Musings, Meanderings, & Loops · Essay
October 6, 2026
In the beginning was the Word, and the Word was with God, and the Word was God.
I have read that sentence my whole life, and this month, of all places, I found it in a book about telephone engineering. James Gleick's The Information is a history of how human beings learned to count, compress, and transmit what we know — drums, alphabets, telegraphs, codes, computers — and Gleick reaches for that same opening line of John's gospel, because by the end of his story the line has stopped sounding like theology. It sounds like physics.
So I have been carrying a cluster of questions around for weeks. If the Word was there first, what does that say about the mind of God? Without the Word, would there be nothing at all? Is consciousness possible without words — and what happens to it if we stop writing our own and let a machine write them for us?
I do not have answers to most of these. But I have found the one place where they all meet, and it is the most interesting sentence in twentieth-century science.
Start with the Greek, because English flattens it. The word John uses is logos, and it never meant only a word. It meant word and reason and account all at once — speech, and the sense inside the speech. Heraclitus, five centuries before John, used it for the rational order running through the cosmos. Philo of Alexandria, a Jewish philosopher writing in Greek in John's own century, made the Logos the blueprint through which God made the world.
John takes that inherited idea and does two audacious things with it. First, he opens with the same three words that open Genesis — in the beginning — so that every reader hears the echo: God said, and there was light. Creation, in this telling, is an utterance. The world exists because it was spoken, and it is intelligible because what was spoken has sense in it. All things were made through him, the third verse says, and without him was not any thing made that was made. So yes, by John's account: without the Word, nothing.
Then, at verse fourteen, the scandal: the Word became flesh. The ordering reason of the universe did not stay a principle. It walked around in sandals.
Whatever else that means about the mind of God, it means this: meaning is not decoration painted on a dumb universe afterward. It is the floor. The world is the kind of place that can be understood because it was said before it was made.
I believe that, as best I can tell. And I believe something else, which is where the trouble starts.
Claude Shannon did something I can only describe as creating reality out of his words — twice.
In 1937, as a twenty-one-year-old master's student at MIT, he noticed that the true-and-false algebra a Victorian logician named George Boole had worked out on paper mapped exactly onto the open-and-closed relays of telephone switching circuits. Logic could be built out of switches. Switches could perform logic. The thesis has often been called the most important master's thesis of the century, and it is hard to argue: every chip in every device you own is that observation, multiplied by a few billion.
Then, in 1948, he did it again. "A Mathematical Theory of Communication" defined the bit, measured information as entropy, and proved how much of it any channel could carry through any amount of noise. The digital age is the long consequence of those two documents.
I do not think Shannon invented any of this the way you invent a novel. I think the structure was already there — Boole's logic was true before Boole, and a relay was always going to be either open or closed — and Shannon was the man who saw it and wrote it down precisely enough that other people could build it. Discovery, not invention. Which is exactly what I think human beings are for, and I will come back to that.
But first notice where he learned to see. During the war Shannon worked on cryptography at Bell Labs, and his classified 1945 memorandum on the mathematics of secrecy holds, in embryo, the 1948 theory. That makes sense to me. A codebreaker is trained to look at a message he cannot read — to study its shape, its frequencies, its redundancies, while the meaning stays locked away. Spend a war looking at messages that way and you might begin to suspect that the shape is the part you can do mathematics on.
Which brings me to the sentence.
Early in the 1948 paper, Shannon explains that messages frequently have meaning. Then he sets it aside in one line:
"These semantic aspects of communication are irrelevant to the engineering problem."
— Claude Shannon, 1948
That is the founding move of the information age. To make information measurable, Shannon had to throw out what it was about. A bit does not care whether it carries a love letter, a stock price, or static. The theory works because it is blind to meaning — the one thing John's gospel puts at the beginning of everything.
So we have two accounts of the foundation of the world, and they look like opposites. John: in the beginning was the Logos, word charged with sense. Shannon: the sense is irrelevant; count the symbols. And here is the thing I cannot get around — both accounts work. John's has carried two thousand years of people through their lives. Shannon's runs the device you are reading this on.
I do not want to resolve that, because I think the honest position is that both are true at different depths. But the tension became urgent for me the moment I followed Shannon one step further, to the part of the 1948 paper almost nobody remembers.
To show what statistical structure looks like, Shannon generated fake English. He began with random letters. Then letters chosen by how often they occur in English. Then by which letter tends to follow which. Then whole words, picked by which word tends to follow the one before. With each step the gibberish grew more convincing, until his samples read like fragments of sentences — grammatical-ish, meaningless, unmistakably English-shaped.
That was a language model. In 1948. Built on a principle that meaning is irrelevant.
I wrote in August about why machine prose often feels dead — that next-word prediction produces text with the shape of thought and nothing standing behind it. I did not know then that I was describing Shannon's experiment carried to its conclusion. Today's AI writing systems are vastly more sophisticated than his tables of letter frequencies, and I use them every day, but their ancestor is a page of convincing nonsense from the man who declared semantics irrelevant to the engineering problem.
Which sharpens the question I started with. If consciousness lives in the struggle to find the right word, what happens when we hand the struggle to a system descended from a theory that excluded meaning by design?
I do not think the answer is that AI makes us stop thinking. I think it is subtler. The cognitive scientists still argue about whether thought requires words; there is good evidence for a deep, wordless layer of mind, the kind an infant or a chess player or a carpenter has. But the reflective layer — the part that knows what it believes — gets forged in the act of writing. Joan Didion put it in a sentence I keep taped to the inside of my head: I write entirely to find out what I'm thinking. The finding out is the point. A machine can hand me a fluent paragraph. It cannot do my finding out for me, because the finding out is not the paragraph. It is what happens to me while I make it.
I build software for a living, and lately the work has me living inside Shannon's world. One of the products I am building watches how work actually gets done — the stream of a working day, windows opening and closing, what gets entered and copied where — and tries to turn that analog activity into a clean digital description of the procedure underneath. It is an information-theory problem from top to bottom. I have spent months borrowing his tools: measuring entropy, gating out noise, separating signal from clutter.
And here is the thing the work taught me that the theory never quite says out loud.
Entropy can tell you where the surprise is. It cannot tell you which surprise matters. To gate out noise, you first have to decide what the signal is — and that decision is not mathematical. It is a judgment about what the work is for, what the person was trying to do, which pause was thinking and which was a sandwich. The moment I define the message, meaning walks straight back in through the front door. Shannon could set semantics aside for the engineering problem only because someone, somewhere upstream, had already decided what the message was.
That is not a flaw in Shannon. It is the boundary of his gift. The bit can carry anything. It can mean nothing. Meaning is the part that was never assigned to the channel.
So whose job is it?
I think Genesis answers that before John does.
In the second chapter, after God places the man in the garden to work it and keep it — I wrote about that verse a few weeks ago — the very next task is this: God brings the animals to the man to see what he would call them, and whatever the man called each one, that was its name. The first human occupation in Scripture, before farming has even had time to go wrong, is naming. God speaks the world into being; the creature made in his image is invited to say what the parts of it are.
That is the role I keep circling. The structure is already there — I believe that, the way I believe Boole's logic was true before Boole. Our work is not to invent it but to find it and say it truly: to look at what was spoken and name it accurately enough that other people can live in it, and build on it, and occasionally rewire the world with it the way a twenty-one-year-old did with a thesis about switches.
I asked at the start whether we must write our words down to bring things into being. The physicist John Wheeler — whose phrase it from bit imagined information as the root of matter itself — spent his last decades speculating about a universe that needs observers asking questions to become definite. That remains speculation, beautiful and unproven, and I hold it the way I hold most beautiful unproven things. But I know a version of it that is simply true. Shannon's thesis did not reorganize matter by quantum magic. It reorganized matter because people read it, understood it, and built what it described. Written words recruit hands. That is how a thought gets into the world, and it is why the writing has to be ours.
The machines can now produce words in any quantity. They are very good at the shape. What they were built, from their very first page, to leave out is the thing John put at the beginning of everything — and the thing Adam was given as his first job.
The bit was never going to carry meaning by itself. That was always going to be our part. In the beginning was the Word; and in the middle, where we live, someone still has to say what things are called, and why they matter.