Ah Christmas! That special two-month stretch in a music analyst’s life when 80% of our workload includes sleigh bells and mentions of holly and mistletoe. While the occasional holiday track trickles into our queues throughout the year, sometime around October first, the slow drip precipitates into a flurry, then a blizzard, culminating into a complete white out right about now. Personally, I love it. Just like spending a day analyzing Hawaiian music can feel like a miniature tropical vacation, total immersion in Christmas songs extends my season of joy and goodwill manyfold. And with the annual return of attention to our pine-scented catalog, come the perennial questions we’ve been mulling for as long as I’ve been here. The top of the list (the low-hanging ornament, so to speak) is a discussion of the instruments that most commonly define the Christmas sound. We all know about sleigh bells, glockenspiel and tubular bells, but what other obvious signifiers should we include in our hunt, if any? Harmonized trumpets? Synthetic string pads? Wood blocks and that whip-crack sound from Sleigh Ride? Moving along to some of the old chestnuts that seem to get trotted out with a clipity-clop each year: If something is a children’s Christmas song, should it be siloed into a children’s station or a Christmas station? Well, when it comes to Christmas, who doesn’t want to hear “Rudolph the Red Nosed Reindeer”? We’re all kids at heart this time of year, so that one’s easy. But what about if it’s a funny song, like the Elmo & Patsy classic, Grandma Got Run Over by a Reindeer, Spike Jones’ beloved, “All I want for Christmas,” or my personal favorite, “I Want a Hippopotamus for Christmas”? Comedy music or Christmas?  (Extra bonus aside: is making a joke about Grandma getting run over by a reindeer offensive?)  The guiding star we always follow, year after year: Christmas wins. Christmas always wins. Moving on to some tricker issues to unwrap, classical music presents its own set of challenges. Handel’s Messiah and The Nutcracker are Christmas staples, but they’re also canonical works that listeners might enjoy on playlists all year round. Conversely, the Ode to Joy from Beethoven’s Ninth Symphony has become associated with the holiday, but the rest of the Ninth? Not so much. And we’ve had to invent novel ways to best capture that a purely instrumental version of “Come Oh Ye Faithful” or “Silent Night,” has religious connotations, without the use of lyric tags. Speaking of instrumental music, what about new age tracks that allude to Christmas in the title, but are otherwise indistinguishable from general new age music? This kind of stuff inevitably lights up a conversation about how strongly, if at all, a distinction needs to be made between secular (Rockin’ Around the Christmas Tree) and non-secular (Away in a Manger) Christmas songs? Or is all Christmas music intrinsically non-secular because of… ya know… it’s Christmas? A few other loose questions jingling around: Is it truly a Christmas miracle that we have almost 3,000 versions of Bing Crosby singing “White Christmas” in our system? Is Eartha Kitt purring Santa Baby too naughty to play on a children’s station?  How many human hours have we spent checking playlists twice to weed Christmas songs out of non-Christmas playlists? Are Christmas lyrics inherently nostalgic? Is a reference to drinking mead considered offensive? Deck the hall with boughs of holly, Fa, la, la, la, la, la, la, la, la! 'Tis the season to be jolly: Fa, la, la, la, la, la, la, la, la! Fill the meadcup, drain the barrel, Fa, la, la, la, la, la, la, la, la! Troll the ancient Christmas carol. Fa, la, la, la, la, la, la, la, la! Inevitably, we find ourselves several meads deep at a company holiday party, engrossed in a straight-faced academic debate about whether or not Frosty the Snowman qualifies as a monster, or if Santa can be described as a mythical creature (for any children reading, obviously not, because he is 100% real). Things get foggy parsing the sledful of songs that have been magically transformed into Christmas music simply by sheer force of repeated association with the season, like Greensleeves, O Tannenbaum, My Favorite Things and Hallelujah (which the Osmonds even went so far as to change the lyrics, shifting themes from Old to New Testament). There are, of course, those wintery ditties that don’t even mention Christmas, but you just KNOW, like Let It Snow, Baby it’s Cold Outside, Winter Wonderland. Heck, even Jingle Bells isn’t technically about Christmas! Real talk; is there any song that uses the word “jolly” that isn’t a Christmas song?  Lastly, what about songs that take place during Christmas time or mention it, but you might not stuff your stocking with them? Is Joni Mitchell’s heartbreaker, “River,” a Christmas song, with its Jingle Bells quote and mention of Christmas in the opening line? Howsabout “Christmas Card from a Hooker in Minneapolis,” by Tom Waits. Even if you think so, they surely don’t belong on the same playlist as that aptly-named icon, “The Christmas Song.” There’s so many of these, that it’s almost a genre unto itself. I thought I’d end here with a mini Christmas-Tinged Non-Christmas songs playlist. This is by no means a comprehensive list, so please add your favorites in the comments!! ... View more

As I mentioned in my previous blog post celebrating Pandora’s 20th Anniversary, Steve Hogan, the Senior Director of Music Analysis (and my boss), happens to be the most senior member of the Pandora staff, clocking in at the quarter century mark, making me look like a noob with my mere twenty-one years. According to his memory, he was the 18th music analyst hired to the company, back in September of 2000, roughly nine months after the founding of the company.   Steve leading an important meeting   So, in addition to having steered Pandora’s ambitious Music Genome project from the beginning, Steve is basically the company’s de facto oral historian. I thought it would be interesting to share some of his memories from the early days. First, some context: January, 2000 was the beginning of the end of the initial Dot-com Bubble. Right as founder Tim Westergren was starting to ramp up his new company, Savage Beast (our name before Pandora), the oceans of venture capital were starting to dry up. In fact, right around the exact time that Steve was hired as a Music Analyst, the Dot-com I was working for (Modo) was going under, as were many, many others.   Photo by the author   But none of this was on Steve’s radar at the time. He was just a musician in his twenties, scraping by in the Bay Area music scene, when he came across a flyer on a telephone pole in Berkeley. “It said something like, ‘Hey Musicians! Looking for the perfect day job?’” The original “office” was Tim Westergren's apartment, but by the time Steve got hired, they were set up in an actual office in a quiet corner of Downtown Oakland, a couple blocks from the Ask Jeeves offices. “We were analyzing songs with paper and pencil,” Steve recalls, “then we’d hand the pages to temps who would manually enter them into spreadsheets.” The Music Analysis crew grew to about 32 employees, and eventually they got a computerized system in place to record the analyses. In 2001, with about 15-20 thousand songs analyzed, Savage Beast was hired to build a jazz discovery interface for a major company’s website. Things were looking up. Then the bottom fell out of the Dot-com economy. Westergren went into personal debt to float the company as long as possible, but according to Steve, “The money ran dry and paychecks became spotty.” One by one, analysts dropped off, until the staff was reduced to just one person. “It was just me,” said Steve. I’d come into the office once a week, I’d check out the Billboard 200 chart and I’d look at any new albums that popped up. Then I’d drive over to Amoeba records, buy the CDs and analyze a track off each one.” Then finally, late in 2002 just before Christmas, Steve recalled Westergren firing the entire company. “I remember him saying, ‘I am plum out of money.’ Before I left, he pulled me aside and said, ‘I still think I can do this. I’m going to give you a call.’ I had to go and get a temp job over the holidays. But right after the new year, I got a call from Tim and he’d gotten more money and he brought me back.” The lore is that Westergren pitched Pandora 360 times and received 360 no’s from venture capitalists. Then finally he landed a multi-million-dollar funding round. Steve remembers that day well. Westergren called a company meeting, “and he immediately came in with a stack of envelopes and paid us back for all the hours we’d deferred. There were about 50 people who had worked for little or no pay for eighteen months.” Westergren doled out one-and-a-half million dollars in salary that day. I asked Steve what had motivated him to stick through those many months without pay. “I just liked the idea. Tim was very convincing. I figured, let’s take a risk. Maybe it’ll work out and it’ll be a job that I like for a long time.”   ... View more

As you may be aware, Pandora is currently celebrating its 20th birthday! Before Pandora’s historic launch as the one of the first personalized online radio platforms, we were a company called Savage Beast. And Savage Beast’s big idea was something called The Music Genome Project.  The Music Genome Project has been extensively covered here and elsewhere, but, in a nutshell, Savage Beast’s founder, Tim Westergren had the idea that if you analyzed music on a song-by-song basis, that information could be used to create playlists that would help listeners discover new music they loved. The Music Genome Project worked, and here we are twenty years later, one fifth of the way to building the hundred-year company that Westergren envisioned.   San Francisco Chronicle/Hearst Newspapers/Contributor/Getty Images   One of the amazing secondary impacts of the Music Genome Project is that it created an engaging, sustainable day job for a bunch of serious musicians. Not only could they earn more than a couple drink tickets to put their musical training to use, but they were helping lesser-known music reach people who might actually enjoy hearing it. In fact, it was such a satisfying and rewarding job, that Music Analysts invested long-term in the project. Our small team of 22 accounts for a whopping 321 years of Pandora experience (that’s an average of over 14 years per person!!), many of us surpassing the twenty-year mark, actually predating the launch of the product, with the department head of our little family, Steve Hogan, recently hitting the quarter-century mark; the most senior member of the entire Pandora staff. Steve Hogan, Senior Director of Music Analysis   Over the coming weeks, I’ll bring you some early Pandora and pre-Pandora memories from a handful of us Music Genome long-haulers. The history of this company stretches all the way back to 2000, the tail end of the first wave of the online tech industry. It’s a story of innovation and perseverance and the transformation of how we all experience music in the 21st century. As for me, I recall being among the very last Music Analysts interviewed directly by founder Tim Westergren, in a small corner office in a very quiet downtown Oakland, a world away from the buzzing tech hub just over the bridge in SF. For my interview test (administered by newly minted manager, Steve Hogan) I had to analyze Sugar Magnolia by the Grateful Dead. I was left alone with a CD player, a pair of headphones, a Casio keyboard and a pencil and paper to work out the musicological details.       My start date was 10/18/2004 along with Jeffrey Burr who currently oversees our priority pipelines… and me (he’s also my manager). According to our databases, the first album I analyzed was Once by Nightwish, though I can’t say I recall the experience.     I do, however, recall the giant demo kiosks that occupied a corner of our cramped offices. One of the original pre-Pandora ideas was to set up these hulking machines at places like Best Buy where you could type in the name of a band or album you liked, and our Music Genome intelligence would recommend a different album for you to purchase at the store, based on your personal tastes. Needless to say, we took the business in a different direction. ... View more

As Music Analysts here at Pandora, we hear a lot of music. Like… a lot. Almost exclusively new releases, spanning a wide spectrum of genres. There’s an in-house tally where fellow employees share their total Pandora listening hours for the week, and let’s just say, any one of our analysts would be top contenders if it included our work-listening. This might sound like bragging, but no more so than a dentist telling you they see a lot of teeth. The long and short of having all that music pass between your ears, is that you might occasionally spot certain trends as they emerge in real time. One of our analysts, Hannah Glass, sensed a pattern developing among some recent hits. See if you can identify what Hannah noticed these three tracks having in common?   Espresso by Sabrina Carpenter   Not Like Us by Kendrick Lamar   Anxiety by Doechii   If you guessed sampling, well, we’re impressed. Watch your mailbox for a commemorative Honorary Music Analyst iron-on! (not really, sorry). Hannah was curious to see how pervasive this trend might be, so she checked out some charts around the stream-o-sphere and discovered that nearly every track on the Soundcloud All Music charts contained samples. More of the same over on Spotify and Tik/Tok. Some tracks, like the Doechii example above, feature samples of tracks that themselves are built on a sample. Samples stacked on samples.   Somebody That I Used To Know by Gotye   When I spoke to Hannah, she pointed out that, “sampling is something you expect to see in rap, but now it’s everywhere. Like trap drums. I remember way back when Britney Spears and Muse started using dub step. That’s when you knew it was a thing.”   Hold It Against Me by Britney Spears   The 2nd Law: Unsustainable by Muse   The more Hannah looked for sampling, the more she found it.  #1 Soundcloud Hip Hop chart: MulaManDam - The Answer #4 Soundcloud Electronic John Summit, Delirium - Silence #4 Spotify Pop Hits: Sombr - Back to Friends   Even The Cure is jumping on the bandwagon: All I Ever Am by The Cure and Mura Masa   While it’s beyond the scope of Music Analysis to determine why this might be, it is fascinating to think about the cause. Obviously, using an already recognizable hook gives you a head start on catching people’s attention, but not all samples are taken from popular back catalogs. And while widely available digital tools make isolating and transforming already recorded material just a few clicks away, the same can be said for generating new sounds with those equally accessible tools. We had an interesting conversation at our department meeting about the possible reasons for the uptick in sampling when Hannah presented her observations to the team, but we’d be super curious to hear you weigh in on why this might be happening! Looking forward to your thoughts.   ... View more

In response to my recent post about identifying songs that create a Scary Mood, someone asked me what the hardest or most time-consuming songs for us Music Analysts to deal with are. It’s a great question, or more precisely, two great questions, because time-consuming and hard aren’t necessarily the same things.     The assumption (even by me) would be that the more complex a track is, the harder it will be to analyze. Not quite true. At a certain point, complexity becomes simple to identify and capture. If a song has 15 different sections, we have straight-forward ways of indicating that the form is complex and that the structure plays a significant role in the identity of the song. It’s easy to hear when a track has extreme metric variations and odd time signatures, or hyper complex chord progressions, though it might take a while to describe it accurately. If a song has extensive instrumentation, like this Caetano Veloso gem, it can be fatiguing to identify them all and say what each one is doing, but we know what instruments sound like. That’s our thing.       It is, however, time-consuming. It can sometimes feel like you’re a paleontologist excavating hundreds of tiny bones of a skeleton with a set of tweezers. Slow, methodical labor, but not heavy mental lifting. A 26-minute prog rock epic takes FOREVER, but it’s not necessarily hard, and I can guarantee that it’s a labor of love for someone on our team. What makes something hard to analyze is ambiguity and the defying of expectations. Ambiguity can come in many forms. Sometimes it’s hard to tell if a piece of music is in a major or minor key, like this intentionally ambiguous piano piece by Bruce Wolosoff. Or perhaps the lyrics are so open to interpretation that you have no idea how to begin describing them. Beck’s early work was infamous for this. And sometimes, ambiguity occurs when the music is simultaneously more than one thing, like how this Anthony Braxton composition flickers between being an homage to John Philip Sousa and an experimental big  band extravaganza.      It can also be a real challenge when an artist known for a particular type of music, releases something completely different from anything they’ve done before. Especially when it can’t be easily categorized, like Andre 3000’s foray into… what exactly? There are some tracks that cross so many creative boundaries that they require specialists from multiple different genres to weigh in, like this avant jazz world fusion track that incorporates traditional Korean influence and instrumentation.      Then there are some tracks that confound on multiple levels, eluding existing definitions, like this genre-morphing abstract emo mind-bender by Lecx Stacy where it’s hard to tell if the sounds are real instruments or synthesized and half of the lyrics are inscrutable. While these kinds of tracks might push our analytical ears to the limits, they’re also likely to be the ones we add to our collections and revisit on our own time. ... View more

As some of you may know, there’s a handful of us here at Pandora who have this one-of-a-kind job where we listen to individual songs and identify the musical elements present. We’re called, “Music Analysts,” though that makes us sound either like music industry speculators or white-coated lab techs monitoring oscilloscopes with medical-grade headphones.     We’re neither. We’re just a bunch of musicians (I’d say that makes us the coolest department in the company, but so many of the other people who work at Pandora are musicians too; our company jam sessions are ridiculous). That’s not to say we’re not a serious bunch. Music Analysts have to pass a listening test when applying, and we all go through a rigorous training before we’re set loose on the catalog. Each song we analyze is gone over with a fine-tooth comb, multiple times, and we’re responsible for knowing and choosing from among thousands of descriptors when tagging a song.   I was about to write, “some parts of a song are easy to describe and some are hard,” but then I couldn’t come up with an example of something that’s easy. Figuring out exactly how fast or how slow a track is, can be very easy, but you’d be surprised about how much debate it can sometimes spark (where the heck is the downbeat?!). Determining exactly what instruments are present is often a no-brainer, although… are those acoustic drums or programmed drums?  A few of us on the team have been arguing about the distinctions between varieties of “twang” for over twenty years. Our Slack channels are filled with musicological minutiae that read like PhD thesis excerpts, but with emojis. And don’t even get me started about interpreting lyrics.   We thought, over the next few months, it might be fun to let you join in on some of the challenges that we on the Music Analysis team deal with hundreds of times a day. To kick things off and celebrate the season, we thought it would be fun to try to identify some Scary tracks.      To earn full-blown Scary, tracks need to consistently hit a certain threshold of eerie, creepy, disconcerting, terrifying or even horrifying. Which of the following tracks would you give the Scary Mood tag to?     Skeletal Remains by Bohren and Der Club of Gore Thriller by Michael Jackson The Disintegration Loops by William Basinski “The Munsters” Theme by Los Straitjackets The Becoming by Nine Inch Nails Main Theme (from A Nightmare on Elm Street) by Mark Ayres Grim Grinning Ghosts (From “The Haunted Mansion”) by The Melomen, Bill Frees, Betty Taylor, Bill Lee, Thurl Ravenscroft Witchy Woman by The Eagles you should see me in a crown by Billie Eilish Serial Killers by Gucci Mane   Comment with your Scary Mood tags below, and let us know your go-to scariest piece of music! ... View more