In January the BBC Radio 4 programme Desert Island Discs celebrated its 80th anniversary. The programme is considered by many to be a cultural touchstone, with an invitation to be castaway a sign of recognition in your field. The list of past guests on the show reads as a who’s who of celebrated figures from the last 80 years. From Marlene Dietrich to Louis Armstrong to David Attenborough, film stars, musical and cultural icons have all featured on the show.
With over 3,200 episodes recorded to date, the longest of any British radio programme, the show’s back catalogue offers a rich dataset to explore trends over its history. For the latest instalment in our Turing Data Stories’ project we analysed 3,211 episodes to discover the developments of people and music on the long-running radio programme. Our project uses a mix of open data, code, narrative, visuals, and knowledge from the Turing to shed light on different topics in the world around us.
This story began with the discovery of an open dataset, compiled by Andrew Gustar from the BBC Radio 4 archive which covers episodes in the series from 1942 – 2020. Our analysis goes one step further and combines this with data on an additional 60 episodes gathered from Wikipedia and the BBC website (Feb 2020 – Aug 2021). We also used Spotify and Wikidata for supplementary information on music and guests. By using these additional sources of data we were able to do more in depth analysis on genres chosen and guest demographics.
The premise for the show is simple: guests share eight song choices along with the life story that accompanies them. At the end of the episode the castaway chooses a book, a luxury, and one song to save from the waves. But how have music choices changed on the show during its 80-year history? And have the guests themselves changed as well?
We found that music chosen on the show has evolved substantially over time. For example, classical music was largely dominant until the mid-1980s, but current choices are spread across a much wider range of genres. The trend for a wider variety of music chosen on the show accelerated after the year 2000 and rock music now represents a larger proportion of discs than classical music. Over the past two decades The Beatles consistently top the charts, joined by Frank Sinatra, David Bowie, Joni Mitchell, and Nina Simone.
We also determined that musicians and singers are more likely to pick classical music. Those, however, not in arts professions, such as athletes and businesspeople, are more likely to choose discs which fall into the genres of modern, pop, rock, vintage. Further information on how we defined these genres can be found in the full story.
There has also been a change in the professions of the castaways. In the early days of the programme, about half of the castaways were actors; while that percentage has decreased over time, the primary professions continue to be in the media and creative industries. Other influential professions in British society, such as politicians, athletes, academics and people involved in business, health, and law, still make up very small percentages of castaways.
There has also been a clear change in gender balance on the show, with male castaways greatly outnumbering female castaways. From the beginning of the show, until the early 2010s, about three out of four castaways were male and one out of four were female. Our investigation indicates that there was an abrupt shift to a more representative gender balance which only occurred about 10 years ago. The suddenness of this change is a notable finding from our analysis.
Our work points to some of the trends throughout the show’s long history. The combined data from this story is freely available at Humanities Commons - so you can have a look and do your own analysis on the data. Here’s to another 80 years of Desert Island Discs!
Read the full story or read past Turing Data Stories. We are working on a number of data stories on topics ranging from e-games to air pollution. One story on the horizon is looking at the UK Doctoral Thesis Metadata from EThOS, hosted by the British Library’s Research Repository.
The analysis from this data story was recently covered by the Mail Online, read the article.