A lot of The Accidental Guerrilla concerns ideas of terrain, space, and time. In fact, quite a bit of it could be considered an architectural approach to counter-insurgency. This is not surprising; a major theme is the idea that the conflict environment - the state of being at war or potentially at war, the disrupted social and political structure, the faltering infrastructure, the global black market - is the enemy. After all, it is one of the reasons people seek survival through certainty by calling on the deliberate guerrillas to influence their other political relationships.
One example of this is the one I've already written up - the armoured patrol vehicle as urban submarine, a self-defeating machine that itself divides the counter-insurgents from the people in an ironic reversal of their own thinking.
Kilcullen goes almost New Urbanist on this; discussing the Iraq experience, he argues that a huge flaw in the US strategy was that they had to commute to the battle, travelling in monster armoured vehicles, without contact with the civilian population, but still vulnerable to IEDs and ambushes on the over-predictable road routes between their camps and their areas of operation. The answer was to redeploy into the cities and move into positions that let them walk to work; I tell you, Richard Florida got nothing on him.
Similarly, a major aim of his campaign plan was to control access to Baghdad, counterattacking the NOIA encirclement strategy and preventing insurgent "commuters" from the Sunni semi-urban belt getting into the city. You could almost call it a critique of suburban warfare.
This concern with space is also a major theme of the case study on Kunar and road building. The construction of a road was intended to get access and control of the narrow flood plain at the bottom of the valley, which is where everyone lives, rather than up on the mountains. Nothing much grows on the tops and it's tough to get up there or back down, so the only important places up there are a few tactically important hilltops.
Road access meant that it was easier to force the Taliban to go quiet, either by climbing into the mountains or by going underground. More importantly, it made it possible to keep them there, and to deliver economic benefits. But perhaps the biggest changes it provided were as follows:
Firstly, it changed the topography so that the government side were in the villages, looking out, and the Taliban were outside, looking in. The US or Afghan government fire was outgoing; the Taliban's, incoming.
Second, it made it worth arguing where different groups' authority ended; without the road, it was bounded by the difficulty of travel. Once they had to argue about it, the government or the traditional authorities could be called in to arbitrate the dispute, boosting their authority and making them useful. In a sense, the Kunar case study is all about creating a demand for government, or at least competing with the Taliban to supply it.
An interesting question, though; the whole paradigm of The Accidental Guerrilla is based on experience in places where the state is absent, illegitimate, or never established. But many of the same phenomena happen in places where the state, or the structure of traditional authority, once existed but has broken down.
Further, the international jihadis are trying to move (as Kilcullen says) from expeditionary terrorism, where their operations are set up in the home base and carried out remotely, to a guerrilla model where they are set up by sympathisers recruited in the target state. This implies that the process will have to take place in an environment where the state exists here and now.
I'm less convinced by his arguments regarding this; obviously, the naked city has as many possible base-areas as it has people, but as Daniel Davies pointed out, the current European takfiris seem to have less access to firearms than a typical criminal gang, and one of the most worrying possibilities in this line is indeed that they cross-fertilise with ordinary decent criminals. Kilcullen's practical recommendations in this line are mostly commonsensical, although he is very keen on Cold War analogies with efforts to start non-communist unions and the like, and the other activities of the Blearsministerium.
However, despite the technological implications of auto-immune warfare, he also believes that "biometric reconnaissance" is a strategically important capability. I rather suspect that we've already been seeing the effects of this advocacy without knowing what was behind it.