Wrotebook

Chapter 2 of 11 · 9 min read

The Black Box

From The Market for Your Mind by Wrotebook

In 1942 Arthur C. Nielsen put a box beside the radio in about 800 American homes, and the box wrote down what the radio was doing. It recorded, on a slowly moving strip of paper tape, when the set was on and which station it was tuned to, and at intervals the tape was sent back to the company, read, and added up. It did not record whether anybody was listening, which turned out not to matter nearly as much as it should have.

The problem the box solved was an old one with a new urgency. A newspaper could tell an advertiser how many copies it had printed and roughly how many it had sold, and on that basis a page could be priced. Radio could tell an advertiser nothing of the kind. A programme went out into the air and disappeared into an unknown number of kitchens, and the industry's evidence for who had heard it consisted largely of letters, telephone calls and the opinions of the people who sold the time, who were not disinterested. A station that wanted to charge more for its evening hours had to assert that more people were listening, and an advertiser who doubted it had no means of checking. What Nielsen offered, from 1942, was a number that came from outside the room, produced by a machine, from a sample chosen by someone with no programmes to sell.

It was called the Audimeter, and when television arrived in American homes in large numbers it followed the aerial. From 1950 Nielsen adapted the device to the television set, and within a few years the figures that came out of it were the most consequential numbers in American broadcasting. A programme with good ratings could charge more for its advertising breaks; a programme with poor ones was cancelled, often quickly and occasionally in the middle of a story. Careers were made and ended on the morning the figures arrived, by people who had never met any of the households concerned and never would.

The box had an obvious blind spot. It knew that the set was on and which channel it was showing. It did not know whether the room was full or empty, whether the family was watching or eating or arguing, or whether the set had simply been left on by someone who had gone to bed. Above all it did not know who was there, which was increasingly the question advertisers wanted answered, because an advertisement for razor blades is largely wasted on a nine-year-old and one for breakfast cereal on her grandfather. So a second instrument was added, and it was considerably less precise than the first. A separate panel of households was sent small printed booklets and asked to write down, for a week, what they had watched, when, and which members of the family had been watching it.

The diary is a wonderful document, because it asks people to do something they are notoriously bad at, which is to give an honest account of how they spent their time. Anyone who has kept a food diary for a doctor or filled in a timesheet at work knows how it goes. The entries are made late, from memory, by whoever in the household has been landed with the job. Programmes that the family is slightly embarrassed to watch have a way of failing to appear. Programmes it believes it ought to watch turn up more often than the set could possibly have shown them. Children's viewing is estimated by parents who were in the kitchen, and the adults' viewing by whichever adult was still awake. Here is a page of the kind that was posted back in large numbers, invented for the purpose but faithful to the genre, for a Tuesday in the late 1950s in a house somewhere in Ohio:

Tuesday. 7.00–7.30 News — Dad. 7.30–8.00 (left blank). 8.00–9.00 the Western — Dad, Mom, Gary, Linda (Linda bed 8.30?). 9.00–9.30 Mom's programme — Mom. 9.30–10.00 set on, nobody really. 10.00–10.30 News — Dad (asleep?).

The question marks are the best thing on the page. They are the moment at which a household, having agreed to turn its evenings into data, finds its conscience getting in the way. And if the receipt had been pencilled into the margin, beside the blank half-hour, it would have read something like this: Paid out: a Western, the news twice, Mom's programme, company in the evening. Taken in: three and a half hours, the family's habits and who sat where. Measured: a set tuned, and the family's own account of itself. Left behind: a Tuesday that now exists chiefly as a row in somebody else's table.

The industry knew that its numbers were imperfect. What it learned, and what made the business, was that this hardly mattered. The value of a rating did not depend on its being true. It depended on its being agreed. As long as the networks that sold advertising time and the advertisers who bought it accepted the same figure as the basis for the price, the figure could do its work, and the market could function. In this respect a ratings point was less like a measurement than like a banknote, which is valuable not because of what it is made of but because everybody accepts that it is, and which would be no use at all if each shop decided for itself what it was worth. Nielsen's real product was not information about audiences. It was agreement about audiences, which is rarer and pays better.

It also made Nielsen the only party in the transaction that was paid either way. The networks subscribed to the figures because they needed to know what to charge. The advertisers and their agencies subscribed because they needed to know what they were paying for. A programme that collapsed in the ratings was a disaster for the network, the producers, the actors and occasionally the sponsor, and a perfectly ordinary week for the company that had reported the collapse. The referee was paid by both sides, set the terms on which the game was played, and could not lose. Within a generation it had become one of the most powerful institutions in American broadcasting without having made a single programme.

The arithmetic underneath all this was, in its way, the most remarkable part. A few hundred homes, and later a few thousand, stood in for a country of tens of millions of television sets. Statisticians can defend that, and did, at length and with justice: a properly chosen sample can tell you a good deal about a population many times its size, which is how opinion polls work and why nobody needs to taste every spoonful of the soup. The striking thing is not that the method worked but that everybody consented to it. Networks that disliked their figures complained, as networks always do, about the sample, the methodology and the weather; then they went on selling their time on the basis of the figures, because the alternative was to sell it on the basis of nothing. An advertiser who doubted that a programme had really been watched in the homes the panel said it had could commission a study of their own, at great expense, which nobody else would accept. Agreement was cheaper.

Once the audience could be counted to the half-hour, the half-hour became the product, and programmes began to bend towards the count. Dramas learned to build a small crisis before each advertising break, so that the people being sold would still be in the room when the selling began. Schedules were arranged to carry an audience from a strong programme into a weaker one after it, on the reasonable theory that people who could not be bothered to change the channel could at least be bothered to be counted. In local markets, the weeks when audiences were measured most closely acquired a reputation for the season's most dramatic episodes, its most heavily trailed specials and its most surprising guests, rather as a restaurant improves remarkably on the night the inspector is expected. And networks began to sell advertisers not slots but people, in specified quantities and of specified kinds, at a guaranteed price. If a programme delivered fewer viewers than promised, the network made up the shortfall with extra advertising slots, the way a coal merchant might top up a short delivery. Attention had acquired a delivery standard.

The box, meanwhile, got cleverer. In the 1970s meters began reporting down the telephone line, so that the national figures were ready by the next morning, and the overnights became a ritual of their own, read over breakfast in network offices with the expression of people opening examination results. In Britain the broadcasters set up a joint body of their own to do the counting. The Broadcasters' Audience Research Board, which began reporting in August 1981, was established by the BBC and the ITV companies, with the advertising agencies at the table, an arrangement which had the tidy feature that the people being measured owned the tape measure. Advertisers themselves did not get a seat on its board until 2008.

The most ambitious refinement came in the United States in 1987. From the 1987–88 season the national television market moved to what Nielsen called the People Meter, a set-top box with a remote control on which every member of the household had a button. When you sat down to watch, you pressed your button, and when you got up to make a cup of tea you were supposed to press it again. The machine could now report not only what the set was tuned to but who was, in theory, in front of it. In practice it had replaced the honesty problem of the diary with the memory problem of the button, since a person who has gone to answer the door does not always remember to tell the television, and a teenager who has been asked to press a button every time he enters a room does not always see why he should. The figures carried on being agreed, which was the main thing.

Diaries survived longer than anyone might have guessed. For local television markets, where meters were too expensive to install in enough homes, the little booklets went on being posted out and posted back for decades, and it was not until mid-2018 that Nielsen finished replacing the local TV diary with electronic measurement. By then several generations of American households had sat at kitchen tables on Sunday evenings trying to remember a Tuesday, and several generations of coders had sat at long tables trying to decide what their question marks meant.

The rest of this book is full of numbers more precise than the ones Nielsen produced: clicks, searches, likes, seconds of viewing, the number of times a phone is picked up in a day. Every one of them inherits the arrangement the box established. Somebody counts the audience. Everyone else agrees to pay on the count. And the count stands in for something nobody can see directly, which is whether the people in the room were paying attention, enjoying themselves, ignoring the screen or asleep. The figures could say that a set was tuned and, later, that a button had been pressed. They could not say whether a single advertisement had worked, or whether anybody had liked the programme rather than merely failing to turn it off. The gap between the measured thing and the real thing was there from the beginning, and everybody involved knew it was there. The market ran on the measured thing anyway, because the measured thing was what could be counted, and what could be counted could be sold.

Panels are designed to turn over. Households are recruited, fitted with their meters or sent their booklets, kept for a period, and then thanked and replaced, so that no family's habits harden into the nation's. Over seven decades an enormous number of homes passed through the arrangement, each one briefly standing in for millions of others, each one eventually returned to private life with its evenings once again its own. The families changed constantly and the box did not, and neither did the invoice, which went out every month to the people who made the programmes and to the people who paid for them, and was settled by both.