Heritability of human life span is ~50% when heritability is redefined
A blog post dissects a recent 'Science' paper that claims the heritability of human lifespan is around 50%, a significant jump from traditional estimates. The author meticulously explains how this new figure is derived from redefining heritability within a simulated, hypothetical world free of 'extrinsic mortality.' This detailed critique of scientific methodology and publishing practices resonates with the Hacker News audience's appreciation for technical scrutiny and skepticism towards overblown claims.
The Lowdown
This article provides a critical analysis of a "Science" paper that redefines the heritability of human lifespan, claiming it to be around 50% rather than the commonly accepted 23-35%. The author, dynomight, unpacks the paper's methodology, highlighting how its grand conclusions rely on a mathematical model simulating a world without non-aging-related deaths.
- Traditional estimates for lifespan heritability, derived from twin studies, typically range from 23% to 35%.
- The analyzed paper constructs a "twin simulator" that models a hypothetical world where individuals do not die from accidents, murder, drug overdoses, or infectious diseases (termed "extrinsic mortality").
- In this simulated environment, with extrinsic mortality reduced or eliminated, the paper calculates lifespan heritability to be significantly higher, between 46% and 57%.
- The author argues that the paper misrepresents this finding, implying a higher heritability in the real world, whereas it applies only to a theoretical, adjusted scenario.
- A strong critique is leveled against "Science" for its vague writing style, lack of mathematical detail, and disjointed explanations, making the paper difficult to understand and reproduce.
- Upon closer examination of the paper's figures, the author contends that even when adjusting extrinsic mortality to more modern, real-world levels, the simulated heritability estimates are closer to 35-45%, not the headline-grabbing 50%.
The piece serves as a potent reminder that heritability is an observational statistic highly dependent on context and definition. While acknowledging the cleverness of the paper's model, dynomight ultimately faults it for ambiguous presentation and a misleading interpretation of its own findings, particularly within a prestigious journal like "Science."