Five Problems with Dawkins’ Weasel Program
Five Problems with Dawkins’ Weasel Program
A clever simulation — but one that quietly smuggles in everything it claims to do without.
“Cumulative selection, by contrast, is a totally different process… it is the kind of selection that is cumulative, that builds up… It works by gradual degrees.”— Richard Dawkins, The Blind Watchmaker (1986)
In The Blind Watchmaker (1986), Richard Dawkins introduced a computer simulation that became one of the most cited illustrations in evolutionary biology. Starting from a random 28-character string, the program converged on Shakespeare’s phrase “METHINKS IT IS LIKE A WEASEL” in just 43 generations. He called it cumulative selection — proof, he argued, that complexity can build itself without a designer.
It’s a compelling demonstration. But there are five serious problems with it as an argument for evolution.
White = correct letter position · Dark green = wrong position · Target phrase known in advance
WDLTMNLT DTJBKWIRZREZLMQCO P — no functional proteins, no biological role whatsoever. In the real world, natural selection would eliminate this organism immediately, before it ever reproduced. In Dawkins’ simulation, it becomes a parent without question. Every intermediate stage survives automatically. The very mechanism that supposedly drives evolutionary progress has been silently removed.
“The evolutionary trees that adorn our textbooks have data only at the tips and nodes of their branches; the rest is inference, however reasonable, not the evidence of fossils.”
— Stephen Jay Gould, Natural History, Vol. 86 (1977)Gould wasn’t rejecting evolution — he was proposing punctuated equilibrium to explain the gaps. But his acknowledgment stands: the fossil record does not show the unbroken chain of gradual transitions that the Weasel model predicts. And right now, the living world should also be populated by far more incomplete organisms than complete ones — creatures with half-formed wings, eyes in mid-development, hearts not yet connected. We simply don’t observe this.
A process converging on a predetermined goal is not evolution — it’s optimization. And optimization requires someone who set the objective. In attempting to show that complexity can arise without a designer, the Weasel program actually demonstrates exactly why a designer is necessary to generate complex, specified information.
Meanwhile, a 2025 Nature study (Yoo et al.) using whole-genome sequencing found that humans and chimpanzees differ by approximately 448 million nucleotides — roughly 14–14.9% of the total genome. This overturns the widely cited “1% difference” by a factor of 14. A biased, hotspot-concentrated mutation mechanism cannot plausibly account for divergence of this scale.
| Problem | Weasel Program | Real Evolution |
|---|---|---|
| Natural selection | ✗ All variants survive | ✓ Most variants eliminated |
| Junk DNA | ✗ Predicts vast non-functional sequences | ENCODE: ~80% of genome is functional |
| Fossil record | ✗ Predicts continuous gradual change | Shows stasis + punctuated change (Gould) |
| Direction / goal | ✗ Fixed target from step one | Claimed to be completely undirected |
| Mutation realism | ✗ Uniform random at all positions | Biased, hotspot-concentrated |
The Bottom Line
The Weasel program is mathematically elegant. But it works precisely because an intelligent programmer set the target, defined the selection criteria, and ran the process. Strip out the programmer — and you have no target, no selection basis, no reason for any letter to be preserved. Dawkins intended to show complexity arising without a designer. His program shows exactly what it looks like when a designer is generating complexity.
Read the Full Deep Dive