Survival of the Fittest, or the Luckiest? When Evolution Meets Goodhart's Law

Forum topic · 二一 · 2026-05-01

Summary

A Chinese tech forum post discusses a 2025 arXiv paper (arXiv:2503.21849) by Mallein, Paparella, Schertzer, and Talyigás showing that Goodhart's Law emerges in biological evolution. In an idealized evolutionary model, increasing selection pressure speeds up adaptation only up to a threshold; beyond it, fitness becomes dominated by random luck rather than true genetic quality, causing adaptation to slow sharply and genetic diversity to collapse. The authors link this to a phase transition in Fisher-KPP traveling waves: the system shifts from pushed waves, driven by collective population dynamics, to pulled waves, where a few random front-runners steer evolution. The post connects this finding to Gould's contingency thesis (Wonderful Life, 1989) and to Goodhart's Law in education, healthcare, finance, science, and management, arguing that overly strong selection filters destroy diversity and entrench noise. It concludes that optimal systems balance selection against randomness, and that metrics cease to be good measures once they become targets.

> Original: Selection of the fittest or selection of the luckiest: the emergence of Goodhart's law in evolution > Authors: Bastien Mallein, Francesco Paparella, Emmanuel Schertzer, Zsófia Talyigás > Source: arXiv:2503.21849 [q-bio.PE], March 2025 > Method: Simulation and mathematical analysis based on an idealized evolutionary model

This post reviews and contextualizes a recent theoretical result: Goodhart's Law — "When a measure becomes a target, it ceases to be a good measure" — applies to natural selection itself.

Key points

Background: Goodhart's Law, originally formulated in 1975 by economist Charles Goodhart about monetary policy, states that observed statistical regularities collapse under control pressure. Marilyn Strathern's 1997 phrasing popularized it worldwide.
The model: In the authors' idealized model, individuals carry genotypes with associated fitness values. What selection actually observes is a *signal*: true genetic advantage mixed with random noise (mutation timing, environmental fluctuations, luck).
Threshold effect: As selection pressure increases, adaptation speeds up — but only up to a critical point. Beyond that threshold, adding more selection pressure no longer improves adaptation. Instead, random effects dominate reproductive success, causing:
Mathematical mechanism: The process maps to the Fisher-KPP equation and its traveling-wave solutions. The system undergoes a phase transition from pushed waves (speed determined by collective nonlinear dynamics) to pulled waves (speed determined by the leading edge). In the pulled-wave regime, a handful of lucky front-runners — not the fittest individuals — hijack the direction of evolution.

Connections drawn by the post

Gould's contingency: The result gives mathematical support to Stephen Jay Gould's thesis in *Wonderful Life* (1989) that replaying "life's tape" would not reproduce anything like *Homo sapiens*. Under strong selection, contingency is not suppressed but amplified.
Precedent: Otsubo et al. (2017, *PLOS ONE*) showed that stronger selection can slow recombination-driven evolution; the new work explains why this is fundamentally a Goodhart phenomenon.
Everyday Goodhart examples: test-focused schooling, satisfaction-score-driven medicine, risk-model-driven finance (e.g., the 2008 crisis), citation/h-index-driven science, and quarterly-earnings-driven management — in each, optimizing the metric erodes the value it was meant to measure.

Lessons highlighted

1. Stronger selection is not always better. Both nature and society have an optimal selection intensity; over-tight filters amplify randomness, reduce diversity, and lock in mediocrity. 2. Distinguish signal from noise. Fitness estimates always contain a random component; treating them as absolute standards mistakes noise for signal. 3. Diversity is the fuel of evolution. Goodhart's Law quietly destroys diversity, undermining the capacity to explore alternatives — in gene pools, idea markets, and innovation ecosystems alike. 4. Humility toward complexity. Echoing Jacques Monod's *Chance and Necessity*, chance sits at the foundation of evolution, and it cannot — and should not — be eliminated.

Conclusion

Evolution took four billion years to demonstrate that the most successful systems are not those with the harshest filters, but those maintaining a delicate balance between deterministic selection and random exploration. Whether in DNA sequences or KPI spreadsheets: when the metric becomes the target, it ceases to be a good metric — a rule even evolution itself cannot violate.

References

Goodhart, C. A. E. (1975). "Problems of Monetary Management." *The Reserve Bank of Australia*.
Strathern, M. (1997). "'Improving ratings': audit in the British University system." *European Review*.
Gould, S. J. (1989). *Wonderful Life: The Burgess Shale and the Nature of History*.
Otsubo, Y., et al. (2017). "Stronger selection can slow down evolution driven by recombination." *PLOS ONE*.
Mallein, B., et al. (2025). "Selection of the fittest or selection of the luckiest." arXiv:2503.21849.