Introduction

As a history graduate, I have always been extremely interested in how our past shapes and informs our present social, political, economic and cultural contexts. A large part of understanding history and applying it to the present comes from understanding relationships between state and non-state actors; in other words, networks. As relationships are critical to understanding historical significance, I thought our networks lab presented a unique opportunity to further my understanding about some of the relationships I found most fascinating during my time as an undergraduate. I chose to focus on military alliances in 20^th century because that is my specialization. Despite having studied the topic for three years, I actually found it very hard to visualize and contextualize the global entanglements that developed and informed the wars and their different natures throughout the 20^th century. I was also intrigued by military alliances because the very nature of these alliances shifted dramatically over the course of the last century; from colonial and self-determination alliances to ideological and geographic-led alliances. By assessing a network of military alliances, we can begin understanding:

• How countries, at varying stages of their history, may have allied themselves according to different priorities;
• How colonialism often brought geographically distant countries together in war;
• How military alliances have actually changed over the last century, and whether the same countries or regions tend to be repeatedly allied;
• How these military alliances may have informed non-military contexts in the present, e.g.: are our present economic and political alliances effectively military ones by another name?

Visualizations that Informed My Design

Surprisingly, I found few network-specific visualizations about historical military alliances and historical networks more generally. Given the amount of available research and data in this field, I assumed this topic would have been explored further in data visualization. As a result, I considered network designs that did not rely solely on historical network/military data to inform my designs.

The “Histomap” (1931) – a small section shown below in Figure 1 – is an impressive and ambitious attempt at illustrating the relative power of states, nations, and empires over a four thousand year history. I felt this visualization was relevant as it approached the topic from the perspective of comparison. We are able to clearly see the rise and decline of civilizations over time and their relation to one another. It informed my design specifically in highlighting the importance of color, size, and perspective to quickly recognize country differences. While color and line demarcations make comparisons clear, it is also misleading that the same color was used for more than one civilization, despite there not being a stated connection between these two anywhere on the map. Therefore, I understood that while colors should be used to highlight country relationships, similarly related and connected countries are the only ones who should share colors.

Another visualization that informed my design was a network illustrating the global arms trade (Figure 2). This was helpful as it showed that networks based on geographic location do not necessarily need to be superimposed on a map to be highly effective. The thickness of the lines alone gives a clear and quick indication of the biggest arms exporters. It also informed my design by showing how vital label spacing is in visualizing geographic information. Furthermore, having different geographic regions grouped by color also proved useful to convey information quickly. These are all visual elements that I sought to incorporate in my network visualization:

Materials

To produce this visualization, I used Gephi. The data was compiled by myself, using this timeline from Wikipedia as a starting point. In order to create my own dataset and ensure it was clean and readable in Gephi, I used RStudio and OpenRefine to sort, organize and format the data to a network-specific table. Once the visualization was finalized, I downloaded it as an SVG into Adobe Illustrator to improve label display readability.

Methodology

I chose to create my dataset for the following reasons:

• It gave me more flexibility to pursue a topic I was interested in;
• It gave me a chance to practice creating my own dataset, with all its pitfalls, ahead of the final project – in case I chose to go down this route again;
• It allowed me to run an R script for the first time and start getting acquainted with the language and its intricacies;
• It gave me more confidence in my data because I knew that I had created it and cleaned it according to best practice parameters discussed in class.

Once I found the data I wanted in Wikipedia, I copied the timeline into an excel document. I then removed all group headings, dates and descriptions about the military alliances, leaving only the treaty name and countries involved. I then transposed the data so that each military alliance had it’s associated countries as comma-separated values in the same row. I then replaced all of the treaty names with a numbered list (Figure 4) in preparation for running the R script:

I then permuted the network data in RStudio to co-relate the multiple countries stored in a single column. After running the script, I imported the file into OpenRefine to remove any countries that had been associated with numerical IDs, trim leading spaces, and generally ensure the data was clean and correct. Once completed, I imported the CSV file into Gephi to build the network data visualization. When the visualization was ready, I exported it as an SVG file into Adobe Illustrator where I moved labels to improve readability.

Data clarifications:

• As several countries underwent name changes during the 20^th century, these have been kept separate in the network. For example, where Russia entered a treaty as the ‘Soviet Union’, this has been kept as a separate label, with its time-contextual relationships, to the label ‘Russia’.

Results and Interpretation

The results largely confirmed my assumptions – that military alliances would closely follow geographic lines. Western European, Eastern European, South America, the Caribbean and Asian countries and largely grouped together.

However, by keeping historical name labels separate for the same country, we are able to see a single country’s military interests change over time. The Soviet Union is grouped together with several South American states, suggesting that the nature of its military alliances were largely designed to spread geographic influence and bolster a contingent of Soviet proxy states close to the USA in its sphere of influence. However, once the country transforms from the Soviet Union into Russia, it’s military alliances shift to groupings more closely aligned to Eastern and Western Europe.

The network also illustrates the effects of colonialism, with India being placed in groupings largely with Western Europe, likely the result of entering military alliances as part of the British Empire. Nigeria and Ghana, both former British colonies as well, also appear to have stronger military alliance relationships than former French colonies, such as Senegal and Niger.

One surprise from the results was that despite the high amount of interconnectedness, there were still geographical regions that managed to remain entirely removed from military entanglements in other regions. Despite several of the Caribbean countries being colonies in the 20^th century, militarily, they are only related to each other. Similarly, Central Asian countries have an isolated military relationship, largely consisting of each other (with the exception of the Russian Federation and People’s Republic of China).

Reflections and Future Directions

I thoroughly enjoyed this lab – by getting to create my own data set and answer a question about historical military relationships that I had often considered, I felt more invested in my data and fulfilled with the final result. I did find working with Gephi challenging. As with any new tool, there is an exploratory phase where I wanted to understand its limits and how it could help me accomplish my goals. It was awkward to have to create several versions of the same network to try out different functionalities. Nevertheless I think the overall result illustrates my intentions well.

I think there are several future directions that would be interesting to take this data in. It would be interesting to superimpose the network onto a map, to perhaps more clearly illustrate the importance of geographic proximity in dictating alliances. I also think having a time-based component to this network would be incredibly interesting. By seeing how different countries, empires, and regions evolved through the 20^th century, we can also see how their strategic and military interests changed over time. It might also be interesting to do a similar visualization of political or economic relationships to see if these mirror or are influenced by military alliances.

Network Visualization