Jottings 6: How the network works, it seems

[ by Charles Cameron — often puzzled & shocked at the way some algorithms behave in public ]

.

I was on Silobreaker this evening, looking at a page called Dead Terrorist Tamerlan Tsarnaev Channeled Islamic Videos, and what I found curious was this little network diagram, presumably algorithmically-generated from the Silobreaker database:

I have to tell you, I like network diagrams — but what are we supposed to infer from that?

Do you see the same diagram when you visit the site — or is this display just for me, knowing what books I buy, what movies I see, what friends I have — and tailored to sell me something, or get my vote?

  1. zen:

    Hi Charles,
    .
    This is what the diagram tells me; Tamerlan’s interests in Islam – at least in evidence of his surfing online – were keyed to cherry-picking justifications or moral and political pretexts for acting out violently in his own social context rather than an immersion in the breadth of Islamic studies or piety.  He was already angry and already violent in general but wanted to scale up for reasons of a “higher purpose” rather than just being seen as a nut. The Unabomber, Ted Kacyzinski was very concerned that he not be perceived as crazy or a lone nut and wove an elaborate eco-radical justification for his terrorism. A process that, if I recall correctly, began after his first acts of terrorism

  2. Charles Cameron:

    Hi Zen:
    .
    I think it’s a bit like the effort to use “big data” that JM Berger is trying on Intelwire — the data doesn’t really support much of anything by way of conclusion, but it’s fun to watch and perhaps a useful Rorschach for eliciting insights — which would then need to be verified by other means.
    .
    If you have a reference for the Unabomber bombing first and (ideologically) radicalizing afterwards, that would be a very useful addition to current discussions of the Brothers Tsarnaev. 

  3. JM Berger:

    Silobreaker’s native (i.e., free unprocessed) data isn’t really big enough to generate very sophisticated graphs. They need to work on a strong automated entity creator. 

  4. Charles Cameron:

    Agreed…