Was it Marshall McLuhan (or Winston Churchill or John M Culkin) who said: ‘We shape our tools and thereafter our tools shape us’? After last week’s post on what the end of the silent movie era tells us about inflection points in the creative industry, subscriber Tom Goodwin dropped me a note commenting on how streaming had completely changed the parameters by which TV and Movies are made (length of episode, number of episodes in a season, how episodes are written). It sent me down a whole bunch of new rabbit holes relating to how streaming has changed our relationship to the consumption of music and film.
One of the most fascinating examples of this is popular music. Music streaming has impacted music creation in some interesting ways, arguably both good and bad. In his peerless analysis of how music has changed since the 1950s, Daniel Parris notes how the rapid advancement of form factor and storage capacity (vinyl to cassette to CD) led to an increase in average track length with song durations peaking in the mid-1990s. As streaming became more popular, song length begins to decline – a phenomenon which Daniel notes is likely down to the decoupling of songs and albums, and artists focusing on maximising individual tracks rather than cohesive album concepts, and the pay-per-play model which rewards the number of listens rather than total listening time (‘streaming ten 31-second tracks pays more than a single stream of a 310-second song’).
Source: Daniel Parris
Similarly, researchers Parth Sinha, Pavel Telica and Nikita Jesaibegjans analysed a dataset consisting of 211,000 songs initially released between the years 2000–2021 and also discovered a relatively steep decline in song duration beginning around 2010.
Parth Sinha, Pavel Telica & Nikita Jesaibegjans, average duration of tracks per year
The researchers indexed song popularity against song duration and further discovered that the most popular songs produced and released on Spotify also show a definite trend towards shorter duration.
Daniel Parris notes another trend that has risen alongside streaming. Song tenure (the average weeks in the charts) has declined while the diversity of artists on the charts has increased. In another words more artists make the charts for shorter periods of time. Streaming may have resulted in sharper spikes of popularity for individual tracks, but it also seems to have contributed to greater diversity in listening. An analysis of Spotify data and music charts across 39 countries revealed an upward trend in music consumption diversity that started in 2017 and spans across platforms. This suggests that, rather than homogenising global music tastes, streaming platforms may be facilitating a divergence with different regions embracing a wider array of local and international music. Another longitudinal study found that users exposed to a diverse range of digitally-enabled music recommendations developed greater openness to the genre over time, even if they were initially unfamiliar with it. Of course there may be other factors involved here (like TikTok), but the sheer accessibility of so much diverse music seems to be having an impact. Spotify alone has around 675 million monthly active users, more than 100 million tracks and over 6,000 genres of music (you can see them all here).
Streaming has impacted music composition in interesting ways as well. Spotify’s pay-per-play model means that a song needs to be played for 30 seconds to be considered a stream, incentivising artists to write tracks with shorter intros and more immediate hooks. As an example, this study used a deep learning framework to analyse song structures and found that modern songs often feature choruses and hooks earlier in the track to capture listener interest quickly. It’s perhaps little surprise that tracks are being written with this in mind when research has shown that listeners are more likely to skip songs during less engaging sections, such as extended intros, and that songs with immediate hooks tend to retain listeners longer.
And let’s talk about another significant impact of streaming. The pressure to generate streams and to be included into as many playlists as possible has led to the creation of a whole new type of music, dubbed ‘Spotify-core’. A genre-defying blend of mellow, mid-tempo, lo-fi or acoustic-tinged music, these tracks are engineered for passive listening and for pleasing the algorithm, and designed to blend seamlessly into multiple playlists, to have broad appeal, and to avoid triggering skips (basically any Spotify-compiled and recommended playlist with the word ‘chill’ or ‘vibe’ in the title will be full of it).
Spotify-core requires little effort on behalf of the listener, but serves to provide background for busy, anxious and stressed people whilst they are working or doing other tasks. As John Harris of The Guardian wrote recently, Spotify has no direct involvement in the creation of Spotify-core music but it’s algorithms and models are incentivising a whole production-line of suppliers who are specialising in so-called ‘perfect fit content’. Technology has always reshaped artistic creativity, he writes, but ‘what sets Spotify apart is something much more insidious: it goes beyond alterations of music’s forms into what we think music is there to do’.
According to journalist Liz Pelly, who recently wrote a book on the topic (Mood Machine: The Rise of Spotify and the Costs of the Perfect Playlist), Spotify realised as far back as 2012 that active listening was a much smaller proportion of listening than all the hours that users were clocking up using music as a background experience. This led the company to start promoting playlists as the background to a diverse set of activities including work, fitness, studying and even trying to get to sleep. In doing so they broke down traditional genre boundaries to define and group music by mood and function. But in pushing music as background, and to fit a particular mood, Spotify somehow managed to disrupt a bond which has existed since the dawn of popular music – it detached the music from its maker. As AI generated music offers up a tantalising opportunity for volumes of cheap, playlist-filling content, this phenomenon is unlikely to disappear anytime soon.
A while back I wrote about whether social media algorithms flatten culture, and whether innovation and diversity is stifled when creators are incentivised to produce content that aligns with algorithmic preferences. Where I ended up was to suggest that algorithms in fact play a dual role in both in cultural homogenisation and cultural diversity – they simultaneously encourage certain approaches and a rapid dissemination of global cultural phenomena whilst also providing a platform for cultural diversity and uniqueness.
I suspect the same may be true of streaming – it homogenises at the same time as diversifying. Of course it’s tricky to establish direct causal links when there are other factors that are likely involved, but streaming seems to have had a diversified impact, whether that be on song duration, song structures or even musical genres and the way that we think about music itself.
As we move into an era of far more ubiquitous use of AI tools it will be interesting to see how AI will reshape our expectation and behaviour.
We shape our tools and then our tools shape us.
A version of this post appeared on my weekly Substack of AI and digital trends, and transformation insights. To join our community of over ten thousand subscribers you can sign up to that here.
To get posts like this delivered straight to your inbox, drop your email into the box below.
Leave a Reply