Thursday, December 27, 2007
Dismantling Able Danger
Fig. 1-1: Never let a conspiracy theorist play connect-the-dots.
In October of 1999, US General Hugh Shelton asked US Special Operations Command (SOCOM) to come up with a novel way of gathering data about terrorists.(1) This new format of profiling is known in the intelligence community as “holistic” intelligence, defined as a process of “fusing and integrating previously unlinked platforms, technologies and resources”(2) – in English, getting as much intelligence as you can about as many suspects as you can, no matter how trivial.
In Shelton’s words, “It was just kind of an experiment.”(3)
Readers who aren’t familiar with Able Danger can get some more background on in the opening pages of the Department of Defense report on it, available here (also discussed below). The rest of you readers know where this is going. However, many of us are unfamiliar with what exactly Able Danger did. We all have a rough idea of its purpose. But did Able Danger complete its mission? Was it even capable of completing its supposed task?
The answer is no. The fact that the answer is no will help explain why the conspiracy theories surrounding Able Danger are just so much noise.
The research and categorization methods used by the Able Danger team involved a series of techniques bundled under the umbrella term “data mining.”
To data mine, analysts aggregate large laundry lists of possible parameters that could define a population.(4) In the case of Able Danger, the population would be “potential terrorists around the world,” and parameters would theoretically include simple quantitative questions like “how often does this person visit extremist website X?” or “how many times does this person visit radical mosques Y and Z?” (Bear in mind that I am postulating here as Able Danger’s content was entirely classified, but you get the idea). Then, because data mining should theoretically be useful in making predictions, these variables’ values are assigned certain risk factors and are used to predict the likelihood that someone will engage in terrorist activity.
Unfortunately for the conspiracy theorists, another key element of any data mining operation is having a large population to choose from. In the business world where data mining is a practical tool, often an entire “set of all consumers of X” are analyzed – potentially, millions of people.(5) Unfortunately for the Able Danger crew, their target population (extremists who rely on the Internet for dissemination of propaganda and for organizational coordination) had only had access to the Internet for about two and a half years, since 1996 when the Internet first exploded on the global stage.
A total of 1,269 terrorist attacks occurred worldwide between 1996 and 1999,(6) but it’s impossible to know how many people and organizations ultimately planned all of them. Prior to 2001, the State Department had named twenty-eight international cadres as terrorist organizations, and almost all of them conducted their activities in and against Israel and its border states, Basque Spain, Northern Africa, Turkey, and Iran. With the exception of four (Kach, a Jewish radical group; Aum Shinrikyo, a Japanese cult; 17 November, a Cyprus nationalist movement; and the Tamil Tighers, an ethnic group with Hindu underpinnings), all of these terrorist organizations were built on some permutation of Islam or Marxism - or, in the case of many Palestine-based organizations, a bastard child of the two.(7)
That means that most of these groups would follow similar patterns of recruitment and training; espouse similar doctrines; and even recruit from among the same populations – at least, this is how it would appear to a program with the extremely broad objective of extrapolating predictive patterns from the entire Internet. For those not catching my drift: the differences between these groups can only be seen under the microscope, and Able Danger was a birds’-eye view of a foggy digital landscape that had never been mapped or even fully comprehended until that point. Indeed, as the FBI report on Able Danger indicates, much of the impetus to try Able Danger at all rested on the fact that the government’s previous most powerful software, a similar method using software called Statistical Influence Assessment Module, couldn’t do the job either.(8)
Readers of this article with an eye towards legal studies will also have noticed a glaring flaw with the statistics-oriented foundation of Able Danger’s “data mining” methods: just because someone is likely to do something doesn’t mean that they will do something. No rational court of law would even pretend to have sound knowledge of a person’s activities based solely on what websites they visit (one shudders at how such a judge would come to view the typical American teenage male). As many of my skeptical friends know, in the course of debates we often end up having to peruse the same garbage websites that conspiracy theorists take as holy writ. The two most frequent viewers of Islamist propaganda material are holy warriors and CIA agents.
So Able Danger (once again, at least theoretically) has two fatal flaws: it seems obvious that such a program in 1999 could never develop sufficiently complex heuristics to predict the behavior of small terrorist cells, much less individuals in the vast and insane sea of the Internet; and no sound policy maker would even try to arrest, imprison, or even engage in costly monitoring of a suspect based solely on the evidence it could provide. Able Danger as a program was too shallow a system for too specific a subject, and its information was too cursory to be used as a solely preventative program.
And conspiracy theorists expect us to believe that Able Danger had precisely predicted exactly who Mohammed Atta was; where he had been every day between 1999 and 2000, who he had spoken to, when, and about what; and what he was going to do in the future, down to how many times he would blink between waking up and dying on September 11, 2001.
The claims about Able Danger all center around this key impossibility: “Did Able Danger identify Mohamed Atta in January 2000? Yes” (The “Able Danger Blog”). “A US Army intelligence program called Able Danger identifies five al-Qaeda terrorist cells... includes 9/11 ringleader Mohamed Atta, and three other 9/11 hijackers” (“Cooperative Research;” a conspiracy site with the good graces to admit that this claim is based on shoddy evidence, as shown below). In short, thousands of conspiracy theorists believe that not only did Able Danger outperform to an impossible degree and precisely pinpoint the activities of the 9/11 hijackers, but that the CIA, the FBI, the DoD, the DoS, and your grandmother all willfully refused to act on those findings.
As should be evident by now, this claim is silly on its face. What’s even better is the home source of these claims.
Army Reserve Lieutenant Colonel Anthony Shaffer approached the Able Danger staff at a conference in January of 1999 with a nifty new gadget, some data mining software that a doctor friend of his has provided to him (whose name is still classified). Evidently, the big draw of this particular software, farmed out to the now-notorious Orion Scientific Systems, was that it could draw nifty, easy-to-read charts. The doctor herself was eventually let go because, in the words of one of the investigators, “they were not happy with her ability to get along well with others.” Shaffer stayed on, chiefly in the role of liaison to the CIA leadership.(9)
Exactly as one would predict from the above analysis, Able Danger was not the silver bullet Shaffer and his anonymous friend claimed it was. “It helped,” a one ‘MG Lambert’ told the DoD investigation, but “we didn’t get the mission accomplished... The final product was just a framework, you know[,] it was... just a template.” Another anonymous Captain said that, overall, by October 2000 the program “had made very little progress.” The “experiment” ended in January 2001, when it was decided that “its work product was limited to the development of a ‘Campaign Plan’ that formed the basis for follow-on intelligence gathering efforts.”(10)
In his career, LTC Anthony (“Tony”) Shaffer attended three Able Danger meetings and was largely considered a “negotiator;” he did an awful lot of “communicating” and “assisting Able Danger team members in receiving special authorization that enhanced their ability to access various World Wide Web Sites” and “coordinating…” but evidently very little concrete work on the project. One “key participant” on Able Danger said he was “basically a delivery boy.”(11)
That is why when, shortly after its termination, he announced that Able Danger had successfully identified Mohammed Atta and his sinister plans, he should have expected the claims against his credibility.
We must bear in mind that it was Shaffer who, along with “the female doctor” as some conspiracy theorists know her as, who first proposed using data mining. He came late to the experiment with a silver bullet, and after it was found to be nothing but empty promises, began crying conspiracy. But yes, it gets better: despite his numerous claims, he also admits he never saw any of his own supposed evidence himself.
As I wrote on the otherwise hopelessly conspiracy-leaning Wikipedia article about Able Danger, the DoD report on Able Danger revealed that Shaffer’s entire claims are based on hearsay. In his testimony, Shaffer says he heard the claim that Able Danger had positively identified Mohammed Atta from an anonymous Captain who says he viewed the charts “for four seconds, maybe five.” That Captain later added, "[LTC] Tony [Shaffer] was relying on my recollection, I think, 100 percent. I mean, I think a lot of people are.”(12)
During the hearings, Shaffer’s lawyer, Mark Zaid seemed eager to defend his clients by speaking on their behalf in a way that didn’t make them sound like the kooks we know Shaffer’s type to be. Early in his testimony, he made several points clear:
“• At no time did Able Danger identify Mohammed Atta as being physically present in the United States [contrary to Shaffer's claim].
”• No information obtained at the time would have led anyone to believe criminal activity had taken place or that any specific terrorist activities were being planned [also contrary to Shaffer]. Again, the identification of the four 9/11 hijackers was simply through associational activities. Those associations could have been completely innocuous or nefarious. It was impossible to tell which, and the unclassified work of Able Danger was not designed to address that question.”(13)
One imagines the pre-trial preparation for this testimony: “Okay, Tony, let me do all the talking so you don’t say anything stupid.” However, Zaid did point out that there were serious pre-9/11 communications problems between the CIA and independent information gatherers, and those complaints are discussed on the same Wikipedia page.
I don’t want to take up any more of your time. Able Danger lacked the ability to even grasp at its supposed objective; the source of its concomitant conspiracy theories had no personal knowledge of the information he garnered from someone else’s backwards glance at a picture that clearly didn’t exist at the time nor could even have theoretically existed at the time, given what we know about Able Danger. It is impossible in both theory and practice for Able Danger to have identified Mohammed Atta. There are about a dozen layers of conspiracy theory BS you have to swim through in order to get to the truth about Able Danger. Frankly, it was just an experiment.
1. Office of the Deputy Inspector General for Investigations. Alleged Misconduct by Senior DoD Officials Concerning the Able Danger Program and Lieutenant Colonel Anthony A. Shaffer, U.S. Army Reserve. Washington: Office of the Inspector General, 2006. p. 12. Retrieved December 26, 2007 from http://msnbcmedia.msn.com/i/msnbc/sections/news/Able_Danger_report.pdf.
2. “Intelligence for the Twenty-First Century.” Dupont, Alan. In Wark, K. Twenty-First Century Intelligence. Routledge: 2004. p.28. Preview retrieved December 26, 2007 from http://books.google.com/bkshp?hl=en&tab=wp .
3. Rosen, James (2005, December 12). “General tells of pre-9/11 intel: Says he never heard Atta’s name.” Newsobserver.com.
4. Much better, more in-depth descriptions of these processes are available at Data Mining Techniques. 2006. Retrieved December 27, 2007 from http://www.statsoft.com/textbook/stdatmin.html .
5. Alexander, D. Data Mining. Retrieved December 27, 2007 from http://owl.english.purdue.edu/owl/resource/560/10/ .
6. The Johnston Archives. Statistics on terrorism. Retrieved December 27, 2007 from http://www.johnstonsarchive.net/terrorism/intlterror.html .
7. U.S. Department of State. Background Information on Foreign Terrorist Organizations. Retrieved December 27, 2007 from http://www.state.gov/www/global/terrorism/fto_info_1999.html#ano .
8. Alleged Misconduct by Senior DoD Officials Concerning the Able Danger Program and Lieutenant Colonel Anthony A. Shaffer, U.S. Army Reserve. p. 13. Retrieved December 27, 2007 from http://msnbcmedia.msn.com/i/msnbc/sections/news/Able_Danger_report.pdf.
9. Ibidem, p. 13.
10. Ibid., p. 20.
11. Ibid., p. 21.
12. Ibid., p. 24.
13. Testimony of Zaid, M. in U.S. Senate Committee on the Judiciary. Testimony of Mark A. Zaid: Attorney at Law: September 21, 2005. Washington, 2005. Retrieved December 27, 2007 from http://judiciary.senate.gov/testimony.cfm?id=1606&wit_id=4668.
-By John Ray