Next in his essay on what works in counterinsurgency, what doesn't, and how to tell the crucial difference, David Kilcullen turns to the question of measuring the performance of the host government.

Significantly, this is a long section. That's appropriate, I think, because the single biggest problem we face in Afghanistan isn't how to contain the Taliban, it is how to alter the rapacious behavior of the Karzai government. 

• "Assassination and kidnapping rate." Well duh, I hear you say. But the devil is in the details. Don't just look at high level officials. Track the sub-district governors, mayors and police chiefs. Are they getting killed? Are they just quitting? These may be indicators of a concerted insurgent push, he warns. But stability and lack of violence might not necessarily be a good indicator, because, he adds, if the area also has a low rate of voluntary reporting, it may be "an enemy district that is stable under insurgent control."

• "Civilian accessibility." Can government officials move around without an escort? Do local people avoid a given area? Even if there are not high levels of violence, this may indicate insurgent control.
• "Where local officials sleep." I really like this one because it is so simple, but it never occurred to me. In fact, I have never seen it listed before in works on metrics in warfare. But it makes sense. DK writes that, "A large proportion of Afghan government officials currently do not sleep in the districts for which they are responsible." He recommends looking into whether they fear for their safety, or perhaps are outsiders not really welcome in the districts. Both reasons are important, but have far different significance for your operations. 
• "Officials' business interests." Map them out, he says. The locals know about them, and you should make it your job to do so also. For example, he says, when there is violence against a local construction company working on an aid project, does a local official own a rival company? More insidiously, he offers the example of an official who engages enthusiastically in opium crop eradication, but has his own opium fields elsewhere. He may simply be eliminating the competition. Try to compile and regularly update a "register of officials' assets" -- and keep it in mind as you try to understand violent incidents in a given area.
• "Percentage of officials purchasing their positions." This is a warning sign. The more people are buying official positions, the higher the likely rate of corruption will be, as they have to re-pay their funders or recoup their investments, and so the more likely that abused locals will play ball with the Taliban.
• "Budget execution." Dull But Important, as I said the other day about another article on Afghanistan. Be careful of using CERP funds a quick fix to get around budget roadblocks-you may just be cementing in those roadblocks. Think beyond the length of your rotation, and consider whether your fixes are going to make life harder for your successor.
• "Capital flight." This is as close as Kilcullen comes to criticizing the Obama Administration. During the Great Afghan Policy Dither of 2009 (my phrase, not his), he notes, "we saw millions of dollars leaving the country on a weekly basis."
• "Rate of anti-insurgent lashkar formation." Another novel observation and metric. Kilcullen says these local militias tend to be indicators of districts that distrust both the government and the Taliban and are going autarkical.
• "Public safety function." Do the locals call the Taliban's 911 line (they effectively have them, he writes, and that is news to me) or the government's?