Was Ft. Hood (or Aurora) Preventable?

I’ve written that Ft. Hood (and Aurora) were indeed preventable, noting:

Now, basically, the question he is posing is this: Service members take an oath when then join, and oath breaking is frowned upon. So he’s asking whether members of the armed forces can still be considered Shahid, if they kill fellow servicemen in the name of Islam despite breaking their oath? Shahid is basically jihadi-speak for, do I still get the 72 virgins?

Now, I’m sorry, but that has to trigger a full investigation. It just has to. What the hell is the point of spying on people and collecting all this email traffic if an Army officer can write to a radical cleric on our watch list and ask whether it is permissible for members of the armed forces to kill fellow servicemen and women without triggering a full investigation?

We have a massive surveillance regime in place, and yet here is a case of a perfectly preventable attack that was missed. Depressing.

A couple of recent blog posts have also address this. DGR and Lauren Morgan wrote to challenge the notion that Hasan’s communication with Awlaki was “fairly benign.” They argue:

The initial defense of Hasan’s emails as “fairly benign” is simply not defensible. The only explanation for why officials might reach that conclusion is that they simply did not know what they were looking at.

To review, Hasan repeatedly expressed what can only be described as a school-girl crush on Awlaki, who was known as a radicalizer in actual cases where Americans were driven to violence. Hasan even expressed a desire to send Awlaki money, tried to set up a $5,000 essay contest on the topic of “Why is Anwar Al Awlaki a great activist and leader,” and inserted Awlaki’s name in a laudatory manner into a Qur’anic verse. Hasan clearly expressed the view that Western forces were at war with Islam. And he sought Awlaki’s counsel on such questions as whether suicide bombings were acceptable, whether collateral damage was permissible in the course of a suicide attack, and–in his very first email–whether Hasan Akbar, who murdered fellow U.S. soldiers, might have been considered a martyr.

Um, yup. But over at Intelwire, J.M. Berger counters:

Anyone who has spent time with raw intelligence knows that you have to triage leads. For every intercept that states “I’ve bought the fertilizer and I’m loading it in the truck tonight down by the lake,” there are literally thousands and thousands that say “I hate America,” “I love Al Qaeda,” “I’d like to kill those bastards,” “I swear to God it’s time for action and not words,” “If I was James Holmes, here’s how I would have done it better,” “You need to start mentally preparing now to kill your neighbors when the revolution comes,” and “Would you consider someone like Hasan Akbar to be fighting jihad, and if he died would he be a martyr?” (All of the above are paraphrases of actual extremist content I’ve reviewed.)

But when you place the content of Hasan’s messages alongside all the other raw intelligence that counterterrorism investigations generate, it’s extremely hard to argue from a subjective, non-psychoanalytical reading that they represented a red flag.

Yes, that is true as well. The issue, of course, is that we don’t seem to have a better way to fuse these yellow flags into red flags. We still seem to largely rely on some sort of bright, blinking indicator to give warning. And worse, we don’t seem to have a system of relatively low-cost options to check out threats that don’t seem to set off alarms. Nor do we have a way to shut off those alarms when an investigation turns up little or nothing.

The issue, it seems to me, is that we’re still not good at putting pieces together. Think of Aurora. So Holmes has psychiatric issues. Well, you can’t have full time surveillance of everyone who is having psychiatric issues. Nor would you want to. But he was also buying vast amounts of weaponry. Well, you can’t have full-time surveillance of everyone who is buying weapons legally. But the overlap of those two sets… well… even if you can’t do full-time surveillance, maybe you can have a police officer drop by and knock on the guy’s door just to check in. I mean, isn’t that a principle of community policing? Getting to know potential threats in your area?

Same with Hasan. You can’t do full-time surveillance or everyone who goes onto a Jihadi website. And you can’t do full-time surveillance of every military officer. But, you know, officers who go onto Jihadi websites maybe should prompt a little personal attention? Not full investigation maybe, but something.

And yeah, maybe that isn’t enough. Maybe the cops knock on Holmes’ door, and he comes off fine. Maybe a superior has a chat with Hasan, and then he still goes out and shoots up people. But at least some things will be prevented and some other deterred.

Nothing is 100%, but we do clearly need to do a better job at fusing together low-priority indicators and using existing policing and supervisory channels to check out and head off threats.

The flip side of this is that when we do focus on someone, we seem relentless in making a case against them, even if, you know, they are just a bunch of clowns who probably should just be given a warning and otherwise forgotten about.


2 comments to Was Ft. Hood (or Aurora) Preventable?

  • But the overlap of those two sets…

    It’s theoretically possible, but it would require two databases plus the authority to thoroughly collect the information to fill them. We’d have to give the government authority to mandate that certain things be reported (and define the precise reporting criteria – not easy for mental health issues), the authority to vacuum up all sorts of electronic information, the authority to analyze that vast amount of data to find suspicious patterns and then the authority to send someone out to check out the leads. And then we’d have to hope we have sufficient oversight to prevent abuses. Not sure how all that is possible under our present Constitutional system and our present notions of privacy. I think that is probably a case where the cure would be worse than the disease.

  • mclaren

    I have to second Andy’s reservations here. As well, the computational issues here are just staggering. You get a comnbinatorial explosion when you try to do a branching search for associations like AlAwlaki -> Hasan. All 7 billion people on earth are connected to one another through no more than 6 intermediaries (Microsoft researchers just proved it mathematically: see “Proof! Just six degrees of separation between us,” The Guardian, 2 August 2008), so you’re facing such a factorial explosion in database search trees that it’s not realistically practical to do these kinds of blind searches on random elint traffic without a computer program that actually grasps the meaning of the messages, instead of just doing string-parsing.

    But computer programs able to understand meaning involve hard AI. And that’s nowhere on the horizon. Hard AI is one of the great failed research projects of recent history. The NSA’s gigantic new data center in the Nevada desert is a gigantic monument to unworkable technology. We can archive as many trillions of emails and cellphone intercepts as IM messages as we want, but searching ‘em for useful clues to terrorist activity is never going to become practical without the kind of AI breakthrough that the last 60 years of computing history says simply isn’t in the cards.

Leave a Reply