There is talk from both candidates about the need to move troops out of Iraq and increase the number of troops in Afghanistan. ... Is there a lot of discussion within the military about the need or the opportunity to move troops?

There's a consensus within the military that more troops are required in Afghanistan, at least a brigade or two, possibly more. It's difficult to really know what military advice about troop levels in Afghanistan would be if it were unencumbered by the reality that only a limited number of brigades is actually available in the foreseeable future.

Talk about those constraints.

There are now as many combat troops in Iraq as there were before the surge. U.S. troops in Iraq will begin to draw down this autumn and into the winter, but at a pace that will limit the number of brigades available for immediate deployment to Afghanistan.

For one thing, the brigades that have been in Iraq have been on particularly long tours, 15 months in duration. They need to be refitted and rested before they can flow immediately into Afghanistan. And the number of stateside brigades that are available for immediate deployment to Afghanistan is quite limited.

So it will be possible to flow one or two brigades into Afghanistan in the first months of 2009. Beyond that, I think the military is uncertain how many soldiers and at what pace they can redeploy to Afghanistan.

So we're talking about 5,000 to 10,000 soldiers, a small number given the needs and the deterioration of Afghanistan.

Yes. And also, there's a sense in which an entirely new strategy needs to be conceived in Afghanistan. ... If U.S. policy is based on the premise that U.S. troops are going to fight this war for five or 10 years, then that policy is likely to fail. What's required is a combination of short-term U.S. deployments and effective nation-building in Afghanistan.

The problem with policy now is that the combat forces available are inadequate. But the broader project of standing up the Afghan army is also behind schedule. ...

The troops are bringing a set of skills out of Iraq. ... Have they learned things that are going to be useful in Afghanistan?

Tactically, yes. This is the most combat-tested American military at every level in generations. ... Many of the soldiers and officers who will end up in Afghanistan in 2009 and 2010 have either already been there or have fought in very challenging, irregular conditions in Iraq. ...

Strategically, Afghanistan, however, is a big challenge, and one that I'm not certain the military has fully come to grips with. This is a war that has been of secondary importance since the invasion of Iraq. It has not received the strategic attention, the top-level attention, that it has required, and that's why it has deteriorated.

In Gen. [David] Petraeus, you have a very ambitious commander who's coming out of a successful tour in Iraq to take control of CENTCOM [United States Central Command], and in that role will oversee strategy in Afghanistan. He brings with him a consensus and a brain trust that is brimming with confidence on the basis of their recent experience in Iraq and has a set of principles and ideas that they're going to apply in Afghanistan out of conviction that they know how to engineer change. ... But I think Gen. Petraeus recognizes that Afghanistan is quite a different environment from Iraq and that it's not necessarily going to be amenable to the same tactics and techniques that worked in Iraq.

How is it different?

First of all, it's a much larger geographical territory. It's a much more difficult terrain. The enemy is more dispersed. The enemy is of a different character. The Afghan state is much weaker than the Iraqi state. It doesn't possess the oil revenue and national traditions that have allowed the Iraqi state, although it's weak, to nonetheless play a rising role in the war. ... Additionally, in Afghanistan, the national forces, the army and the police, are far behind where the Iraqi army looks to be at the moment.

And the nature of the war is quite different. You have insurgency that is rooted in 20 or 30 years of continuous conflict, that takes succor from a sanctuary across the border in Pakistan, and that has a call upon rural populations and an ability to intimidate rural populations that is distinct. It's not unique, but it's different from the Iraqi insurgency. ...

The idea that you can walk over to Afghanistan and try to arm the tribes against Al Qaeda or against the Taliban is a fallacy. … All counterinsurgencies are born from local conditions and local challenges and local problems. You're going to have to relearn, from the ground up, in Afghanistan how to change security conditions there. …

So an Awakening won't work.

An Awakening is an interesting analogy. … Are there reliable tribes that are already ideologically opposed to Al Qaeda and the Taliban and who can be strengthened, as part of a local approach to counterinsurgency? Possibly. But there is no broad, easy play available to turn some extant tribal structure against the Taliban.

The tribal structure in Afghanistan has been scrambled by 25 years of war. It is not as solid or coherent a social structure as the tribes in Iraq, particularly Sunni Iraq, were. The tribes in Iraq had been instruments of statecraft, continuously, right back to the colonial period. They were an extant entity. So when they turned, they turned with strength.

You mention deterioration on the ground in Afghanistan. ... What's gone wrong?

... The Taliban now controls large swaths of Afghanistan effectively, in the sense that they can deny movement on the roads. They control the nights. They can intimidate populations. They can deny the delivery of aid, and they can challenge the government for control of the delivery of services and justice.

These are the core definitions of governance, so in substantial parts of the south and the east, and increasingly around Kabul, the Taliban are either partially controlling territory or moving in that direction.

And they have, in particular, been able to shut down transport on a national basis. They control not only the main highways in the south and east to a significant degree, or they threaten those highways, but they are also able to challenge NATO supply lines across the border from Pakistan to the extent that NATO has now had to reconsider its own geography and try to truck more material down from the relatively quiet northern, former Soviet republics. ...

How did they gain strength, though? ...

The Taliban have had this plan for insurgency from the very beginning, from the time that they lost power in late 2001. Their leaders declared: "We'll be back, and we're going to follow a script that we have followed before. We're going to slowly build an insurgency. We have patience. Our clock is a lot longer and more elastic than that of the Americans and NATO. And we will gradually build our way back."

And that's precisely what they've done. They've done it from several sources of supply and finance. They have taken advantage of permissive territory in Pakistan to connect themselves to smuggling and other rackets to develop revenue bases. They have attracted international finance from Islamic proselytizers in the Persian Gulf, and elsewhere [they have] revived old financing networks that they used to exploit.

As the opium economy revived, they have inserted themselves into that as taxers, as transport carriers, and they have now multiple sources of revenue in the many tens of millions of dollars, whereas in 2003, 2004, they were struggling to meet payroll.

They have a political program that, while ruthless and to some extent not very promising to Afghan civilians, is nonetheless coherent and enforceable by the writ of violence and execution.

And they have been quite ruthless in building out their control through executions of rival leadership and opponents. They started this in Pakistani territory, in the Federally Administered Tribal Areas [FATA], where they essentially took control by executing many of their local opponents. And they've extended those same methods across the border into sections of Afghanistan as well.

So who's running the Taliban now?

Mullah Omar still leads a shura that is honored by many of his followers as the supreme command. So there is a national command. Its full extent and daily operations are not known to NATO and the United States. Otherwise, presumably, they would attack that headquarters. But they know it exists. ...

At the same time, the Taliban has always been a kind of federation of local allied Islamist leaders who choose the Taliban flag to fly under because it is a source of finance and a source of identity and coherence, but who operate in their provinces as regional leaders. And there are many examples today of either Taliban leaders who are essentially quasi-independent, like Jalaluddin Haqqani and his family, along the border near Khost, or Taliban-allied Islamists, like Gulbuddin Hekmatyar, who may not even strictly be Taliban but who find it convenient to fight alongside Mullah Omar in a national cause.

And where does Omar operate from?

Most likely still in Pakistani territory. He was clearly operating from Pakistani territory up until 2007, when the United States increased pressure on the Pakistan government to do something about the impunity that Mullah Omar seemed to enjoy. Where he is now, nobody can say for certain. But he's probably up in the Federal Administered Tribal Areas, perhaps moving back and forth between Pakistani and Afghan territory from time to time, perhaps still enjoying some sanctuary in the neighborhood of Quetta from time to time.

What power does [Afghan President Hamid] Karzai have?

President Karzai runs a national government that receives a great deal of international support. ... He's a sort of doorkeeper between the international community and the territory of Afghanistan.

But as the head of the Afghan state, his power is waning. The territory that he effectively controls as a president has been shrinking over the last few years. And his ability to govern in a meaningful way from Kabul seems also to be waning, partly as a result of his own ineffectiveness and partly as a result of the Taliban's increasing ability to control territory as insurgents.

President Karzai is not the popular figure who has galvanized Afghans' national hopes for revival and for reconstruction as he was when he was first elected president. However, he is still regarded as a credible national leader, albeit a weak one. ...

One of the problems that the United States will confront in Afghanistan as it attempts to change the narrative of this war is a national election, scheduled for late 2009, that is going to be challenging on a number of counts: first, just the security problem of pulling it off and preventing voters from being killed on their way to the polls; and secondly, the question of what kind of Afghanistan is going to emerge from this election, what kind of leadership is going to emerge.

Is President Karzai going to be re-elected? If so, how is he going to improve on the performance that has been so heavily criticized in the last couple of years? And if there is new leadership, as is sometimes discussed, will it have the national reach that Karzai's presidency, at its best, has enjoyed?

What do you make of the allegations that he and his family are corrupt?

There are credible allegations from the United States government, including from the recently resigned head of counternarcotics at the State Department, who has said publicly that the evidence of corruption reaches into the Afghanistan government at the highest levels.

There's a perception that President Karzai is afraid to act against those, particularly in the poppy-growing areas of the south and the east, who are raking off profits from the drug economy, because his re-election is perceived to be at stake. These are the people who control the provinces where President Karzai has historically been strongest politically, and so the perception is that he is, at a minimum, passively tolerating this corruption, because those carrying it out seem to be crucial to his re-election. At worst, he or his closest allies are engaged in this for entirely selfish reasons.

Can you describe Afghanistan in terms of the fronts where the war is being fought and where it's becoming more dicey?

The field of combat in Afghanistan has been expanding steadily over the last couple of years. When the insurgency first began to revive, it was concentrated primarily in the deep south, around Helmand and Kandahar provinces, which is the historical heartland of the Taliban and also one of the centers of the opium economy.

Gradually the war has expanded along the Pakistan border, almost in its entirety, from the southern areas around Ghazni and Paktika provinces in Afghanistan all the way up to the mountainous northeastern territory of Kunar.

The most important recent developments are the encroachments that the Taliban are making around the capital of Kabul, in Wardak and Logar provinces. They are gradually squeezing the territory that the federal government can operate in, in the environment of Kabul. That means that national transportation is pressured. It means that the political reach of the government is constrained.

There are still large sections of the north and northwest that are inhabited by non-Pashtun allies of President Karzai, where the war is essentially not present except in occasional suicide bombings or assassination attempts. Those territories promise to be stable for an indefinite time, because the Taliban simply do not have the ethnic identity and language identity required to operate effectively in those areas. But that is the only part of Afghanistan today that is untouched by the war.

So essentially, we're looking at a Pashtun front against NATO and the Americans.

The Taliban are a product of Pashtun identity, Pashtun grievances, Pashtun aspirations to be sure. They are not, on the other hand, an ethnic insurgency. They are a religious insurgency located in a Pashtun setting. ... Their ideology is primarily that of an international Islamist religious movement, and so they are able to attract, for instance, non-Pashtun volunteers from Pakistan in substantial numbers. They are able to attract non-Pashtun volunteers from the Arab world in increasing numbers. And all of that speaks to the ideological character of the movement. ...

And their brotherhood with the Pashtuns of the tribal areas in Pakistan is very significant.

There are more Pashtuns living in Pakistan than in Afghanistan. And many of the tribal and other sources of historical sort of solidarity that the Taliban draws upon spill across that border without reference to the distinction between Pakistan and Afghanistan.

So going back to this idea that they're following a script, that they're patient, ... American troops have been there fighting. Are they just being passed by, by the gathering strength of the Taliban?

Not in every case, but in strategic terms, the success of American soldiers on the battlefield, in battles, at a tactical level is not adequate to overcome the gathering strategic strength of the Taliban. That's the essential problem. We're winning the battles, but we're not yet winning the war.

However, ... Afghan popular opinion is still with the international community on the whole. Afghans do not want to return to Taliban rule. They are waiting for the international community to finally deliver on its promises and to create space for the revival of an independent Afghan nation of the sort that brought so many back from exile after the Taliban fell. So the United States does have potential in Afghanistan to reconnect with Afghans' national wish to be rid of the Taliban.

I don't think that it is within the capacity of the international community to eradicate the Taliban as a factor in Afghan national life, but to roll back the territory that they control, to return to an era when the Taliban is more nuisance than strategic threat, that is an achievable goal, in my judgment, if the time horizon is five to 10 years and not one to two years.

Let's talk about the tribal areas. What are we looking at there?

A new group has claimed control of the tribal areas in Pakistan: the Pakistani Taliban. They announced themselves to the world in the autumn of 2007, and a year on, they're trying to build themselves as a force in national Pakistani life. They're more wish than fact as a national movement, but they are gathering strength, and they control territory.

The Pakistani Taliban effectively control South Waziristan, one of the tribal agencies; North Waziristan, a second large one; Bajaur; and parts of other agencies as well. They have a leadership shura that is constructed in the image of Mullah Omar's Afghan Taliban, and they have stitched together a coalition of local leaders who are willing to fly under the Taliban flag in Pakistan.

They have gradually come down from the hills, as it were, into settled territory in Pakistan and now menace the city of Peshawar in the North-West Frontier Province and a number of significant towns around Peshawar. They control territory, not just in a military sense, [but] in a sense that they control the roads, tax transport, intimidate outsiders.

They also are administering territory. They mete out justice in mosques. They substitute themselves for the Pakistan state, for its court system. They decide who lives and dies, as judges. But they also settle more ordinary disputes about boundary walls and land and grazing rights.

They are displaying the characteristics of a classic insurgency, and I think this is a new feature in Pakistani national life. Pakistan has endured much political violence. It has endured a number of separatist insurgencies in the past. It has never experienced a religious insurgency of the Taliban's character so advanced that it is actually substituting itself for the state in important parts of Pakistan's west.

There were Pakistani Taliban who were running cross-border raids in 2004. There were even peace deals early on. You say that this is a new phenomenon that begins in 2007. What's the difference? ...

In 2004, you had the Afghan Taliban in exile on Pakistani territory starting to set down roots in the Pakistani context. The new development in 2007 was the formation of an entirely new entity with Pakistani ambitions. ... And that organization has been waging war against the Pakistani state -- attacking the Pakistani army, attacking the Pakistani government in places, and also negotiating with that government for truces, for peace deals and the like.

So there has been a development gradually leading to the formation of this Pakistani organization with ambitions to control territory and indeed, ultimately, to overthrow the government of Pakistan. That's a wish that is well beyond their capacity. However, their ability to control a substantial chunk of the west of Pakistan looks more realistic, just based on their recent activity.

So how do we connect the events that we saw in the Shikai Valley in 2004, for instance, with the events of the peace deal in '06 and coming forward?

Since 2004, the Taliban in Pakistan ... have been contesting each other to control territory, and then, as they won that territory, gradually building a larger organization that could sustain conflict with the Pakistani government.

So characters like Nek Mohammed and Baitullah Mehsud started out essentially in control of their valley or their region or a section of their tribal agency. Today, Baitullah Mehsud is operating in a comfortable alliance with other Taliban leaders in a broader cause, in a bigger cause, than his valley and his region.

Now, this is still more aspiration than fact, but it is building. And that's what the Pakistani Taliban is all about. ...

Who is Baitullah Mehsud?

Baitullah Mehsud is a local tribal radical of no particular social standing who has forced his way into the leadership of Pakistani Taliban by violence, primarily by vanquishing his enemies. He's essentially an Islamist smuggler who is rooted in South Waziristan and has gradually built up a militia that became the most powerful of its type in South Waziristan.

The Mehsuds are a very powerful tribe in the territory that Baitullah Mehsud now controls. He is not a tribal leader. He has made himself in this sort of pattern of warlordism that is all too common in Pakistan and Afghanistan these days. He's made himself a leader at the point of a gun. ... And he controls smuggling and transport brackets that provide steady revenue apart from his war-fighting activities.

So it's gone from a bunch of mini-warlords, if you will, to a guy who's able to get other little fiefdoms under his control in a kind of federation of Pakistani Taliban.

Yeah. The mechanism for leadership is a shura, a committee, essentially, in which local tribal traditions of egalitarian debate and discourse are honored, but with Baitullah Mehsud aspiring to the supreme position of leadership and building alliances with other local leaders to territories that are outside of his tribe's area.

So he has built this federation from South Waziristan, down near Quetta in the southern areas of the tribal areas, all the way up gradually now beyond North Waziristan, into Bajaur, into Swat and all the way up toward the Himalayan mountains. So you now have an organization, at least in name and to some extent in fact, that is operating across the entire Afghan-Pakistan border, on the Pakistani side for the purposes of overthrowing the Pakistani government. That is its stated ambition.

As well as continuing to attack NATO forces.

Yes. And in fact, the Pakistani Taliban have entered into agreements with the Pakistan army that seem calculated to permit the Pakistani Taliban to support the war in Afghanistan while taking a break, in effect, from certain Pakistani targets.

There has been a narrative of negotiations and violence between the Pakistan army and the Pakistani Taliban that has gone zigzagging back and forth since 2004.

The Pakistan army, left to its own devices, would seem to prefer a targeting agreement with the Taliban, whereby the Taliban concentrate on NATO and American forces in Afghanistan and leave the Pakistani state alone.

The Taliban, on the other hand, are not always ready to accept the hegemony of the Pakistan army. And the United States is putting pressure on Pakistan to actually accept that it has to control the Taliban and can't simply permit them to run across the border and hit the United States and NATO forces. That's not an acceptable partnership for the United States.

What are the links between Baitullah Mehsud's Taliban and the Pakistani intelligence services?

In the case of Baitullah Mehsud, his relationship with the Pakistani intelligence services is a bit foggy and probably not as well developed as some of the other Taliban leaders'. But Mehsud clearly has contacts with Pakistani intelligence. He has negotiated with them in the past. He has held press conferences unmolested in South Waziristan where journalists have been able to come and go to hear him speak. So the Pakistani state clearly has a complicated relationship with him that includes elements of accommodation.

But Mehsud is a relatively young figure in the war. He's an upstart, a radical, a smuggler. There are other allies in the Pakistani Taliban who have even deeper ties to the Pakistani intelligence services. Jalaluddin Haqqani comes to mind, his son Siraj. They have been clients of the Pakistani intelligence service collaborators since the anti-Soviet war of the 1980s. There's recent evidence cited in the open sources that this collaboration has continued to the point where the Pakistani intelligence service apparently facilitated, supported the attack by Haqqani's group against the Indian Embassy in Kabul, the capital.

What is the relationship, then, between Haqqani and Mehsud?

Haqqani and Mehsud are notionally partners in this project of the Pakistani Taliban. And they control complementary territory, Haqqani in North Waziristan and across the border in Khost, Afghanistan, and Mehsud in South Waziristan. So their local clan identity, their networks, their Islamist allies, their connections to international Al Qaeda finance and to the Pakistani state are complementary. They're not competing for the same territory. They're neighbors.

And what's the relationship between the Pakistani Taliban and Al Qaeda?

Elements of the Pakistani Taliban, including Haqqani and Mehsud, have historical ties to Arab Al Qaeda leaders. They have provided refuge. They have provided territory. They have collaborated with them in fund raising in the past. This goes all the way back to the formation of Al Qaeda in 1989.

Haqqani in particular has historical ties to Osama bin Laden and to the early leadership team of Al Qaeda. Mehsud more recently, after 9/11, was collaborating with Uzbek and Arab Al Qaeda, after they fled to Pakistan to flee the American invasion of Afghanistan. ...

So what's brewing here is a showdown eventually between NATO, the Americans and this Taliban state, if you will.

The problem is that ... Al Qaeda, the organization, has announced ambitions to [attack] American citizens and interests in the United States, in Europe, wherever they can find them --

And has done so.

And has done this repeatedly -- that organization is located primarily on the Pakistan side of the border, where NATO and American forces cannot operate openly and where the Pakistani army is ambivalent at best about fighting the Americans' war. ...

Now, the United States has made it plain, both in its declaratory policy, in its actions and its messages to the Pakistani government, that this statement cannot long endure. It is not acceptable.

On the other hand, there is no easy fix. The Pakistani army, even if it had the will, lacks the capacity to defeat the Taliban along the border anytime soon. And it lacks the will to do so by military means alone.

And the Pakistani army is correct when it argues that there is no military solution to the problem of the tribal areas. There is no purely military solution. The only way Al Qaeda can be marginalized, the only way that the Taliban can be subdued in the long run is to change the conditions in which they thrive by incorporating the tribal areas into a modernizing, successful Pakistani state. But like the revival of Afghanistan, that is a daunting and, by any realistic measure, long-term project.

What progress has been made since 9/11 to bring the tribal areas into Pakistani society?

Very little. By and large, since 9/11, the tribal areas have gone in the other direction. They have essentially separated themselves from Pakistan through a Taliban takeover. ...

What's required now -- and what the Pakistani government, by and large, has in mind -- is a different project of ... economic development and political change that would gradually change the constitutional status and the economic circumstances of these tribal agencies. There is a national consensus, more or less, in Pakistan -- including among residents of the tribal agencies -- to change their relationship with the government of Pakistan, to eliminate this special status that's a hangover from the British colonial period.

But that's a 10-to-15-year project. And right now, the Taliban are in control of this territory, and they're not about to hand it over to Pakistani political parties or to some new system of constitutional commissioners who are there to reincorporate the tribal areas into Pakistani national life.

It's been a tough year in Pakistan. We've seen a lot of changes.

It's [been] a year of rising instability and political change of a sort that the country hasn't been through, really, for 15 or 20 years. ... 2007 opened with President [Pervez] Musharraf firmly in charge. The year ended with him out of uniform, gone as head of the army and so weakened politically that he was no longer an effective president of the country.

His departure has left Pakistan in a state of constitutional uncertainty, political instability, and also confronting rising economic problems -- inflation and other sources of chronic instability. This essentially was a result of Musharraf's overreach. Musharraf began to believe his own clippings, and he proved that while he had a role to play in the immediate aftermath of 9/11, that he was not a talented enough politician to sustain his own ambitions. ... He failed to broaden his government beyond the narrow group of military officers around himself. And he failed, even, to bring the army along with his own project.

So by the end of 2007, he was exposed as isolated from his own generals, not to mention the Pakistani people and its longstanding civilian political parties. ...

Wasn't he also seen as a puppet of the United States?

This was part of his problem, though in the end probably not the most significant problem that he faced. He failed in Pakistan on Pakistani terms, but he also failed because he was unable to navigate his very close alliance with the United States at a time when American policy in Pakistan was increasingly unpopular. ...

He had allies in the United States. ... Our calculation was to stand by him during this period.

The United States was concentrating on Iraq after 2003 almost to the exclusion of other complicated foreign policy problems. ...

We essentially outsourced our Pakistani policy to Musharraf after 2003. We asked him to try to create the conditions for Afghanistan to succeed. We asked him to try to keep Pakistan calm and stable and to keep it on a modernizing path, to marginalize the Islamists. And this was a mistake, because first of all, it was implausible for any one man to rule Pakistan in that way; but secondly, it fed Musharraf's own ego, his sense that he was the indispensable man in Pakistan.

Was there an alternate policy?

Yes. ... Pakistan is a troubled democracy, but it is a democracy with a 50-year constitutional history. It is an open society with a free media, with vibrant political parties and with diverse political leadership. This is not a country without a history.

The interests of the United States lie in promoting a stable, modernizing, democratic, constitutional Pakistan. And President Musharraf was not, in the end, an instrument of a stable, democratic, modernizing, constitutional Pakistan. He was a usurper. And he sought to be an instrument of constructive progress and constitutional reform, but in the end, he proved not to be a politician or a national leader who was selfless enough to see that transition through.

Some Pakistanis say to me: "Musharraf was a good guy. He was not corrupt. And the alternatives -- [Pakistani President Asif Ali] Zardari, [former Prime Minister Nawaz] Sharif, Bhutto -- are corrupt politicians."

Democracy in Pakistan is the worst of all possible systems, except for all the others. ... Pakistan's civilian politicians have a lousy track record. They are greedy. They have amply documented records of corruption while in the office and while out of office.

The army is the most competent, less corrupt institution in Pakistani national life. It is a source of stability. It is now and must be in the future one element of the construction of a stable, modernizing, democratic Pakistan. But it can't do it all. And that's what Musharraf learned. He stretched the army beyond its capacity and endangered the army's own position as a source of stability in Pakistani life.

But you just made the argument for why the United States supported Musharraf.

But we made this mistake over and over again in our policy in Pakistan. We have overestimated the capacity of the army to be the only instrument, the dominant instrument of politics in Pakistani national life.

Pakistanis won't stand for that. They have demonstrated that over and over again. They would prefer their flawed democratically elected politicians to the dictate of generals, like people in many other countries. And the United States has repeatedly erred by allying itself with well-intentioned generals who nonetheless stood in opposition to the national aspirations of ordinary Pakistanis. ...

What do you make of the allegation that it was Baitullah Mehsud who assassinated Benazir Bhutto?

The evidence is that the suicide bombers who killed Bhutto came out of South Waziristan and probably were dispatched by Baitullah Mehsud. Whether he was acting alone or as part of some other collaboration with elements of the Pakistani state I think is a mystery that may never be resolved. ...

Why does he deny that he was involved?

He doesn't want to end up in Guantanamo Bay. He's trying not to make himself a target of American or Pakistani retaliation. He's moved on from the death of Benazir Bhutto. He's trying now to negotiate his own ambitions in South Waziristan, and that involves accommodating at times the Pakistani state. He doesn't want to be public enemy number one.

But he's declared himself as an enemy of the state already.

He's an enemy of the state who nonetheless negotiates with elements of the state, who makes money from corrupt collaboration with elements of the state, and who holds press conferences with Pakistani media under the watchful eye of the Pakistani state. So he is a complicated public enemy. ...

The Pakistanis are continuing to come up with these peace overtures with the Taliban.

You've got a new government that was elected on the promise that they would fight a different kind of war than Musharraf had fought, that they wouldn't just do whatever the Americans told them to do. And they're anxious to address the considerable sources of popular unrest that have nothing to do with the war against the Taliban; that have to do with the deteriorating economy, a falling stock market, high unemployment and a general sense that the country is adrift. So if the civilian government is going to deliver anything, they're going to deliver in the spheres of the economy, employment, patronage and civil administration. And what they're looking for, essentially, is some time to buy some time, to establish themselves after a long period of exile.

So they will somehow get a peace deal with the Taliban to leave them alone for a while. Is that the idea?

That was the idea. But this is, I think, proving not to be a sustainable idea. First of all, the United States government is not going to sit by and provide many hundreds of millions of dollars of aid to a Pakistani government that is explicitly cutting peace deals with militants who are killing American soldiers. That's not going to prove sustainable.

We've done it in the past.

... During the beginning of this problem, there was sort of a willful blindness about the role that the Pakistani state was playing in collaboration with the Taliban. I think that period has now yielded to a more clear-eyed recognition that this is a problem that is directly leading to battlefield deaths and instability in Afghanistan and must be addressed.

Having come to that recognition, however, it's also the case that there is no button to push. There is no meeting to have that is going to instantly reverse this accumulating structural problem along the border.

So what do we do?

In the long run, American policy in Pakistan ought to be clear. We're invested in the success of a stable, democratic, constitutional Pakistan. There is every reason to be hopeful about Pakistan's success, looking out 10 or 20 years.

India is rising. In 50 years, India may be one of the most prosperous and significant countries on the planet, and Pakistan is right next door. It has every reason to succeed as India succeeds if it is able to organize its political and constitutional affairs to benefit from this historical change that's going on in South Asia.

The United States, in its own interests, ... ought to be investing in a stable, democratic, constitutional, strong, modernizing Pakistan at the level of civil society, at the level of democratic politics, at the level of media, at the level of economics and meeting the basic needs of the many tens of millions of Pakistanis who live in poverty. That, along with security and stability in Afghanistan, has to be a part of American strategy in Pakistan.

It has been a neglected aspect of American policy. Virtually all of the financial investments that we have made in Pakistan since 9/11 have been military investments, security investments. Those investments now have to be rebalanced by a broader approach.

Having said all of that, we have to be clear-eyed that there is a short-term threat to American lives and interests in the form of the Taliban and Al Qaeda operating on Pakistani soil. ... Addressing that threat is not going to be easy at the same time that you're building this long-term strategy to support Pakistani democracy.

We have a disconnect between interests, it seems. They have an interest in having the Taliban stop attacking Pakistani targets, we have an interest in them stopping attacking us, and the two don't really meet.

It is true that the interests of the Pakistani army and of the United States are not aligned as fully as the United States would wish them to be. But they are, in the long run, more aligned than the Pakistani army has yet been willing to embrace.

The Pakistani army supports the Taliban partly for rational reasons, but also out of a fear that I think is irrational, and that fear is located in the belief that the United States is collaborating with India in Afghanistan to essentially encircle and weaken the Pakistani state, and that, in fact, the American goal is to break up Pakistan or undermine its military strength. This fear animates a great deal of the Pakistan army's collaboration with the Taliban. ... The United States has been unable to dissuade the senior commanders of the Pakistan army from this view.

But the Pakistani political class is not as infected by this paranoia. And I think there is the potential, in the long run, to bring American interests and Pakistani interests into better alignment than they have been in over the last five or six years.

... Is it not a legitimate fear that Afghanistan will be used as a way to strengthen India's pressure on Pakistan?

It's a legitimate concern, but it's vastly overinterpreted by the Pakistani army. Just because you're paranoid doesn't mean India isn't out to get you. India does have its own national interests. It does see itself on the receiving end of mischief making by the Pakistani intelligence services -- Islamist insurgencies in Kashmir, terrorist groups blowing up civilians on trains outside of Bombay. ... But by and large, the Indian government recognizes that it has no solution to its Pakistan problem other than a stable, modernizing, democratic Pakistan.

The question that India has is, what is the role of the Pakistan army in the long run? How can the Pakistan army be persuaded to stop using Islamist radical violence as an instrument of foreign policy?

That's a question that the United States also has to ask: How can the Pakistan army recognize that its use of Islamist radical violence as an instrument of foreign policy is not sustainable and in contravention of Pakistan's own national interest? To date, that has not been an argument that anyone has persuaded the Pakistan army to accept.

[What's happening with the cross-border raids into Pakistan?]

It would be impossible to defeat the Taliban without denying them sanctuary in Pakistan. That's the lesson of the anti-Soviet war. The Soviet Union could never control Afghanistan because it could never control the supply lines of the rebels, and it could never control the refugee camps and the sanctuaries to which they return when they were exhausted by war.

A similar pattern is now developing for the United States and NATO. ... The problem of Pakistani sanctuary is central to the sustainable strength of the Taliban. And you see this in the form of the Taliban's insurgency -- their ability to bring weapons across the border, ... their ability to infiltrate men under arms for short-term attacks and then withdraw before the United States can identify them. All of this is at the heart of their tactical approach at the moment.

The Taliban are in a position now, and are likely to be in a position for an indefinite time, where they cannot mass on the battlefield. ... They have to fight an irregular guerrilla-style war where they minimize their vulnerability to rapid retaliation by air or by artillery. The best way to do that is to attack and retreat, and to retreat to a place where they cannot be easily attacked from the air. That's the role of the tribal territories. It is a sanctuary in a quite literal and physical sense. ...