Metrics in COIN


Force Recon
Verified SOF
Mar 15, 2008
Tun Tavern

When David Kilcullen is at his best, he is unexcelled at discussing how to wage a counterinsurgency campaign. And I think the Australian infantry officer turned political anthropologist/COIN guru is at his best when he gathers field observations, boils them down to distilled principles, and then describes those rules in a clear, practical manner.

So I want to take some time to go through a paper he wrote recently in Afghanistan. (I didn't get it from him, by the way.) While it ostensibly is about metrics in COIN campaigning, it amounts to a thorough discussion of what works in such warfare, what doesn't, and -- especially -- how to tell the difference. It is written about the current campaign in Afghanistan, but clearly has broader applications. ...

After some initial throat-clearing (one of my rules when I was an editor was to see if I could cut the first three pages of any long article), Kilcullen's first major section is about metrics to be avoided. These are:

"Body count." As he says, when you have 100 enemy and kill 20 of them, you may wind up with 120 live enemies, because you just created 40 more. It's more algebra than arithmetic. ‘Nuff said? Sure, but as Sean Naylor's excellent reporting in Army Times lately has shown, there are still some Army commanders who disagree with this basic point.
"Military accessibility." Yes! One of my many peeves in Iraq was when a battalion or brigade commander would say that a route was "green" because his up-armored Humvees generally could use it without getting blown up. That may have been true, but it also was irrelevant to the security experience of the average Iraqi on the street. When I asked about that, he just didn't seem to know, or care. So I was pleased to see this high up on Kilcullen's list of don'ts. His point is a bit different from that here. It is just because you don't get hit on a road doesn't mean it is under your control. Rather, it may just mean that the enemy doesn't care to engage you there. This may be because it lacks support there, or conversely because it doesn't want to fight in an area where it is popular. Why risk blowing up your own peeps?
"SIGACTs, especially those involving violence against the coalition." This is a related point. Be wary of SIGACTs trends. Violence may be low in an area simply because it is in the uncontested control of the enemy.
"Dialog with the enemy." An interesting point, because there has been so much talk lately about various parties talking to the Taliban. In our tradition, we stop talking to the enemy when the fighting starts. Not so the Afghans, Kilcullen notes. Also, he adds, "the mere fact that our local partners are in dialog with the enemy is not an indicator, in and of itself, of disloyalty to the government."
"Any input metric." Megadittoes. This was another thing that used to drive me nuts in Iraq, listening to Americans boast about money spent, projects initiated, patrols conducted, and such. "These indicators tell us what we are doing, but not the effect we are having." Rather, he advocates, look at outcomes, and especially the effect on the population. How to measure those will be the subject of our next installment on this insightful essay.
He's right, it's all rhetoric to make you feel good and successful about a situation you are really unsure of 90% of the time. Considering that one Private First Class having a bad day or feeling impish can turn a whole village against you in the space of a few hours, it's never a good idea to delude yourself that you are winning.

Sgt Smith gets a Dear John letter from his wife. On routine patrol through a "friendly village" that afternoon he loses his cool and shoves an old man up against a wall, or kicks a kid, or tugs somebody's beard or whatever...that's all it takes. Everbody in that ville will know the story. And half the people in that ville have a relative, friend, acquaintance, 3rd cousin who's running with the opposition. So the next time Sgt Smith comes through that ville he gets a surprise. Metric input doesn't mean shit. It doesn't matter if you spent x amount of dollars building them a well or a school house...what matters is you insulted and offended and degraded a human who will react the same way you and I would react if somebody threw us up against a wall or kicked our kid. They'll fucking kill you and why not? You're a foreigner and you had it coming. Problem is, when they ice Sgt Smith they also ice half of Sgt Smith's men who had no control over what Sgt Smith did that afternoon.

We used to have a saying: All it's takes is one cowboy to fuck things up.
The sad thing is our modern military is fascinated with metrics.

Body count was the big thing back in my L/Cpl days, so it really hasn't changed much, just the buzzwords. The metrics justify the existence of the REMFs & fobbits who have to translate them into bullet points to present during all the video conferencing. To me, metrics translated into a case of beer at the end of the week if we racked up the most silhouettes on the COs Ear Board.

Some more from the same article:

Kilcullen isn't out to attack all metrics, just bad metrics. Which leads us to the point of today's post. Yesterday, he told you why he dismisses certain metrics as unhelpful. Today, he discusses how to tell what effect your operations are having on the people:

"Voluntary reporting." How many tips are you getting from the population? And how many of them are unsolicited? He warns that this metric must be assessed in the context of how many tips pan out. The more accurate the tips, the more confidence the population has in your and your allies in the host government.

"IEDs reported versus IEDs found." This one took me a moment to get my mind around. "Accurate reporting indicates that the population is willing to act voluntarily to protect the security forces." Variations in this rate may be a good indicator of local support for security forces and the government, he says.

"Prices of exotic vegetables" and "Transportation prices." Now we are getting into the nitty gritty. Anything that embarrasses your S-3 as he discusses it in the briefing probably is a good metric. Until now most of DK's recommendations have been more or less rooted in common sense. But to understand this weird one, you need to understand local conditions. What people are paying for vegetables grown outside their district is a quick indicator of road security. Trucking companies factor in the risks they face, as well as the cost of bribes and other forms of corruption. So variations over time may be a helpful indicator of trends in public perception of security conditions and the corruption level of government security forces.

"Progress of NGO construction projects." A better indicator than government-sponsored works, which, he notes, "the insurgents may attack on principle." NGO projects go well when materials prices are stable, the labor supply is adequate, and security problems aren't interfering.

"Influence of Taliban versus government courts." If the locals trust the Taliban-run courts more than the government's, you have a problem. How many cases are each handling in a given district?

"Participation rate in programs." Both the government and the Taliban have a variety of economic and community programs. Which are more popular?
"Taxation collection." What is the compliance rate with government taxes, vs. Taliban taxes?

"Afghan-on-Afghan violence." Unlike sigacts against coalition forces, he says, this is a good measure of public security.

"Rate of new business formation and loan repayment." A good indicator of public confidence. He notes that Afghans tend to have a low rate of business formation but a high rate of repayment.

"Urban construction new start rate." Another good indicator of confidence in a given area.

"Percentage of local people with secure title to their house and land." This one really surprised me. Kilcullen says that the Taliban has used land disputes adroitly, sometimes settling them justly to further their influence, and at other times exacerbating them to gain the allegiance of one side. The higher the percentage of secure titles in a given area, the less chance for the Taliban to step in and exploit the situation. Can you imagine being a new battalion commander in the area trying to keep up with this stuff? Tribes, women, feuds, land disputes, religion -- it is just too hillbilly for me. Where is Andrew Exum when you need him? Probably off writing up the new policy for Afghanistan.