point. But do not add them into one total, and do not compare a figure with one made under a different version of the method: the version is part of what the number means.