cryptopath on Nostr: Information Wants to Be Paid Stewart Brand said information wants to be free. He also ...

Information Wants to Be Paid

Stewart Brand said information wants to be free. He also said, in the same breath, that it wants to be expensive — and he didn't think he was contradicting himself. The internet has spent forty years failing to resolve that paradox. AI is about to force the answer.

In 1984, at a gathering of hackers, Stewart Brand said the line that became the whole internet's founding scripture: information wants to be free. The hackers took it literally, and a movement was built on it.

What almost everyone forgets is the sentence he said right next to it: information wants to be expensive. And Brand did not think he was contradicting himself.

Information wants to be free, he explained, because it's so easy to copy and move that the price tag falls off — a copy costs essentially nothing, so the cost races toward zero. But it also wants to be expensive, because the right piece of news or data or advice, at the right moment, can be beyond price. Both are true at once. The tension between them, Brand predicted back in 1988, would never go away — it would only get worse with every new device.

He was right. And the newest device wants to swallow all the information there is. Let me trace why that breaks something, and what the only real fix turns out to be.

The tension that never resolved

Watch the paradox play out across forty years and you see the same fight, over and over, with the names swapped.

Napster versus the music industry. The web versus newspapers. BitTorrent versus the studios. Linux versus the software giants. Each one was the same collision: information that was valuable meeting a world where it was suddenly free to copy and move. And each time, the free-to-copy side won on the technology, so the valuable side had to invent a new way to get paid. Spotify. Netflix. Paywalls. Software as a service.

Notice what every one of those "solutions" actually is: a workaround for a missing piece. None of them priced the information itself. They priced access to it, or convenience, or wrapped it in a subscription — because there was never a way to charge for the thing directly. The copy was free; the only question was where to build the tollbooth around it.

Which points at the real hole, the one from an earlier essay: the internet could move information brilliantly and could never move value. It had no native way to put a price on a thing at the moment someone actually used it. So the whole history of digital media is improvised tollbooths bolted onto a road that was built with no way to charge for the trip.

The truce was attention — and AI just broke it

The biggest improvised tollbooth of all was advertising. It let information be free to humans while capturing the value somewhere else — in your attention, sold to advertisers. Publishers, creators, retailers all ran the same deal: give the information away to people, get paid by harvesting the eyeballs that came to read it. An uneasy truce between free and expensive, held together by ads.

AI breaks that truce cleanly, and here's the mechanism. A large language model needs an almost unimaginable amount of text to train — by one founder's estimate, more than a hundred times every book ever published. The only place that much information exists is the open web. So the labs send bots to harvest it.

But when a bot takes the information and feeds it to a model that then answers the human directly, the attention never reaches the creator. The eyeballs don't arrive. The ad never loads. The tollbooth is bypassed entirely. The information was "free to humans" only because the attention it generated was being sold — and AI consumes the information while routing the attention away. It takes the free side of the bargain and skips the expensive side.

Which is why publishers started blocking the bots. And why a whole business sprang up to un-block them — protocols that disguise the labs' bots by routing them through ordinary people's home internet connections, borrowing a slice of your bandwidth so the scraping looks like it's coming from a residential address that no one blocks. It's the FedEx driver asking to borrow your car for the delivery while his own truck idles at the curb. And its founder, naturally, waves the old cypherpunk flag over it: let's free the internet.

That's not freeing anything. It's the free-to-copy side of Brand's paradox, weaponized, sneaking past the valuable side's only defense. It's a symptom — and the symptom tells you exactly what's missing.

Information wants to be paid

Here's the resolution, and it's the one Brand's paradox was quietly pointing at the whole time. Information doesn't want to be free or expensive. Those were never the only two options. Information wants to be paid — to move as freely as copying allows, and to carry a price at the point where it delivers real value.

Both of Brand's halves come true at once, with no contradiction, the moment you can price a thing at the instant of use. The copy stays free to make and move. The use — the right data at the right moment, the thing that's beyond price — is where the value gets captured. You stop building tollbooths around the road and start charging, tiny amounts, for the actual trips.

And that requires the thing the internet never had: a native money layer, able to move value in amounts small enough to match a single query. Which is exactly where the machine economy was always headed. An AI agent that needs data doesn't need to sneak through your bandwidth disguised as a human. It needs to pay — a fraction of a cent per token of training data, streamed directly to whoever produced it, settled by the kind of micropayment rail the ad-funded web could never build. The labs scrape because paying was impossible, not because paying was refused. Make paying possible, at the right granularity, and the entire disguise-the-bot industry has no reason to exist. Information flows freely; value flows back. Free and expensive, finally reconciled by a price.

The honest edge

Two cautions, because this doesn't resolve into a tidy utopia.

The tell that we're at the breaking point is that the labs have been forced to argue both sides at once. They claim the freedom to train on everyone else's information scraped from the open web — and simultaneously insist that competitors must not be free to train on the output of their models. Information wants to be free when it's someone else's, and expensive when it's theirs. They might even be right both times. But a position that depends on the fence running exactly around your own property is not a principle. It's just power, and it shows the old truce has collapsed with nothing yet built to replace it.

And a price layer doesn't automatically mean the creator wins. Build the rail badly, or let the labs own it, and "pay for data" could just mean a handful of giants paying pennies they set themselves, to a world of creators with no leverage to refuse. A market needs both sides to be able to say no. The missing money layer is the necessary fix. It is not, by itself, the just one.

The one line to keep

Information wants to be free, because a copy costs nothing. And information wants to be expensive, because the right piece at the right time is beyond price. Brand was right about both, and the internet spent forty years unable to hold them together — because it could move information and never move value, so it papered the gap with ads and called the attention it sold "free."

AI tears that paper. It takes the information and routes the attention away, and the sneaking-through-your-bandwidth business is the symptom of a web with no way to simply pay for what it takes.

So the resolution was never free versus expensive. It's that information wants to be paid — to flow freely and be priced where it's used. That needs a money layer the web was built without. And whether that layer, when it finally arrives, pays the creator or just launders the theft is the fight that comes next.

Information wants to be free. And expensive. It turns out what it wanted, all along, was a price.