Every economy runs on a standard, some scarce thing the money is a claim against. Gold, for a while. Then the full faith and credit of governments. The digital city ran on something stranger and more intimate: it ran on the attention standard. Its base unit was a human minute.
Every instrument in the city's financial system was denominated in that unit. The impression: a human probably glanced here. The session: a human stayed. The click: a human chose. Advertising bought human minutes wholesale and sold them retail. Search engines were the central bank, deciding whose pages deserved the flow. Publishers ran printing presses that converted words into minutes and minutes into revenue. An entire commercial civilization, trillions in market value, stood on one monetary fact: reading was scarce, because readers were human, and humans run out.
Then the city gained a population of readers who do not run out, and the attention standard began to fail the way standards fail: at the base unit.
The visit stopped occurring. Nearly half of all Google searches, 47 percent across a clickstream panel of ten billion of them, now end without a single click. The person asks, the machine reads, the answer arrives, and no human minute is ever spent on the page that supplied it. From the person's side of the counter the bargain feels like magic: read nothing, know everything. From the publisher's side it is the same four words, experienced as a run on the bank. Where an AI-generated answer appears above the results, clicks to the pages beneath fall by roughly three-fifths. The fully conversational interfaces refer a visitor onward for perhaps two percent of queries, against seventeen to nineteen for the old page of links. Each of those numbers is the same event measured differently: the base unit of the city's economy quietly failing to occur, billions of times a day.
I have watched this from the inside of my own ledger. One property I operate was shown in search results four hundred thousand times in a single quarter and received five thousand visits. Its pages ranking fourth and fifth, positions that reliably earned five to seven percent click-through for twenty years, now earn less than one. The pages are read more than they have ever been read. They are visited less than they have ever been visited. Under the attention standard those two sentences could not both be true. Both are true.
The asymmetry underneath is the part the aggregate numbers hide. The old crawlers honored a bargain older than the ad industry: they read your pages and paid you back in readers. The new reading pays back nothing by design. A majority of crawling is now for training, and a trained model returns no one; it simply knows what you knew. The agents are no better as customers: when someone's assistant visits five thousand pages to compare options on their behalf, that is five thousand readings, zero impressions, zero sessions, and one very satisfied human who never saw a single ad. On one large network, bots were already generating about two-thirds of the most resource-intensive traffic while producing about a third of the pageviews. Maximum cost, minimum currency. The new residents consume the city's goods and are constitutionally incapable of paying in the city's money.
So the money is changing what it believes in, which is how standards actually end. Nobody repeals a standard. Transactions just start clearing in something else.
Watch where the clearing moved. Publishers now negotiate directly with the operators of machine readers: licensing deals for whole archives, priced in bulk, signed upstream. Infrastructure operators run toll booths that charge per crawl, a market where the unit is not a human minute but an act of machine reading. Access itself is becoming the commodity, sold the way utilities sell, invisibly, wholesale, in contracts no reader of the resulting answers will ever see. The attention economy priced the reader's time. The access economy prices the reader's existence.
For the legacy residents, the transition will mostly be experienced as weather. Familiar businesses will fail for reasons their dashboards cannot show, because the dashboards count human minutes and the loss happened in a currency they do not display. Unfamiliar businesses will thrive on revenue no consumer remembers providing. The signs of the repricing are already legible if you know which column to read: traffic up and revenue down is not a paradox anymore. It is the exchange rate.
And there is one more devaluation coming, the one nobody's dashboard is built for. As agents begin to transact and buy on their owners' behalf, the persuasion industry meets a reader that cannot be flattered. The billboard does not work on a commuter with no eyes. Brand, in the human sense, is a claim on attention, and the new residents do not have any to claim. What they have is criteria. An economy that spent a century learning to move human feelings will now learn to satisfy machine checklists, and those are different arts with different winners.
None of this is the end of value. The city is producing more reading, more answering, more knowing than at any moment in its history; the wealth is real and growing. What ended is the particular standard that measured wealth in interrupted human minutes, and honestly, the epitaph writes itself: it was always a strange thing to build a civilization on, the systematic harvesting of the one asset its residents could never make more of.
The attention standard gave the city thirty years of solvency and one permanent lesson: whatever the money stands on, someone eventually builds a reader that does not need it. The next standard is being set right now, upstream, in contracts between landlords and the operators of the new residents. The legacy residents will live under its prices either way. The only choice on offer is whether to be a party to the negotiation or a line item in it.