Tuesday, February 27, 2007


Intelligence Failures

Intel channels are buzzing today over Ralph Peters’ searing indictment of their community in this NY Post editorial.

He starts his critique with an anecdote of questionable import, describing how American troops “mistakenly” detained, searched and “jerked around” Amar al-Hakim, the son of the leader of the Supreme Council for the Islamic Revolution in Iraq (SCIRI), presumed to be a cooperative Shia partner to the US and Coalition efforts in Iraq.

Peters relates that he reacted first with annoyance, then with indulgence, about what he perceives as an apparent lapse in tracking such an important personage. Why annoyance?
SCIRI certainly has some dark connections with Iran. The party's a dubious ally, at best. But jerking the boss' kid around was, in diplo-speak, "unhelpful." Even if Hakim Jr. was smuggling money, or worse.

I'd be shocked if he wasn't. It's the Middle East, folks. We're just betting we can handle the least poisonous local snakes.

As a former Military Intelligence officer, my first reaction to teaching Little Hakim the perp walk was: "Who's responsible for tracking this guy?"

A sound intel effort would monitor all of the male family members of Iraq's key leaders 24/7. How did Hakim Jr. slip off the reservation?
I think Peters would be right on target here, presuming of course that US Intel assets weren’t tracking the SCIRI lad, and US forces weren’t in fact tipped off to his travel. Offset, of course, by the high probability, suggested by Peters himself, that Hakim Jr. was conducting business at the least at cross-purposes to US interests, and quite possibly anti-US and Coalition.

But Peters relents from his initial stance, and what follows is an exaggerated and unhelpful indictment of a specific piece of Intelligence processing software.

Peters sets up a straw man in his essay, the better to knock it down with the bat he will introduce. The problem is, he oversimplifies in building the straw man, and mischaracterizes the size and importance of his bat.

Peters makes the claim that those pushing an emphasis on Human Intelligence (HUMINT) were ignored, in favor of high priced technical intel solutions:
While a minority of us had argued since the mid-1980s that the human factor would be paramount in our future conflicts and that technology couldn't replace the human mind, the MI establishment just went on buying platinum-plated junk that never delivered a tenth of what the contractors and apostles of hi-tech promised.

Appropriate technologies can help us - but no database or collection system is a substitute for seasoned human judgment. The key task in intelligence is understanding the enemy. Machines do many things, but they still don't register flesh-and-blood relationships, self-sacrifice or fanaticism.

Forgetting that tech is supposed to support people, we wasted talented people supporting worthless technologies.
This calls to mind several Intelligence technologies and capabilities that proved dramatically successful, at least until the NY Times and Washington Post disclosed classified intelligence and tipped our enemies off. I also can’t help but recall any number of dramatic capabilities that have emerged and matured during my over two decades in Military Intelligence. This sounds a lot like, “Yeah, but what have you done for me lately?” We don’t get credit for wins, only blame for losses, and frankly, we don’t want a lot of publicity anyway. But Peters stance here strikes me as the kind of insiders lament that any of the INTS always indulge, at least when speaking of rival Intelligence sources. (HUMINT against IMINT, COMINT against ELINT, SIGINT against everybody. Look ‘em up, it will be good for you.)

Peters then hefts to his shoulder his preferred bat against this straw man, a bat in the form of what Peters mischaracterizes as a “Rube Goldberg contraption”:
The cardinal example of this corrosive mentality was the purchase of a multibillion-dollar, Rube Goldberg contraption called the All Source Analysis System (ASAS). Under development for more than two decades, ASAS never worked. But a generation of senior MI leaders made rank pitching the system as the answer to every intelligence need.

ASAS was going to fuse the data from every classified intel source and give the commander instant, perfect answers. Early on - in 1984 - a self-assured technocrat in uniform told me that, within 10 years, human analysts would be irrelevant.
I can certainly understand the overall perspective, and perhaps Peters is right that ASAS was grossly over-hyped and over-sold, but I think he unhelpfully conflates ASAS with a lot of other intelligence technology components. Peters makes the ASAS sound like some ludicrously complex piece of hardware, but it’s actually a suite of intelligence analysis software. As such, the ASAS is the software tool at the user end of a very long and multi-threaded string of intelligence collection and processing capabilities. It’s actually one of the lesser expenses within the overall system that I believe would more accurately be the target of Peters’ ire. We’ve been training with it for over a decade, with admittedly poor results. But focusing on ASAS as the loci of one’s righteous indignation is kind of like beating on the ATM when the bank somehow misplaces the monies in your account.

What ASAS hoped to replace was an analytic toolset of acetate, grease pencil, Sharpie pen, cabinets full of paper files records, and great multitudes of redundant situation maps and overlays, with data out of date before the ink was dry. ASAS was intended as a toolset that made it (theoretically) possible to filter all the disparate volumes of raw intelligence data into an electronic data store, supported by automation, which created a unified picture of battle spaces. Peters is 100% right that all of this software solution could only be in support of the most crucial element of any intelligence process or system: the analysts who sorted through the data and made analytic judgments and predictions. But that doesn’t make the ASAS software unnecessary, nor the effort unwarranted.

At every Warfighter I ever attended, ASAS was always part of the analytic toolset. And always, the ASAS naysayers outnumbered the proponents. We surely never gave the system a real chance, nor properly invested in the training commitment that would have been necessary to create a Fusion process truly supported by automation. It should be no surprise that MI reacted the same way in Iraq, when the stakes were higher and the frustration greater.

That being said, I do think it obvious from many years working Intel with and without ASAS (often with, but not used properly and thus ineffectively), that the underlying problem ASAS exposes is not a flaw in this or any other processing system, but the extreme magnitude of data, and the essential problems of how to find and extract the needles of interest, and once extracted, how to build anything resembling sensible understanding of what you're looking at.

Calling the MI bureaucracy bloated, stupid, and corrupt has a place, and I won't mind joining in on a chorus or two, but after we've had our fun with the Bronx cheers, where does that leave us?

If Peters knows how all these battle-wised "Dueces" [Division G2, and Brigade and Battalion S2, staff officers responsible for Intelligence and Security] are answering the root problems, he doesn't give a glimpse of what the answers are. Here’s what he does proclaim:
MI's combat veterans understand what intelligence must do, and they realize that satellites can't pierce the human soul. There's a powerful reform effort underway, from Iraq and Afghanistan back to the Army Intelligence Center and School.

Today's Military Intelligence personnel are a damned sight better than my inept, physically slovenly and intellectually lazy generation was.

On a recent visit to the Intel School at Fort Huachuca, Ariz., I found that the last ASAS nonsense had been swept away. Captains returning from Iraq and Afghanistan for the Advanced Course - a training staple - had no patience with yesteryear's bureaucratic approach to intel. They know that commanders need results, not just data dumps. The lives of our soldiers depend upon the quality of our intel.

There still isn't nearly enough money for language training (Congress would rather pay contractors, as usual), and there isn't sufficient classroom time to make up fully for the lost years. But it was reassuring to see commanders, students and faculty discarding the old faith in technology's divine powers and coming to grips with the rigors of real intel work.
ASAS training, and it’s prominence at the schoolhouse, would have been more noticeable on the enlisted side of the house, as enlisted MI soldiers are the ones who are trained to do all the heavy ASAS lifting, not the MI officer consumers of product who receive its results. But I won’t challenge Peters premise that once, attendees of the MI Advanced course spent considerable time and effort on ASAS. I know they did in the MI MOS and NCOES courses on the enlisted side, so I’ll accept that MI practitioners spent a lot of effort learning ASAS, with little practicable result.

I am curious about all that good, combat-based training now that yields so much helpful insight into the Jihadi soul. (Sorry.) I also think efforts to introduce greater automation as a means of sifting through mountains of data can’t be rightly described as intellectual laziness, either. But, in focusing on the results commanders need versus dumps of raw data, Peters and I are in complete agreement.

Frankly, as to Peters aside about the need for more translators, I'm skeptical that having thousands more translators gives us anything more than a lot more raw data we can't patch into coherent understanding. That, and we're sure to gain a lot greater understanding of all manner of partisan, tribal, and old fashioned political machinations, and likely start paying too much attention to what the players say, rather than what they do. (Which in my opinion has been the kind of errant attention that bedeviled US Middle East foreign policy for over three decades.)

Linked at Mudville Gazette's Dawn Patrol.


<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]