Here is Alanna’s entry for Task 9: Network Assignment Using Golden Record Curation Quiz Data:

Alanna Carmichael – Task 9 – Golden Record Network Assignment

And here is my entry:

Task 9: Network Assignment Using Golden Record Curation Quiz Data

In the Task 9 Network Assignment, Alanna and I were group ‘friends’. The program algorithm had decided we deserved to be lumped together based on the song choices we had each made. We had some similar thoughts about this:

• In our group we all shared three selections, and several other selections were shared between the rest of us in pairs or sets of three.
• Most of each of our choices were not shared between all the members of our group.
• Each of us had at least one selection not shared by anyone else in the group.

From there we took different tactics to analyze the results of the class. I examined the linkages to the one song I had chosen that was not linked to anyone else in our group, and Alanna looked at the popularity of each track among the overall class. Alanna identified interesting patterns of potential cultural bias among the song selections, and I looked at how must people in the class are linked to each other through just one additional connection.

We also both identified the issue with the small sample size of the class when attempting to draw conclusions, and the lack of context that the data itself provides. We both identified that qualitative data is necessary to really make any judgement about why people made their choices and what those choices say about them. Without knowing the reasons behind the selections, any algorithmic grouping is problematic, flawed, meaningless, and arbitrary. Reading through our individual curation processes confirms this – Alanna and I had very different methodologies for choosing the songs that we did, and because of that, grouping us together based on the choices we made ultimately makes no sense if the grouping is meant to convey some sort of similarity between us beyond the superficial of the selected songs themselves.

The fact that we are grouped together by this algorithm shows the power and influence, as well as the problems with, opaque algorithms. It also highlights the problems with the prevalence of social network algorithms. Modern social networks group us together like this all the time, and the assumption is that it is meaningful, and that people in those groups are similar in interests, taste, personality, etc. This may not be the case at all, depending on the data used. It also shows the need to understand the algorithm and its purpose. Part of this may just be an incorrect assumption being made based on incomplete or incorrect knowledge of the algorithm. For example, by implying that the algorithm in this task is meant to group people together to imply that we should be friends is an assumption I am making, but the intention of the algorithm might be to simply form the largest groups possible based on shared song selections, without putting people in more than one group. It may not be intended to say anything about the people in the groups or any similarities between them that may or may not exist. As a text technology, algorithms are not well understood or well used – we are assigning meanings to the semiotics that they produce without fully understanding the intention and meanings that produce them.