SPREAD
How Information Learns to Move
I keep noticing the same three pieces in everything that spreads.
Payload. Replication. Routing. Every time I look, it's these three.
In 1971, a program called Creeper displayed a message and hopped to the next machine. Nobody told it to. It had a payload (the message), a copy mechanism (self-replication), and an address book (the ARPANET).
By 2003, SQL Slammer infected 75,000 hosts in ten minutes. Same three components. The payload shrank to 376 bytes. The replication got faster. The routing got random.
Code was doing distribution before anyone called it that.
Before writing, before printing, stories spread by mouth. The ones that survived weren't the most detailed — they were the most retellable. Structured for the copy mechanism: a human voice. Routed by the only network available: whoever was sitting around the fire.
The Odyssey survived 2,700 years not because it was good. Because it was built for how people actually remember things — rhythm, repetition, a skeleton you can't forget. The details drifted across every retelling. The structure didn't. That's what "retellable" actually means.
I was reading Ogilvy and it hit me. He called it a headline — that's the payload. He called it placement — that's the routing. Different words, same thing.
"On the average, five times as many people read the headline as read the body copy."
— David OgilvySame pattern again. If people scan, the first line has to survive scanning. If people share, the first line has to survive sharing. If an AI agent is reading your README, the first paragraph has to be useful or it moves on.
The best headline works like SQL Slammer — small enough to fit the transmission medium, potent enough to change what happens next.
This one took me a minute. The new reader has a context window of 1M tokens and no patience. It doesn't skim — it ingests. It doesn't bookmark — it either loads you into context or it doesn't. So the question becomes: did you fit in the window, and were you useful once you got there?
A good README gets ingested into a context window. It changes what the agent does next. If it's useful, the agent recommends it to the human — who puts it in more context windows. Same three pieces again. The payload is the docs. The replication is usefulness. The routing is whether the agent can find you at all.
I didn't plan this table — I just started listing things that spread and the columns kept being the same. The meme row is the most obvious: it doesn't pretend to be anything other than a joke that mutates on copy. But look at any row. Same three pieces. Different targets, same structure.
I keep expecting this site to have the reach of something that's been around for years. It's been months.
The tools are new. The platforms are new. The audience is new. A six-month-old blog doesn't have the distribution of a six-year-old newsletter. That's not a failure — that's just how time works.
"Working with AI is parenthood training. Can't ask a 1 year old to do things a 5yr old can. You have to gradually increase responsibilities."
— something I wrote about patience with agents. Applies here too.There's a fourth thing to be patient with, beyond agents and humans and yourself: the audience that doesn't exist yet.
You write the thing. You put it somewhere. You choose who sees it first. And then you wait — because the audience is brand new and you're expecting it to already care.
I can see the pattern now. Worms had it. Stories had it. Ads had it. Three components, every time. Cool.
None of that helps me get anyone to read this.
The writing might be fine. The replication mechanism is "post it and see." The routing is "whatever the algorithm decides." That's not a distribution strategy. That's a lottery ticket.
Noticing a pattern doesn't mean you can use it. Knowing how a worm spreads doesn't make you contagious. The question is still: did anyone actually read it?
I can see how every piece connects.
I still can't make it go.
Worms didn't read about spreading. They spread. Stories didn't theorize about retelling. They got retold. Ogilvy didn't admire distribution. He bought the placement. I'm still in the reading-about-it phase.
Stop hoping, start building
"Post and hope" is not replication. I need to make things people forward because the structure demands it — not because I asked nicely.
Pick where it enters the network
Random scanning worked for worms because they had infinite patience. I don't. I need to choose where it lands.
Make the shareable part smaller
SQL Slammer was 376 bytes. The Odyssey was built on six-beat lines. Ogilvy spent weeks on a headline. The thing that actually spreads is never the whole thing.
Give it time
This site is months old. I keep comparing it to things that have been around for years. That's not strategy, that's impatience.
Actually do it
I keep studying the pattern instead of running it. This is the step I keep skipping.
I keep finding the same three pieces in everything that moves.
I still haven't figured out my own.
Working on it.