Human Capability Index · Scoring rule sensitivity · Build v0.1

Who wins depends
on a rule nobody
argues about

Synthetic population
Real method · Simulated data
No human subjects

A broad-capability sport has to compress eight incommensurable domains into one number. That compression is a choice, not a measurement. Change the normalization or the aggregation and the leaderboard reorders — same athletes, same performances. Move the controls and watch the top ten.

Standings

Δ vs. open · arithmetic · equal
Bars: Phy · Cog · Adp · Dec · Soc · Emo · Res · Cre Red bar = below 50th pct

Field collapse, round by round

What actually eliminated people

The binding domain — each competitor's weakest — is what ends their run. If one bar towers, the sport has one real event.

Who this rule selects — top 50

Profile of the current leader

Domain share of composite variance

If one domain explains most of the spread in the composite, the index is mostly measuring that domain wearing eight labels.

What this is, and what it is not

Real method, synthetic data. No person was measured. A population of 50,000 synthetic competitors is generated from a three-factor model (physical, cognitive, socio-emotional) with domain-specific loadings, then shifted by age and sex effect sizes set as stated parameters below. The scoring, normalization and aggregation code is the real thing; the inputs are invented.

The effect sizes are assumptions, not findings. They are set to plausible magnitudes so the mechanics are visible. The physical sex gap in particular is a single composite standin for measures that in reality range from near-parity to well over 2 SD depending on whether you test relative endurance or upper-body power. Do not read any number here as an empirical claim.

What is real is the structural result. Under any factor model where one domain has a large group-level mean difference and the aggregation is additive, that domain dominates the composite and the eligibility rule quietly determines the podium. That conclusion does not depend on the specific constants.

Age curves are directional. Physical and fluid-reasoning peaks are placed in the twenties; decision quality, social and emotional capacity peak decades later. This is the standard shape in the literature, rendered here as smooth quadratics rather than fitted curves.

DomainLoading: physLoading: cogLoading: soc-emoAge peakSex effect (SD)

Sex effect is signed male-minus-female in standard deviations, applied as ±half to each competitor. Loadings are on orthogonal standard-normal factors; unique variance fills the remainder.