a transformation similar to the data-driven shifts occurring at Target and elsewhere. Just as retailers were using computer algorithms to forecast shoppers’ habits, music and radio executives were using computer programs to forecast listeners’ habits. A company named Polyphonic HMI-a collection of artificial intelligence experts and statisticians based in Spain-had created a program called Hit Song Science that analyzed the mathematical characteristics of a tune and predicted its popularity. By comparing the tempo, pitch, melody, chord progression, and other factors of a particular song against the thousands of hits stored in Polyphonic HMI’s database, Hit Song Science could deliver a score that forecasted if a tune was likely to succeed. [206]
The program had predicted that Norah Jones’s
When executives at radio stations ran “Hey Ya!” through Hit Song Science, it did well. In fact, it did better than well: The score was among the highest anyone had ever seen.
“Hey Ya!,” according to the algorithm, was going to be a monster hit.
On September 4, 2003, in the prominent slot of 7:15 p.m., the Top 40 station WIOQ in Philadelphia started playing “Hey Ya!” on the radio. It aired the song seven more times that week, and a total of thirty-seven times throughout the month. [207]
At the time, a company named Arbitron was testing a new technology that made it possible to figure out how many people were listening to a particular radio station at a given moment, and how many switched channels during a specific song. WIOQ was one of the stations included in the test. The station’s executives were certain “Hey Ya!” would keep listeners glued to their radios.
Then the data came back.
Listeners didn’t just dislike “Hey Ya!” They hated it according to the data. [208] They hated it so much that nearly a third of them changed the station within the first thirty seconds of the song. It wasn’t only at WIOQ, either. Across the nation, at radio stations in Chicago, Los Angeles, Phoenix, and Seattle, whenever “Hey Ya!” came on, huge numbers of listeners would click off.
“I thought it was a great song the first time I heard it,” said John Garabedian, the host of a syndicated Top 40 radio show heard by more than two million people each weekend. “But it didn’t sound like other songs, and so some people went nuts when it came on. One guy told me it was the worst thing he had ever heard.
“People listen to Top 40 because they want to hear their favorite songs or songs that sound just like their favorite songs. When something different comes on, they’re offended. They don’t want anything unfamiliar.”
Arista had spent a lot of money promoting “Hey Ya!” The music and radio industries needed it to be a success. Hit songs are worth a fortune-not only because people buy the song itself, but also because a hit can convince listeners to abandon video games and the Internet for radio. A hit can sell sports cars on television and clothing inside trendy stores. Hit songs are at the root of dozens of spending habits that advertisers, TV stations, bars, dance clubs-even technology firms such as Apple-rely on.
Now, one of the most highly anticipated songs-a tune that the algorithms had predicted would become the song of the year-was flailing. Radio executives were desperate to find something that would make “Hey Ya!” into a hit. [209]
That question-how do you make a song into a hit?-has been puzzling the music industry ever since it began, but it’s only in the past few decades that people have tried to arrive at scientific answers. One of the pioneers was a onetime station manager named Rich Meyer who, in 1985, with his wife, Nancy, started a company called Mediabase in the basement of their Chicago home. They would wake up every morning, pick up a package of tapes of stations that had been recorded the previous day in various cities, and count and analyze every song that had been played. Meyer would then publish a weekly newsletter tracking which tunes were rising or declining in popularity.
In his first few years, the newsletter had only about a hundred subscribers, and Meyer and his wife struggled to keep the company afloat. However, as more and more stations began using Meyer’s insights to increase their audiences-and, in particular, studying the formulas he devised to explain listening trends-his newsletter, the data sold by Mediabase, and then similar services provided by a growing industry of data-focused consultants, overhauled how radio stations were run.
One of the puzzles Meyer most loved was figuring out why, during some songs, listeners never seemed to change the radio dial. Among DJs, these songs are known as “sticky.” Meyer had tracked hundreds of sticky songs over the years, trying to divine the principles that made them popular. His office was filled with charts and graphs plotting the characteristics of various sticky songs. Meyer was always looking for new ways to measure stickiness, and about the time “Hey Ya!” was released, he started experimenting with data from the tests that Arbitron was conducting to see if it provided any fresh insights.
Some of the stickiest songs at the time were sticky for obvious reasons-“Crazy in Love” by Beyonce and “Senorita” by Justin Timberlake, for instance, had just been released and were already hugely popular, but those were great songs by established stars, so the stickiness made sense. Other songs, though, were sticky for reasons no one could really understand. For instance, when stations played “Breathe” by Blu Cantrell during the summer of 2003, almost no one changed the dial. The song is an eminently forgettable, beat-driven tune that DJs found so bland that most of them only played it reluctantly, they told music publications. But for some reason, whenever it came on the radio, people listened, even if, as pollsters later discovered, those same listeners said they didn’t like the song very much. Or consider “Here Without You” by 3 Doors Down, or almost any song by the group Maroon 5. Those bands are so featureless that critics and listeners created a new music category-“bath rock”-to describe their tepid sounds. Yet whenever they came on the radio, almost no one changed the station.
Then there were songs that listeners said they actively
One night, Meyer sat down and started listening to a bunch of sticky songs in a row, one right after the other, over and over again. As he did, he started to notice a similarity among them. It wasn’t that the songs sounded alike. Some of them were ballads, others were pop tunes. However, they all seemed similar in that each sounded exactly like what Meyer expected to hear from that particular genre. They sounded
“Sometimes stations will do research by calling listeners on the phone, and play a snippet of a song, and listeners will say, ‘I’ve heard that a million times. I’m totally tired of it,’ ” Meyer told me. “But when it comes on the radio, your subconscious says, ‘I know this song! I’ve heard it a million times! I can sing along!’ Sticky songs are what you
There is evidence that a preference for things that sound “familiar” is a product of our neurology. Scientists have examined people’s brains as they listen to music, and have tracked which neural regions are involved in comprehending aural stimuli. Listening to music activates numerous areas of the brain, including the auditory cortex, the thalamus, and the superior parietal cortex. [211] These same areas are also associated with pattern recognition and helping the brain decide which inputs to pay attention to and which to ignore. The areas that process music, in other words, are designed to seek out patterns and look for familiarity. This makes sense. Music, after all, is complicated. The numerous tones, pitches, overlapping melodies, and competing sounds inside almost any song-or anyone speaking on a busy street, for that matter-are so overwhelming that, without our brain’s ability to focus on some sounds and ignore others, everything would seem like a cacophony of noise. [212]
Our brains crave familiarity in music because familiarity is how we manage to hear without becoming distracted
