.Pandora’s Box

Today’s lab team: one drummer, two saxophonists, two guitarists, one bassist, two pianists, one violinist. Today’s specimen: Eminem’s rap anthem “Lose Yourself.” Today’s mission: crack all 235 of the song’s genes.

Yes, genes.

• 

• 

• 

• 

• 

• 

Welcome to the Music Genome Project, the most awesomely and exhaustively nerdy cataloguing endeavor in pop-music history. Founded by Stanford alum Tim Westergren, a pianist and former film composer, the Oakland-based project breaks down songs according to their component traits, or genes, just as the Human Genome Project has mapped nature’s blueprints. “Genes collectively make a person tall or short, black or white, fast or slow, with freckles or not,” Westergren says. “It’s kind of like your building blocks — and we think of this as the same thing for music.”

“Lose Yourself” thus has a gene describing whether the bass is played ostinato or as a riff, and another for whether the kick drum sound is tight or booming. There are genes for handclaps, turntable scratches, and organ solos — times 235. In fact, the four genres, or “genomes,” these music analysts have scrutinized to date — jazz, hip-hop/electronic, rock/pop/country, and world music — contain a total of about four hundred genes. Some are genre-specific — hip-hop, for example, has no need of the jazz gene that counts improvised sax licks.

This is far from an academic exercise. As chief strategy officer of Pandora Media, a company he created to oversee the project, Westergren is attempting no less than an egalitarian revolution in music marketing — one he believes can reach a vast, neglected cadre of fans who love music but have aged out of the crate-digging demographic, or are uninspired by MTV. Pandora’s goal is to introduce these people to new music based on what they already like, and while it’s too early to draw conclusions, the strong initial response to Pandora’s unique product suggests its effort, or something similar, could help transform a recording industry battered by illegal downloading, flagging CD sales, and listener ennui.

Westergren’s idea was simple: Compile hundreds of thousands of songs into a gigantic database that can match those you love with structurally similar tracks, thereby creating the ultimate music-recommending device. That was Phase One, which took six years to finesse.

Phase Two, which launched last fall, is where Westergren and his colleagues hope to recoup their startup costs and turn their musicology into a profitable online venture. Pandora.com, named in honor of the curious, is the Music Genome Project’s public interface. Type in a song or band that you like, and Pandora combs the database to create a personal Web radio station that plays “genetically” similar songs — ideally, it will churn up tracks you already like and introduce you to some pleasant surprises along the way. Enter “Pixies,” and Pandora launches the band’s live cover of “Wild Honey Pie,” along with a note: “We’re playing this track because it features grunge recording qualities, a subtle use of vocal harmony, mild rhythmic syncopation, repetitive melodic phrasing, and extensive vamping.”

Maybe “extensive vamping” is a big deal to you, maybe not. The point is that all the songs that follow on your new “Pixies Radio” station should sound pretty familiar without being exactly alike: Mission of Burma, Black Flag, Mudhoney, the Fall, plus a half-dozen bands you’re guaranteed to never have heard of. One mouse click will net you a list of the traits that made them musicological dead ringers. Not that you’ll necessarily need it: At least some of the matches are so right-on it’s uncanny.

Because Pandora’s stations are generated on the spot, they represent a radical departure from other Internet radio stations, which often rely on prechosen playlists or are mere simulcasts of terrestrial stations. Pandora’s selections are based on sheer musicology, not sales rank or purchases made by like-minded customers, which is how sites like Amazon and Netflix recommend products. “That’s a popularity contest,” Westergren says. “Our system doesn’t actually know how popular a piece of music is. It doesn’t know who else liked it. All it’s concerned about is the musical content.”

In other words, Pandora gives equal weight to megahits, deep-catalogue cuts, obscurities, and indie tracks. “You can be a completely unknown garage band, and if your song is the right recommendation for the latest Mariah Carey hit or Wilco record, then you’re going to end up there,” Westergren says.

The site has obvious appeal to record geeks, but so far its greatest response — this is key to Pandora’s business model — has been from people well past their music-buying heyday who are hungry for a pain-free introduction to music they might like. It doesn’t hurt that Pandora is free to anyone willing to tolerate a few on-screen ads, or about $3 a month for an ad-free subscription. Since the free version debuted in November (the subscription site launched last August), Pandora has become a tech-blog darling — it’s been Slashdotted (twice), Farked, and made the subject of about three thousand blog entries. The site’s execs are loath to reveal subscriber statistics, but insist that site usage is on a sharp ascent: Four million stations have been created to date, three million of them in December alone. As the site’s popularity rises, so do its revenues. Besides subscriptions, Pandora earns its money by hosting ads, and via paid links to Amazon or iTunes, where listeners can purchase a song or album directly.

The underlying music genomes are also undergoing a building boom. Pandora employs more than thirty analysts, all trained musicians and musicologists, to fill its database. It typically takes at least fifteen minutes to score a track, ranking each gene on a scale of zero to five. The analysts usually work in libraryesque silence, headphones firmly clamped over ears, plugged into computer banks that randomly serve up a constant stream of tracks to decode. To ensure consistent scoring, they must go through five group-training sessions in order to master a new genome. These are far noisier endeavors filled with good-natured bantering and people playing imaginary tabletop pianos or pounding out rhythms as they deconstruct the music together.

Today, enough musicians to form a respectable jazz ensemble have descended on a Pandora conference room to debate the finer points of Eminem’s “Lose Yourself.” Music operations manager Steve Hogan and Chris Horgan, the project’s rap/dance specialist, take their seats, and everyone shuffles their paperwork. Their ultimate goal: to reach a consensus on every gene, within a half-point.

Some genes prove pretty easy: The rap analysts nail down Eminem’s accent (medium-strong, urban Midwestern), his main vocal timbre (“nasal,” as opposed to, say, “gritty” or “breathy”), and his register (three and a half octaves). They rank his shouting level a two (audible vocal-cord rattling), and declare his rhyming pattern moderately consistent (mostly on the second and fourth beat, occasionally coming in early on the third). They give the song a high score for chord patterning, and dub it highly repetitive with a memorable chorus hook. There is a slight skirmish over whether or not anyone can hear swung sixteenth notes.

But the group gets stuck on what may or may not be a piano. The song has a lengthy keyboard intro that sounds like an acoustic upright, but several analysts think it’s a synthesizer or, at the very least, an electric Yamaha. They have this problem a lot, actually — they have to hear the music not only as professionals, but as everyday fans. “We have to try to put ourselves in the shoes of the casual listener,” Steve Hogan explains. After all, if listeners like what they think is an acoustic piano on “Lose Yourself,” then Pandora should dish them other rap tracks with acoustic piano.

“Our litmus test is: What would a soccer mom call that piano?” Chris Horgan adds. “Would she call it an acoustic piano or an electric piano?”

This piano is tricky, though. It’s not overtly fake — analyst Ava Mendoza nails it by dubbing the sound “plasticky.” Ultimately, against their better judgment, the analysts render a verdict: Soccer mom’s going to call it acoustic. Eminem has piano genes.

On to lyrical content. All agree that “Lose Yourself” is pretty much this generation’s “Eye of the Tiger,” but they disagree on where it falls on a scale of lyric to narrative. Is it an abstract cautionary tale about success, or strictly autobiographical, Eminem’s own rise from trailer park obscurity to hip-hop fame?

Hogan, who’s leaning toward abstract, Googles the lyrics and reads them back in a professorial deadpan that reduces his colleagues to giggles:

A normal life is boring, but superstardom’s close to post-mortem

It only grows harder, only grows hotter

He blows us all over, these hos is all on him

Coast to coast shows, he’s known as the globetrotter

Lonely roads, God only knows

He’s grown farther from home, he’s no father

He goes home and barely knows his own daughter.

The lyrical semantics are now pondered with an intensity rarely seen outside graduate English lit seminars. “To me it seems like the narrative to this whole story follows him at different points in his career — he’s coming up, he’s throwing up mom’s spaghetti on his shirt, then he’s successful but not seeing his daughter,” analyst Tom Griesser concludes. Hogan argues that Eminem seems distant, particularly in this verse, as if he’s describing somebody else.

“Eminem always talks in the narrative third person,” Pat Greene counters.

“It’s sort of royal,” Alan Lin agrees.

“I felt it had a narrative quality but it wasn’t a slam-dunk story, like a ‘Rocky Raccoon,'” Hogan concedes, and agrees to change his score.

At the other end of the table, Lin is reconsidering his tally for offensive lyrics. “Is one ‘motherfucker’ and two ‘hos’ not enough to score a four?” he wonders aloud.

People chew pens and debate that for a while. It goes on like this all day.

---

Across a vast record library, wedged into an office that was clearly once a hallway, sits the guy who feeds music into the maw of the analysis beast. Mild-mannered guitarist and singer-songwriter Michael Zapruder has, very simply, the best job in the world. He buys records. All day. With someone else’s money.

As Pandora’s buyer, Zapruder has purchased fifteen thousand albums for the library in the past year, doubling its size, and his spree is only speeding up. He’s a regular at Amoeba. He haunts merch booths at shows, and places massive orders with mail-order outfits like Insound, Forced Exposure, and San Francisco’s Aquarius Records. He buys anything that tops industry pop charts, whatever’s got buzz on respected review sites like Pitchfork, and the entire catalogues of record labels he trusts — kranky, Mute, Warp, Stones Throw. His credit card is on the verge of melting. “It’s constantly getting turned down for fraud warnings because there are so many transactions,” he says wryly.

Zapruder has three basic rules: 1.) Pandora must own original hard copies of everything it plays. 2.) He’d rather pay than hassle labels for free copies. 3.) The music can’t totally suck. “Essentially, I have to feel like somebody would be happy to discover this music,” he says.

Pandora’s staff, composed as it is of musicians and record geeks, is firmly anti-illegal-download and anti-bootleg, so Zapruder spends a lot of time tracking down obscure originals, such as certain J-pop albums that are rarities outside of Asia, or, say, the hard-to-find Piracy Funds Terrorism, a collaboration between British rapper M.I.A. and American deejay Diplo, which many fans undoubtedly pirate out of sheer frustration.

But since the free site launched, Zapruder says, the record labels are now looking for him. They’re sending him back catalogues along with current hits, since they know Pandora will serve up songs that haven’t been on the charts for years. “My desk is radically transformed over the last two months,” he admits.

Music fans also are deluging the site with suggestions. Pandora employees, Westergren in particular, maintain a torrential correspondence with users — on any given weekday, he has around three hundred messages in his inbox, urging him to look into everything from Voice of the Beehive to more Russian bands or Celtic music. Some of the missives include incredibly detailed discographies, band histories, and personal essays about why a group is so beloved, and this fills Westergren with unabashed glee. “They’re so invested in this,” he says proudly, clicking open a letter from a subscriber asking if he can send in albums by his favorite local bands. “Goddamn right you can submit that stuff,” he beams. “Send it in — that’s exactly what it’s for.”

Pandora has five computers running full-time simply to rip mp3s from the twelve hundred CDs that come through its doors each month. Its album library, about the size of a recreation room, is large enough for a pool table and a stage for staffers’ Friday afternoon jam sessions. It looks roomy now, but this time next year it’ll be three sizes too small.

Zapruder and the analysts may be the human side of the process, but once they’ve cracked a track, says Etienne Handman, a jazz trumpeter and former new wave deejay turned chief operating officer, “what’s actually going on is very cold and mathematical.” Given a request, Pandora’s software measures the relative distance between that song and all others in the system based on its hundreds of genes, then returns the closest matches. “You can imagine the whole universe of all the songs that we’ve analyzed so far as being like stars in space,” Handman says. “Some are near each other in little nebulas, and some are far apart.” So tracks by the Beatles and the Beach Boys might share a nebula, while avant-jazz saxophonist John Zorn would exist in an entirely different galaxy.

Pandora’s genius, though, lies in how it shuttles you from star to star. If it played songs in order from the closest match to the most distant, you’d get a deadly dull stream of near-identical songs and a sudden-onset case of listener fatigue. Instead, the database builds a series of short playlists around common elements, much the way a real-life deejay puts music together into sets, first establishing one mood, then breaking it for another one. “In a club, it turns over people on the dancefloor and gets them to the bar,” Handman says. “On the radio, it gives you something to put station IDs and commercials in between.” On Pandora, the site plays a set tightly connected to the original request in some way, then moves on to another set that’s related a different way.

As an example, Handman pulls up a flowchart showing possible playlists generated from “What Would You Say?” by the Dave Matthews Band. One of its defining traits is an acoustic guitar riff, so Pandora might first find matches for the riff and pull up tracks by the Stones, the Guess Who, and John Mellencamp. The next set might focus on a harmonica gene, leading to matches with Blues Traveler and the like. A third set might focus on bands with similar subtle funk influences, and so on, creating an endless procession of tracks in the right vein for a Dave Matthews fan, but which reflect different facets of the band’s sound. Because the site is constantly shuffling between hits and obscurities, it’s a relatively painless way of slipping in new material amongst old favorites. “Ideally you want to make it familiar, but occasionally you make it interesting by having something unexpected happen,” Handman says. “You want to let them kind of enjoy that new mood for a little while, and then give them a new mood.”

Listeners can affect the stream by clicking “thumbs up” or “thumbs down” icons that steer Pandora toward or away from tracks like the one that’s playing. To diversify their stations, people can use several songs or artists as starting points. (The database delivers a mix of songs representing each of the originals, not songs that reflect the common qualities.) A Pandoraphile can maintain up to one hundred stations at a time, and can also type in friends’ e-mail addresses to hear their stations, a little like sharing iTunes playlists.

You can’t rewind, however, and while you can skip forward, you can do it only so many times per hour. This is to prevent listeners from just finding certain tracks and bootlegging them. Pandora CEO Joe Kennedy, who loves Top 40 pop but has a background in classical piano, says the site’s music licensing agreements also prohibit playing tracks on demand. “If you want to hear Madonna’s latest track, you can’t enter that track name and have it play that song right away for you,” he says. You’ll never know if, or when, it’s going to come up. “The only other really significant constraint is that we can only play three or four songs by an artist in any given time period, so it’s not like you can have Coldplay after Coldplay after Coldplay,” he adds. “That’s not the actual benefit that we offer. The benefit we offer is: Yes, you love Coldplay, but we’ll also introduce you to five other artists that have a wonderful sound and are in that same part of the music universe.”

Because the database is blind to an artist’s popularity, Pandora execs love to point out that well-known records can easily become launching pads to indie ones. “You can start from Coldplay, which is at sales rank 3 [on Amazon], and immediately on song two link to sales rank 432,000,” says chief technology officer Tom Conrad, a singer and diehard record collector. “We don’t know anything about whether the songs are popular or not, or what the social conventions around the music are. We couldn’t make recommendations based on those things if we wanted to.”

That means these fame-to-obscurity segues will also take place — often hilariously, and sometimes nerve-shreddingly — in reverse. Everyone in the office has stories of typing in a highbrow personal favorite only to find it quite accurately matched with something embarrassingly cheesy — Zapruder calls it a “you got vapid pop in my indie rock” moment. “The third time I started my Wilco station up,” Conrad admits, “it played something and I thought, ‘Oh, this is a great song,’ and I went to see what it was and it was Cheap Trick.” After that, he says, “I just stopped looking. My preconceived notions are clearly too strong, and I just need to embrace the musicology.”

Some other fine office legends: Certain Indigo Girls songs can trigger a Metallica cover of the Hank Williams ballad “Tuesday’s Gone,” and several Sarah McLachlan-based stations call up Britney Spears doing “I’m Not a Girl, Not Yet a Woman.” This annoys the bejeezus out of McLachlan purists, but tough luck: “No matter what you think of Britney Spears, it happens to be a heck of a match for Sarah McLachlan,” Kennedy says. “If you’re prepared just to kind of surrender a little on artist stereotype, it’s pretty eye-opening.”

---

Opening consumers’ eyes is pretty much what Tim Westergren had in mind all along. A jazz-trained pianist who spent eight years trying to make it as a rock musician, he was frustrated with how hard it was to get noticed. “It’s like being a needle in a haystack,” he sighs. He doesn’t blame the record industry: “I think labels do what they can do. It’s a difficult industry to succeed in, and you can’t afford to put bets behind a thousand bands — you’d lose your shirt.”

The problem, Westergren concluded, is an embarrassment of riches. “There’s all this great music out there without an audience, or with an audience that can’t find it,” he says. In the late 1990s, he hoped Internet music sales would make it easier for unknown bands to find an audience, but he feels it never delivered on its potential. “All these new indie music sites, and you could go online and browse this catalogue of a million CDs,” Westergren recalls. “It had so much promise that it was going to revolutionize the business, and pretty quickly, it became a replica of the traditional record business, which is, a very small number of records sell a lot and the vast majority hardly sell anything.”

An even more fundamental problem, he says, is that music lovers tend to become estranged from it as they age. Trading in school for work and family leaves less time and energy for record hunting, not to mention less access to the kind of music-friendly social networks fostered on school campuses. “You ask your average adult listener, 24 or 25 or older, do they love music? Yes. Do you listen to music a lot? No. Why not? Because I can’t find stuff I like. “ Westergren pauses thoughtfully. “It’s an amazing thing,” he says, shaking his head.

What everyone needs, he thought, is a guide. While Westergren had struggled to market his own music, he turned out to be pretty adept at pointing others toward music they liked. He spent five years as a composer for independent film scores, a job that required him to anticipate a director’s tastes. “They’ll say, ‘I like this song and this song — write me one like that,'” he recalls. “Or they’ll do what’s called temping, where they’ll take a piece of music and slap it onto the film and say, ‘Here’s what I had in mind.’ So as a professional I started thinking about music taste in terms of musical attributes, because I would have to go back and write a new piece of music I thought they would like.”

From there, it wasn’t a huge leap to the Music Genome Project, nor, in the late ’90s, was it difficult to find angel investors willing to back such a venture. The company launched in 1999 as Savage Beast, Inc., with a plan to market music-recommending software to major retailers. But it foundered during the dot-com crash and went through severe layoffs. It wasn’t until 2002 and 2003 that Westergren’s company landed several big-name clients — America Online hired it to build software for AOL music sites, while Best Buy and Borders commissioned it to design information kiosks that could recommend products. Then, in 2004, the company got an infusion of venture capital, and a whole new direction.

CEO Kennedy was convinced that the strongest application of the Music Genome database was in reaching out to individuals, not retail chains, and, more radically, that attempting to sell music was yesterday’s business model. “People who start digital music companies I think are typically the folks who like to go to Virgin Megastore or Rasputin and spend a sunny Saturday afternoon under the fluorescent lights browsing the used CD bin,” technology chief Conrad agrees. “It’s tempting for young companies to think the whole world is like that.”

In fact, most people rarely buy albums: Americans thirteen and older spend an average of only $35 a year on music. The Recording Industry Association of America reports that CD purchases are in steady decline — in the first half of 2005, Americans bought 307 million CDs, down from 369 million for the same period in 2002.

While this is often blamed on illegal downloads, Conrad also notes that most people spend only about an hour a week listening to music they own, compared to about eighteen hours a week spent listening to the radio. “The interesting thing that Joe pointed out to us is that helping people buy more CDs is probably not where the real opportunity is,” Conrad says. “The real opportunity is helping people with listening, and listening in the mass market is about radio — about free access to free music.”

Internet radio, in particular, is a huge sleeper hit, Conrad says. A survey by Jacobs Research in early 2005 concluded that more people listen to Internet radio than own mp3 players — and 2005 was the year of the iPod. Combined with Arbitron Radio Ratings’ recent finding that Internet radio listeners are twice as likely to own mp3 players and make purchases online, it’s clear how guiding these listeners to vendors of new music might be a lucrative proposition. “We very much see ourselves as the complement to the iPod,” Kennedy says. “I think a lot of people who own iPods have kind of hit the wall of frustration — yeah, it’s great that I’ve got all these tracks on here, but I know them all. Yeah, I made playlists but I’ve played it ten times and I’m tired of it. I need something new, something that surprises me.”

So the once-flagging company transformed its business model from selling software to selling site subscriptions and ads, and taking a cut of iTunes and Amazon purchases driven by Pandora’s personalized radio stations.

Because Pandora is aimed at the masses, its designers resisted the urge to belabor the site with discographies, biographies, or other features that might appeal to collectors. While it provides the artist’s name, track title, and an image of the album cover, it ignores such liner-note essentials as album title, record label, and release year. “It has to be simple and it has to be about listening,” Conrad insists. “It shouldn’t be about browsing and clicking and album marketing and band histories and who played the cowbell on this Blue Oyster Cult album.”

Oddly, nobody at Pandora imagined it would be so hard to convince people that the sets they were hearing were created just for them. During beta testing, users who requested a U2 station, for example, thought Pandora was just linking them to another Internet station already playing U2. They thought the thumbs up and down functions recorded user feedback that would be given to record companies, and that playlists were generated based on other users’ purchase histories, or because of a band’s relationship to other groups, based on common bandmembers, for instance.

So Conrad and the design team stripped the site down further and added pop-up windows explaining why each track was chosen as a musicological match. That seemed to clear up the confusion. “That was the moment, I think, when the real Pandora was born, when it went from being a kind of a consumer-electronics gadgety kind of experience — little buttons that did things — and it really became a conversation,” Conrad says. “It became this dialogue between the listener and the music genome.”

---

As it turns out, people have a lot to say about their interactions with Pandora, and while it’s too early to really read the tea leaves, the database seems to be doing what its founders intended. One of the most adored bits of fan mail floating around the office simply states: “Since I discovered Pandora, I have not left my computer. It is awesome. Please send food and water.” (Not to worry — Westergren sent him a bottle of water and a protein bar.)

What’s remarkable about this letter is not its ardor, but its author. John Weyer, 34, is exactly the kind of guy Pandora hopes to reach. Chicago futures trader and father of three is a huge music fan — he has seen the Grateful Dead more than fifty times, and Bruce Springsteen more than thirty — but he just doesn’t buy albums like he used to. “I think some of that is just the age I am at,” he muses via e-mail, adding that he also isn’t inspired by radio. “Commercial radio these days is overprogrammed and cookie-cutter,” he says.

Weyer credits the site with introducing him to the likes of the Stranglers and the Jazz Butcher, and reawakening his interest in new discoveries. “I will definitely be buying more music because of Pandora,” he says.

Even some younger users say the site has rescued them from musical anomie. “I’m 24 years old, and I haven’t had ‘that person’ in my life who turns me on to new music since high school,” says Ryan Thibodaux, a San Leandro resident and student at Cal State East Bay. “My wife will tell you that I listen to the same crap over and over again. My music life for the past few years has been extremely dull.”

When he heard Westergren interviewed on Air America a few weeks ago, Thibodaux went straight home and booted up the site. “I was in a sad-bastard music mood, so I typed in ‘Damien Rice.’ Within seconds, Pandora was playing me Damien’s ‘Delicate,’ followed by dozens of amazing songs from artists I’d never heard of,” he recalls — Geoff Farina, the Bevis Frond, Josh Ritter. Now once he hears something he likes, he bookmarks it to buy later. “My Amazon wish list has gone from zero items to more than twenty in two weeks,” Thibodaux says. “I may need a second job soon.”

Through their partnerships with iTunes and Amazon, Pandora execs know the site is boosting music sales, but they can only guess what effect it’s had on illegal downloading — whether people who hear music on Pandora are then seeking pirated versions elsewhere. Nevertheless, they get occasional e-mails from someone like Wendell, a systems administrator and inveterate file-sharer from Perth, Australia (no last name given, for obvious reasons), who wrote in to proclaim, “Pandora has completely gotten me off illegal mp3s!” Wendell says he used to download whole albums at a time to “preview” them, throwing out the ones he didn’t like and eventually buying those he did. Now, he says, Pandora lets him do much the same thing legitimately.

Right now, Kennedy says, it’s clear most Pandora listeners use the site largely at work — one of the company’s next goals is to make it available on portable platforms, the way radio is. And while Pandora’s music library is respectably large, expanding it further is another priority. The company’s world music genome is just launching, starting with the Latin category, but it will probably be a while before all those Celtic and Russian music fans get well-stocked stations. Classical, too, is still a ways down the road. Even in the well-established pop/rock genome, die-hard listeners can easily name missing favorites: Thibodaux, for example, couldn’t find local band From Monument to Masses, and Weyer wonders why there’s no Poi Dog Pondering. “You can certainly find some thin areas of the genome, and we’re trying to patch those as quickly as we can,” Westergren concedes.

But here is something they’ll never do, despite many fevered queries to the contrary: mine the database for the blueprint to the perfect pop song. “Anyone who tells you they’ve figured that out is full of crap,” Westergren snorts.

Yet this suggests a niggling worry about the whole databasing concept, a sense that art can’t be totally reduced to its component parts, and that something can be similar without really being the same. Anyone who’s had the experience of typing in Fabulous Band A and getting back Third-Rate Band B knows there’s something — virtuosity, perhaps, or originality, or spirit — that makes great bands great, and imitators only imitators. But what is it, exactly?

“What if you went into the studio and you exactly mimicked a Pixies song and it was a perfect mimic, doing the same guitar sound, same tempo, same everything?” Zapruder muses. “Even if you could make the exact same sounds … I don’t know. There is something else. It’s more than the sum of its parts. That is one thing that even we can’t lock up or quantify.”

That’s okay. The Music Genome Project isn’t claiming that musical tastes are rigidly predictable. Pandora is meant to be a bottomless gift box that keeps on giving, adjusting its approach until it finds something to make your heart sing. And maybe that’ll happen because it turns out that you really do respond to extensive vamping and mild rhythmic syncopation, and the machine has somehow got your number. Or perhaps it’s that mysterious something else. After all, isn’t there something truly ineffable about the music you love, something your ears and your heart instantly recognize, yet can’t be described for a database, the same way the person you are can’t be totally described by your DNA?

“Absolutely,” Westergren says. “And God bless it.”