What This Has to Do with Life
Two full chapters in, and I haven't mentioned biology once — on purpose. It's time to fix that, with the same insistence on numbers as before.
First, a word to anyone who knows evolutionary theory, because I know what you're going to say before I've said anything. Nobody, you'll tell me, thinks a protein ever popped into existence all at once, by a single lucky throw. Evolution works by slow accumulation: one small change at a time, each one tested by natural selection, kept if it helps and discarded if it doesn't. So all these calculations about random strings are beside the point.
I know. Those answers are listed at the end of this chapter, and the chapters that follow deal with them, one at a time. But you can't judge whether an answer is strong enough until you know the size of the problem it has to answer. That's all this chapter does: it frames the problem in plain arithmetic. And the heart of the problem is this — the space of possibilities life would have had to search is vastly larger than anything life could ever actually have searched. Whatever happened, it happened after exploring only a vanishingly small corner of what was possible. Any answer, slow accumulation included, has to explain how that could work.
So stay with me. This is a long chapter, and in places a complicated one, but every step is just arithmetic, and the payoff comes at the end.
The basic unit of life is the cell. Every living thing, from a single bacterium to a blue whale, is built from cells — either a single one, doing everything on its own, or trillions of them, each specialized and working in concert. A cell isn't just a container; it's a tiny, self-sustaining chemical factory, taking in raw materials, converting them into energy and structure, copying itself, and responding to its surroundings, all without any outside instruction telling it what to do next. Whatever "life" actually is, this is where it lives.
But what actually makes a cell do all of that, at the level of molecules? Strip away everything else, and at bottom a cell runs on two kinds of molecule doing two different jobs.
One kind carries information. This is DNA, and its job is to store instructions — a kind of archive, or a blueprint, or a recipe book, depending on which metaphor you prefer. The other kind does the actual work. These are proteins, and they're the machines: the things that build, break down, transport, signal, defend, and drive every chemical process a living cell needs to survive.
Between the two sits a third molecule, RNA — a close chemical cousin of DNA. When a cell needs a particular protein, it copies the relevant stretch of DNA into RNA, and that copy carries the instructions to the cell's protein-building machinery. RNA does some real work of its own as well; parts of that protein-building machinery are themselves made of RNA. Some researchers think RNA may even have come first, before either DNA or proteins. We'll meet it again in later chapters. For now, DNA and proteins are enough.
Both of these molecules are, in a very real structural sense, languages. Both are built from a small alphabet of simpler pieces — four kinds of "letter" for DNA, twenty for proteins — strung together, one after another, into long chains, the way letters string together into sentences. Change the sequence, and you change what the molecule does, or whether it does anything useful at all, exactly the way changing a letter in a sentence can turn sense into nonsense. This isn't a loose poetic comparison I'm reaching for to make a nice chapter opener. It's a structural fact about how these molecules actually work, and it's why everything we've spent this book building — odometers, drums, search spaces, winning tickets — turns out to apply here almost without modification.
DNA, though, is hopelessly big for where we want to start. The human genome alone runs to something like three billion letters — so far past anything we've dealt with that it would swamp our odometer before we'd even gotten our bearings. So let's set DNA aside for now and start with proteins instead. Proteins are smaller and more manageable, and — even though a protein is never built independently of the DNA instructions that specify it — it's the protein itself, and its own size, that determines whether it can actually do the job it's needed for. A recipe is only as useful as the dish it produces.
So: how big is a typical protein?
Last revised 2026-11-17