Can your Facebook network finance your dream project?

November 06, 2012

Two weeks ago I started playing with Gephi, an open source network graph generator. Here is the first network graph I generated in a Crowdsourcing Discovery progress report posted two weeks ago:

 

 

This simple 10-person network is comprised of 9 contributors to Crowdsourcing Discovery, and yours truly, A. What I find most interesting are the three shapes that emerge from our connections.

 

The first and simplest is the line, linking me to an “orphan” friend, F, who’s not connected to anyone else. The second is the triangle, which is formed between me and two friends who are also friends. Two examples are the triads A-B-C and A-D-E. The third and more complicated structure is the modified diamond formed by A-G-H-I-J, which is centered around H.

 

What happens as the network grows to include supporters K, L, M, N, and I tweak the color scheme to specify how I know these friends?

 

Blue indicates friends from college, red indicates science friends from grad school, yellow indicates non-science friends, and gray are other scientists I met through H.

 

The triads A-B-C and A-D-E are unaffected, while the modified diamond grows into a more complicated structure anchored by two mini-hubs, L and H. L is cool because he/she forms a triad with N and M, but a diamond with K, H and G.

 

What does this mean for the final two weeks of our crowdfunding campaign in terms of marketing strategies? Short answer: I’m still figuring it out! In the meantime, I decided to take a look at a graph of my entire Facebook network, which may provide clues how to convert more friends from likers to contributors in the home stretch.

 

Because the Internet is awesome, a tutorial on how to generate such a graph has already been crowdsourced! Start by downloading and running the suggested FB app, which exports a .gdf file that can be opened in Gephi as an undirected graph. (Undirected means mutual in graph theory jargon. An example of a directed graph is your Twitter network, where A may follow B but B may not follow A).

 

After you load the data file, use the algorithm called Force Atlas, as explained in the tutorial, to expand the initial hairball of connections into something with meaningful architecture. Here again is my network, which is comprised of 669 friends (nodes) and 3,227 friendships (edges):

 

 

The size of each node is proportional to its connectedness, and are most highly connected nodes are blue. The orphans and lowly connected nodes are more red, and green/yellow/orange nodes are somewhere in between. I’ve taken the liberty of labeling majoring clusters for ease of interpretation.

 

What stands out the most is that my college friends are the least cliquish of all my friendship clusters. In other words, I was friends with a lot of people who were each part of separate social groups. This more dispersed organization is unique to college. By contrast, from middle school to grad school and beyond to my postdoctoral years, distinct clusters of mutual friends emerge. I’m not sure what this means in the absence of comparisons. It could be a function of my attending Columbia College, or my early adult angst, or both. (Oh, and Burners are really tight).

 

If you’re interested in seeing what your FB network looks like but you’re a Luddite when it comes to network graphs, send me your .gdf file and I’ll render your network!

 

Related Posts

Google Analyze This
Crowdsourcing Discovery
The Academia-Pharma Complex
  • http://twitter.com/debivort Benjamin de Bivort

    I think I know who H is.

    • http://twitter.com/eperlste Ethan Perlstein

      shhh