Lifespan Heritability Hits 50% Only After Redefining What Heritability Means

A recent Science paper claims human lifespan is roughly 50% heritable, far above the 23-35% figure that traditional twin studies produce. The dynomight author argues this reframing is technically valid but misleading: the authors built a simulator that models a hypothetical world with no accidents, murders, overdoses, or non-age-related infectious disease, then computed heritability on that simulated population. Strip out the random non-genetic ways people die young, and of course genetics explains a bigger share of what’s left.

The model works by fitting an ‘extrinsic mortality’ parameter to historical twin data, then dialing that parameter down. When extrinsic mortality matches reality, the simulator reproduces the standard 23-35% estimates, which suggests the math is sound. The 50% figure simply describes a counterfactual universe, not our own. The author’s broader point is that heritability is always context-dependent: redefining the trait or the environment can push the number anywhere between 0 and 100%.

A secondary complaint targets Science itself. The paper is fundamentally about math but contains almost none, scatters its methodology across the body and appendix, and includes at least one undefined symbol that survived editing. The author reads this as a systemic failure of Science’s professional editors rather than the authors’ fault, and a reason to be skeptical of how the headline result is being framed in coverage.