In March 1996, a Stanford graduate student named Larry Page wrote a piece of software that crawled across the internet. He called it BackRub. It started at a single page, the Stanford computer science department homepage, and followed every link outward, and then every link on every page those links reached, mapping the whole tangle of connections. Page was chasing one idea, and it would shortly make him one of the richest people who has ever lived. He had realised that a link was a kind of judgement.
Our Larry
When one page links to another, a human being has decided that the second page is worth pointing to. So if you could count the links, weighting each one by the importance of the page it came from, you could measure what the collective judgment of millions of people held to be worth looking at. He and Sergey Brin called the method PageRank, and it became the engine of Google.
There is a deeper thing buried in that idea, and it is worth excavating, because everything that has happened since depends on it. PageRank was built on a proposition that would have been recognised, and cheered, by the philosophers of the eighteenth century: that no single authority should be trusted to tell you what is true, and that the better guide is the distributed judgment of a multitude of free individuals, each deciding for themselves what is worth their attention.
Google did not hand down verdicts. It tallied millions of small human choices and showed you the standings. Time magazine, at the time, called it the first search engine to treat the internet as a democracy. The most pointed-to pages won the ballot and rose. And then, crucially, it stopped, and left the last step to you.
That is what the ten blue links really were. Not a piece of technology, but a small daily ritual of judgment. You typed a question, and the machine handed you a ballot: here are the ten pages the crowd rates highest, now decide. You opened three of them. You distrusted one.
You cross-checked another. You clocked that the third was trying to sell you something and you weighted it accordingly, half-consciously, the way you learn to read a market trader's patter. You reached your own conclusion. It took ninety seconds, and a small effort of discrimination, and you resented the effort, because the page was cluttered and some of the links were junk. But the effort was the thing. The effort was a person practising making up their own mind many times a day.
In 2022, ChatGPT arrived, and within three years, the ritual inverted. Ask a question now, more and more, you are not handed a ballot. You are handed the verdict. Google prints an AI summary above a growing share of its results; millions of people skip the search box altogether and put the question straight to a machine that answers in a single, calm, complete paragraph. The Pew Research Centre properly measured what this does to us, tracking nearly 69,000 real searches from 900 real people rather than asking them to describe themselves. When the AI summary appears, people click through to an actual source 8% of the time, against 15% when it does not. They click a link inside the summary 1% of the time. And in 26% of cases, they read the summary and stop there, the session over: no source opened, nothing compared, nothing checked. More than a quarter of the time, the answer is received, and the matter is closed.
Look at what has been reversed. PageRank counted the votes and showed you the tally so that you could judge. The machine now reads the same web and returns the judgment already reached, in a voice with no visible author. Brin said in 2000 that he dreamed Google would one day "return answers, not just documents," that it would become "your interface to all the world's knowledge."
He got his wish exactly, and the wish contained the trap. A machine that returns documents leaves the judging to you. A machine that returns answers has taken the judging away, and done it so smoothly that it feels like a gift.
And it is a gift, in the way the most expensive things are. When ten links tried to sell you something, you could see the selling. The single answer removes the seam. It arrives sourceless and serene, in the voice of the machine itself, and it does not read as a recommendation because it wears none of the marks of one. But it was assembled from somewhere, and what these systems lean on, over and over, is not a company's own page but the third-party sites they treat as neutral: the review pages, the comparison tables, the lists of the ten best.
A great many of those are paid by commission, on what they recommend. Their whole purpose is to be the trusted source the machine quotes, and to be paid when the trust pays off. So the calm verdict, this is the one to buy, reaches you having passed through a layer of commerce with every fingerprint wiped from it. You are no longer able to judge the seller, because you can no longer see that there is one.
This does not stop at shopping, or at search, because search is only the most recent thing to make the move. The same inversion has already run through most of modern life, everywhere a recommendation has stepped in front of a choice. The record shop, where you flicked through the racks and gambled on a cover, became the app that assembles the playlist for you. The map, which laid the whole city out before you and made you work out where you stood in it, became the single blue line you followed without ever learning how anything connected.
The room full of strangers you might have talked to became a deck of profiles pre-sorted by a machine. Each time, the system optimises for the satisfaction it can measure, the click, the play, the swipe, and discards the serendipity it cannot: the wrong turn down a street you come to love, the dreadful record that becomes yours for life, the person you would never have picked who becomes the reason you would congratulate God. The people who study these systems report the same finding across all of them. Inside the curated feed, satisfaction scores rise while people describe a growing sameness, a blandness they can feel and cannot name. The machine keeps handing us what we asked for, and a little less than we wanted, every single time.
That gap, between what we ask for and what we want, is the whole of the matter, and someone described its mechanism with complete precision two hundred and forty years ago. In 1784, Immanuel Kant tried to define what the Enlightenment actually was, and landed on this: it was mankind's emergence from "self-imposed immaturity," which he defined as "the inability to use one's own understanding without the guidance of another." Self-imposed, he stressed, because the cause was not stupidity but something more ordinary, "the lack of resolve and courage" to think without a guide. The whole modern world, in his telling, was the long project of daring to know, of throwing off the authorities who had always been so happy to do your thinking for you. And then Kant gave an example of what surrendered judgment looks like in practice, and reading it now is a small shock, because he is describing a product that would not be built for another two and a half centuries:
"If I have a book to have understanding in place of me, a spiritual adviser to have a conscience for me, a doctor to judge my diet for me, I need not make any efforts at all. I need not think, if only I can pay: others will readily undertake the irksome work for me."
I need not think, if only I can pay. That is the subscription. Kant named the precise transaction, and the two things that drive us into it, laziness and cowardice, and warned that the danger of the arrangement was not only that your thinking was done for you but that, in time, you lost the capacity to do it at all.
The person who is never allowed to make the attempt, he wrote, becomes "really incapable of using his own understanding, because he was never permitted to make the attempt," and worse, "grows fond" of the guardians who spared him the trouble. This is the part that the cheerful language of convenience is built to hide. A faculty that is not exercised does not stay intact and unused, waiting to be picked up again. It wastes. Every question that ends at the summary is one more repetition of a muscle going slack, and a civilisation is only ever the sum of the faculties its people still bother to use.
None of this is fated, and the reason it isn't is just how people are. We have handed tasks to machines before and, on finding the machine's version subtly worse than the thing we wanted, taken them back. Convenience seduces, and then it cloys. You can watch the counter-current running now, in the return of the independent bookshop, the fatigue with the algorithmic feed, the deliberate small turn back towards the thing you had to look for and the human who pointed you to it. The instinct to think for oneself is stubborn, and it has survived worse than this.
Kant, though, saw the catch in that hope, and it is the hardest thing in his essay. Immaturity is comfortable. It is convenient to be told. And the faculty you would need in order to notice that your judgment has withered is the very faculty that is withering. You can rail against a bland playlist because you can still feel the blandness. It is a great deal harder to rail against the slow loss of the habit of finding out, because the loss erases the thing that would have sounded the alarm.
Larry Page built a machine to count what millions of free people judged worth seeing, and to show you the count so that you could make up your own mind. A quarter of a century later, the thing has become increasingly to make up your mind for you, in one voice, from no visible pulpit, and does not show the count at all. The Enlightenment was the long, difficult business of learning not to accept the single answer handed down from on high. We appear to be paying, monthly, to have it back.
Kant's motto for the whole project was two words, borrowed from Horace. Sapere aude. Dare to know. It was never an instruction to acquire the answer. It was an instruction to do the work of reaching it yourself.
The great awakenings of the Western mind, the Renaissance and the Enlightenment, were both, at bottom, the same act: people deciding to see and judge the world for themselves rather than take the answer handed down to them. It would be a strange kind of progress if the next age were remembered as the one in which, offered the answer to everything and the effort of nothing, we chose to hand the faculty back.
See you on the next one.
Renascentia secunda: the second renaissance.
