Evolution and Arithmetic

Chapter 3: A Panorama of Life

← One Sonnet a YearReality Check →

Every Bacterium a Computer

So let's ask the obvious next question: how many "computers" does life actually have available to do this work? Not the impossible swarms we built out of blown-up universes a chapter back — the actual number of living things that have ever existed to do any random generating at all.

Before we count them, there's a hard physical limit worth setting up first, because it gives us a ceiling that can't be moved, rather than just another staggering estimate.

Every living thing that has ever existed — bacterium or blue whale — needs water. Not figuratively. Literally, as a physical ingredient, present inside every one of its cells. So how much water does a cell actually take? A single ordinary bacterium — an E. coli cell, the standard yardstick biologists use — holds, by direct calculation from its known mass and water content, about twenty billion water molecules. Not "a lot of water" in some vague sense. Twenty billion molecules, in one cell you can't see without a microscope.

How much water does Earth have to work with? Add up the oceans, the lakes, the rivers, the groundwater, and — being as generous as the geological literature allows — a large allowance for water locked away deep in the Earth's mantle, and round up. You arrive at about ten to the forty-eighth water molecules on the entire planet, surface and deep interior alike, counted with generosity toward the high end of every published estimate.

Divide one number by the other — total water molecules on Earth by water molecules per cell — and you get an absolute ceiling on how many bacterium-sized living things could possibly exist on this planet at once, no matter how favorable the conditions. That ceiling comes out to about ten to the thirty-eighth: thirty-eight drums. Not an estimate of what does exist — a hard limit on what could ever exist, if every drop of water on Earth were somehow devoted to nothing but building the smallest living cells. (The arithmetic: The Water Ceiling.)

That's the outer wall. Now here's the actual figure, from biological research rather than a physical limit: how many living cells does Earth actually hold today? The most widely cited estimate — from a landmark 1998 study that other researchers have cited many thousands of times since — puts the number of bacteria and other single-celled organisms of their kind on Earth at about five thousand billion billion billion. In our notation, about five times ten to the thirtieth: thirty-one drums.

Notice that: thirty-one drums of actual living cells, against a hard physical ceiling of thirty-eight. We are nowhere near the water limit. Whatever holds the number of living things down, it isn't water.

Five times ten to the thirtieth is such an overwhelmingly large number that no higher organism comes close to competing with it. Even if all ten million or so currently existing higher species had populations as large as the single most abundant higher organism on Earth today, their combined total would still fall short of the bacteria by several orders of magnitude — and bacteria, on top of that, reproduce far faster than any higher organism. So the calculation we're about to do, with bacteria as its basis, would barely move if higher organisms were added in. It's like adding a penny to a hundred thousand dollars.

That isn't to say higher organisms are insignificant by every measure — only by this one. A single person carries trillions of cells; a tree, even more. The cells of higher organisms also tend to be far larger than the average bacterium. Taken all together, the total weight of all higher organisms on Earth, plants above all, far exceeds the total weight of all bacteria. But the trillions of cells making up one human being take part in reproduction only through the occasional offspring, while a single bacterial cell can divide in a matter of minutes or hours, depending on conditions. What counts here is not how many cells an organism contains, but how many separate organisms exist and how often each one reproduces. Cell count simply isn't the right yardstick for this purpose.

So let's put the real number — five times ten to the thirtieth living cells — up against the ninety drums of tries we need for even one working protein.

If you divided that requirement evenly across every living bacterium on Earth today, each individual cell would need to make something like ten to the fifty-ninth random genetic "attempts" — a sixty-drum number. Not the population as a whole. Each cell.

To see how far outside the possible that is, go back to computers. Think of every bacterium as a computer whose only job is to turn out brand-new protein chains, one after another. Make it a fast one, by bacterial standards: one brand-new, never-before-seen three-hundred-letter chain every twenty minutes — roughly the fastest a bacterium can divide under ideal laboratory conditions — nonstop, for the entire three-and-a-half-billion-year history of life on Earth. That's about ten to the fourteenth chains per computer.

Now give life the whole fleet. Assume, generously, that Earth has held a steady five times ten to the thirtieth bacteria throughout that entire span — hard to imagine, since bacteria mostly live off organic matter, and there was none of that to speak of when life began, but we'll grant the number anyway. Let every one of them turn out a new protein chain every twenty minutes, for three and a half billion years. The total number of chains produced across all of life's history comes to about forty-five drums' worth. (The arithmetic: Every Bacterium a Computer.)

That's forty-five drums short of the ninety we need — for one working protein, not six hundred million of them. That isn't a near miss. It isn't even in the same conversation as a near miss. Fill every drop of water on Earth with bacteria, right up to the thirty-eight-drum ceiling, and run them the same way, and the total still comes to only about fifty-two drums: thirty-eight short.

And this is the generous version of the story. Bacteria, at least, exist in staggering numbers and reproduce fast. Move up the tree of life to anything larger and slower — plants, insects, mammals, anything that takes months or years or decades between generations, and exists in millions or billions rather than in numbers thirty-one digits long — and the same arithmetic gets dramatically worse. A population of ten or a hundred thousand slow-breeding animals, spread across a few million years, isn't running anywhere near the same number of "trials" as a swarm of bacteria dividing every twenty minutes. Whatever room the raw numbers might seem to offer at the bacterial level evaporates almost entirely once you're talking about the organisms we usually picture when we talk about evolution.

Here's where we stand. The size of the search space — twenty possible amino acids, three hundred positions, three hundred ninety-one drums — is a fixed mathematical fact. We cut it down dramatically by counting as a success any chain in which, at every position, any of half the amino acids would do: an assumption far kinder than the truth, which leaves us needing ninety drums' worth of tries instead of three hundred ninety-one. The number of living things available to do the searching is, likewise, a measured and published figure, not something we get to adjust upward to make the numbers work.

The real unknown in this whole argument is the size of the target — how many different three-hundred-letter chains would actually work, fold properly, and do something useful, and how many wouldn't. Our fifty-percent assumption is a stand-in for it. Nobody knows that number precisely, and probably nobody ever will. As we shall see, there are estimates for the number of folds — the different basic three-dimensional shapes a protein can take — but there are no clear estimates of the ratio between chains that fold into something that works and chains that don't, or of how many fail to fold at all, or fold into something entirely useless.

But even when we make the most favorable assumptions, the gap between what's needed and what's available is so enormous that it holds up no matter which reasonable guess you plug in for that one unknown. It doesn't matter whether the gap is forty-five drums, or more realistically fifty, fifty-five, or even seventy or more — none of these is close. All of them are extremely far from what life would need to have had a reasonable chance of doing what it's done. Anyone who wants to maintain that this happened by chance needs a very good answer.

Last revised 2026-11-17

← One Sonnet a YearReality Check →