How the simulation works
The complete method, so you can judge how much weight to put on any single run.
1. The mortality curve
Everything starts with one number per country: life expectancy at birth, and how many babies die before their first birthday. Both come from the World Bank and both carry the year they were observed.
The first year of life uses the country's actual infant mortality rate. From age one onward, the model uses a hazard curve fitted to the country's own life expectancy, so that a draw from the curve produces a lifespan that matches the published national figure. The model was checked against all 217 countries: the average absolute deviation between the model's expected lifespan and the published life expectancy is under one year, and no country deviates by three years or more.
This is the part that matters most, and it is worth being precise about what it means. The model reproduces the distribution of lifespans in a country, not the circumstances of any one life. When you see a character die at forty-four, that is a draw from the real risk curve — not a statement that forty-four is typical, and not a statement about why.
2. Your starting position
Before the first year, you set where you start: the household's income tier, its structure, whether you live in a village, a town, a city or the capital, how many siblings you have, and how much schooling your parents finished.
These are not difficulty settings. They map to measurable quantities. The income tier scales the household's starting wealth against the country's GDP per capita and its poverty rate, so a “poor” household in a rich country is poorer in absolute terms than a poor household in a poor country. A city raises income and raises the cost of living. A parent's years of schooling affect how much the household pushes for the child's own schooling, and how likely the child is to be sent out to work before twelve.
3. Why decisions are not a stat menu
A choice in LifePath does not add points to a score. It changes what is possible later.
When a decision is made, its consequences are planted with a delay. Leave school at fifteen and the model removes the schooling path, sets a flag that closes off the jobs that require qualifications, and schedules the specific consequence that lands two or three years later. Those consequences can be nested: one consequence can plant another, which plants another, so a decision made at sixteen can still be shaping the situation at forty.
The catalogue contains 83 events and 46 decision groups, covering infancy through retirement, plus 73 occupations with promotion paths. A job you can enter today opens jobs you cannot enter tomorrow.
4. What the model does not do
- It does not predict your life. It runs a model built on national averages, and no national average can describe an individual.
- It does not rank countries, groups or lives. A shorter life expectancy is described as a difference in mortality conditions and the structures behind them, never as a judgement about the people who live there.
- It does not apply differences between demographic groups without a published statistic behind them. Where no such statistic exists, the group carries no modifier at all.
- It does not send anything anywhere. The entire simulation runs in your browser.
- It does not give advice about health, money, family or migration.
5. Reproducibility
Every run is driven by a seed you can see and set. The same seed with the same starting choices always produces the same life, down to the year of death. If two runs with different seeds produce very different lives, that is the range of outcomes the model contains, not a bug — it is also a fair way to see how much of a life is determined before it starts.
6. The indicators used
16 indicators from the World Bank Open Data, each with its World Bank code:
| Indicator | Unit | World Bank code |
|---|---|---|
| Life expectancy at birth | years | SP.DYN.LE00.IN |
| Infant mortality rate | per 1,000 live births | SP.DYN.IMRT.IN |
| Under-5 mortality rate | per 1,000 live births | SH.DYN.MORT |
| Total fertility rate | births per woman | SP.DYN.TFRT.IN |
| GDP per capita, PPP | current international $ | NY.GDP.PCAP.PP.CD |
| Adult literacy rate | % of adults 15+ | SE.ADT.LITR.ZS |
| Government expenditure on education | % of GDP | SE.XPD.TOTL.GD.ZS |
| Urban population | % of total population | SP.URB.TOTL.IN.ZS |
| Poverty headcount at $2.15 a day | % below $2.15/day | SI.POV.DDAY |
| Youth unemployment rate | % of labor force 15-24 | SL.UEM.TOTL.ZS |
| Total population | people | SP.POP.TOTL |
| Current health expenditure | % of GDP | SH.XPD.CHEX.GD.ZS |
| Crude birth rate | births per 1,000 people | SP.DYN.CBRT.IN |
| Secondary school enrollment | % gross enrollment | SE.SEC.ENRR |
| Female labor force participation | % of female population 15+ | SL.TLF.CACT.FE.ZS |
| Maternal mortality ratio | per 100,000 live births | SH.STA.MMRT |
7. Known limits
- National averages hide enormous variation inside every country. The variance within a country is much larger than the difference between most countries.
- The model has no wars, no pandemics and no economic shocks beyond the events in its catalogue. A real life contains shocks no model anticipates.
- Conflict risk is included for countries with published conflict data, but it is applied as an annual risk, not as a timeline.
- Health, happiness and agency are modelled as simple 0–100 scales that drift with age and change with work. They are not clinical or psychological measures and should not be read as such.