What Does the Network Graph Tell Us?

The network graph displays which songs each student (curator) picked for the previous assignment. Some songs were more popular than others, but each song was chosen by at least 3 curators. Track 25 was the most chosen with 15 edges connected to it while Tracks 8, 17, and 27 were the least chosen with three edges each.

Table that lists Golden Record Tracks along with number of edges (connections).

The network graph can also be sectioned by communities. I noticed that the .json file had data that pre-assigned curators into the six communities. I asked Ernesto how he came up with these communities and he let me know that using a tool called Gephi, he applied a network modularity algorithm to the data. Network modularity groups nodes into modules (i.e., communities) based on the density of edges in the group. So, if a group of nodes have a higher than usual number of edges (i.e., higher than what would be found randomly) the algorithm will group those nodes into a community (Wikipedia, 2023).  However, modularity algorithms are not fool proof. They sometimes find communities in completely random graphs and cannot detect small communities (Wikipedia, 2023). Using the Table function in Palladio, we can see that all the curators belong to at least four or more communities.

Table that lists curators’ names along with which communities (1-6) they belong to.

What Does the Network Graph NOT Tell Us?

The network graph does not give us context. In this scenario, we do not know why curators picked the songs they picked. There were some trends with reasons curators picked certain songs such as songs with or without lyrics, songs with certain sounds or instruments, or choosing songs based on geographic origin. In my case I picked eight of my songs using a random number generator. Therefore, the reasoning behind curators’ choices is quite varied and cannot be inferred from the graph. We are also left in the dark about why curators cut songs from their list. Tracks 8, 17, and 27 were the least chosen but we really don’t know why.

The communities generated by the algorithm also do not tell us much about the relationship between the curators. Yes, we picked some of the same songs but again this does not mean that we have similar musical taste or used the same reasoning process when creating our lists. We just happen to pick the same songs.

In the picture below, both Simon and I chose the same four songs as part of this community (#2), but I highly doubt we listen or appreciate the same songs on the Golden record, especially since most of my choices were random.

Network graph showing four songs that both Olivia and Sam chose.

How Does this Relate to the Internet?

If we applied this model to the internet, we would expect that sites (in this case nodes) that have more edges would get more traffic than nodes sitting on the outer edge of the graph. Curators who belong to more communities would receive more traffic as well. But should these nodes (i.e., curators and songs) receive more traffic and be higher up on the result page list than other, less connected nodes? In this case, I would say no because music is subjective, and the songs and curators should have equal weight in this network. Just because a song or person has less connections (edges) does not mean they are not worth finding in a search result. However, when I am searching for information online, I want to find information that is relevant to my search and is also secure (i.e., not lead me to a site full of spam and viruses). In this case, I can understand why algorithms are used to weigh connections and create communities.

The issue I notice when I’m searching online is that it feels like I am only given access to very small portion of the internet. When I search for certain topics, it is the same few sites that come up in the search results, often coming from American sources. I’ve also noticed when shopping online, the same sites come up and it is quite difficult to find local or small businesses. In the end, this vast network which is the internet has been narrowed down significantly to limited communities.

References

Wikipedia. (2023, October 4). Modularity (networks). https://en.wikipedia.org/wiki/Modularity_(networks)