§ Method — v1

How the index is computed.

Every axis score is a weighted mean of normalized public evidence, computed against the full weight of the axis: missing inputs contribute zero, not undefined. Confidence, shown alongside every score, reports what fraction of the axis is actually backed by evidence. The overall is the straight mean of seven axis scores. Scale ceilings are aspirational — 100 indicates elite-human, AGI-level performance, not today's best-in-class. A real, well-evidenced leader should sit in the 40–70 band; anything above 85 deserves scrutiny. Weights and scales are inspectable in src/scoring/ and expected to iterate.