Monday, July 03, 2006


Hindsight Analysis

(UPDATED and brought to top)

Newsweek offers the latest in a series of variations on the theme of bad intelligence and exaggerated threat, with Michael Hirsh’s story on The Myth of Al Qaeda.

Hirsh runs through a litany of what he construes as mis-identifications and mischaracterizations, largely based on a few Libyan exiles and Ron Suskind’s new book, "The One Percent Doctrine."

I’ll skip the Jihadi gossip, read Hirsh’s piece for the flavor of it. But here’s how he concludes:

The ultimate tragedy of the Iraq war was not only that it diverted the U.S. from the knockout blow against Al Qaeda in Afghanistan and Pakistan—the deaths of bin Laden and Zawahiri would likely have persuaded most jihadis it was wiser to focus on the near enemy—but that Iraq also altered the outcome of Al Qaeda's internal debate, tipping it in bin Laden's favor. "Iraq ended that debate because it fused the near and the far enemy," as Arquilla puts it succinctly. America ventured into the lands of jihad and willingly offered itself as a target in place of the local regimes. And as a new cause that revived the flagging Al Qaeda movement. It is, no doubt, bin Laden's greatest victory.
Except, of course, that Bin Laden and his Al Qaeda managers think the opposite, based on intercepted documents and communications.

I am really curious about the credentials of these foreign policy and security “experts” who come out of the woodwork with fortuitous 20/20 hindsight assessments that, incidentally, back up what seem like absurd assessments when they sat in positions of influence.

Zarqawi was a strawman, Al Qaeda wasn’t really a threat, we created international terrorism, Al Qaeda was a bunch of bumbling, failing would be Jihadi James Bonds, with no appreciable following or organization. We made them into the threat they became.

Except, of course, that these failing bumblers managed to orchestrate a massive, coordinated terrorist strike over several years, including getting operatives into the US, through flight schools, and coordinated horrific attacks against the World Trade Towers, the Pentagon, and a probable target of either the White House or Congress. (Only this last effort unsuccessful, save the destruction of Flight 93 and every soul on board.

Western Intelligence misread or misidentified the exact roles and importance of individuals operating in a clandestine, tightly controlled, and impenetrable terrorist organization. This underscores the great difficulties of gaining intelligence against such a non-state adversary, rather than diminishes the threat posed by the organization.

Hirsh seems to fall victim to that most common of analytic blunders, assuming that what small proportion of information available should somehow be construed as representative of the (missing) whole.

A quick example. On a battlefield, intelligence sensors might pick up evidence of scattered reconnaissance elements across an area wide enough to suggest the presence of a Division sized enemy. An analyst can take limited intercepts and indicators, compare it against a template of what such an organization would look like, extrapolate the missing information, and posit the Division.

If he mistakes unit boundaries, he may be lumping disparate elements into a whole, and he instead has several Divisions in front of him, rather than one. In the same manner, if a mistake has been made on type of units in evidence, he could be seeing elements of a Regiment, Brigade or even a Battalion.

Intelligence Analysis is as much art as science, and much like the reporters who try to glean the “real story” out of isolated data points, the analyst can exaggerate what data he has or overextend patterns of analysis into the area of the unknown (the “gaps”) in ways contrary to what is really the case.

Could the West have exaggerated the threat? Possibly. But more likely, armchair critics with an agenda, with 20/20 hindsight, can always point at what didn’t happen, and say, “it wouldn’t have.”

An alternative explanation is, that aggressive analysis and interdiction prevented it, as one might argue, we have done in preventing another Al Qaeda attack on US soil.

Arrests and reports of broken plots would seem to suggest such efforts were underway. Rather than point to the absence of attacks as proof that the threat was exaggerated, it is rather more likely that the demonstrable examples of broken plots and failed attempts reflect success against an enemy that was all too capable, were we to go back to ignoring the threat.

UPDATE: I meant to reference this and lost my train of thought. I first posted on some of the reasons for presistant analytic error, in the context of how our Intelligence Analysts displayed similar or at least analogous patterns of analysis. At the time, John Schroeder of Blogotional also added some insight into that discussion.

Here's what I said then, which I still find applicable:
Any time you try to analyze a trend, or develop an overall macro definition of a large number of small, discrete events, you run into a problem of methodology, even perspective. And there's no easy resolution.

Imagine you are looking at overall violence in Iraq. You could look at a range of violent acts by Anti-Iraqi Forces (AIF), from small arms fire to assassinations to improvised explosive devices (IED) or various kinds to coordinated attacks and ambushes.

Zero in on IEDs, they pretty much do in command channels here. Next, imagine that you might create some kind of plotting over time or geography, look for hot spots, chart patterns, trends, increases, decreases, etc.

Now try to describe what you're seeing, and be careful of the assumptions that are part of your descriptions.

1. First assumption to acknowledge, you picked one type of violence over another. The fact that you pay attention to that one factor will influence how and whether your enemy changes tactics to change how you see the data. (That is, if they’re good. Viet Cong? Soviet Union? Capable of that. Serbia? North Korea? Fat chance. Al Qaeda? If not in the beginning, probably by now.)

2. Also, what's the context? Was that a type of violence that exists elsewhere, you can compare levels to? Or Pre-War? Are these isolated, spectacular media driven events (or can they be), how frequent, how many people or what proportion of people (soldiers or civilians) does this event affect?

3. How does this level or type of violence compare with othert situations, regions, trouble-spots, historical precedents? Is Iraq safer than Columbia? Less violence prone?

4. Where you focus. Obviously, in a "data scatter" type diagram or model that will be used by the Press, you will look at the "black" data points and only see those. But you could also note or evaluate all the "white" spaces in between (the absence of the event). Or, do you create some kind of average, in effect blending black data points and white empty space, and create a degree of gray that you then evaluate?

I am beginning to think this paradigm hold true for reporting in Iraq. Liberal news media that had an agenda to begin with (we were right about the war and the incompetence of this administration), seek out only individual discrete events that support their template. And in purely statistical terms, there's a value to that way of looking at data. (Projections and forecasts, for one thing.)

You could evaluate the same data but concentrate on the empty spaces, the cessation or absence of violence. There's utility in that, too. Is the threshold of violence-to-safety beyond some point that populations change their behavior significantly in response to it?

Blending the data could give you the closest to what some news organizations attempt in highly subjective "mood" or environment pieces (the "raise questions" and "the mood was tense" type reporting), but usually committed without any hard data to support whatever the subjective "impression" of the reporter doing the hit piece.

My own gut suggests that there is truth in all three degrees or areas of focus. But there is no doubt at all, that to focus on just the little black dots is to not just miss the forest, but to run headlong into some tree as well, from searching out each root.
(Cross-posted at Milblogs)

Links to this post:

Create a Link

<< Home

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

Subscribe to Posts [Atom]