Six Degrees: The Science of A Connected Age by Duncan J. Watts
The Social Organization of Conspiracy: Illegal Networks in the Heavy Electrical Equipment Industry by Wayne E. Baker and Robert R. Faulkner; American Sociological Review, Vol. 58, No. 6 (Dec., 1993), 837-860; Available at: http://links.jstor.org/sici?sici=0003-1224%28199312%2958%3A6%3C837%3ATSOOCI%3E2.0.CO%3B2-V; I wrote a summary about this: Writing in the Sciences – Writing Assignement 1
In this review I summarize „The Social Organization of Conspiracy: Illegal Networks in the Heavy Electrical Equipment Industry“ by Wayne E. Baker and Robert R. Faulkner from 1993 .
The paper studied how the structure of „secret societies“ involved in price-fixing effected the convinctions of participants. For this, the researchers used the conspiracy in the Heavy Electrical Equipment Industry in the 1950s, called „the most carefully studied corporate offenses in the history of the United States“ (Geis and Meier 1977, p. 68). The case involved 40 manufacturers (i.e. General Electric, Westinghouse) with more than 20 product lines included. Three major conspiracies were analysed: Switchgear and transformer with low information-processing and turbine with high information-processing inside – each with the same need of secrecy and avoidance of judicial dangerous direct contact with other competitors.
To understand the structure and characteristics of „secret societies“, let us first take a look at the economic theory under efficency constraints: Organizations with high information-processing, like turbines, work best decentralized. Ones with low information-processing are more efficient when hierarchical, like switchgear and transformer production.
The result shows major differences between „secret societies“ and legal activities. The turbine conspiracy with high information-processing was the most centralized one with lower possibility of sentencing. More difficult, ambiguous and complex tasks and decisions needed more centralized communication. The switchgear and transformer conspiracies were organized in a more decentral, two-tier structure. Easy tasks and regular orders led to self-organizing systems with a buffer between the command levels. Changing the perspective from the whole network to the different protagonists, four characteristics emerged, which significantly increased the probability of a guilty verdict. In descending order, these are: „1) Being in the thick of a conspiracy 2) participating in a decentralized conspiracy; 3) occupying a top-executive position and participating in a centralized conspiracy (turbines); 4) occupying the oppressed position in the middle of an organization“ (Baker & Faulkner, p. 854).
The research revealed, „that the structure of intercorporate secret societies does not follow the same underlying efficency logic as the organization of legal business activities“ (Baker & Faulkner, p. 854). The data for analyses of criminal activities like this are rare and often not reliable; so this first quantitative analyses of an intercorporate conspiracy breathed new life into to the development of network theory, industrial economic theory and organizational crime theory.
 Baker, Wayne E. and Faulkner, Robert R.: The Social Organization of Conspiracy: Illegal Networks in the Heavy Electrical Equipment Industry, 1993. American Sociological Review, Vol. 58, No. 6 (Dec., 1993), p. 837-860.
Short sum up of the first week homework in the Social Network Analyses Course.
First, we had to download our Friends Network out of Facebook: For this the NetGet Application was used to save the Network in a GML File.
The GML File was opened in Gephi, which looked meaningless like this:
Network after importing GML file
Then I applied the Force Atlas 2 Layout Algorithm for about 20 seconds.
Network during running of the Force Atlas Layout Algorithm
To give the network some informations – represented in color and form – I was computing this methods:
Average Degree: 30.8
Connected Components: 26
Average Path Length: 2.84
After the computing, the visualization was completed with this settings:
Partition -> Node: Modularity Class
Ranking -> Nodes -> Size: Betweenness Centrality
Adjust by Size: yes
Repulsion Strength: 5000
Size Mode (Text): Node Size
Network after computed and visualized Modularity Class and Betweenness Centrality
Finally, I’ve exported 4 different Layouts. The pictures below do not show all elements in the network. The original vector files with all nodes and edges are very big, so they are just linked as SVG files.
An ethical question arose, when I was thinking about releasing the Facebook Friends data with names or not (as I finally did). The Network has been anonymized, cause the informations about my friends on Facebook were not given for publishing purpose, so it’s not ethically correct to do it afterwards, cause the data is there and I can see it.
Some important steps
The GitHub repository was taken down after a request from the course managers before the second run to do it.
All the data for this exercise and upcoming ones are available on my GitHub Repository for Social Network Analyses. They are my first commits to the open source community ever, so made a very important step this weekend. There are all raw files, the gephi file, the generated images and also the reports made in gephi.
Deeper analyses will follow up. Right now getting in touch with gephi is more important. But one short note from viewing on the data with real names: Seemed, like my past and my social live was mapped pretty good, and it’s always a huge step to see something like this visually out of a data file from a social network. Amazing, but also alarming.
Thanks to Beatriz Patraca Dibildox for the tutorial, which helped me in doing some nice stuff like Modularity Class. Another nice tutorial how to visualize facebook data in different softwares (also gephi) is made from Luca Hammer at his blog.
Here’s a tutorial video in german:
And last but not least: Would love to see some other networks. Contact me, if you want to share your network anonymized or if I should do that for you.