In 1985, the best-selling song in the UK was Jennifer Rush's The Power of Love. Thirty years on, it was Uptown Funk, by Mark Ronson, featuring Bruno Mars. From soft-rock power ballad to dance track, these were very different.

It's obviously difficult to take a sample of two songs and draw sensible conclusions about changes in popular music. But what about a sample of 500,000 songs? That's what scientists at the University of California Irvine have done, to track trends in the success of different kinds of song between 1985 and 2015.

The researchers made use of the burgeoning availability of large datasets, in this case the crowd-sourced online music encyclopedias Musicbrainz and Acousticbrainz. They analysed half a million songs released in the UK in that 30-year period and correlated chart success with the songs' acoustic features.

These were broken down into variables such as tonality, mood, danceability, and genres. The findings suggest there is a broad trend for fewer happy songs and more sad songs, while there has been an increase in the number of danceable songs. Yet while this kind of "big data" study can reveal insights about what we listen to, it's also important to look at how we listen.

Advertisement

Categorising songs as "happy" or "sad" depends greatly on social context and interaction. Take the example of one that topped the charts twice, 16 years apart, Queen's Bohemian Rhapsody. It's a complex production, not straightforwardly danceable, and sung from a murderer's perspective. Yet it's the source of much joyful group participation.

The way we consume music, and how that consumption is measured, has changed in 30 years. The charts are a lot less important now that the sheer amount of music available to listeners is greater than in 1985. Then, audiences relied on a comparatively small number of radio stations to hear new music. The charts were selected from a limited number of available singles and were much more prominent in people's everyday listening.

Today, listeners have the history of recorded music in their pockets and increased control over how it's playlisted and ordered to taste. The technology we use to listen to music has even altered our relationship with it, simultaneously expanding the parameters of musical choice and making the listening experience more intensely private.

Even though the charts have adapted over the decades, incorporating downloads in 2004 and streaming in 2014, they no longer represent the same measure of cultural dominance they once did. As psychologists Raymond MacDonald, David Hargreaves and Dorothy Miell note, there has been a "democratisation of musical styles in that the previous association of certain styles with 'seriousness' and others with 'popularity' no longer exists to anything like the same extent".

While the charts record mainstream success, they also interact with and are fed by musical subcultures that are often defined in opposition to that mainstream. They initially grow because they're different to what's in the charts but can eventually achieve success by building on that status, creating tensions with the original fans.

The limits of big data

The recommendation algorithms of large tech companies are increasingly a part of the process of musical and cultural choice, and the massive datasets associated with this are a huge resource for researchers. But the "popular" in popular music is more than a quantitative measure of consumption, and we can't just reduce it to aesthetic and stylistic components. We also need to take account of its social functions. And that means researchers from different disciplines — both arts and science — engaging in dialogue to analyse and interpret the data.

Music encoded as digital data is now feeding into the broader matrix of economic and political decision making, such as the Bank of England using it to take the economic temperature. So it's more important than ever that the social aspect of music's use does not get buried beneath the numbers.

• Adam Behr is a Lecturer in Popular and Contemporary Music, Newcastle University.

- The Conversation