The argument is simple. (Really, really simple.) It's stupid to say that you can't kill your way to victory in a counterinsurgency, because if you kill enough of a certain cohort of people, then they'll stop fighting you. "If you are really killing the enemy day on day, with captured weapons supplying the proof of a legitimate and armed threat," writes William Owen, "then sooner or later the enemy's will to endure will be broken. No enemy has unlimited manpower." Also, the population in which an insurgency operates are merely "spectators to armed conflict."
Where to begin. Sometimes, as in nearly all counterinsurgency fights, the counterinsurgent cannot easily distinguish the insurgent from the civilian. That's not always because of poor tactical intelligence or ignorance of a foreign culture. It's because the guy who gives his old cellphone to his cousin so his old neighborhood friend can use it to construct IEDs for the guy paying a good going rate -- quick, is he an insurgent or not? If you can't immediately answer, Owen's argument falls apart.
Even if you unflinchingly decide the guy's an insurgent, killing the guy can easily inspire the whole neighborhood to rally to the insurgents' cause. Quick: do you kill the guy so you can approach the Magic Number of dead insurgents that assures you victory? Or does not killing the guy take you further away from the Magic Number?
I know, I know. Counterinsurgency is OVER. Whatever context, wisdom or experience led people to consider it a least-bad option ought to be ignored. Its unsuitability for Afghanistan has rendered the entire enterprise inert. What, you didn't read that National Journal piece?
So look: there are many good arguments against counterinsurgency. Take up its co-mingling with imperialism; check out its dubious predicate relationship with counterterrorism; consider that most counterinsurgencies are no place for American troops; whatever. Just don't mug yourself by asserting that you can know exactly who to kill to prevail over an insurgency.