1. Manual Network Analysis

Here I’ve mapped out Act I of Shakespeare’s _Much Ado About Nothing _using a simple network and chart. Note the importance of Leonato (a character that has very little bearing on the story itself), and the relative _insignificance _of Hero, a driving force in the plot of the work.

Lab 5 NetworkLab 5 Dialog Charts

  1. Network Analysis with Gephi

Working with Gephi demonstrates the importance of being familiar with a text before one starts poking it with digital tools. This lab in particular makes an excellent case (at least to me) for only visualizing networks that are already familiar through traditional reading practices and hands-on interaction.

To make what I am talking about a little more clear, I propose comparing the first four graphs of the lab with the last one. In working with the former, I struggled to get the interface to make the network say anything. I played around with various styles and colors in order to make the information in the network emerge from the lines and dots, but I still had very little idea what exactly the network should look like because I had no prior relationship with the data. That said, I realize that the Grandjean tutorial is just that and the point was learning the interface. Still, working on those first four problems left me feeling a bit alienated from the steps I was taking. Tweaking this color or running that procedure was completely meaningless to me. I simply followed the steps.

By contrast, the network I created from my own data was far less opaque. I participated in its creation from the csv files to the actual manifestation of the nodes and edges. Too, because I actually drew a network myself on paper that represented the data; I already had a solid idea of what I wanted the visualization to look like. One might argue that this runs counter to the notion that digital tools should be investigatory. I will offer a slight correction and say that visualization is investigatory. In that light, knowing what I wanted my network to look like was more appropriate because I had already experimented with visualizing it using pencil and paper. At any rate, being familiar with the material made coming up with a combination of colors, lines, and patterns far more engaging. Too, I felt my network actually demonstrated the hypothesis that came out of crafting it in the first place (Leonato’s understated significance).

This last point leaves me wondering about the nature of network analysis in the larger sense. Literally I am wondering whether or not larger networks are truly helpful for understanding the data they hope to visualize. While it might be true that the appeal of digital tools is their ability to operationalize and process large swaths of data that traditional methods cannot hope to tackle, I wonder what is actually coming out of these gigantic analytical schemes. Yes, analyzing networks does provide a clearer picture of the relationships between potentially massive numbers of agent. But what does one know without getting closer to the data? Without zooming in to see the local collections of nodes and edges as they feed into the larger whole? I think what I might arguing for is a closer approach to network analysis that attends to these local realities before moving out to discuss broader relationships. This is not really a critique of the Grandjean tutorial (it couldn’t be because that’s not a complete project and he doesn’t present it that way) but more of a commentary on the Cordell article on reprinting in antebellum America. I will say that the So and Long piece on 20th century poetry does a better job of getting down to the local nitty-gritty before doing a comparative analysis and drawing conclusions about the larger relationships. Cordell, by contrast, is after something that cannot be provided by close reading and therefore never really bothers getting closer. I do not mean to overreach and say that distant reading is a completely invalid method or that Cordell’s work reveals nothing. I merely wish to point out the benefits to remaining closer to the data being visualized. I might have liked to see a breakdown-by-region (just for example) with a discussion of how those local dynamics shape the larger whole. I think this plays up what networks have to offer analytically because it demonstrates how the smaller patterns permeate the whole picture. Falsification is ambitious and that ambition ultimately alienates the data from the truth.