Photo by Markus Spiske on Unsplash

(and Drake is taking over the world)

Like Rob Gordon, I think that if we can understand popular music, the soundtrack of our lives, then maybe we can understand ourselves a little better.

With that goal, I’ve done an even longer analysis¹ of Billboards Hot-100 data, Spotify track info, and song lyrics to figure out what trends we can observe in music over the last two decades. The results are fascinating and point to some big changes.

Attention Spans are Shortening

In the age of Instagram and instant gratification on the internet, it’s not surprising that we have less patience than we used to. What is surprising is that this trend has been going on for over 18 years.

Using Spotify data to determine the length of a song and then weighing that data by how many weeks it spends on the Hot-100, we can see that the average duration of a track has been steadily decreasing each year.

From well over 4 minutes to now around 3 and a half, the average hit song is getting markedly shorter.

Similarly (as I showed in my last post) the length of song titles is changing. New hit songs are more likely to have 2 words in the title instead of 3

While the number of words in a track title has gone down, the number of words comprising a song’s lyrics hasn’t changed much. As such, there is even more lyrical content per song. Lyric density, or the number of words/second in a song has increased, especially in the last few years.

Hip Hop Isn’t Dead

Whatever Nas may say about the commercialization of hip hop, rap music is more popular than ever.

By analyzing Spotify and Wikipedia data, we can determine what genre best represents a musician. Cross referencing this with the songs on the Hot-100, we can see that the charts are currently dominated by rap artists.

Of course, since genre is assigned by artist and not by track, many of the most popular rap songs sound more like pop or R&B. So for example Ja Rule “singing” is considered a rap track since he’s a hip hop artist. With this consideration in mind, we can see that rap artists have been steadily gaining in popularity, while rock and R&B artists have been in decline.

The other big gainers on the charts are pop and electronic dance music. Since 2007, electronic music and pure pop have gained steam. At the same time country music has declined somewhat.

Music is Getting Sadder

In addition to providing us with a song’s length, Spotify’s developer API provides data on some qualitative features of a song. More info on the definitions of these track features can be found here.

One of the most interesting features to look at is a song’s valence, or its positivity. Looking at the weighted mean of track valence per year (by weeks on the charts), we see a clear downward trend. It seems that songs have been steadily getting sadder since the 90s.

Similarly, the overall energy of a song has been on a marked decline since about 2011.

Finally, despite a general downward trend in positivity and energy, danceabiltiy in a hit song has gone up since 2011. It seems that people like to dance more than ever.


As we can see, the last two decades of popular music have been dominated by a few observable trends. Songs are getting shorter and sadder. Hip hop and dance music have gotten more popular while older genres like rock and country have declined. Since 2011 especially, the average energy of a song has gone down while its danceabiliy has gone up.

One of the fun parts of analyzing all of this data was looking at the very specific genres that Spotify creates to categorize artists. In recent years, as the monoculture has splintered, these genres have proliferated. One genre that I saw over and over again was “canadian hip hop”. This genre represents one very popular artist in particular. He has dominated the Hot-100 charts over the past few years and is largely responsible for the rise of songs that are danceable, but sort of sad and low energy. Next time I’ll talk about how he exemplifies both popular music and popular culture in the 2010s.


1- All code used to collect data and generate plots is available on github at

Married engineer in San Francisco. Interested in words, networks, and human abstractions. Opinions expressed are solely my own.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store