A whole generation of instant experts on counterinsurgency and counterterrorism have spent the last several years proclaiming how these struggles are real wars, not police matters. Being people with more volume than knowledge, they've missed the point. Of course, counterinsurgency and counterterrorism are forms of warfare. However, victory depends on turning these wars into something that looks like police matters.
At the end of a counterinsurgency war like the one in Iraq, political and military measures will have reduced the insurgents from revolutionaries to criminals, from a movement to a gang. Robbed of political legitimacy, the ability to intimidate or persuade, larger numbers of active members and supporters, and a free range of action, the insurgents dwindle to something that should be treated as a police matter.
Long before the war has reached this point, what happens to suspected insurgents while in custody matters greatly. Unspeakable acts that fall under the polite term "human rights abuses" are the persistent mistake that governments under siege make. Form a combination of impatience, arrogance, and fear, regimes see torture, murder, indefinite detention, and other forms of "abuse" (another polite term) as the short, urgent road to defeating the insurgents. The political backlash is rarely worth the limited, unreliable information gained, or the number of potential insurgents jailed.
People with successful track records at counterinsurgency counsel a much different approach. (For an example, see Stuart Herrington's recent op-ed piece.) Therefore, stories about the treatment of Iraqi prisoners are central to the counterinsurgency story. If war outlasts the bad memories of Abu Ghraib, jails will still be an important stage on which Iraqi drama plays out. Prison statistics may be far more important than casualty numbers.