Oh, you know this band, too?
Reference: https://vrs.amsi.org.au/wp-content/uploads/sites/75/2018/04/tobin_south_vrs-report.pdf
Ever meet someone who has the same taste in music as you? You’ll be talking, rattling of artist after artist, song after song, and everything seems to just click. Well, this connection with someone based on their music taste might stem not just because you two like the same things, but because of the actual collaborations, connections, and networks that musicians have between each other.
An analysis completed by Tobin South of the University of Adelaide called, Network Analysis of the Spotify Artist Collaboration Graph, uncovers complex networks, also known as graphs, as we learned in class, between difference genres and subgenres of music. The analysis breaks down the Spotify ecosystem. We can see that in this case node or vertices are artists that have published songs on the platform. In terms of edges, we can see that these are represented by instances where artists have collaborated on the same song, album, or EP as one another.
To further tie in concepts from INFO 2040, we can see that there are concentrated ‘components’ within the larger Spotify artist network. These components are conglomerations of nodes (artists) who are internally linked to each other (i.e., have collaborated in some way). However, we also know that there are such things as ‘strong’ and ‘weak’ ties or edges between nodes/vertices. The paper shows that we can consider ‘appears-on’ connections as edges where one artist includes work from another artist (like a track or a sample) but doesn’t take credit as being ‘featured’ on the song (like Doja Cat and Post Malone in their newest song).
An interesting feature that the article dives deeper into is the popularity of certain songs. It breaks down popularity as the number of total streams relative to others. This is an interesting case that we haven’t necessarily talked about in depth in class, but could look into further with delineating the types of ties artists have with each other, and whether there is a fixed popularity or if it changes based on who is the most popular at the time. Overall, the article seems to uncover that there are trends where certain genres are more concentrated than others.
We can see that these networks show many concepts learned in class. First is unbalanced and balanced triadic structures as I have labeled below in the example. We know that three ‘strong’ or positive ties creates a balanced structure and so does two weak ties and one strong tie; however, there can be examples of unbalanced triadic structure such as that of Anderson Paak., Kanye West, and Paul McCartney – where Anderson Paak. and Kanye are both featured on someone else, and Paul McCartney has songs directly with Kanye and Anderson Paak., separately. We see that rap (as stated in the paper) has a relatively high concentration since there are a lot of artists who collaborate with each other in rap. This is its own component. The other component is what is generally ‘not rap’ or pop for simplicity’s sake. My drawing is not perfect, but if I were to continue we would surely see these components more prominently.
While I am not making any sweeping assumptions, it would be interesting to further research how these networks translate to the listening behaviors of people. For example, if I like 21 Savage, I might find myself also liking artists that he collaborates with (Drake, Young Thug, Post Malone, J.Cole). It wouldn’t be surprising if at the same time, a fan of Young Thug also liked 21 Savage (and others like Drake, and Post Malone). Conversely, those who like non-rap such as songs by Paul McCartney, might not know or listen to a lot of these rap songs, since he is buried more deeply in the non-rap component. Music is the universal language, so at some point, there is bound to be overlap. So, before you start judging someone based on their music taste, ask them, “Hey, do you know this band, too?”