Generally, the future is so chaotic that there's no way of guessing what will come next. But one area is so obvious, so repetitive, so completely and utterly predictable that it's a breeze to guess what's coming next: pop music.
Scientists have been trying for a while now to come up with a way to predict the next big pop song. It might seem like a trifling matter, but it's a chance to predict group behavior in an easily quantifiable way. Sure, it's not exactly psychohistory, but it would still be pretty cool to have a computer algorithm that could accurately predict the future, even if it's just the pop music charts.
Dr. Tijl de Bie and his team at the University of Bristol's Intelligent Systems Laboratory believe that is now within their grasp. The researchers built up a "hit potential equation" that analyzed twenty-three different features of a given song, including tempo, time signature, song length, loudness, harmonic simplicity, complexity of the chord sequence, and the "noisiness" of the song, among others.
The next step was to look at the UK pop charts through time, giving different weight to the various parts of the equation depending on what sorts of songs were popular at any given time. The researchers could then take a given song and predict how successful it would be in a given year based on how similar it was to the songs that were charting.
They found that the algorithm could predict a song's success with an accuracy rate of 60%. Not perfect, maybe, but considering how unpredictable pop culture's shifting tastes can be, 60% is pretty damn impressive. The team has also set up a site that puts their algorithm to work, predicting the next big hits and flops while also breaking down how well their equation predicts the success of songs throughout the history of pop music.
Dr. de Bie points out that song tastes change over time, which means the equation has to as well. The evolution of that equation reveals some intriguing trends in popular tastes. For one thing, the equation was least successful in predicting hits around 1980, while it was at its best in the early nineties and after 2000. The researchers suggest that this means the late 70s and early 80s were a particularly creative, innovative time for pop music, making it harder to guess what would be a hit, while the early 90s and last decade were more uniform and, implicitly, a bit boring. I'll leave you all to draw your own conclusions on that one.
The researchers also found that slower musical styles were most likely to be a hit in the eighties, which does line up nicely with the golden age (if you can call it that) of the power ballad. Until the early nineties, pop hits tended to be simpler harmonically than other songs from the era, which then shifted to songs with simpler rhythms. Songs are also getting louder over time, and the public preference for louder and louder songs seems to be increasing. I must say, as a lover of reasonable volumes, I'm very upset with all of you for that last one.
Via the University of Bristol. Image by motumboe on Flickr.