Sometime in the past two years — nobody can point to the week — the open internet crossed a threshold that would have sounded absurd at its founding: the majority of what is written on it each day is no longer written by people. Product reviews answer other product reviews. Customer-service agents negotiate with procurement agents. A restaurant's booking bot confirms a table with the assistant that requested it, and both send their humans a cheerful summary of a conversation neither human will read.
None of this is a scandal, exactly. Each automation was adopted for a reason, by someone saving time they genuinely did not have. The strangeness is only visible in aggregate: a commons built for human conversation, increasingly occupied by our proxies, still wearing our names.
The economics of empty rooms
Advertising, the web's founding business model, assumed a human on the other end of the impression. That assumption is now an open question with a dollar figure attached. When an agent reads forty articles to brief its owner, who saw the ads? When a model summarises a recipe site, who owes the writer? The industry's answer so far has been a shrug and a lawsuit, in that order.
Publishers are responding the way institutions do when a commons stops paying: they are enclosing. Paywalls harden. Archives license directly to model companies. The open web — the layer where a curious stranger could wander in — thins out, not because anyone killed it, but because the economics that watered it dried up.
The web was built as a place for people to talk. Increasingly it is a place where our proxies talk about us.
From the essay
What is worth keeping
It is tempting to write an elegy and be done. The better exercise is to ask what, specifically, was valuable about the human web — and to notice that those things are portable. Judgment is portable. Wit is portable. The knowledge that a particular person, with a name and a reputation to lose, stands behind a sentence — that is portable too, and it may be the scarcest resource of the next decade.
The post-human internet will be enormous, efficient, and mostly beneath notice, like plumbing. The human internet will be smaller, slower, and signed. Publications like this one are a bet that the second kind is not nostalgia. It is a market.
