After spending last month in Afghanistan on my fourth trip this year, the situation can best be described as a glass half full. A multifaceted effort in the south, led by a "surge" of U.S. and Afghan troops, has increased security in the southern Pashtun heartland this year. But a steady drumbeat of high-profile attacks, including a brazen assault on the U.S. embassy and assassinations of key Afghan officials, has had an outsized impact on the population by eroding already weak confidence in the Afghan government and the forces supporting it. As I was ending my trip with a few days in Kabul, an SUV packed with 700 kilos of explosives rammed an armored "Rhino" bus, killing a dozen Americans and inflicting horrific burns on others in a targeted, planned attack reminiscent of the worst days in Baghdad. The U.S. military has adopted the line that these "spectacular attacks" are in fact signs of the Taliban's weakness, and points instead to a notable decline in the number of enemy initiated attacks.

But numbers matter less than perception in these wars -- and shaping perception is precisely how terrorist groups fight and win. The dominant narrative in Afghanistan is: the Americans are leaving, the government is weak, the Taliban is still strong, and Pakistan is the problem. And though the details of the cross-border NATO bombardment Saturday, which killed 24 Pakistani soldiers, are still murky, it will undoubtedly send U.S.-Pakistan relations into a further downward spiral. Unless the reality underlying these perceptions about the United States in the region are addressed, sustainable momentum on the government side cannot be generated. The U.S. surge may have started to turn around the momentum that is currently favoring the Taliban, but Afghan political and military actions are required to create a perception -- and the reality -- that the Afghan government is capable of surviving and prevailing. Addressing several factors bedeviling the Afghan campaign between now and next summer, when the traditional fighting season resumes and more troops go home, might turn perception and thus momentum in the Afghan government's favor. Six are most critical:

Promptly reaching a strategic partnership agreement in which the United States pledges to continue providing security and economic assistance after 2014 could go a long way toward reducing Afghan fears of abandonment. The U.S. and Afghan governments have been negotiating such an agreement for over a year, and the general idea of a partnership (albeit with a 10-year limit on American forces in Afghanistan after 2014) was approved at President Hamid Karzai's Loya Jirga this week; both sides hope to finalize it before the Bonn summit next month. The sticking points are night raids, which Karzai would like to end, and the transfer of detainees to Afghan control, despite a recent U.N. report cataloguing abusive practices in Afghan intelligence detention centers. The first issue can be postponed or finessed, but the United States is bound by international convention not to turn over detainees if it believes they will be tortured or mistreated. In the run-up to the Bonn Conference, Karzai will hopefully stay focused on the need to reassure Afghans of ongoing U.S. support and assistance as coalition troop levels decline over the coming years.

Forging a common front with his main backer, the United States, would help Karzai repair some of the image of weakness that plagues his government. Veteran U.S. ambassador Ryan Crocker and the International Security Assistance Force (ISAF)'s top commander Gen. John Allen have worked hard to "reset" the relationship with Karzai since they arrived this summer. One of his frequent interlocutors notes that Karzai "is exhausted, and he knows he is irritable and unpredictable." Nader Nadery, a commissioner of the Afghan Independent Human Rights Commission, chides Karzai for his "lack of firmness. There is no clarity on where he stands." And, like many Afghans I spoke to, Nadery sees the president's tolerance of corrupt activity as a major weakness. Yet despite these failings, many Afghans I've talked to still see him as currently the only leader who can balance the myriad rivalries and swirling tensions that beset Afghan national politics, in particular the concern that Tajik leaders may be girding for an all-out civil war once the U.S. departs. The assassination of former president Burhanuddin Rabbani, the Tajik head of Karzai's High Peace Council, significantly ratcheted up those concerns.

Afghan security forces need to move demonstrably into the lead. The U.S. plans to push Afghan forces out front in more and more operations, if only to determine just how ready they are to take the lead and where the weak links are. To get them ready, U.S. forces must make partnering with and mentoring their Afghan partners its top priority, which it is not. By the military's own count in the latest DOD report mandated by Congress, 95 Afghan units in critical areas have no coalition partner whatsoever, and many others are only loosely mentored.

Afghans will have to step up if the three stated objectives of the military campaign are to be accomplished -- namely, to secure the south, the populated corridor along Highway One up to Kabul, and key areas of the east, especially the crucial provinces of Paktia, Paktika, and Khost. The south has been greatly stabilized over the past year, but the job is not yet done. I spent ten days in Kandahar province, where the longtime Taliban strongholds of Zhari, Panjwayi and Maiwand are still heavily contested. Returning from Maiwand, which is mostly still Taliban dominated, I passed through the Panjwayi district center, where a car bomb detonated a short while later. We stopped in Kandahar city for dinner -- one of the few times I have been with a U.S. military unit that felt comfortable dining at night in an Afghan city. The company commander ordered roast chickens, pilaf and delicious nan bread for all his men, and we picnicked amid the city's bustling nightlife, surrounded by curious Afghan men and children. Yet that reality coexisted with a frontal assault, hours earlier, on the Kandahar provincial reconstruction team, located just across from the provincial governor's office.

The current priority is to expand the security bubble beyond Kabul to Wardak, Logar and then Ghazni and Zabul provinces, thus securing the country's major highway all the way down to Kandahar. After that, the campaign plans to shift to the east, but it needs to tightly focus on the key population areas. There is a danger of focusing too much on sparsely populated provinces and neglecting the Khost to Gardez corridor, which has a large population and has been both an insurgent stronghold and transit route from the beginning of the war. Attempting to cut every ratline in the country is a sure fire way to fritter away forces. The only way that the campaign objectives can be achieved is if coalition forces focus on the important areas, and if Afghan security forces take more of the lead.

The expanding self-defense initiative may increase security. Security out in the countryside, where the insurgency is based, is starting to get a lift from the year-old Afghan Local Police program, but it is too early to judge the net effect of the program. It now has 8,500 recruits, mostly farmers who have volunteered and been trained to do checkpoint security and patrols in their villages. I've visited a half dozen of the 57 sites, where Special Operations Forces train and mentor the ALP and work with the district police chief, who is the Afghan official charged with their formal command and control, as well as providing them with salaries, trucks, weapons and ammunition. The program, strongly backed by ISAF, is funded and approved to grow to 30,000 in 100 districts, but the painstaking process of gaining community support and ensuring that the government exercises proper oversight means it will be another year or more before the program is fully implemented.

Thus far, RAND quarterly assessments done for the Special Operations Forces command in Afghanistan say that the program is broadly supported by local residents and that security in the sites has improved. The principal concern expressed by the program's critics is that these small groups (limited to 300 per district) will become militias, which historically have been armies of 30,000 or more under the command of a single warlord. The ALP program aims to prevent that by creating a strong link to district and provincial police chiefs from the outset. A recent Human Rights Watch Report catalogued allegations of abuse, disputes and ethnic tensions in some areas, although U.S. officials dispute some of the claims, which are under official investigation.

The ALP program features greater oversight and vetting mechanisms than earlier similar efforts, but continuing vigilance is necessary, as with all security forces. The Special Forces team in Maiwand told me about an ALP member who went AWOL, and along with his brother, started holding up cars for bribes on Highway One, the highway linking Kandahar to Kabul. The Taliban shot him dead. ALP commanders have fired numerous members they feel are not up to snuff. One of the sites, in Baghlan, is a unique case in that the Afghan government asked SOF to form a local police unit out of former insurgents from the small Hezb-e Islami faction; insurgents are not the recruiting base for ALP, unlike the Sons of Iraq program. The tensions in the area are increased by the fact that the area is a pocket of Pashtuns surrounded by a larger Tajik population. The Afghan interior ministry official who has national oversight of the program, Brig. Gen. Ali Shah Ahmadzai, told me that he has put the ALP commander, who is under investigation, on notice that he will be arrested "if he does not behave."

Promises to rural Afghans should be moderated but honored. The reality is that the dysfunctional Afghan bureaucracy cannot deliver funds to individual provinces, let alone the district level where 76 percent of the Afghan population ekes out a subsistence living. At present, most district governments are fig leafs for the distribution of international aid projects. I sat in the Khas Kunar district governor's office as he talked to the provincial governor's secretary about getting funds for a backhoe to dig out canals damaged in recent floods. A U.S. Agency for International Development (USAID) official had suggested he seek to tap the $25,000 "performance based" fund that USAID provides each month to provincial governors. The district governor smiled and nodded as the conversation concluded, then turned to us and said, "It will take a year for the necessary paperwork and to get approval from Kabul." Yet there has been progress - in many areas district governors did not even visit, let alone live in, their districts last year. Many areas have now been pacified sufficiently for them to return. During the week I spent in Maiwand, I saw the governor preside over four meetings, although he still returns to Kandahar city each weekend where his family lives. Prioritizing needs is essential, since USAID funds and staff are slated to decline significantly, and the civilian state-building effort will focus on the provincial level rather than the districts. The district delivery program, for example, will end next year after reaching only 45 of the country's 398 districts. This program, which supports training and salary increases for district officials, is current under way in only 12 districts. A number of civilian officials and nongovernmental organizations believe dispute resolution, primarily through the traditional local mediation practices system, should be the highest priority in conflict-ridden areas where land and other disputes often fuel insurgent violence.

Pakistan is the glaring problem that has dramatically worsened and must be addressed. A senior U.S. official in Kabul declared flatly to me, "We will not succeed unless the Pakistan safe havens are reduced." He does not believe Pakistan will come around, despite countless U.S. pleas for more action. Pakistan balks at getting tough on the Afghan insurgent groups that are the central cards in its hedging strategy -- particularly the deadly Haqqani network, based in Pakistan's North Waziristan tribal agency. And U.S. officials recognize that an assault into the agency simply is not going to happen.

The United States and Afghanistan will have to neutralize the threat from Waziristan through their own efforts, and indeed these efforts have stepped up dramatically. Operation Knife Edge rounded up many Haqqani operatives last month. A senior Haqqani facilitator, Mali Khan, was nabbed as he crossed the border, and another, Janbaz Zadran, was killed by a suspected CIA drone attack on October 13. During the week I spent in Paktika, U.S. forces dropped 33,000 pounds of bombs on the border to close mountain passes that the insurgents use. A U.S. commander there pointed out a dozen "POOs" or points of origin, of artillery or rockets fired from Pakistan into Afghanistan. They were all within sight of Pakistani Frontier Corps outposts - leaving little doubt about active or passive support for militants among some Pakistani forces. U.S. and Afghan forces are under direct attack from Pakistani territory, and Lieutenant General Curtis Scaparrotti assured me that when they take "effective" fire, U.S. forces shoot back. The dilemma, however, is that Pakistan holds a critical trump card, as it demonstrated this week by its ability to close major commercial crossings in retaliation for Saturday's cross-border strike. When I visited that part of the border area last month, Special Forces soldiers told me that an insurgent camp near the two border posts was a major source of attacks in the region.

If the Afghan and U.S. governments can agree on a common plan, backed by a lower but sustainable level of U.S. troops, they just might start winning the battle of perceptions. Given the insurgents' lack of popularity -- they rely on intimidation -- the Afghan government should be able to prevail in either defeating the insurgency or forcing them into a political accommodation if the United States remains willing to lend a hand. It might take five or ten years. But the alternative is not a pretty one to contemplate: the U.S. departs, Afghanistan crumbles, and the war fought to avenge the 9/11 attacks is perceived to be a failure. As I wrote in my last book on the Iraq war, Americans need to be prepared for wars to last a decade. The Afghan war, which the United States attempted to fight largely through a counterterrorism approach for the first seven years, has only begun in earnest in the last three. There is a third choice between the current large-scale U.S. counterinsurgency campaign and reverting to a counter-terrorism approach, and that is an Afghan-led counterinsurgency effort that that the United States can support in a sustainable way.

Linda Robinson is Adjunct Senior Fellow for U.S. National Security and Foreign Policy at the Council on Foreign Relations. She is writing a book on the war in Afghanistan.