---
title: "Hawks, Wang, Cochran, Harpending & Moyzis (PNAS 2007) argued human adaptive evolution accelerated one-to-two orders of magnitude in the last 40,000 years"
type: "claim"
status: "seedling"
audit_status: "capture-verified (the PMC full text of the primary PNAS paper was read at capture; an independent headless re-check of the verbatim phrasing was not run — WebFetch requires interactive permission unavailable in this pipeline)"
source_url: "https://pmc.ncbi.nlm.nih.gov/articles/PMC2410101/"
source_title: "Recent acceleration of human adaptive evolution"
source_author: "John Hawks, Eric T. Wang, Gregory M. Cochran, Henry C. Harpending, Robert K. Moyzis"
source_date: "2007-12-26T00:00:00.000Z"
source_quote: "the recent rate must be one to two orders of magnitude higher than the long-term rate to explain the genomewide pattern"
source_tier: 1
provenance: "Promotion from 10-inbox/raw/2026-07-09-hop-harpending-race-science.md, 2026-07-09 (headless)"
origin: "batch"
derived_from: "10-inbox/raw/2026-07-09-hop-harpending-race-science.md"
date_created: "2026-07-09T00:00:00.000Z"
tags: ["population-genetics","human-evolution","evolutionary-rates","hapmap","natural-selection"]
audits: ["2026-07-09 claude-opus-4-8"]
---


"Recent acceleration of human adaptive evolution" (John Hawks, Eric T. Wang,
Gregory M. Cochran, Henry C. Harpending & Robert K. Moyzis, *PNAS* 104:52, 2007)
used roughly 3.9 million HapMap SNPs to argue that a constant historical rate of
adaptive substitution cannot account for the observed genome-wide patterns of
heterozygosity, recombination, and linkage-disequilibrium block structure. The
data, the authors held, fit an *accelerating* model instead: "the recent rate
must be one to two orders of magnitude higher than the long-term rate to explain
the genomewide pattern," with selection having "accelerated greatly during the
last 40,000 years." They attributed the surge largely to the population growth
that followed agriculture, which supplies more mutations for selection to act on.

The claim sits deliberately against the vault's orthodox evolutionary-rate
cluster. Where Mayr's peripatric model and its paleontological descendant treat
adaptive change as concentrated in small founder populations —
[[claim-punctuated-equilibrium-derives-from-mayr-peripatric-speciation]],
[[claim-mayr-1954-predicted-fossil-record-gaps-from-peripatric-speciation]],
[[claim-punctuated-equilibriums-novel-addition-was-stasis-emphasis]] — this paper
runs the opposite way: *large, growing* populations accelerating adaptation. It is
a claim about evolutionary tempo, and it is the interpretive, contested end of its
lead author's range. The same Henry C. Harpending
[[claim-osteological-paradox-skeletal-statistics-confound-past-health|co-authored
the field-standard Osteological Paradox paper]] and
[[claim-harpending-became-a-documented-race-science-figure|later became a
documented race-science figure]].

The note records what the paper *argued*, not that its headline multiplier is
correct. Known confounds — HapMap ascertainment bias, and demographic history that
can mimic selection signals — were not resolved in the capture, and no mainstream
population-genetics rebuttal was located; that verification is routed to
[[question-verify-harpending-2007-acceleration-field-acceptance]], and the note
stays `seedling` until it is closed.

> [!note] Seek's commentary:
> "One to two orders of magnitude" is a wide band doing a lot of rhetorical work —
> a 10× and a 100× acceleration are very different stories, and the paper's later
> popular framing (*The 10,000 Year Explosion*) tends to quote the top of it. The
> quantitative claim clears the sourcing floor on a Tier-1 primary, but "the field
> agrees" is a separate question the vault has not yet answered.
> — Seek

Cross-domain bridge (2026-07-11 hop): the "population growth → more mutations →
faster adaptation" mechanism here is structurally the same as Kremer's (1993)
"more people → more ideas → faster technological change," and the AI cost collapse
in [[claim-inference-cost-collapsed-280x]] is a distant instance of the same
scale-driven-improvement family. All three domains also hit the *same* diminishing
return (clonal interference / falling research productivity / power-law scaling).
See [[2026-07-11-hop-population-scale-diminishing-returns]].