The achievement-well-being divergence of neurotraits and the moderating role of harness: six self-management pathways and propensity-score evidence from 5,432 historical creators

We test a variance-amplifier-plus-regulator model on 5,432 cross-civilization historical creators (achievement n=5,347; well-being n=3,909) using a six-dimensional Bayesian P_ND, independently coded achievement and well-being, a six-pathway harness indicator (CPSA), and g/controls. Achievement peaks in the high-ND bin (3.735; pure quadratic non-significant) while well-being follows a robust U-shape trough at 2.412 (linear -3.18, quadratic +2.25; t=-10.97/5.58). Harness moderates with a larger marginal benefit for ND (well-being CPSA beta=1.39, t=15.56; P_ND x CPSA beta=2.55, t=5.06). Within the ND group (P_ND>=.5, n=106), propensity-score matching gives +31 percentage points in top-achievement rate (72.4% vs 41.4%) and a well-being IPW effect of +0.607 (vs +0.135 database-wide; max SMD 0.649->0.265). Externalized, routinized pathways (external container 3.29, survival prosthesis 3.41, ritualized routine 3.47) yield high well-being, while pure interest (2.47) and faith/mission (2.57) yield high output but low well-being. Subtype well-being: ASD 3.26 > AuDHD 2.81 > ADHD 2.34. Because harness is post-outcome-coded, all effects are comparable non-randomized differences, not randomized causal effects; causal claims await a prospective living cohort. Labels are most-plausible hypotheses, not diagnoses.