Information flow analysis of a small community

Social media and the Internet have had a profound effect on developing and sustaining fan communities in the past 20 years. Advances of technology have allowed for cheaper and instantaneous communication between members of fandoms, allowing them not only to talk about their favorite shows, but also to present their creative work – videos, art, and fan-fiction, easier than ever. Moreover technology allowed for faster spread of new information about shows, and that is exactly what this post is about.

I’ve been writing a paper about fandom and the way they use technology to communicate, and thus I was tracking pretty much all the alerts/tags related to the series of interest. During the weekend also happened to be the annual MCM Expo Birmingham (16-17 March 2013), where actors from the shows I was following were attending.

Happily for me, I was able to witness how new information travels through a small community, who were able to learn new information about their favorite shipping couple1 from Warehouse 132. Since the show is scheduled to return in a month, the fans have been deprived of information about WH13 for some time. Furthermore, their favorite ship – Bering/Wells has just been heavily subtext, without being canon, so the fans were eager to know what will happen. During autograph signing on the MCM Expo, one fan asked Jaime Murray (actress playing  H.G. Wells – one part of the “couple”) about  Bering/Wells situation and recorded it. Soon enough it was uploaded on Youtube, from where that fan shared it on Tumblr:


This was posted on 16.03.2013 and of course just several hours later it was reblogged hundreds of times and another fan had made a gif that contained absolutely the same information, but with pics from the video. Both posts were reblogged and liked respectively 5853 and 1009 times by the time I decided to collect the user activity data4 from those posts.

The information that Jaime Murray provided and was published by the fan on Tumblr, generated massive engagement

Although I am aware that the text message and the gif one have different characteristics, I decided to incorporate the activity data for them and generate one analysis for this particular blog post today. My reason for this is that they essentially contain the same information (verbatim) and thus do not provide new and insightful info. What I did was only to add an edge between the original text Tumblr post and the gif one, signifying that the gif post found the information through the initial one.

Here, each Tumblr user(who is also shipper/fan) is represented by a node, connected through links to all other nodes it has reblogged the information from. Tumblr has played fundamental role in the sustainability of this particular community, which finds new information by either following another fan of the show, or by tracking “Bering and Wells”, “Warehouse 13″ or “Jaime Murray” tags on the blogging platform.

The collected data represents 727 nodes with 946 edges. The graph is directed. 


I also calculated Eigenvector centrality, to establish node influence in this particular network. As can be seen from the table, there are several nodes with very high eigenvector centrality measure. Due to the nature of Tumblr it is impossible to say, whether they have many connections and therefore have influence in this particular community, or just were reblogging the content in the right time to be shown on people’s feeds.


I also calculated betweenness centrality, because I suspect there are many users who act like bridges, connecting people throughout this particular network with a lot of important information passing through them. What came as a surprise for me when I looked through the data is that there are only 5 people with extremely high betweenness centrality, while the rest of the user base varies between 0-24.

Betweenness Centrality Distribution

Finally, I also wanted to measure how well this particular community can be broken down into modular communities. In this case, there were 23 smaller modules from the original community:



I was also able to see there are a lot of hubs in the network, connecting the smaller ones to a larger structure. Not all nodes from a sub network community reblogged the post from the original, instead they reblogged it most likely from people they have already followed, who reblogged the posts on the first place. Furthermore many community members have overlapping connections with other members (reblogged posts from various users). Generally, looking at the data, 1/3 of the users reblogged the content from a 2nd or 3rd hop on the network in isolation with the central players.

There are many studies focusing on network diffusion processes, focusing on spread of epidemics, decisions and innovation. I wanted to focus on how new and exciting information travels through a small but extremely dedicated and active community. In other words, the nodes in this particular network are socially aware and active, they are ready to forward the message, which they actually did, adding along the way funny comments and observation about the actress. Many of the users additionally started blogging only parts of the gifs, or created their own art about the quote from Murray. 


This post is just reporting the initial analysis of the community. For the paper I would expand the analysis and include how this information was reshared on Twitter and Facebook, as well as through the forums, dedicated to Bering/Wells ship. I did not publish the usernames of fans on Tumblr here even though they are publicly accessible. I decided against it because of the small fall out of someone posting the video on SpoilerTv.

Update – 20/03/2013. I saw that two of the tables are mistaken, so I am changing them to the correct ones. This happened due to an error I did while copy/pasting the data. I know it is a newbie mistake, but it can happen every once in a while. Fortunately, I caught it today, looking through the data with fresh eyes. I also played a little bit more with the graph, so there are some more pictures (the radial graph is a new one).

  1. The term “Shipping” — derives from “Relationship” and is the belief of series fans that two of the characters are (will become, should be, are offline) in a romantic relationship
  2. Warehouse 13 is a sci-fi show about Secret Service agents collecting artifacts with various powers
  3. the original post, as well as the video was removed Sunday night because of the fall back
  4. The data is publicly available in the notes section (where likes and reblogs are recorded), along with the date of publishing