Song formats, tempos and structures have changes to suit the market through the ages of Motown, Mariah Carey and Drake. Credit: Herbie Knott/REX/Shutterstock
Spotify is expected to float on the New York Stock Exchange in the coming weeks. Shares sold privately last year value the Swedish music streaming service at between $6bn and $23bn. Dominant in the streaming market yet still making losses of $1.5bn last year, Spotify’s financial worth is a riddle.
Its musical value is also contentious. Last year Hubert Léveillé Gauvin, a US doctoral student in music theory, published research into song intros. Looking at US top 10 hits between 1986 and 2015, he found their average length had dropped from more than 20 seconds to as little as five. The report received widespread coverage, mostly threnodies for what the BBC called the “dying art of the great song intro”. Streaming was collared as the villain: Spotify pays out for a song after 30 seconds of play, an incentive for songwriters to cut to the chase.
Format has always been critical to pop. The three-minute length of the archetypal pop song is linked to the amount of music that vinyl records could contain, a limitation dating back to the first days of the phonographic industry. Motown Records designed songs so as to sound best on car radios (the concept of “drive time radio” emerged in the 1960s). Today’s hits are similarly engineered for headphones and mobile phone speakers.
Cassandra-like commentators warn that Spotify and its ilk are promoting a low-attention-span, algorithmically-led musical culture, the malign overflow of Big Data into our listening habits. The fears chime with broader feelings of foreboding about our tech-driven world. But is streaming really changing the structure of songs?
Last year Mark Bannister, a London-based data analyst, published analysis of every US number one in the Billboard charts from August 1958 to April 2017. The data were taken from Spotify’s publicly available store of data about its song library. The results are catnip for pop’s version of the trainspotter, the chart obsessive. “I’m a big geek about that as well,” Bannister says.
His findings do not support the idea that streaming is radically reshaping songs. The shortest hits, averaging under three minutes, were in the 1960s, while the longest ones on average were in the 1980s, the decade when “attention deficit disorder” first appeared in the American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders. During our own supposedly attention-degraded times there has been a gentle dip to just under four minutes.
Tempo has fluctuated since 1958, but not in a way mappable to notable format changes such as the arrival of CDs in the 1990s or streaming in the 2010s. Broader musical trends are at work. “In the 1960s and 1970s when you had rock and disco, it climbed up,” Bannister says. “In the 1990s when number one singles were dominated by Mariah Carey and Boyz II Men, it dropped.”
Loudness — compressing audio so as to make it more blaring, or “hotter” in studio-speak — climbed steeply at the turn of the century when digital music recordings supplanted analogue. But the trend rate since Spotify launched in 2006 has been broadly flat. The one area of dramatic change is swearing, the use of which has shot up in lyrics recently. Freed by streaming from relying on radio airplay, the likes of Migos’s “Bad and Boujee” appear to have ushered in the age of the profane chart-topper.
DJ Marshmello performs at Spotify’s ‘Best New Artist Party’ in New York in January 2018 © Roy Rochlin/Getty
New York consultancy Hit Songs Deconstructed take a microscopic approach to breaking down hits. A resource for music labels, courses and songwriters, the company conducts in-depth analyses of songs in the US charts. At the FT’s request, they provided a breakdown of compositional trends in the US top 10 since 2013, when their database begins.
Intros have not shrunk in the past five years, remaining broadly constant at about 13 seconds. The average time taken to get to the first chorus has been steady at almost 40 seconds. The average number of choruses in songs has not significantly changed. The only ammunition for the hypothesis that songs are being front-loaded to reach the Spotify 30-second-payment mark is an upwards tick in the proportion of songs in which the chorus, the hookiest part of a song, is placed before the first verse.
The main music trend of the 2010s is the supremacy of hip-hop. One-third of hits last year were by rap acts, including half of the US number ones. Over the past five years, hip-hop has doubled in prominence as an influence on other genres in the US top 10. The most streamed performer last year was the Canadian rapper Drake with 6bn streams. He was also the most streamed of 2016.
Hip-hop’s pre-eminence has been over 30 years in the making: streaming is not the cause. The real transformation that the likes of Spotify have brought about is to sabotage traditional measures of a song’s popularity, its sales figures.
Chart rules are constantly contorting themselves in order to incorporate streaming. The latest change, introduced this year in the US, privileges paid-for streams over free content such as YouTube. Meanwhile albums have swollen to unimaginable sizes, like the 45-track behemoth released by Chris Brown last year, so as to game streaming figures.
We can therefore blame Spotify for hastening the death of the album and killing the charts. But hit songs themselves, pop’s traditional unit of currency, are proving more resilient.