When President Obama relieved General Stanley McChrystal over comments published in Rolling Stone last week it was big news, but it was not the first time a political leader fired his most important commander in Afghanistan. In late 2007, Quetta Shura Taliban leader Mullah Omar fired his commander for southern Afghanistan, Mullah Mansoor Dadullah because "he disobeyed the orders of the Islamic Emirate." The command failure was serious. Mullah Omar had never trusted Mansoor Dadullah and worried that he would be as independently minded, and brutal, as his older brother who he replaced. Mullah Omar's fears were born out when Mansoor Dadullah struck out on his own (described variously as negotiating with the Afghan government and refusing to negotiate with the Afghan government), even threatening attacks outside Afghanistan. Such an approach would have been a major policy shift and directly undermined Mullah Omar's carefully crafted political position. In a war without a stark division between killing and negotiation, Mullah Omar would leave no doubt he was in charge.

Obama's dismissal of McChrystal was warranted for similar reasons. The war in Afghanistan can be lost just as easily in diplomatic conference rooms in Washington, Kabul, and Islamabad as in Marjah's poppy fields. Indeed, success on the battlefield in Afghanistan, achieved via counterinsurgency (COIN) or some other approach, can only create the conditions for the political deals that will ultimately determine the war's outcome. Achieving and sustaining those deals will require Pakistani support, and the president is the only American capable of cutting the deals with allies and enemies alike. There can be no doubt among any of them that he is the ultimate authority.

Counterinsurgency advocates rightly argue for the integration of military and political strategy in Afghanistan. The problem is that COIN supporters often fail to acknowledge that many of the political deals fundamental to positive results inside Afghanistan must be made outside Afghanistan.

COIN in Afghanistan without a reasonable plan to convince Pakistan to confront all the militant demons in its midst is an irresponsible half-strategy. Counterinsurgency will not defeat the Quetta Shura or the Haqqani network, the two most dangerous militant networks in Afghanistan, if Pakistan continues to harbor these networks. And al Qaeda will not be defeated, dismantled, or destroyed if Pakistan continues to resist destroying the Haqqani network, the jihadists' most important protector and closest ally in Afghanistan. Pakistan is not Vietnam's Laos or Iraq's Syria -- a place where Afghan militants train, travel, and regroup -- it is part of an integrally connected political and cultural system that fosters the insurgency in Afghanistan and, more importantly, the globally focused terrorists who seek to attack targets in the West.

Obama rightly recommitted himself to a COIN strategy in Afghanistan after dismissing General McChrystal; despite the difficulties, it is one component of a strategy that offers the United States the best chance for a positive outcome. But that effort must be coupled with deeper Pakistani support to induce certain Taliban components to the negotiating table. If such support is not forthcoming, U.S. strategy should change dramatically.

What does a political strategy for Afghanistan and its environs look like? These are four interconnected and rather distasteful ideas that reflect the hard and bitter choices the United States must make, an admittedly incomplete list:

1. Prioritize efforts to convince Pakistan to attack the Haqqani network and its al Qaeda allies in North Waziristan and the Federally Administered Tribal Areas while reducing pressure on Pakistan to attack the Quetta Shura. The Haqqani network is more closely aligned with al Qaeda than is the Quetta Shura, and its defeat is more important to Western security interests. Unfortunately, the Haqqani network is probably also dearer to Pakistani officials than is the Quetta Shura. The United States must quickly and forcefully demonstrate that incorporating the Haqqani leadership into an Afghan government is unacceptable.

2. Quietly support Afghan government efforts to negotiate with the Mullah Omar-led Quetta Shura and assure Pakistan that doing so means that the United States acknowledges its legitimate interests in Afghanistan.

3. Stoke existing tensions between the Quetta Shura and the Haqqani Network, which swears allegiance to Mullah Omar but operates independently. An Afghan accommodation with the Quetta Shura is unacceptable to U.S. security if it remains allied with the Haqqani network because of the latter's ties with al Qaeda.

4. Prioritize building and supporting Afghan security forces that are capable of denying Taliban groups victory in Afghanistan after a U.S. withdrawal, even if that means decreasing the overall number of Afghan troops. Right now, the Quetta Shura and other militants believe that the coalition will withdraw, the Afghan government will wither, and it will ultimately seize control. Thus, the Quetta Shura and other potentially reconcilable elements have no incentive to cut a deal. Critics argue that President Obama's 18-month timeline explains the Taliban's confidence that the United States will withdraw -- and no doubt the deadline reinforced the point -- but Afghans know invaders have always come and gone. Afghans know the Taliban and warlords will stay, not to mention the Pakistanis and Iranians. They have nowhere to go. But if the United States cannot stay, it must convince the Taliban that U.S. withdrawal will not deliver Taliban victory. Building an elite corps of reliable Afghan troops that can be supported from abroad is more likely to send that message than a large, unmotivated force.

Gen. David Petraeus was the right choice to succeed Gen. McChrystal. But even Petraeus, with his knowledge of COIN and a regional perspective earned while Centcom commander, would be wise to approach Afghanistan with a healthy dose of humility about what the U.S. public understands about Afghanistan. Such humility offers its own lessons, chief among them that the United States is easily manipulated by Afghans and Pakistanis who understand Afghanistan better. The United States needs a political strategy in South Asia, but it must incorporate this epistemological disadvantage, which means clearly delineating a short list of policy red lines (like excluding the Haqqanis from reintegration) and leveraging the military situation so that all players have an incentive to negotiate. Beyond that, the permanent political players in South Asia will have to work things out for themselves.

Brian Fishman is a counterterrorism research fellow at the New America Foundation.