1in8b Engine · for research and teaching
Methods and validation
The 1in8b Engine is the life simulation behind 1 in 8 Billion. This page says what it rests on, how it is fitted to the data, how close it comes, what it leaves out, and how to cite it. The lives it makes follow real statistics, but they are simulated: not real people, records or forecasts. Say so wherever you use them.
What the engine does
It is a microsimulation that lives one person at a time, a year at a time, from birth to death. A birth is drawn by the real share of births in 172 countries and territories (UN World Population Prospects 2024), with a sex, a town or village, a family and its place among the country's income fifths. Each year, what can happen (illness and death, school and its end, work, pay and its changes, a partner, children, a move abroad, fleeing, disasters, the history of that year in that place) happens at the rates the data gives for that country, that age and that person's situation. Choices are made by the person (on the site, by the reader; in the API, by the caller or by a policy of what most people choose); a choice moves the odds of what follows, it never decides it. Past years (1950 to 2025) live each year in the country as it was then; the coming world follows the UN's projections.
How it is fitted
- Mortality. Death rates by age, sex and cause start from the WHO Global Health Estimates and the UN's life tables, and each country's levels are fitted so that simulated life expectancy at birth meets the UN's for that country (the table below). A life's own circumstances (its income fifth, smoking, where it lives) move its risk up or down around that level.
- Schooling. The chance of finishing each stage comes from UNESCO's completion rates (UIS) and UNDP's education figures by country, and from the family's income fifth and town or village where the data says how much they matter.
- Work and income. Jobs are offered by the country's sectors (World Bank WDI, ILO) and the person's schooling; incomes are fitted so that a working life's place in its country's income distribution comes out as the data's.
- Families and moves. Marriage and births follow each country's ages and rates (UN WPP); moves abroad follow the UN's migrant stock, and fleeing follows the wars and disasters of each place and year.
These checks are run again whenever a change touches what they measure, and a change that moves the outcomes away from the data is fixed in the mechanism, not in the check.
How close it comes
Measured on engine 0.9.52, 2026-11-06, with choices made as most people would.
Life expectancy at birth (years), 400 lives each
| country | simulated | UN WPP 2024 | difference |
|---|---|---|---|
| Nigeria | 54.9 | 54.5 | +0.4 |
| Ethiopia | 68.5 | 67.3 | +1.2 |
| India | 72.2 | 72.0 | +0.2 |
| Egypt | 71.2 | 71.6 | -0.4 |
| Brazil | 75.3 | 75.8 | -0.5 |
| Mexico | 73.9 | 75.1 | -1.2 |
| China | 77.0 | 77.9 | -0.9 |
| United States | 78.5 | 79.3 | -0.8 |
| Germany | 82.5 | 81.4 | +1.1 |
| South Korea | 83.6 | 84.3 | -0.7 |
| Japan | 85.1 | 84.7 | +0.4 |
| Lesotho | 56.3 | 57.4 | -1.1 |
Schooling finished, of 2,000 lives drawn by today's births
| stage | simulated | data (births-weighted) |
|---|---|---|
| primary or more | 82.9% | 82.8% |
| lower secondary or more | 70.9% | 72.0% |
| upper secondary or more | 51.0% | 51.3% |
| university or more | 24.5% | 25.0% |
Where the working lives worked, by sector
| sector | simulated | data (births-weighted) |
|---|---|---|
| agriculture | 31.5% | 35.1% |
| industry | 21.9% | 19.9% |
| services | 46.6% | 45.0% |
Of the same 2,000 lives, 4.2% ever lived abroad and 2.1% ever had to flee. These numbers move a little from run to run; a sample of a few hundred lives differs from the data by chance alone.
What it leaves out, and where to be careful
- It is not a forecast and not a measurement. A finding about simulated lives is a finding about the model; compare it with real data before saying anything about the world.
- Rates are mostly national averages, split by sex, age, income fifth and town or village where the data allows. Differences within a country (its regions, its groups) are only as fine as the data the engine has.
- The links between events (how a loss changes the years after it, how a choice changes the odds) are kept simple and drawn from published studies where they exist; many are estimates.
- Some figures are told in a life's record as facts with their source; those marked as estimates are estimates.
- The engine changes almost every day as readers report what does not match the world. For work that must be run again, use a research edition.
Research editions
Each May and November the engine as it is then is frozen and kept for a year. Add engine= and its name to a call of /api/v1/life or /api/v1/simulate, and the same calls with the same seed give the same answers to anyone, as long as the edition is kept. The answer's meta says the edition and the line to cite. The record comes in English or Korean.
engine=2026-11: engine 0.9.52, kept until the end of the month before 2027-11
curl -H 'Authorization: Bearer k8b_…' 'https://1in8b.life/api/v1/simulate?country=NGA,KOR&n=200&seed=7&engine=2026-11'
Research at a university or an institute, teaching and other uses that make no money can have a larger plan free for a year: write to contact@1in8b.life from the address the key goes to, saying who you are and what the work is.
How to cite
1in8b Engine, research edition 2026-11 (engine version as given in meta.engine). 1 in 8 Billion. https://1in8b.life/en/methods/ (accessed on the date you used it).
@software{one_in_8_billion_engine,
title = {1in8b Engine: the life simulation behind 1 in 8 Billion},
author = {{Jin's Earth}},
year = {2026},
version = {research edition 2026-11},
url = {https://1in8b.life/en/methods/},
note = {Simulated lives following real statistics; not real people, records or forecasts}
}
For teaching
In a class, students can live a life each in the classroom edition, with a guide, a lab (five children born the same day in five countries, a child born in the same country in five decades, hundreds of today's babies beside the real figures, a guess from the start of a life against how 300 such lives went, and the class's own lives gathered from short codes) and a worksheet: the guide for teachers. For a seminar, the API's free key is enough for a class (about 600 lives a month), and its answers can be compared with what students predicted.