Evolution and Arithmetic

Chapter 8: Weasels

← Back to the Human LineThe Paradox of the Lock →

It's All a Game

So, in sum, the Weasel program does not inform us of how evolution works. If anything, it tells us how it doesn't work — and thinking about its defects leads to a better understanding of the kinds of hurdles evolution actually faces. So, thank you for that.

But there is one thing it cannot be thanked for, and that is how misleading it is.

Dawkins, and every version of the Weasel program since, has used the word "generation" to label each step of the process, and the language around the program is full of other words — "mutation" among them — that make the whole thing sound like a description of evolution itself, rather than what it actually is: a demonstration of how easily a program reaches a goal once you've written rules that guarantee it will.

After everything shown in this chapter, it should be clear why that word doesn't hold up. In the program, a "generation" is just one round of mutation and selection — a single copying event, immediately followed by a keep-or-discard decision. A real generation requires an organism to actually reproduce, and for a mutation to matter at all, it has to spread — climbing from a single individual to a meaningful share of a whole population, the same climb the first citrate-related change had to make in Lenski's flask, one real generation at a time, long before the second, harder change ever arrived. What the program calls a generation is really a single round of copying. An actual generation is only one link in the much longer chain a mutation needs to survive before it can become the foundation for anything further.

To be fair to Dawkins, he said as much himself. In the same book, he warned that the demonstration is "misleading in important ways": real selection has no distant target to aim at, only whatever helps survival or reproduction right now. But that warning rarely travels with the program. Online, in classrooms, and in countless videos, Weasel is shown again and again as if it were proof that evolution works — forty generations, problem solved.

So the conclusion is this: the result you get out of a demonstration like Weasel depends entirely on how you choose to play the game — what you allow to keep changing, what you don't, how many candidates you let compete each round, and how you define success. That much is fair; it's a game, after all, built to illustrate a principle, not to reproduce biology. But how the result then gets reported — as if "forty generations" were some kind of settled fact about how evolution works — is not a game at all. I would call it a travesty. It's either a failure to understand what the demonstration actually does, or something considerably less innocent than that. It would seem that "weasel" is a very apt name for the game.

This isn't the first time this book has run into a demonstration that isn't quite what it claims to be. Chapter 6 told the story of Jesse Anderson's "virtual monkeys write Shakespeare" experiment, and showed how badly that story was mangled on its way to the public: uncritical coverage from the BBC and others, a mathematician's "not a chance" wrapped in a soundbite that made the impossible sound merely slow, and a label ("an evolutionary approach") that didn't describe what the program did.

Weasel turns out to be the same story wearing different clothes. An original program nobody can produce, a term ("generation") that quietly misleads, and a "forty generations" figure repeated for forty years without anyone checking what it would take to make that number honest.

The internet can't be trusted to get this right. Neither, evidently, can some rather famous books, or even professors quoted by the media. And the public can easily be fooled. That's worth remembering the next time a tidy number shows up dressed as settled science, even when it is widely accepted.

Just playing games — weasels.

Last revised 2026-11-19

← Back to the Human LineThe Paradox of the Lock →