Evolution and Arithmetic

Chapter 3: A Panorama of Life

← How Different Are Two Species?Every Bacterium a Computer →

One Sonnet a Year

Here, finally, is the assumption I want to build on. It is deliberately, unrealistically generous to the idea we're testing. Let's assume that, on average, each species differs from its nearest evolutionary relative by the equivalent of just one new protein's worth of genetic instruction — one new three-hundred-letter chain, or something functionally similar to it, whether or not it's literally a protein. Given that even the most cautious count gives each step in our own line more than a hundred proteins' worth of change, "one protein's worth of difference" per species is an almost absurdly small allowance. But that's precisely the point: we want the smallest, most defensible number, so that nothing that follows can be dismissed as unfair.

We already worked out that the pace comes to one new species, on average, every year across the whole six-hundred-million-year span we're using for higher life. Combine that with the one-protein-per-species assumption, and the implication follows immediately: evolution would need to have produced something on the order of one brand-new, three-hundred-letter, working genetic innovation — a full working "sonnet," in the language of this book — every single year, on average, for six hundred million consecutive years.

How hard is it to hit one working protein by chance? Let's be exactly as generous as we were with sensible strings in Chapter 2. There we assumed that at every position, half the alphabet would keep the string sensible. Do the same here: at every one of the three hundred positions, assume that any of ten of the twenty amino acids would do. That's wildly generous — real proteins are far pickier at many positions — but it's a clean assumption, and it doesn't rig the game against the idea we're testing. And it brings back a number we've met before. Half the choices at every position, three hundred times over, is our three hundred coin flips, all heads: one chance in ninety drums' worth. (The arithmetic: Counting Protein Chains.)

Remember ninety drums? Every usable atom in the universe, turned into computers and run for the whole age of the universe, managed only seventy-eight drums' worth of tries. So here is the arithmetic problem. How could life write the equivalent of one "sonnet" a year, year after year, without fail, for six hundred million years, when every usable atom in the universe, turned into computers, couldn't produce even one in the whole age of the universe?

If that sounds like the end of the story, it isn't — it's only the first round. Next we count what life actually had to work with.

Last revised 2026-11-17

← How Different Are Two Species?Every Bacterium a Computer →