Nobody Read the Warning
Evaluated Applications · Published
Every civilisation that collapsed left us a report. We set them all to music.
On this page
Here's a thing that happened, and I want you to sit with how strange it is before I tell you what I think it means.
Around 1177 BC, the eastern Mediterranean fell over. Not one kingdom — all of them, more or less at once. Mycenae, the Hittite empire, the Levantine trade cities, Egypt battered into a shadow of itself. Writing systems vanished. Linear B stopped being written and nobody could read it again for three thousand years. Trade networks that had run for centuries simply stopped, the palaces burned, and afterwards there was a period so dark that historians just call it that — the Dark Age.
The people who came out the other side were, overwhelmingly, the ones who'd never mattered. Hill villages. Herders. Populations too marginal for the palace economies to have bothered integrating. The great houses died with their records, and the nobodies inherited the ground. Then those nobodies wrote things down.
The pattern nobody points at
Once you notice it you can't stop noticing it: it's always the most sophisticated thing in the room that goes first. The most connected, the most efficiently organised — the empire with the standardised weights and the courier roads and the unified administration is the one that dies, and what walks out of the rubble is whatever was too scattered, too rural, too unimportant to have been plugged in. Slaves become masters, not by winning, just by still being there.
And every time, the survivors write an account of what happened. The account gets copied, the copies get venerated, and a few hundred years later the thing's become a hymn people sing without hearing. That last step is the one I actually want to argue about, because I think we've been handed the same report over and over, in text we have every word of, and the reason nobody's acted on it isn't that the message got lost. It's that it got preserved too well.
The tower story is a lab notebook
Everybody knows Babel. Nobody actually reads it.
Go back to the text, because it's doing something odd. God looks at the tower and says — close to a direct quote — the people is one, and they have all one language, and now nothing will be restrained from them. Read that cold and it isn't a complaint about arrogance, it's a statement about capability, and it names the source of the capability precisely: they're unified. One people, one language, everybody pulling the same direction with no friction between them.
Then look at what actually gets done about it. The tower isn't destroyed, nobody's killed, there's no flood or fire or smiting of any kind — their language gets confused and they get scattered. That's the strangest divine intervention in the whole book, and it's strange because it's so specific. If you asked an engineer today to design the minimum intervention that reduces a population's capability without harming a single person, you'd get more or less exactly that: break their ability to coordinate and spread them out.
Somebody watched a unified civilisation get catastrophically capable and then come apart, and wrote down what they saw — that the unity was what made it dangerous, and afterwards it was scattered and speaking different languages. They filed it under pride, which is fair enough, that's the vocabulary they had. But the observation is sitting right there, unedited, in one of the most widely distributed texts in human history, and it says coupling was the cause. We've been singing it for three thousand years.
Why this keeps happening, in one formula
Here's the bit that actually got me thinking about this, and it's just arithmetic — statisticians have known it since 1965. If you ask a thousand people a question, your answer isn't worth a thousand opinions, it's worth however many of them are thinking independently. Ask a thousand people who all read the same paper, went to the same school, and use the same three websites, and you haven't got a thousand data points, you've got maybe eleven.
Push the correlation all the way up, everybody wired into the same sources and tools and assumptions, and you get to the actual punchline: a million perfectly coordinated minds have roughly the decision-making power of one. Not "less than a million" — one, basically. Population stops mattering, and a civilisation running on one effective brain is one bad idea away from the floor no matter how many people are technically in it.
And here's the annoying part. What is progress, mechanically? Shared standards, shared infrastructure, faster communication, everyone reading the same thing at the same time — getting better at civilisation IS the process of becoming more correlated. The same thing that makes you powerful is what collapses your effective size. Which is why it's always the sophisticated one that dies first — not decadence, not moral rot, just that the empire with the best roads is the empire whose provinces all make the same mistake in the same week. The hill villages survive because nobody ever built them a road.
Distance used to save us for free
Obvious objection: Egypt lasted three thousand years, Imperial China was gigantic and standardised and unified and it's still here — if coupling kills, why aren't they dead?
Because until about a hundred and fifty years ago, nobody could actually coordinate anything. Ancient empires were centralised on paper and completely independent in practice — an order from the capital took months to reach the frontier, arrived garbled, and got implemented by a governor who was functionally making it up as he went. The emperor's unity was a legal fiction, every province was quietly running its own experiment and being wrong in its own local way, and that accidental diversity was doing a lot of work nobody paid for or even noticed. Geography was the immune system. Distance kept people independent for free, and we got rid of distance — which is mostly a good thing, and also removed the one thing that was automatically keeping our effective number above one.
The reports we can't read
So here's the actual thesis, the part I'd argue at a pub.
Every one of these collapses produced survivors, and the survivors wrote it down accurately, because they were there — what they left is forensic, a case file on what a civilisation did right before it stopped existing. But the survivors were, by definition, the ones who'd been peripheral to whatever failed. They saw the tower and the scattering; they couldn't possibly have seen the mechanism, because the mechanism is a statistical property of a system they were never inside. So they recorded the event perfectly and explained it wrong — pride, wickedness, forgetting the old ways — and the wrong explanation is just the container the right observation got shipped in.
Then the killer: what do you do with a text about the end of the world, written by the only people who lived through it? You make it sacred. And sacred texts get recited, perfectly, in unison, for centuries — word-for-word transmission with essentially zero loss and essentially zero comprehension, because reciting something is the opposite of reading it. Ask a congregation to say a passage and every one of them nails it. Ask what it means and you get the explanation from the container, not the content. Veneration is the best preservation method humans have ever invented, and the best concealment method, at the same time, using the same mechanism. We didn't lose the warnings — we put them somewhere so safe nobody's opened one in three thousand years.
The one that got through
There's a single exception I know of, and it's my favourite thing in this whole argument.
In the Talmud, a capital case before the Sanhedrin needed a majority to convict — but if the vote came back unanimous, all seventy-one judges agreeing the man was guilty, he walked free. Think about how mad that sounds: your strongest possible evidence gets thrown out for being too strong. Except it's exactly right, and it's the same formula from earlier applied to a jury. Seventy-one judges from one tradition, in one room, hearing one presentation, are already heavily correlated — total unanimity is the warning light telling you the correlation's hit its ceiling, that you don't actually have seventy-one opinions, you have one opinion held seventy-one times, and you don't hang a man on one opinion. That's 1965's statistics sitting in a law book from the second century, still usable as-is.
My guess for why this one survived intact is that it came out of a disaster small enough to leave behind a procedure instead of a scripture — a court that convicted an innocent man, a scandal, a rule. Big enough to hurt, too small to become a hymn. The catastrophic ones don't leave you rules, they leave you stories, and stories get sung.
And now the uncomfortable paragraph
I don't think we can measure ourselves against any of this, because we haven't been tested yet. Every civilisation in the evidence pile has finished its trial; ours is still running, and pointing at our own institutions and saying "look, we're fine" is a bit like a man halfway down from a building saying he feels great. But I'll say what the model predicts anyway, so it's on the record and you can check it later.
We're running the largest coupling experiment in the species' history — an enormous number of people routing an ever-larger share of their thinking through an ever-smaller number of shared systems, same platforms, same rankings, same handful of models. Each individual gets sharper. The population's effective size falls anyway, because mistakes that used to be scattered and cancel each other out now get made once, at the source, and handed to everyone at the same time.
The symptom won't be stupidity — everyone will seem cleverer than ever, and will be. The symptom is that disagreement stops telling you anything. Two people arguing used to triangulate something, because they'd gotten wrong in different ways from different places; when everyone's inputs come from the same upstream, the disagreement's decorative, and you can have a furious national argument where the effective number of positions in the room is one.
If that's actually happening, the fix isn't better tools. It's deliberately keeping some people and some methods and some sources unplugged — not because they're better, but because they're wrong differently, and that difference is the only thing that's ever caught an error the whole system was making together. Which is, more or less, what a court in the second century was doing when it let a man walk because everybody agreed with each other.
The statistics here are real and kind of boring and you can look them up yourself — Leslie Kish, design effect, 1965. The history's real too. The bit joining them together is mine, and it's a hunch rather than a finding — if I'm wrong about it, worst case you just end up reading Genesis 11 more carefully than you did before.