Este tipo de coisa é assustadora... Ma imagino o que andam fazendo por aí e que não é publicado!
Este é o lado negro de pertencer a redes sociais...
What’s Wrong with the ‘Terrorist Facebook’
By Drew Conway, on August 21st, 2009
The Independent has a new article describing a new U.S. military program to compile and analyze massive stores of social network data on terrorist, which the editor delightfully coins a ‘terrorist Facebook.’ The thrust of the program is described as follows:
The idea is to amass huge quantities of intelligence data on people – no matter how obscure or irrelevant – and feed it into computers that are programmed to make associations and connections that would otherwise be missed by human agents, scientists said.
This statement is eerily derivative of the so-called ‘mosaic philosophy,’ which is both morally and scientifically abhorrent. To the author’s credit, he highlights this connection, citing the criticisms. Beyond this, however, there are several methodological problems at the core of the research described in this article. First, the view that collecting massive amounts of data increases the clarity of a social network is flatly incorrect. At best, such an approach would illuminate areas of the network for which the collectors (in this case, intelligence officers and interrogators) have bias or former knowledge. More likely, blindly increasing the number of social ties simply adds noise, which ultimately confuses the analytical process.
The collection process described in the article also breaks one of the fundamental rules of network analysis. According to the article, the analysis combines data “from thousands of people [who] have been arrested and interrogated for information,” with “mining the vast amounts of telecommunications data collected from emails and telephone calls.” What’s problematic is this method funnels many different types of ties into a single data set; unfortunately, analyzing them in a single dimension leads to errant conclusions. The question that must be addressed is how are social ties formed over telecommunications mediums different from those created and maintained in person? This is critical to all social network analysis, but of utmost importance when dealing with adversarial structures, wherein social ties are heavily monitored and actively fabricated.
Finally, the ‘fallacy of prediction’ seems to be well built into what the researchers of this program hope to deliver. Dr Ian McCulloh, a US Army major at West Point Military Academy, states, “Before a terrorist event is going to occur there is usually a change in that organisation as it begins to prepare and plan and resource the event. In that context I can monitor a network in real time and monitor the change in behaviour before an event occurs.” As is often the case, however, this perspective only repackages traditional network analysis as prediction—despite the best intentions of the researchers. Reporting changes in organizational structure is merely a descriptive statistic. To approach dynamics, we must have a theory about how one one network formation leads to another, and at present the theory simply does not this.