You May Also Like (16 page)

Complexity aside, why do we seem to prefer what is familiar? With food, the familiar is evolutionarily adaptive: What did not kill you last time is good for you this time. We face Paul Rozin's “omnivore's dilemma”: Like rats, we are not restrictive in our food choices, but as a consequence, writes Michael Pollan, “
a vast amount of brain space and time must be devoted to figuring out which of all the many potential dishes nature lays on are safe to eat.” In music, and with taste in general, we face a similar omnivore's dilemma: more songs than you can listen to in a lifetime. The early promise of the digital music revolution
was, as the musician Peter Gabriel noted, freedom to choose. As hard drives overflowed and the cloud began to burst with music, we suddenly needed freedom
from
choice.

So we fall back on exposure: Why should we not like what we know (even if there might be something we would like more)?
It saves us time and energy, versus foraging in the great musical wilderness for things that are difficult to process. This may be why people seem to most like the music that they heard during “a critical period of maximum sensitivity,” as research by Morris Holbrook and Robert Schindler has shown—an age they peg at 23.5 years. This too could be familiarity.
It would be strange for people in their 70s to
not
prefer the Mills Brothers' “Smoke Rings” to Peter Gabriel's “Sledgehammer,” if only because they will be more familiar with the former.

There may be more than mere exposure and familiarity, however, to explain why we hold a special place in our hearts for the music of our early adult years. Holbrook and Schindler raise the idea of some kind of Lorenzian imprinting, a “biologically fixed” period in which we form parental attachments or learn language (
although the long-held idea of an age-based “critical period” for language acquisition has been more recently challenged).

I think something simpler is also going on. The college-age years are when we typically have the most time to search out and consume music. I still feel a vestigial crick in my neck from hunching over record bins. Now I barely have time to scroll through a playlist.

During a period of life when most of us do not have fancy watches or cars, music becomes a cheap, socially important signal of distinction. We are trying on, like silk-screened T-shirts, various identities. My high-school notebooks were filled with band logos, while an old cigar box held countless concert ticket stubs, like fetish objects, clues to my soul. Arguments over bands were arguments over who we wanted (and did not want) to be. How could these fierce attachments survive the transition to adulthood? In the documentary film
Rush: Beyond the Lighted Stage
, Matt Stone, the creator of
South Park
, talks about being the sort of person who would try in vain to impress upon his skeptical peers—whose taste ran to more “critically accepted” acts like Elvis Costello—the virtues of the Canadian progressive rock trio. “Now it's
like we're all so old,” he said, “even if you hated Rush in the '70s and '80s, you've just got to give it up for them. You've just got to.”

And indeed, when I now hear a song like Rush's “Spirit of Radio,” damned if I do not derive a certain pleasure from it. Was I wrong about Rush all those years ago? Is my new appreciation itself unadulterated or leavened with a dash of ironic distance? Or is it that not only do I not have time to figure out what music to like (all over again), I do not
even have time to maintain my dislikes
.
*
⁷
I am “losing my edge.” I am suffering “
taste freeze.” What is the opportunity cost of hunting down the latest band when it sounds to your ears like some rough derivative of a band you heard in your youth? It is hard to escape what has been called the “reminiscence peak,” described as follows: “
The events and changes that have maximum impact in terms of memorableness occur during a cohort's adolescence and young adulthood.”

By this analysis, one suspects the reason Woodstock loomed so large in the culture is not the music itself but an almost
statistical
outcome of the largest birth cohort in American history suddenly hitting that age bracket of maximum impact. But why do we all—and not just
The Big Chill
generation—seem to insist that the music of our youth was
better
? As Carnegie Mellon's Carey Morewedge points out, because everyone basically has this experience, it cannot be objectively true. He suggests that in the same way we tend to remember positive life events more strongly than negative events, only the “good” music from our past tends to survive in our memory. In the raw and unpolished present, meanwhile, we hear music we think we like
and
music we know we do not like. Memory, as he describes it, is like a radio station that only plays what we want to hear. Given that we had devoted so much time to thinking about the music, it is no surprise that it still so easily fills our memory and that we seem to have a hedonic soft spot for it.

So how do we move beyond the safe perimeter of our typical foraging ground into promising, if terrifying, new vistas, filled with new, unknown pleasures? We look for someone to take us there. As Westergren had joked to me, “We allow the lazy middle-age man to get back in the game.”

Before coming to Pandora, I had a rather anguished back-and-forth with the people in its public relations department. The sticking point seemed to be the word “taste.” They wanted me to know they were not “tastemakers in any sense of the word.” Rather, they strove to “provide each and every listener with a unique experience.” It seemed another hint of how much the notion of Taste with a capital
T
had fallen since mid-century, as if prescribing taste were some old, outmoded habit, like drinking martinis at lunch.

People now talk in softer terms, floating words like “discovery” and “curation.” The Book-of-the-Month Club, upon its launch in 1926, promised that any title that survived the “differing tastes” and “good judgment” of its panelists was “bound to be an outstanding book.” Nearly a century on, the club more demurely promises “our favorite new titles that we know you'll want to read” (note the shift in focus, from objective standards from above to playing upon your personal preference).

But is this not still imposing some kind of selective criteria? Westergren tried to sound wholly catholic about Pandora's playlists. “We don't want to be judgmental about any of this,” he said. “Some people want to hear the same ten songs over and over again; that's what we should give them.” That raises the question of whether they need an army of music analysts and fancy algorithms. A moment later, he added, “We curate our collection. We turn down an awful lot of music, because it makes the other side more satisfying.”

I put the question to Michael Zapruder, for many years the company's head music curator. His job was to pick which recordings—most of which he was unfamiliar with—would enter the Music Genome.
He had wrestled with the problem of democratic inclusion versus the elitism of curation, comparing it to being a judge at a baby beauty contest (a not uncomplicated problem I will return to later in the book).

He called it a “paradox” and paraphrased Orwell: Some songs are more equal than others. There was taste at work here, even if it was not Pandora's per se. But it seemed to be working: As Westergren told me, more than 95 percent of the service's million-plus songs were played every month.

Pandora had created a giant musical sandbox. There was a lot to
play with, a lot to discover, but there were still boundaries. As Steve Hogan, the company's manager of music operations, told me, “That's the reason we have one million songs.” Other services may have eighteen million songs, he said, “but we have human beings making judgments. If a label sends us a bunch of karaoke music, they're going to pass on it.” Rather, says Hogan, Pandora analysts will “try to pick the songs they feel best represent the artist and have the best chance to succeed in their opinion.”

But where radio could only play one song at a time, typically in a format that listeners expected, Pandora was trying to use math and musicology to create an army of invisible DJs, each serving up a mix of what you liked and what you might like. In one story I heard, Tim Westergren was at a town-hall-style meeting where someone told him he had no idea there were so many fans of marching band music in the world. Indeed, Pandora has a marching band music channel, and to that listener that was what Pandora was
for
. Tom Conrad, the company's chief technical officer, told me, “We want people to feel like it's really theirs, and for our musical taste—or other people's musical taste—to not impinge upon that.”

The question of what music people might potentially play was only a small part of the problem. By trying to create stations tailored to individuals, Pandora was opening the Pandora's box of taste. For our liking of music, like food, is open to a staggering amount of influence. Heard too much of one thing in a row? Sensory-specific satiety can set in. What did you listen to beforehand? A sad song might seem less sad after a succession of happy songs. Where are you listening to the music? In college, I used to visit a well-regarded local record shop, a shrine of obscure erudition where you felt vaguely blessed if a clerk gave a nod of tacit approval to your purchase. I soon realized that no music could
ever
sound as good as what those reedy, severe sages had on the turntable behind the counter.

And there is the idea that like food, music may comprise basic “tastes”—instead of saltiness or sweetness, think of syncopation or vocal breathiness or drum snare—but it is the “flavors” we learn to like and discern. A few years ago, Pandora listeners seemed to be lodging particularly negative feedback in the electronic dance music genre. “We had analyzed
about forty-five thousand tracks, and we realized a lot of the club dance music was indiscriminately mixing together,” Hogan said. “To the genome, they all have the same ‘boosh boosh boosh' beat.” But fans were hearing techno on their trance stations. Techno, says Eric Bieschke, “means something very specific if you're into electronic music. If you're my dad, everything I've ever listened to is techno.” So Pandora, Bieschke said, added a dozen or so new “attributes” into the genome. “How much reverb and ambience is on there? What sort of eq'ing effects or filter sweeps are being used?”

Even an individual band can represent many different pathways of taste. Sometimes it's the people who change, while the song remains the same. Take the hit song “We Are Young” by Fun. The year prior, Conrad told me, Fun was “just one of the countless kind of semi-faceless bands that put out a record and get a review on
Pitchfork
and no one in any mass scale hears about them.” The song, he noted, had been on Pandora “for years,” played by a “core of people who felt like they had discovered this band.” Then, suddenly, the song appeared on the soundtrack to the popular television show
Glee
. “Overnight this song had a huge new audience, who I think had a different set of expectations when they came to listen to it on Pandora. They wanted to hear other songs that had been on
Glee
.”

The whole world of recorded music, Zapruder had suggested to me, is like an ocean. “Every recording is an entry point. So you might get into the water at the Beatles, and once you're in the water, you can end up anywhere.” Some people hug the shore; others brave the open ocean. At its most incisive, Pandora might make a serendipitous connection, the way a free-form DJ might, following the Beatles with, say, “Lemons Never Forget” by the Bee Gees. The sound is quite Beatles-esque. But for many people, their mental model of the Bee Gees as a disco act would not permit them this connection. Bieschke says the “holy trinity” at Pandora is “variety, discovery, familiarity.” It has a mathematical model of where you sit on that axis—from “active” to “passive” listening. The stations you create are shorthand for the breadth of your taste. “If you've ever created a jazz station, you probably have stations all over the map,” he said. “If you ever typed in ‘Coltrane,' you're likely to have a very wide umbrella of interested listening habits.”

In the end, the thumb rules. In early 2015, someone's up or down vote was the fifty billionth thumb on Pandora. The thumb is the clearest
signal it has, stronger than the skip button. But even here there is room for ambiguity. Are you saying you do not want to hear that right now? Do you not like that band, or is it not quite right for this station? “We actually ran a test,” Bieschke said. “We took half of a percent of people listening to Pandora. When they hit thumb up or thumb down, we'd ask why.” Listeners could list reasons why in a text box. “The tricky part was that the things people wrote in were all over the map. They would write things like ‘I thumbed this up because it was the first dance at my daughter's wedding.' As an algorithm guy, I was like, what the hell am I going to do with that?” The feature was scuttled, and that particular Pandora's box—trying to learn why people liked or disliked something—was closed.

“They say there's no accounting for taste,” Hogan told me. “But we can account for it, en masse. We can say there's an 84 percent chance that this song is going to work for people listening to Rolling Stones radio. It's a good bet; we've accounted for the taste of this big group of people.” He paused, looked briefly into space, then added, “Maybe there's no accounting for why they
didn't
like it.”