It's a simple enough question: What do you read all day long?

For hundreds of years, it was also an utter mystery to the publishing business. In the 1920s, newspapers were so desperate to learn what features were most popular that they hired private investigators to sneak up behind readers to mark the articles they were looking at. They sent people on trains and trolleys to record the exact page that discarded papers were left open.

George Gallup, the man synonymous with modern political polling, was a University of Iowa student who didn’t think that open broadsheets strewn across the floor of street trolleys made for particularly good data. In 1928, he wrote a thesis proposing a new scientific method for measuring audiences. Don’t stalk your readers, he said, and don't snoop through their train trash, either. Meet your readers.

Shared exclusively with The Atlantic, Gallup's 1928 thesis, "An Objective Method for Determining Reader Interest in the Content of a Newspaper," is an eerily prescient work, which reflects on today's discussions about viral media, news analytics, and even native advertising, all while showing that we've been having the same debates in this business for the last 80 years.


In the 1920s, the news industry was going through a technological and cultural shift that felt just as unsettling as today's. Urbanization was encouraging smaller papers to consolidate. The invention of the telegraph and the syndication of news brought national and even international news to local readers. This didn't just change how newspapers worked. It changed what newspapers meant.

In the late-1800s, there were papers for every "class, sect, and political group," Gallup said. Local journalism in that time was easy to do. When each writer was a member of his own audience, he could trust that anything interesting to him would be of equal interest to his readers. But as newspapers got bigger, journalists were suddenly writing for massive crowds that included every class, sect, and political group. Suddenly, journalists were writing for people they didn't know at all.

As newspapers got bigger, they were also getting less news-y, Gallup observed. The role of the press as "the chief agency for instructing and informing the mass of people" had diminished with the growth of public schools. The 1920s marked the beginning of the movement for mandatory public schooling, and Gallup predicted that schools were naturally taking over for newspapers as America's predominant source of serious information.

The other trend pushing newspapers out of the news business, he said, was industrialization and the routinization of work. When somebody is doing the same thing over and over again throughout the day, he doesn't want vegetables when he takes a break to read the newspaper, Gallup said. He wants candies. "If fails to find it in his newspaper, he will look elsewhere," he wrote.

How did Gallup know that audiences wanted candy? Because he went to their homes and watched them read the paper.

* * *

It's ironic that Gallup, the name we associate with polling, became famous for pointing out that readers lie in polls.

In 1920s newspaper surveys, readers reported that they enjoyed hard news on the front page. (That's what we still say in polls...) Gallup thought these readers were lying. (... and we're still lying!) But how to prove it?

Today, it's easy work. Software like Chartbeat, Omniture, and Google Analytics can tell publishers how many people are on a site, where they click, and how long they stay. But in 1928, newspapers were saddled with flawed feedback metrics. Circulation records described the journey of each paper, but not what readers read when they opened the bundle. Complaints and responses—i.e.: letters to the editor, or their modern-day equivalent, the comment under the article—are an honest source of feedback, but they represent an "articulate minority" of readers that's a caricature of the true audience.

Gallup's solution was to treat newspaper readers like an anthropologist. He would observe them in their own environments—in their own living rooms. He sent interviewers into Iowans' homes to sit across a table from subscribers and go through the day’s paper, marking each column and cartoon as read or unread. He called it "the Iowa Method.”

After sorting through all the data, Gallup learned that:

  • People are liars. "The person who believes he has read all of the front page may not have read a fourth of it,” he wrote.
  • Nobody likes serious news nearly as much as they report on questionnaires. Gallup’s interviews reported that front-page stories were actually no more popular than small features in the back of the paper.
  • The most-read thing in the newspaper wasn’t news at all: It was the front-page cartoon by J. H. Darling, read by 90 percent of men compared with just 12 percent reading the day’s local government news.
  • For women, the most-read parts of the newspaper were “style and beauty pictures.”

It's useful to compare these observations from August 1928 with a May 22, 2014, rant by Facebook product director Mike Hudack on "the state of media." Newspapers, Hudack wrote in his post, "are ghosts in a shell." TV news is tabloidish. Even the most promising websites are stuffed with entertainment. "It's hard to tell who's to blame," he said, "but someone should fix this shit."

What Hudack didn't point out is that his own product, Facebook News Feed, which is perhaps the most voluminous fount of digital news in the country, drives many of these trends. Facebook audiences aren't hungry for complicated and depressing updates on the state of the world. They prefer stories that fulfill the very purpose of Facebook's machine-learning algorithm—to give us a break, to entertain us, and to make us feel a moment's touch of connectedness to a group or individual.

But more importantly, the trends that Hudack and other media cranks often observe—the dumbing-down of news, the rise of entertainment, and the power of identity-driven media—aren't new. These are precisely the features of 1920s newspaper audiences that Gallup discovered when, in August 1928, he accidentally invented media analytics. Gallup's most powerful idea was that "people wanted entertainment that they could personalize, that they could relate to their own experiences," according to Susan Ohmer, a film and culture professor at the University of Notre Dame. "They enjoyed narratives with situations where they could see themselves."

We were always "Facebook readers" long before there was a Facebook.

* * *

George Gallup was no idealist. He regarded newspapers as gladiators in the arena of attention. "The problem of the modern newspaper is to fit itself as nearly as it can to the needs of the reading public," he wrote. "Specifically, its problem is to get itself read." Newspapers' competition didn't just come from other newspapers, he pointed out. It came from any single thing that could command the attention of a potential reader.

A newspaper which isn't read is worthless from an advertising point of view; and a newspaper without advertising would be unable to exist under present day conditions. In this sense all means of entertainment—the radio, the moving picture show, the automobile—which take the time ordinarily devoted by a reader to his paper, are competitors of the paper.

Gallup's thesis made him a celebrity in the attention business of the 1930s. The advertising agency Young & Rubicam named him its first research director in 1932. In the depths of the Great Depression, Gallup applied his innovative analytics to marketing and quickly changed the way ad guys thought about white space, surprising fonts, and large images. Using eye-tracking and close interviews with readers, Gallup showed that ads placed "beneath the fold," in the bottom half of a newspaper, were often ignored, and that readers pay more attention to photos than words.

It's easy to see how Gallup's personalized-entertainment thesis applies today, where Facebook's News Feed is cluttered with What Character Are You quizzes and emotional appeals to particular groups. But Gallup's most prescient finding was that advertisements that looked like news articles were the most successful sort of advertisements. Today the line between editorial and advertising in news is blurring. Gallup aggressively advocated for that blurring in the 1930s. As Ohmer reports:

An ideal ad, in Gallup's view, was one that looked so much like the rest of the paper or magazine that readers couldn't tell them apart. In one memo, Gallup praised an issue of Life in which some of the ads were built around groups of photos, with a minimal amount of text. These ads, he pointed out, looked just like the rest of the magazine, and readers paid more attention to them because the layout did not broadcast their identity. As a result of this research, Young & Rubicam clients who had not used photos before began to do so.

Critics complain that the media is too devoted to entertainment, mock consumers for ignoring serious news, worry that newspapers are destroying their own integrity, and cry out for a clearer separation between advertising and editorial. They're raising perfectly adequate concerns about the relationship between the media's civic duty to inform and its business prerogative to be read. But they're asking for things to go back to the way they never were.