Evolution and Arithmetic

Chapter 2: The Answer

← The OdometerHow Many Tickets Would You Need? →

A Universe of Universes

Twenty-five drums got us to a laptop losing to the age of the universe by a factor with twelve digits in it. Our sonnet — three hundred letters, not twenty-five — is a different animal entirely. Before I show you a single number, I want to show you a picture, because the picture will do more work than the arithmetic ever could.

Start with everything. Every atom in the observable universe — every atom in every star, every planet, every grain of dust, out to the farthest galaxy whose light has had time to reach us. All of it.

Now take every one of those atoms, and blow it up into an entire universe of its own — a whole new universe, packed with just as many atoms as the one we started with.

Now take every atom in every one of those brand-new universes, and blow each of those up into a universe of its own too.

Do it again.

And again.

If you've lost the ability to picture any of this — if somewhere around the second or third blow-up your imagination simply threw up its hands and quit — don't feel bad. So did mine. Nobody's mental camera can actually zoom out this far. We're well past the point where "picturing it" is even the right goal.

Now take every single atom, after all four blow-ups, and turn each one into a computer. Set every one of those computers to work, churning out random letters as fast as a computer can go. Run all of them — this staggering, unimaginable swarm of former atoms, now computers — for most of the age of the universe.

Do all of that, and you'd have a real, fighting chance of one of them typing out a Shakespeare sonnet by pure accident.

Not a guarantee. A chance.

Here is something that sounds as if it ought to help. Suppose that covering just one-tenth of all the possible strings is enough to give you a real shot. Your instinct — a perfectly reasonable, perfectly human instinct — is to imagine that a tenth of the job looks like a tenth of everything I just described. A few fewer blow-ups, maybe. A noticeably smaller pile.

It doesn't work that way. Not even close. And to see exactly why, we need to go back to the odometer.

Let's rebuild this whole scenario, drum by drum. Remember the drum rule: every drum you add to an odometer multiplies its range by ten. So to talk about numbers this size, we're just going to count drums instead of writing out endless strings of zeroes.

Our sonnet is three hundred letters long, and each of those three hundred spots can be any of twenty-six letters. So the odometer that could display every possible sonnet-length string would have three hundred drums, each with the twenty-six letters of the alphabet on it.

How many drums would you need on a plain, ordinary, base-ten odometer just to count all of its settings — every random string and every real sonnet alike? Four hundred and twenty-five. Not thirty-five, like our song. Four hundred and twenty-five.

Now let's build our swarm of atoms-turned-computers, the same way, in drums.

The atoms in the observable universe take eighty drums to count. Blowing each of those atoms up into a whole new universe multiplies the count by the number of atoms in a universe — and multiplying by an eighty-drum number adds eighty drums. So after one blow-up, the count needs a hundred and sixty drums. Do it again: two hundred forty. Again: three hundred twenty. One more time: four hundred. Four blow-ups, and we've gone from eighty drums to four hundred — that unimaginable swarm from the picture a moment ago.

Turn every one of those atoms into a computer and set each one pumping out a million random letters every second. A million is six drums' worth (six zeroes), so the whole swarm, working together, produces letters each second at a rate that takes four hundred and six drums to count.

We have a catalog that takes four hundred twenty-five drums to count, and a swarm producing four hundred and six drums' worth of letters every second. Subtract one from the other, and you're left with nineteen drums: that's how many it takes to count the seconds our swarm would need to cover the whole catalog. Nineteen drums' worth of seconds comes to about a hundred billion years — roughly seven times the age of the universe. (The arithmetic: The Sonnet Catalog and the Swarm.) Run the swarm for most of the age of the universe, and it covers about a tenth of the catalog — which is exactly the real, fighting chance the picture promised.

And notice: the time is not the big number here. It pales beside the really big number, the number of computers. Take away just one blow-up and move its eighty drums over to the time, and the time needed would be the age of the universe multiplied by a number eighty-one digits long — a span no time-related words can describe. These are numbers well beyond practical comprehension, however one frames them.

Now back to that tenth of the job. Say, generously, that covering one-tenth of all the possibilities gives a meaningful, if modest, chance of a hit. How many drums does doing "only a tenth of the job" save?

If your gut says something like forty-two — ten percent of four hundred twenty-five — your gut is doing ordinary, everyday arithmetic on a number that isn't ordinary or everyday at all. Ten percent doesn't mean stripping away ninety percent of the drums. It means dividing the total by ten. And dividing by ten costs you exactly one drum. One. Doing "only a tenth of the job" takes our four-hundred-twenty-five-drum catalog down to four hundred twenty-four. That's the entire discount.

With small numbers, this is obvious. Everyone knows that 10 is a tenth of 100, and that 100 is a tenth of 1,000; we deal with numbers like these all the time. In each case, the smaller number is just the larger one with one zero removed. One zero is one drum: adding a zero multiplies by ten, and removing one divides by ten.

But show someone 1080 and tell them that 1079 is a tenth of it, and the mind goes blank. Writing both out in full doesn't help either: one with eighty zeroes, the other with seventy-nine, and the second simply doesn't look like a tenth of the first — even though almost anyone can see that 100,000 is a tenth of 1,000,000. The difference lies in what we're used to, not in the numbers themselves.

So in our case, cutting the job by ninety percent means removing one drum out of four hundred twenty-five. That's not a rounding error in the story — that is the story. At this scale, cutting the job by ninety percent barely moves the number at all. This is the single strangest, most consistently misunderstood fact about numbers this size, and it's the reason nobody's intuition — including mine, including yours, including, I'd bet, most working scientists' gut reactions when they haven't sat down and actually done the arithmetic — can be trusted anywhere near a number like this one without a pencil, some patience, and an odometer to keep it honest.

The same goes for the fact that we'd be happy with any one of Shakespeare's 154 sonnets, rather than one particular sonnet. A hundred fifty-four targets instead of one divides the job by 154 — a little over two drums. Four hundred twenty-five drums become about four hundred twenty-three. It hardly makes a dent.

Before we move on, I want to leave you with one more picture — not of time, this time, but of space — to give you a better sense of the size of what we're discussing.

Where would you even put all this output?

What's the best data storage device that exists, right now, anywhere on Earth? Not science fiction — a working, measured technology. As it happens, the answer is a pleasing little joke on the whole subject of this book: the single densest data storage medium anyone has ever actually built and tested is DNA. In 2017, researchers packed the equivalent of 215 million gigabytes — 215 petabytes — of ordinary computer data into a single gram of synthetic DNA. A gram. About the weight of a paperclip.

So let's be generous to our random-letter swarm and imagine we could turn every single atom of ordinary matter in the entire universe into that same DNA-based storage, at that same record-setting density. How many drums would it take to count the total number of bytes all of that matter could hold?

Seventy-four. The entire observable universe, converted matter and all into the best data storage device humanity has ever built, holds a number of bytes that takes seventy-four drums to write down.

Our swarm's output, remember, took four hundred and six drums to write down — and that's just what it produces in a single second.

Four hundred six against seventy-four isn't close. It isn't even the right kind of comparison to call "close" — the two numbers are three hundred and thirty-two drums apart. There is not enough matter in the entire universe, turned into the best storage medium ever invented, to record even one second of what this swarm is producing.

Maybe you're thinking: fine, we've run out of matter — but space itself is enormous, and almost all of it is empty. What if some future technology could store data directly in that emptiness, with no atoms involved at all? Physicists have actually worked out an answer to exactly that question — an absolute ceiling on how much information can be packed into any region of space, built into the laws of physics themselves, regardless of what device anyone ever invents. Apply that ultimate limit to the entire observable universe, matter or no matter, empty space and all, and you get a number that takes about a hundred twenty-four drums to write down, in bits. Even if you could store one bit in every smallest possible speck of space — far more than physics allows — you'd reach only about a hundred eighty-five drums.

Our swarm's one-second output, converted to the same units, takes about four hundred seven drums. (The storage arithmetic: Where Would You Put It?.)

The universe — every atom, and every last cubic inch of the space between the atoms, filled to the absolute physical limit anyone can conceive of — cannot hold what these computers produce in a single second. Not almost. Not close.

Based on what we just found out, I have to assume that Stephen Hawking never did the actual calculation. He was a theoretical physicist whose mind worked on a far higher plane, and he probably never bothered with mundane arithmetic of the sort we just did. If he did do the arithmetic, and understood what the numbers meant, I would have to assume he had a very odd way of expressing things. I would never call the time it would take a horde of monkeys to type a sonnet by chance "very occasionally"; the most low-key term I could use would be "virtually impossible." So I'll give him the benefit of the doubt — and a failing mark for accuracy and carefulness.

So, in the end, what can we expect from our horde of computers outputting a huge number of letters every second?

Remember the get-rich sonata scheme: the total number of possible songs, twenty-five notes long, drawn from a scale of twenty-five notes, comes out to 2525 — about 8.8 × 1034 songs. Multiply that by the twenty-five notes in each one, and the total number of individual notes needed to produce every single one of them, just once, is a number with thirty-seven digits.

Suppose, instead of outputting a mere million notes a second, we scale up to two billion computers, instead of just one billion, and have each one systematically — not randomly, so nothing is ever wasted on a duplicate — churning out three and a half billion notes every second. You read that right: three and a half billion, not one million. Run that swarm continuously for ten billion years, close to the age of the universe itself, and the total output finally catches up to that thirty-seven-digit number. Every single possible song, twenty-five notes long, on a twenty-five-note scale, generated exactly once. (The arithmetic: Every Song, Once.)

Storage, however, and getting it all safely to the copyright office, would be a bit of a problem. I don't think the local post office is equipped to handle that much. Maybe FedEx would be a better idea.

Psst. Wanna invest?

Last revised 2026-11-16

← The OdometerHow Many Tickets Would You Need? →