Met1,000 to 1,000,000 lives, 20 seeds each
The test at a glance
- What it tests
- Correctness of an effect that needs lives to meet. A property of the engine, not a claim about any country.
- Data
- Simulated worlds of 1,000 to 1,000,000 lives, 20 seeds each, 600 months; UK 2019 death and birth tables, placeholder partnership rates.
- Held out
- Not a held-out test: the predictions were registered before any run.
- Compared against
- The direct answer: each life partners on its own chances, with no one to find.
- Bar
- At 1,000,000 lives men's gap below zero and women's above zero in all 20 seeds; at 1,000 lives men's sign unstable (fewer than 18 of 20).
- Result
- Both met: men -3.4 points, women +1.7 at a million lives; stable for men from 10,000 lives.
Living Worlds shows effects that exist only when lives meet. After ten years in which more boys are born than girls (about 1.18 to 1), the men of those years are 3.4 points less likely ever to have had a partner thirty years later, and the women 1.7 points more likely, in all 20 runs at a million lives. Computing each life on its own chances gives zero. Registered before any run; 20 seeds at every size from a thousand to a million lives. From ten thousand lives the effect holds for men in every run; the size follows the engine's placeholder partnership rates, the direction comes from people having to find each other.
Protocol, result and scorer