Describing a Network - The Creative Act Blog



Background
In the last 3 months I was taking part in The Creative Act class taught by Ze Frank.
Each day we had to post a daily exercise, on a given theme, to the class blog. Once a week we also had to post a pressure project exploring a medium in more detail.
The blog was open and public and Ze Frank directed external (=non-ITP) traffic to it. So anyone could comment on our work if they wish to.
Eventually, though, most of the comments on the blog space were from class members and people we know and not so much from strangers.

We were 21 students in the class, and since we had to post to the blog on a daily basis for 3 months, new connections and relationships started to evolve.
The blog swept us into it, and demanded a lot of time and dedication. Most of us were addicted to it at some point, waiting for comments on our posts or to see what other people are creating. Sometimes we treated it as personal diary.
This whole process developed a network/community of people with noticeable connections between them. A lot of the works were riffing or responding to other students' work, sometimes with the encouragement of Ze Frank and eventually it just happened.
Some examples are Jeff Gray's "compliments for fellow peers" and Eric Fino final project in which he wrote and performed a short poem on each of us.

One other interesting fact to mention is that the class' audience was much varied than the usual, both in age and in background. And that it was a large class in size - 21 students.
The class and the blog were so demanding and dominant and it forced us to get to know the people that are taking the class with us, something you can usually avoid doing... if you chose to.
Here are Lian's final words on the class:




Visualization of the network

For my final project I've chosen to visualize 2 aspects of this network:
1. The entries' posting time:
- Number of posts per hour
- Average Posting Time - Total - September - October - November
- Number of posts per hour divided by users

2. "The comments space" - people commenting on others posts or just having over-night conversations. I think this is where the network was actually forming.

I divided the network's nodes to the individual class members, Ze Frank, anonymous commenters, and non-anonymous commenters that are not class members. (*)

Terminology:

Loose connection - describes a commenter that commented on the same author's work more than 2 times.

Following are some guidelines to understand the graphs:



First I created a few tables that summarize the data of the graphs -> the numebr of comments each autor received and posted.
Another table I generated is the one which presents the most commented entries on the blog.
The first 2 entries on this table are interesting example to the "comment space" and to the formed network - the first one is Eric's post-off competition, commenting on his own work, anonymously and responding to the criticism, and the second one is my late night chat with Kleoni.

I created 3 different graphs:
1. Visualizing the strong connections.
2. Visualizing the strong + loose connections
3. Visualizing all comments (this graph turned out to be huge and almost impossible to read).



The way the graph are generated by Graphviz, will place in the middle the nodes that have more arrows pointing at them or from them. So in the absolute middle of the graph we will find the most active people (Kleoni...) - meaning the ones that commented more and were most commented. On top of it are the people that commented more but got less comments, which naturally consist of the non-author group (Ze Frank, Not a class member and the anonymous). And then in the middle bottom are the people that got comments, but didn't comment that much.
As we go left and right in the graph and away from the middle, means that the amount of activity of these nodes was somewhat lower.
It was interesting to see the differences between the distributions of the nodes in the strong connection graph comparing to the graphs with most of the connections. A lot of nodes moved quite a bit, of course it also has to do with the complexity of the graph when it consist so many connections, but also the clusters are a bit different.
I was expecting to find more mutual strong connections between people (more of the "friend" type in social network and not the "fan" (upcoming.org terminology) one-sided connection).


The process

My first steps were creating a copy of the database on my own server and organize the data in it (a very tedious process). Once that was completed I started thinking which information can be extracted from it and would be interesting and revealing related to the blog structure. I created a few SQL tables summarizing both the times of posts and the relationships between commenters and authors and tried to find a good way to visualize them. Since there were a lot of comments, I felt I need to find a better way to differentiate the connections. Therefore I decided to come up with a factor to distinguish between strong connections and loose connections.


What I would do differently...

I would have liked to divide the comments graphs in to the two periods of class - the first half of the semester and the second half of it. Then I could compare the differences betweens the social maps as we got to know each others work better.

Another thing I am missing, and also been raised up by Ze Frank, is that it would have been interesting to know the qualities of the comments: which percentage of them is related to the work, and which is a teaser (like Eric's post-off competition) or just a chat.

I also wish the graphs were more interactive, so they were linked to the comments or to all that author's entries.

Another thought I had is to print them on a large format that will allow browsing them without scrolling.

I think one of my problems in this project is that I didn't anticipate that the difficult part would be to generate the graphs.

When I planned the project I thought that getting my own database would be most of the work. I was wrong... So when I finally started to get comfortable with Graphviz I was already running of time. But overall I am quite pleased with the results.