Task 9: Network Assignment
Honestly, I found it challenging to answer the questions in this task. So, I began by simply listing my observations:
The entire class:
- I observed that none of the songs received unanimous agreement from all students. Both “Melancholy Blues” and “Johnny B. Goode” had the most connections, totaling 16 each. This could be because these songs are familiar to us as they represent well-known American music genres.
- Interestingly, the Senegalese percussion track received a strong vote of 15, although the reason for this is uncertain. Reflecting on my own choice, I might have been drawn to it because it’s a type of African music I’m familiar with and have heard before, and perhaps others felt similarly.
- Most other songs had varying numbers of votes, indicating a lack of consensus, which is to be expected given the diversity in musical preferences.
Analysis of my group also reveals some interesting patterns:
- We unanimously agree on one song, which is “El Cascabel”, and four of us agree on “The Wedding Song”
- Most songs in our group have only 2 or 3 connections, indicating limited consensus.
- Each of us has chosen at least one outlier song that no one else in the group selected. To me this suggests that choosing a song that no one else agrees with is just as common as selecting a song that we all agree on.
Why are these responses similar?
I’m not entirely sure. Out of curiosity, I peeked at some posts from others in my group and noticed that songs were often chosen for reasons different from mine. Some chose based on personal significance, others for specific musical aspects I wasn’t aware of. Overall, it seems the songs were selected to represent world music. Perhaps we share a similar understanding of what constitutes world music, but without an explanation, it’s unclear how that consensus came about rather than just happening by chance.
Is the visualization able to capture the reasons behind the choices?
No, I don’t think so. My interpretations (above) are based on educated guesses, but choices could be influenced by numerous factors such as shared interests or backgrounds that aren’t captured in the dataset. Without additional information, it’s impossible to make such inferences accurately.
Can the reasons for these “null” choices ever be reflected/interpreted in the data?
Similarly, I don’t think this dataset can explain why there are “null choices”; we need more information for that. However, I think that generally, this information can be gleaned indirectly. For instance, by identifying patterns, one can consider and infer reasons why certain groups didn’t respond a certain way. Understanding the dataset context would be crucial. I suspect the reasons for null choices are likely as complex and significant as reasons for making a choice, but equally difficult to interpret.
Reflect on the political implications of such groupings considering what data is missing, assumed, or misinterpreted.
I think political implications arise from how these visualizations are used or interpreted by policymakers, the media or the public. This is significant because these interpretations can then go on to influence policy and decision making. I find this realization a little concerning because in looking at these visualizations, there is a significant amount of data missing that could easily lead to assumptions or misinterpretations.
For instance, we lack the rationale behind the choices made (or not made). Without this information or additional context, it’s easy to assume that all song choices were based on similar reasons, potentially skewing interpretation. For example, without understanding the task behind this visualization one might assume our group is unified by a shared love of mariachi music and try to cater to our group in that way, or worse yet try to exclude others who don’t “like” mariachi music.
As mentioned in our videos this week (Systems Innovation, 2015), the metric of how connected a node is becomes crucial in assessing its significance within a network. If applied here, one might incorrectly conclude that Mariachi is more significant than other choices for this group, which I don’t believe is accurate.
I also question the nature of this grouping. For example, I’m uncertain about my place within this community because my choices were often impulsive and didn’t reflect strong personal feelings. As a result, I’m unsure if this community truly represents who I am. I anticipated more connections and similarities among members and the one song I truly liked (and which I felt was most representative of me), wasn’t well connected with others in the community. This leads me to reflect on the arbitrary nature of groups and how decisions can be made without a genuine understanding of a community involved.
Reference:
Systems Innovation. (2015, April 18). Graph theory overviewLinks to an external site. [Video]. YouTube.
Hi Steph,
Thank you for your analysis of the data. Like you I quickly noticed that we all selected El Cascabel and most of us selected Wedding Song, but that only led to more questions: what about the other seven people that selected El Cascabel that didn’t end up in our group, or the other five people that selected Wedding Song? I started looking at our non selections and noticed that all five of us did not select Brandenburg Concerto 2, Sacrificial Dance, or Flowing Streams. While this could be a potential reason, as you’ve mentioned it’s impossible to infer accurately.
Another grouping method I thought about was seeing how many of tracks I shared with all participants. I shared five tracks with you and Jonathan, while I only shared three tracks with Carlo and four tracks with Carol. There were three others whom I shared five tracks with that didn’t end up in my group. If I had the motivation and time, this could potentially be done with all participants, and that could perhaps give more data on how the algorithm grouped us.
Your point about the one track you truly liked wasn’t well connected in our group also highlights the deficiencies in our data. If we had to rank our selections and produce a weighted graph (which would have made both this task and task 7 far more difficult), we may have been able to obtain a visualization and grouping that better reflects our personal connects to these tracks.
Finally, like you I shared the same thought about not belonging to this group. While most of the group had diversity as their top selection criteria, I went with having vocals for mine. This further supports your point of the arbitrary nature of algorithm groups; my main take away is that we should be vigilant about trying to understand the processes behind these algorithms rather than allowing it to remain as a “black box.”