Cover photo

Weapons and Flashlights

Before you start: I wrote this essay with Claude, an AI model, over a long conversation. You should know that up front. By the end, you'll see why I think it's part of the point.

I asked an AI to help me title this essay. It offered: "You Didn't Think That."

Good line. It was also Barack Obama's, from 2012, bent by one word. The machine hadn't thought it. It had caught it, twisted it, and handed it to me as if it were new.

I laughed. Then I stopped laughing, because I do the same thing all day long. So do you.

Most of what you think, you caught

Pick an opinion you hold strongly. Now try to say where it came from.

A teacher, maybe. A book you half remember. A friend who said it with conviction at the right moment. A headline. A podcast you were listening to while doing something else. Very few people can trace it back to a moment where they sat down and reasoned it out from nothing.

That's not a flaw in you. It's how minds work. Ideas come in from outside; they stick, or they don't, and the ones that stick start to feel like furniture you picked out yourself.

What Dawkins actually meant

In 1976, Richard Dawkins gave this a name. A meme, in his original sense, wasn't a captioned cat. It was any idea that copies itself from mind to mind the way a gene copies itself from body to body: a tune, a slogan, a way of building arches, a belief about the afterlife.

The unsettling part was the logic underneath. Memes don't spread because they're true, or good for you. They spread because they're good at getting copied. The ideas you've met are, by definition, the ones that were best at reaching you.

Author and carrier

You don't just copy, though. You collide things. A phrase from one place meets a frame from another, and something new comes out. Human history is a long chain of exactly this. Newton said he saw further by standing on the shoulders of giants, and even that line was borrowed from a twelfth-century scholar.

That recombination is real. It's how ideas evolve. A remix is a mutation, and mutation is how anything new gets made.

And then comes the urge. You want to post it, pitch it, explain it at dinner. It feels like pride of authorship, and it is. It's also precisely what a replicator would want you to feel.

You feel like the author. You're also the carrier.

Weapons and flashlights

Ideas earn their copies in different ways.

Some are weapons. You pick them up because they win. Balaji Srinivasan, whose Network School just bought Farcaster, recently described a whole class of them as academic theories shrunk down for phones and sharpened into blades for winning arguments. Arguments are everywhere, so weapons never run short of places to go. That's why the feed feels like a knife fight.

Others are flashlights. "Impostor syndrome." "Technical debt." "Too many cooks." They name something you already felt and couldn't say. Repeating one doesn't win you anything except looking like someone who sees clearly.

The same idea can be either. Ask any security guard with a six-cell Maglite: a flashlight becomes a club the moment you swing it. "Gaslighting" gave people a word for something done to them. Now it often gets swung to end an argument. Same word, different grip.

Human slop

Everyone is worried about AI slop right now. The bland, frictionless content flooding every feed, written by machines that have nothing to say.

But open LinkedIn. Read a corporate all-hands memo. Sit through a keynote. We were drowning in slop long before any machine could write a sentence. Recycled takes, borrowed frameworks, the same five motivational quotes, a whole economy of people saying what everyone else is saying, slightly louder.

Slop isn't about who made it. Slop is a copy that added nothing. A remix changes the idea. Slop just passes it along with a new face on it.

The purest carrier

Which brings us back to the machine that handed me Obama's line.

An AI model is built from almost everything people have written down. Every idea anyone ever caught, compressed into one system. It isn't a new kind of thinker. It's the purest carrier ever built.

That makes it a mirror. When it produces slop, it's showing us the average of what we fed it. When it surprises us, it's doing what we do at our best: colliding two caught ideas into something neither held alone.

The question was never whether the thing writing is a human or a machine. It's whether anything got added on the way through.

You didn't think that

Go back to the line the machine handed me. Obama said "you didn't build that" to argue that nobody succeeds alone: someone built the roads, someone taught you to read. A machine caught it, bent one word, and gave it back to me. I caught it again and built this essay around it.

Three carriers. Not one original thought among us. And yet every pass changed it. A campaign speech became a title. A title became an argument about where your thoughts come from.

That's the whole difference. Slop goes through you and comes out the same. Everything worth reading comes out changed.

You didn't think that. You never will. The only part that's yours is what it looks like when it leaves you.