What makes a #1 hit song? | Did You Know?


It’s a complaint I hear a lot these days. Same effects on the vocals. Same patterns. Same “boots-and-pants” drum beats. Who’s writing quality songs these days? Where’s the next Bob Dylan, or Jeff Buckley, or Bjork, or Tracey Chapman? Who is doing different and inventive stuff? Well, lots of people, actually. According to science, the key to landing a
number one hit in the US is not to copy what came before. Quite the opposite, actually. Some academics looked at nearly 27,000 songs from Billboard’s Hot 100 chart over the past 60 years. They wanted to see what makes a song popular. Is it just your image and how much money your record label puts behind you? Can you just push out the same old crap? Do the actual contents of the songs matter? Can we predict performance just based on these external factors, social media, institutional backing, what record label you are a part
of etc. So what they found was the tracks that hit the top spot in any particular year were actually pretty different from what was considered
an “average” song the previous year. To stand out from the pack, you have to do something new. Yes, this sounds pretty obvious – but it’s
not just any kind of “new”. For the research, they had a computer analyse various characteristics of the songs. There’s basic features, like tempo. And instrumentalness, or how much of the song is wordless. Then there are some more interesting characteristics, like danceability. How easy is it to maintain your groove to
on the dance floor? The more regular the beat, the better. So, for example, if a song had a section where there was no beat, it was penalised. This wasn’t really a problem for hip hop,
where the beat rarely dies. But American Pie waits a minute and a half before the drums come in, so it scores pretty poorly on danceability. Another interesting characteristic was valence, which essentially measures the positiveness of a song. How vibin’ is it? Does it get you in a good mood? In general, songs in a major key sound more positive than those in a minor key. But there are other aspects of a song that
can also change the mood, such as the vocal tone, or the production. Believe, which topped the charts in 1999,
is in a major key, but listen to the lyrics and you realise it’s anything but celebratory. Here’s some other examples from recent years. Uptown Funk was a hit in 2015. It was very positive, very danceable with
high energy – much more so than the previous year’s average. In 2014, you had All Of Me top the charts
– which couldn’t feel like a more different song. The tempo is very similar, but it had much
more of an acoustic feel, with less positivity and energy. The year before that, 2013, there was Blurred Lines. That song – which became the subject of
a major copyright lawsuit for its overall vibe – sounds like there was a raging party going on in the recording studio. It had none of the introspection of All Of Me. But again, both songs have the same tempo. So basically, what the researchers found was that – in academic speak – songs with “optimal differentiation” are the ones
that rise to the top. That means they were different to previous hits, yes. But not too different — they were just distinct enough stand out. So what can this tell us about the future
of popular music? Well, it involves – wait for it – artificial
intelligence. Right now, there are companies developing software they say will help you write better songs. They do this by feeding heaps and heaps of data into a computer to teach it about music – to the point where, eventually, it can
probably whack out a tune that sounds maybe better than a professionally trained musician could. A team at Google created AI Duet, software that will jam with you. “So in this experiment, you play a few notes. They go to the neural net which basically
decides, based on those notes and the examples it’s been given, some possible responses.” Scientists at Sony’s CSL Research Lab went one step better, they produced a whole song – in the vein of The Beatles – using AI,
though with A Little Help From Their Friends (a human composer). It raises an interesting question about authorship. If a computer program helped you pen a song calculated to be a chart-topper, can you really claim credit? Stephen Phillips is the founder of an Australian AI music company called PopGun, and he thinks so. “I don’t see the algorithm owning anything. I see more like it’s going to change and
make high-quality composition more accessible to more people.” That’s not the only way technology is changing music-making. At the other end of the computers-and-music spectrum is a burgeoning genre called “algorave”. People use live coding to make music on the fly. Shelly Knotts, of the UK duo Algobabez, literally builds waveforms as she goes using an audio programming language. “So, basically like writing the sounds from
scratch – so making the wave forms and combining them in different ways to create the sounds from scratch, so I would do the sequencing and pattern generation after that.” Clearly, when it comes to music and tech,
there is a lot going on. But for now, as much as data and AI might aid songwriting, the research suggests you can’t necessarily reverse engineer the anthem of the summer. Pop stars are continuously pushing forward. That’s what makes them stars. “Our analysis suggests that yes, you can certainly listen to what’s around you and try to be different and that will likely serve you well. But there is also an element of randomness and an element of art involved that you can’t just scientifically determine what will become a hit. There’s noise in that process, and for us, I think that’s a good thing.” We’ve included a playlist below to some
of the Billboard #1 songs that were a part of the study – one from every year the study covered. But we’ve also included in the playlist, some songs from a bunch of the worthy successors of Bob Dylan, Jeff Buckley, Bjork and Tracey Chapman that you might enjoy as well. Cheers!

Leave a Reply

Your email address will not be published. Required fields are marked *