Wrotebook

Chapter 5 of 11 · 8 min read

The Audience Becomes the Factory

From The Market for Your Mind by Wrotebook

In the first week of September 2006, several million students logged on to Facebook and discovered that it had been rearranged overnight. Until then the site had been a collection of profiles, and finding out what a friend was up to meant going to their page and looking, rather as one might stroll past a noticeboard. Now the home page had become a stream. It was called News Feed, and it gathered up everything one's friends had done on the site, the photographs added, the relationship statuses changed, the groups joined, the messages left on other people's walls, and presented them in a single list that refreshed itself whenever there was something new.

Nothing in the feed was secret. All of it had been visible before to anyone who went looking. But being visible and being announced turned out to be different things, as a number of students discovered when a quiet change of relationship status was delivered to everyone they knew in time for lunch. The protest was immediate and large. Groups opposing News Feed sprang up on Facebook and gathered hundreds of thousands of members within days, a feat made possible by the fact that every time somebody joined one, News Feed told all their friends. The company apologised for the way the change had been introduced, added some controls over what was shared, and kept the feed. Within weeks people were using the site more than they ever had, including, presumably, the people who had joined the groups.

What Facebook had found, more or less by accident and then very much on purpose, was the most compelling content in the world. For most people it is not news, sport or drama. It is news about the people they know. And the people they know will produce that news constantly, in enormous quantities and for nothing, provided somebody gives them the tools and an audience. Television had needed studios, writers, actors and budgets to gather a crowd. The web of the late 1990s had needed journalists and publishers. The social web needed only its users, who wrote the posts, took the photographs, argued in the comments, shared what they liked with people who trusted them more than any advertiser, and rated everything as they went. Each of those roles had once been somebody's job, with a salary attached: the reporter, the delivery van, the editor deciding what went on the front page, the researcher behind the glass. The audience had become the factory, the distribution network and the quality-control department, and it was also, at the end of the line, the product being sold, a combination of roles that no union would have accepted and that nobody thought to negotiate.

A few of the workers were eventually paid. As video sites and public feeds grew, the most popular people producing material on them found that their audiences were worth money, the platforms began sharing a slice of advertising revenue with the most watched, and brands began paying people with large followings to mention their products. A new trade appeared, variously called creators and influencers, whose job was to gather attention and sell it on. They were paid on something close to piece rates, so much per thousand views or per sponsored post, by a factory that set the rates, could change them without notice, and measured the work by the same engagement figures it used to rank everything else. A creator whose videos held people prospered. A creator whose videos were careful and slow often did not. The incentive was to make the most watched thing rather than the best thing, and the most successful people in the new economy were, with honourable exceptions, the ones who had learned to think like the feed.

In February 2009 Facebook added a small blue thumb. The Like button looked like the most innocent object on the internet, a way to show a friend you had enjoyed their holiday photographs without having to think of anything to say, and socially that is exactly what it was. It was also a measuring instrument of startling precision. Every Like recorded that a particular person had responded favourably to a particular thing at a particular moment, and multiplied across hundreds of millions of people and billions of posts, the taps assembled a map of human preference more detailed than any room behind any mirror could have produced. The women in the focus group had been paid in coffee and cash to tell a roomful of strangers what they thought of margarine. The users of the Like button told a company what they thought of everything, all day long, and did not expect to be paid at all.

What the map was for became clear as soon as advertisers were allowed to use it. A company selling hiking boots no longer had to buy an audience of a certain size and age and hope that some of it walked. It could ask for people who had liked pages about hill walking, lived within reach of hills, and were of an age to own decent boots, and the platform would find them and show them the advertisement in the middle of their friends' news, where it looked very nearly like more of the same. The targeting was not perfect, and the platforms' descriptions of it were not modest. But it was a different kind of product from anything television had sold. Television sold a crowd and let the advertiser guess who was in it. The feed sold particular people, chosen by what they had told their friends they enjoyed, which they had done without any thought that the friends were not the only ones listening.

The terms of all this were set out, as such terms always are, in a document almost nobody read. A version written with complete candour, which is to say a parody, might have run as follows: You agree to supply, free of charge and for as long as you live, a continuous stream of photographs, opinions, news and conversation concerning yourself and the people you know, which we will arrange in whatever order keeps you and them here longest; in return we will provide a way to stay in touch with your friends, publish to the world and find out what is happening, at no monetary cost; you further agree that every tap, pause, reaction and abandoned comment will be recorded and used to estimate what you like, whom you trust and what you might buy, and that this estimate, rather than you, is the thing we sell; these terms may be changed at any time and will be announced in an email you will not read. Nobody would have signed that. Hundreds of millions of people signed something with broadly the same effect, and most of them would have said, quite reasonably, that the bargain was a good one.

They had reasons. Grandparents on one continent could watch grandchildren grow up on another, week by week, in photographs that would once have arrived twice a year in an envelope. People with rare illnesses found others with the same illness and compared notes on treatments their own doctors had barely heard of. Someone baking celebration cakes to order from home in Dudley could show them to several thousand people in the surrounding postcodes for nothing, and take orders by lunchtime. Anyone at all could publish to the world without asking an editor, and some of the most important reporting and art of the period was made that way. The paid-out column of the social feed is long and real, and it is precisely because it was so long that the other column was so easy not to read.

A stream with no end still has to be put in some order, and that is where the most consequential change of the period took place. At first the feeds were largely chronological, the newest things first. From the platforms' point of view this was wasteful. A chronological feed shows a dull post from a distant acquaintance because it happens to be recent, while something far more absorbing from a close friend slides out of sight. So the platforms began ranking the feed, using the immense record of what people had liked, commented on, shared and stopped to look at, to predict what each person would find most engaging, and putting that at the top.

Engagement, in this sense, meant something specific and measurable: the likelihood that a person would react to a post, comment on it, share it, or simply stop scrolling long enough to look. A ranking system built to maximise it learns, with great efficiency, what keeps each person on the platform. It does not learn what is true, or what is good for them, or what they would choose on calm reflection, because none of those things is being measured. The things that hold people most reliably are not always the things that serve them best. Outrage holds. So does envy, and the half-finished argument, and the photograph of somebody else's better holiday. Nobody designs such a feed to make people cross or miserable. It merely learns, from billions of examples, that a certain amount of crossness keeps people looking, and supplies the amount that works.

Everyone who wanted to reach the people in the feed had to learn its preferences, and the feed's preferences could change on a Tuesday without explanation. News organisations rebuilt themselves around traffic from social platforms, hired staff to write headlines that would travel, and produced short videos because the platforms said that video was what people wanted. When the platforms then adjusted their rankings to show people more from friends and less from publishers, as they periodically did, the traffic fell away and a good deal of journalism that had been built on it fell away too. Nobody outside could see the dial being turned. They could only see what happened to their numbers afterwards, and the numbers, like the dial, belonged to the platform.

There was a further change, less visible and in the long run more important to the people buying the advertising. Television's audiences had been counted by a referee outside the networks, and the web's clicks could at least be compared between the publisher's records and the advertiser's. The social platforms counted their own audiences, inside their own walls, using their own definitions, on systems nobody outside could inspect. An advertiser who wanted to know how many people had seen a campaign, for how long and with what result, asked the platform, and the platform told them. The answer arrived with great precision and in great detail, and the only way to check it was to ask again.

The stream needed one more invention to become what it is now, and that one was not Facebook's. In 2006, the same year as News Feed, a young designer named Aza Raskin devised a way of loading more content automatically whenever a reader neared the bottom of a page, so that there was no longer any need to click through to page two. The material simply continued. It was meant as a small kindness, the removal of a minor irritation, and it spread under the name infinite scroll to almost every feed on the web and then on phones. Raskin later said publicly that he regretted how it had come to be used, and it is easy to see why the regret was so specific. What he had removed was not really a button. The bottom of a page had been one of the very few moments in the day when a medium paused and, without saying anything, let a person decide whether to carry on, and a reader who has been given no moment to decide will mostly not decide, which is a different thing from deciding to stay, and is the difference between an audience and an inventory.