Briefing Book

Recipe for Failure

Why Copenhagen will be a bust, and other prophecies from the foreign-policy world's leading predictioneer.

Want to know what's going to happen with climate change? Is the world going to come together this December at the Copenhagen summit, or at some future date, and regulate away enough of the greenhouse gases that are heating up the planet to warm Al Gore's heart? I'm no climate scientist, but I've done my own calculations, and I can tell you the answer: probably not.

Despite the hoopla, the U.N. climate change conference in Copenhagen is destined to fail. Here's what will happen instead: Over the next several decades, world leaders will embrace tougher emissions standards than those proposed -- and mostly ignored -- in the 1997 Kyoto Protocol. But real support for tougher regulations will fall. By midcentury, the mandatory emissions standards in place will be well below those set at Kyoto, a far cry from the targets for carbon dioxide and other greenhouse gases set to be discussed by world leaders in Copenhagen. And by the time 2100 rolls around, the political will for tougher regulations will have dried up almost completely. The reasons are many, but come down to this: Today's emerging powerhouses like Brazil, India, and China simply won't stand for serious curbs on their emissions, and the pro-regulation crowd in the United States and Europe won't be strong enough to force their hands.

How do I know all this? Because in 1979, I learned that I could predict the future.

Don't get me wrong -- I'm no soothsayer and I have no patience for crystal-ball gazers, astrologers, or even most pundits. In my world, science, not mumbo jumbo, is the way to predict people's choices and their consequences for altering the future. I use game theory to do just that for the U.S. government, big corporations, and sometimes ordinary folks, too. In fact, I have made hundreds, even thousands, of predictions -- a great many of them in print, ready to be scrutinized by any naysayer. For instance, I can tell you right now that bribing Kim Jong Il to mothball, but not eliminate, his nuclear program is the best way to handle North Korea, that the land-for-peace formula in the Middle East won't succeed, and that it will take approximately $1.5 billion annually in U.S. aid to Pakistan to keep that country's government fighting the Taliban and al Qaeda.

There is nothing uncanny about my ability to predict. Anyone can learn to use scientific reasoning to do what I do, though I've been refining the model I use ever since I accidentally got into the prediction business back in the last days of disco.

The opportunity initially fell into my lap when a U.S. State Department official called to ask me who was likely to be India's next prime minister. At the time I was a professor of political science at the University of Rochester -- where the application of game theory to political questions originated -- and I had written my Ph.D. thesis at the University of Michigan about winning and losing strategies among India's opposition parties. So the State Department official was asking me to use my "expert" knowledge to speculate about the next Indian government.

It happened that I had just designed a mathematical model for a book I was writing about war, as well as a little computer program to make the necessary calculations. The program provided a way to simulate decision-making under stressful circumstances like those that sometimes lead to war. It calculated the probability that actors would get what they wanted if they chose one course of action (say, negotiations) or another (like war), weighting the probabilities by an estimate of how much the decision-makers valued winning, losing, or intermediate compromise outcomes. Of course, it also recognized that they had to work out how others might respond to the choices they made.

The phone call about India got me thinking that maybe war and peace decisions really aren't that different from everyday political confrontations. Sure, the stakes are higher -- people get killed in wars -- but then any politician seeking office sees the personal political stakes as pretty darn high. Intrigued, I grabbed a yellow pad and listed everyone I thought would try to influence the selection of India's next government. For each of those people (political party leaders, members of India's parliament, and some members of critical state governments), I also estimated how much clout they had, what their preference was between the various plausible candidates for prime minister, and how much they cared about trying to shape that choice. With just one page of my yellow pad filled with numbers, I had all the information the computer needed to predict what would happen, so I plugged it in and awaited the results.

My "expertise" had led me to believe that longtime parliamentary leader Jagjivan Ram would be India's next prime minister. He was a popular and prominent politician who was better liked than his main rivals for the prime minister's job. I was confident that he was truly unbeatable. He had paid his political dues and it seemed like his time had come. Many other India watchers thought the same thing. Imagine my surprise then when my computer program, written by me and fed only with my data, predicted an entirely different result. It forecast that Charan Singh would become prime minister, that he would include someone named Y. B. Chavan in his cabinet, and that they would gain support-albeit briefly-from Indira Gandhi, then the recently ousted prime minister. The model also predicted that the new Indian government would be incapable of governing and so would soon fall.

I found myself forced to choose between my personal opinion -- that Ram would win -- and the logic and data behind my model. In the end, I chose science over punditry. When I relayed my findings to the State Department official, he was taken aback. He noted that no one else was suggesting this result and that it seemed strange at best. When I told him I'd used a computer program based on a model of decision-making that I was designing, he just laughed and urged me not to repeat that to anyone.

A few weeks later, Charan Singh became the prime minister with Y. B. Chavan as his deputy prime minister and support from Indira Gandhi. And a few months after that, Singh's government unraveled, Gandhi withdrew her backing, and a new election was called, just as the computer model had forecast. This got me pretty excited. But had I just gotten lucky, or was I onto something?

I set out to push my model by testing it against wide-ranging questions about politics and economics. I applied it to prospective leadership changes in the Soviet Union, questions of economic reform in Mexico and Brazil, and budgetary decisions in Italy. The model worked so well that it eventually led to a grant from the Defense Advanced Research Projects Agency, a research arm of the U.S. Defense Department. darpa gave me 17 issues to examine, and as it happened, the model -- by then somewhat more sophisticated -- got all 17 right. According to a declassified cia assessment, the predictions for which I've been responsible over the years have a 90 percent accuracy rate.

This is not a reflection of any great wisdom or insight on my part-- I have little enough of both. What I do have is the lesson I learned back in 1979: Politics is predictable. All that is needed is a tool, like my model, that takes basic information, evaluates it by assuming people do what they think is best for them, and produces reliable assessments of what they will do and why they will do it.

However reliable my model has proven, though, it still represents a radical departure from the way most "experts" shape decisions about international affairs. Most diplomats, for example, remain convinced that a country's name is an important variable that helps explain behavior. That's why the State Department continues to be organized around country desks, just as the intelligence community is organized around geographic regions. Leaders of multinational corporations take much the same view. When they have a problem in Kazakhstan, they call their guys in Kazakhstan to find out what to do. That seems eminently reasonable. Yet it is terribly inadequate for solving most problems.

Certainly knowing about places and how different they might be is important, but not as important as knowing about people and how similar they are, wherever they are. I have not arrived at this view lightly nor, I hope, in ignorance. After all, the training that led to my Ph.D. molded me into a South Asia specialist. I even studied Urdu for five years and did field research in India, so I certainly respect and value area expertise. But area studies alone are a poor substitute for the marriage of knowledge about places and the deep understanding of applied game theorists about how people decide. Surely we would think it ridiculous if chemists believed that oxygen and hydrogen combine differently in China than they do in the United States, but for some reason we think it entirely sensible to believe that people make choices based on different principles in Timbuktu than in Tipperary.

This is a controversial stance in many of the circles in which I travel, and many in those circles see my views as foolish at best and dangerous at worst. In 1984, when I predicted that Ali Khamenei and Ali Akbar Hashemi Rafsanjani would succeed Ayatollah Ruhollah Khomeini as the leaders of Iran, regional experts denounced me as a charlatan (five years later, when I was proven correct, the most prominent such critic publicly apologized). Foreign Policy's own Stephen Walt once dismissed rational choice -- the theory of human behavior that underpins much of my work -- as a "cult of irrelevance." Still, I do not shy away from the risk of publishing predictions -- and by and large, those who disagree with me do not do the same.

So why not tackle perhaps the most controversial and consequential question of our time? What does game theory tells us about how -- or if -- we humans will solve global warming? The timing is perfect: After years of debate, there now seems to be broad agreement within the scientific community that Earth's temperature is on the rise. And the political will to do something about it is on the rise, too -- or so it seems.

I'll explain why "seems" is the operative word here, but first, some background. In December 1997, 175 countries, not including the United States, signed the Kyoto Protocol. Kyoto produced a large market in which polluters and nonpolluters could buy and sell "pollution rights." This market has helped rationalize decisions at the level of individual firms, but it has so far failed to result in the magnitude of reductions envisioned by the Kyoto Protocol. Enforcing the 1997 agreement has been virtually impossible.

One consequence of the difficulties encountered since 1997 was a meeting in Bali, Indonesia, in December 2007. The Bali meeting had more modest goals than Kyoto. It was an interim step on the way to Copenhagen, where it is hoped there will be a new international agreement. After considerable resistance, the U.S. representative at Bali agreed to significant concessions at the last minute. This made it possible to set out the "Bali road map" for future climate control. Now the question is: Will these efforts work?

Let's investigate. To address the prospects for controlling greenhouse gas emissions, especially carbon dioxide, we'll start with data that reflect the views of the big players on global warming. These are the governments and interest groups with the most at stake. In all likelihood, any agreement that can be reached will be settled primarily among these few stakeholders. They include the European Union, the United States (where opinion is divided between those who favor regulating carbon dioxide and other greenhouse gas emissions, and those opposed), China, and India. It also includes other relatively large economies such as Brazil, Japan, Russia, Canada, and Australia. For good measure, I have also represented environmental nongovernmental organizations, because they had a significant presence at Bali, and multinational corporations. For each stakeholder I have estimated potential influence in negotiations over an agreement to replace the Kyoto Protocol, position on mandatory emission controls, salience (eagerness to weigh in on emission controls), and flexibility-the extent to which the stakeholder is committed to finding an agreement (even if not the one it favors most) or will stick to its guns under political pressure (holding out for the policy it believes in).

I have rated the stakeholders' positions on a scale from 0 to 100. A position of 50 is equivalent to continuing the greenhouse gas targets that came out of the 1997 Kyoto Protocol. These standards called for rollbacks based on 1990 emission levels. Higher values on the scale reflect tougher standards. For example, 60 is a 10 percent toughening of standards relative to the 1990 benchmark, while 100 is a 50 percent increase in mandatory greenhouse emissions reductions compared with 1990. Likewise, values below 50 reflect a weakening of the terms in the Kyoto agreement. If your position is below 50, as is the case with China and India, that means you're not on board even with the limited emissions targets set in 1997.

Because so much can happen over the next 125 years of my simulation, I have spiced up the model with random shocks to salience and to each stakeholder's interest in building consensus or sticking to its guns. By randomly changing 30 percent of the salience values and 30 percent of the flexibility values in each bargaining round, we can look at a range of predicted futures to see whether the global warming simulations reveal strong trends. That will help us sort out how confident we can be about the toughness or weakness of future regulations of greenhouse gas emissions.

In terms of the big picture, the heavy solid black line in the graph shows the most likely emission standard predicted by the game. The two heavy dotted lines depict the range of regulatory values that we can be 95 percent confident includes the true future regulatory environment according to the simulations. That range of values is pretty narrow, encompassing barely five points up or down through about 2050. After that, as we should expect, there is more uncertainty, but even as far into the future as 2130, the range is only about 10 points up or down, so these are probably pretty reliable forecasts.The most likely value -- the heavy solid line-reflects our best estimate of what the big players might broadly agree to if the global warming debate continues without any significant discoveries in its favor or against it. It tells us two stories. First, the rhetoric of the next 20 or 30 years endorses tougher standards than the ones put forward in Kyoto in 1997. We know this because the predicted value through 2025 is above 50. That's the green part of the story. Second, support for tougher regulations falls almost relentlessly as the world closes in on 2050, a crucial date in the global warming debate. When we get to 2050, the mandatory standard being acted on is well below that set at Kyoto. By about 2070 it is down to 30, representing a significant weakening in standards. By 2100 it is closing in on 20 to 25. There's no regulatory green light left in the story by its end.

The figure shows us that there are some considerably more optimistic scenarios and also some considerably more pessimistic views that fall outside the 95 percent confidence interval. The most optimistic scenario predicts no rollback in emission controls. It never dips below 50. In fact, most of the time in this scenario the predicted level of greenhouse gas reduction hovers around 60, implying a 10 percent or so tougher standard than was agreed to in Kyoto. Only about 10 percent of the scenarios, however, look optimistic enough to anticipate even holding the line at the standard set in the Kyoto Protocol.

In contrast, there are dozens of scenarios in which the standard falls close to 0, indicating abandonment of the effort to regulate greenhouse gases. Typically in these scenarios, some mix of Brazil's, India's, and China's salience rises while the salience of pro-control factions in the United States (mostly liberal Democrats) and the European Union drops well below their opening stance. They just seem to lose interest in greenhouse gas regulations. That decline raises its ugly head especially during global economic slowdowns, so global economic patterns are critical to watch, as they can guide our choice of the scenarios that we should pay the most attention to. Without commitment to change by the European Union and the United States, it becomes much easier for the key developing economies to prevail with the support and encouragement of the anti-control American faction (mostly conservative Republicans).

All of this may be leaving you rather depressed, but perhaps it shouldn't. As for me, I am most optimistic for the future, despite- - yup, despite -- agreements like the ones struck in Bali and Kyoto, or the one to be struck in Copenhagen. These will be forgotten in the twinkling of an eye. They will hardly make a dent in global warming; they could even cause hurt by delaying serious changes. Road maps like the one set out at Bali make us feel good about ourselves because we did something. The trouble is, deals like Bali and Kyoto include just about every country in the world. To get everyone to agree to something potentially costly, the something they actually agree to must be neither very demanding nor very costly. If it is, many will refuse to join because for them the costs are greater than the benefits, or else they will join while free-riding on the costs paid by the few who are willing to bear them.

To get people to sign a universal agreement and not cheat, the deal must not ask them to change their behavior much from whatever they are already doing. It is a race to the bottom, to the lowest common denominator. More demanding agreements weed out prospective members or encourage lies. Kyoto's demands weeded out the United States, ensuring that it could not succeed. Maybe that is what those who signed on -- or at least some of them -- were hoping for. They can look good and then not deliver, because after all it wouldn't be fair for them to cut back when the biggest polluter, the United States, does not. Sacrificing self-interest for the greater good just doesn't happen very often. Governments don't throw themselves on hand grenades.

There is a natural division between the rich countries whose prosperity does not depend so much on toasting our planet and the poor countries that really have no affordable alternative (yet) to fossil fuels and carbon emissions. They have an incentive to do whatever it takes to improve the quality of life of the people they govern. The rich have an incentive to encourage the fast-growing poor to be greener, but the fast-growing poor have little incentive to listen as long as they are still poor. As the Indian government is fond of noting, sure, India is growing rapidly in income and in carbon dioxide emissions, but it is still a pale shadow of what rich countries like the United States have emitted over the centuries when going from poor to rich.

But when the fast-growing poor surpass the rich, the tables will turn. China, India, Brazil, and Mexico will then cry out for environmental change because that will protect their future advantaged position, while the relatively poor of one or two or three hundred years from now will resist policies that hinder their efforts to climb to the top. The rich will even fight wars to keep the rising poor from getting so rich that they threaten the old political order. (The rising poor will win those wars, by the way.)

So how might we solve global warming and make the world in 500 years look attractive to our future selves? My short answer: New technologies will solve the problem for us. There is an equilibrium at which enough global warming -- a very modest amount more than we may already have, probably enough to be here in 50 to 100 years -- will create enough additional sunshine in cold places, enough additional rain in dry places, enough additional wind in still places, and, most importantly, enough additional incentives for humankind that solar panels, hydroelectricity, windmills, and as yet undiscovered technologies will be good and cheap enough to replace fossil fuels. We have already warmed enough for there to be all kinds of interesting research going on, but today such pursuits take more sacrifice than most people seem willing to make. Tomorrow that might not be true, and at that point, I doubt it'll be too late. And, looking out 500 years, we'll probably have figured out how to beam ourselves to distant planets where we can start all over, warming our solar system, our galaxy, and beyond with abandon.

Remember, we're looking out for numero uno.

Briefing Book

The Case for Humility in Afghanistan

A Taliban victory would have devastating consequences for U.S. interests. But to avoid disaster, America must beware the Soviet Union’s mistakes -- and learn from its own three decades of failure in South Asia.

The United States has two compelling interests at issue in the Afghan conflict. One is the ongoing, increasingly successful but incomplete effort to reduce the threat posed by al Qaeda and related jihadi groups, and to finally eliminate the al Qaeda leadership that carried out the Sept. 11 attacks. The second is the pursuit of a South and Central Asian region that is at least stable enough to ensure that Pakistan does not fail completely as a state or fall into the hands of Islamic extremists.

More than that may well be achievable. In my view, most current American commentary underestimates the potential for transformational changes in South Asia over the next decade or two, spurred by economic progress and integration. But there is no question that the immediate policy choices facing the United States in Afghanistan are very difficult. All of the courses of action now under consideration by the Obama administration and members of Congress carry with them risk and uncertainty.

To protect the security of the American people and the interests of the United States and its allies, we should persist with the difficult effort to stabilize Afghanistan and reverse the Taliban's momentum. This will probably require additional troops for a period of several years, until Afghan forces can play the leading role.

However, that depends on the answer to Gen. Colin Powell's reported question, "What will more troops do?" As Gen. Stanley McChrystal wrote in his recent assessment, "Focusing on force or resource requirements misses the point entirely." Instead -- after years of neglect of U.S. policy and resources in Afghanistan and after a succession of failed strategies both in Afghanistan and Pakistan -- the United States, as McChrystal put it, has an "urgent need for a significant change to our strategy and the way that we think and operate." While I cannot endorse or oppose McChyrstal's specific prescriptions for the next phase of U.S. engagement in Afghanistan because I do not know what they are, I do endorse the starting point of his analysis, as well as his general emphases on partnering with Afghan forces and focusing on the needs of the Afghan population. I believe those emphases are necessary but insufficient.

Whether President Obama's policy involves no new troops, a relatively small number of additional forces focused on training, or a much larger deployment, we can be certain of one thing: American soldiers will continue to put their lives on the line in Afghanistan and the U.S. Treasury will continue to be drained in pursuit of U.S. goals there. We know this because President Obama has publicly ruled out withdrawal from Afghanistan as an option. Instead, within the administration and prospectively in Congress, the question seems to be whether to pursue U.S. goals with the resources already invested, or to invest more in tandem with the adoption of a new strategy. It is important, then, to think through what U.S. interests in Afghanistan actually are and what means may be required to achieve them.

General McChrystal and other senior military commanders have apparently recommended substantially increased U.S. troop levels in Afghanistan in order to stabilize what remains a weak and fractious Afghan state; to protect large sections of the Afghan population from Taliban coercion; to build up Afghan security forces; and to prevent the Taliban from forcibly seizing control of the Afghan government.

A number of credible objections have been made to this project. Some argue that the stabilization of even a weak Afghan state safe from Taliban control is beyond the capacity of the U.S. and its allies. Thus, according to Rory Stewart, in recent testimony before a Senate committee, "The fundamental problem with the [Obama administration's] strategy is that it is trying to do the impossible. It is highly unlikely that the U.S. will be able either to build an effective, legitimate state or to defeat a Taliban insurgency ... Even an aim as modest as ‘stability' is highly ambitious." Stewart has extensive direct experience of Afghanistan and his view is shared by some other credible regional specialists.

It is right to be skeptical of the abstract slogans of U.S. counterinsurgency doctrine and the enthusiasms of those in the West who define success in Afghanistan through their own political-science terminology of legitimacy, rights, and development. The Soviet Union defeated itself in Afghanistan by demanding, absurdly, that the country conform to its preconceived theories of revolution and state development. As the editors of a review of the Soviet war composed by the Russian general staff put it, "Despite the Soviet Union's penetration and lengthy experience in Afghanistan, their intelligence was poor and hampered by the need to explain events within the Marxist-Leninist framework. Consequently, the Soviets never fully understood the mujahideen opposition nor why many of their policies failed to work in Afghanistan."

Similarly, the United States should be cognizant of its own potential blinders of ideology and preconceived interpretation. For example, while the development of counterinsurgency capacity and principles by the U.S. Army, as outlined in the recently ascendant field manual FM-34, is a generally positive development in U.S. Army doctrine, and those capacities clearly have a role to play in U.S. military strategy in Afghanistan, it would be self-deceiving to believe that the Afghan war can now be "won" simply by "applying the manual," as the most ardent counterinsurgency advocates sometimes seem to argue.

To succeed, counterinsurgency approaches require deep, supple, and adaptive understanding of local conditions. And yet, as General McChrystal pointed out in his assessment, since 2001, international forces operating in Afghanistan have "not sufficiently studied Afghanistan's peoples, whose needs, identities and grievances vary from province to province and from valley to valley." To succeed, the United States must "redouble efforts to understand the social and political dynamics of...all regions of the country and take action that meets the needs of the people, and insist that [Afghan government] officials do the same."

This will be difficult at best, but it is not impossible. The international effort to stabilize Afghanistan and protect it from coercive revolution by the Taliban still enjoys broad support from a pragmatic and resilient Afghan population. Nor does the project of an adequately intact, if weak and decentralized, Afghan state, require the imposition of Western imagination. Between the late 18th century and World War I, Afghanistan was a troubled but coherent and often peaceful independent state. Although very poor, after the 1920s it enjoyed a long period of continuous peace with its neighbors, secured by a multi-ethnic Afghan National Army and unified by a national culture. That state and that culture were badly damaged, almost destroyed, by the wars ignited by the Soviet invasion of 1979 -- wars to which we in the United States contributed destructively. But this vision and memory of Afghan statehood and national identity has hardly disappeared. After 2001, Afghans returned to their country from refugee camps and far flung exile to reclaim their state -- not to invent a brand new Western-designed one, as our overpriced consultants sometimes advised, but to reclaim their own decentralized but nonetheless unified and even modernizing country.

Despite the manifold errors of U.S. and international policy since the Taliban's overthrow in 2001, a strong plurality of Afghans still want to pursue that work. And they want the international community to stay and to correct its errors.

Then, too, the difficulties facing the United States in Afghanistan today should not be overestimated out of generalized despair or fatigue. Consider, as one benchmark, a comparison between the position of the U.S. and its allies now and that of the Soviet Union during the 1980s.

In a global and diplomatic sense, the Soviet Union failed strategically in Afghanistan from the moment it invaded the country. Nor did it enjoy much military success during its eight years of direct occupation. Neither Soviet forces nor their client Afghan communist government ever controlled the Afghan countryside. And yet, despite these failures and struggles, the Soviet Union and its successor client government, led by President Mohammad Najibullah, never lost control of the Afghan capital, major cities and provincial capitals, or the formal Afghan state. Only after the Soviet Union dissolved in late 1991 and Najibullah lost his supply lines from Moscow did his Islamist guerrilla opposition finally prevail and seize Kabul.

The territorial achievements of the Najibullah government -- no forcible takeover of the Afghan state by Islamist guerrillas, continuous control of all the country's cities and major towns -- might look attractive today to the United States as a minimum measure of success. And there is every reason to believe that the international community can still do better than that.

By comparison to the challenges facing the Soviet Union after it began to "Afghanize" its strategy around 1985 and prepare for the withdrawal of its troops, the situation facing the United States and its allies today is much more favorable. Afghan public opinion remains much more favorably disposed toward international forces and cooperation with international governments than it ever was toward the Soviet Union. The presence of international forces in Afghanistan today is recognized as legitimate and even righteous, whereas the Soviets never enjoyed such support and were unable to draw funds and credibility from international institutions. China today wants a stable Afghanistan; in the Soviet era, it armed the Islamic rebels. The Pakistani Army today is divided and uncertain in its relations with the Taliban, and beginning to turn against them; during the Soviet period, the Army was united in its effort to support Islamist rebels. And even if the number of active Taliban fighters today is on the high side of published estimates, those numbers pale in comparison to the number of Islamic guerrillas fighting the Soviet forces and their Afghan clients.

In other words, the project of an adequately stable Afghan state free from coercive Taliban rule for the indefinite future can be achieved, although there are no guarantees. The next question, however, is whether it should be pursued on the basis of U.S. interests, given the considerable costs, risks and uncertainties that are involved. Here, too, a number of credible objections must be considered.

One is the argument that a heavy U.S. military presence in Afghanistan focused on population security is not the best way to defeat al Qaeda and may even be counterproductive. Counter-terrorism is "still Washington's most pressing task," write Steven Simon and Jonathan Stevenson in the current issue of Survival, but "the question is whether counter-insurgency and state-building in Afghanistan are the best means of executing it. The mere fact that the core threat to U.S. interests now resides in Pakistan rather than Afghanistan casts considerable doubt on the proposition. ... The realistic American objective should not be to ensure Afghanistan's political integrity by neutralizing the Taliban and containing Pakistani radicalism, which is probably unachievable. Rather, its aim should be merely to ensure that al Qaeda is denied both Afghanistan and Pakistan as operating bases for transnational attacks on the United States and its allies and partners."

Apparently, like some in the Obama administration, they recommend a policy concentrated on targeted killing of al Qaeda leaders by aerial drones and other means. They acknowledge that a Taliban takeover of Afghanistan might aid al Qaeda but argue that greater risks would flow from the failure or a U.S.-led counterinsurgency strategy.

This argument misreads the dynamics within Pakistan that will shape the course of U.S. efforts to destroy al Qaeda's headquarters and networks there. Simon and Stevenson, for example, fear that the provocative aura of U.S. domination in Afghanistan would "intensify anti-Americanism in Pakistan" and by doing so ensure that the Pakistan Army would refuse to cooperate with American efforts to root out Islamic extremists previously cultivated by the Army and its intelligence wing, the Inter-Services Intelligence Directorate, or ISI. There are certainly risks along the lines they describe, but something like the opposite is more likely to be true.

The relationship between the Pakistani security services and Islamist extremist groups -- al Qaeda, the Taliban, sectarian groups, Kashmiri groups, and their many splinters -- is not static or preordained. Pakistani public opinion, while it remains hostile to the United States, has of late turned sharply and intensely against violent Islamist militant groups. The Pakistan Army, itself reeling as an institution from deep public skepticism, is proving to be responsive to this change of public opinion. Moreover, the Army, civilian political leaders, landlords, business leaders, and Pakistani civil society have entered into a period of competition and freewheeling discourse over how to think about the country's national interests and how to extricate their country from the Frankenstein-like problem of Islamic radicalism created by the Army's historical security policies. There is a growing recognition in this discourse among Pakistani elites that the country must find a new national-security doctrine that does not fuel internal revolution and impede economic and social progress. The purpose of American policy should be to create conditions within and around Pakistan for the progressive side of this argument among Pakistani elites to prevail over time.

American policy over the next five or 10 years must proceed from the understanding that the ultimate exit strategy for international forces from South Asia is Pakistan's economic success and political normalization, manifested in an Army that shares power with civilian leaders in a reasonably stable constitutional bargain, and in the increasing integration of Pakistan's economy with regional economies, including India's. Such an evolution will likely consolidate the emerging view within Pakistan's elites that the country requires a new and less self-defeating national security doctrine. As in the Philippines, Colombia, and Indonesia, the pursuit of a more balanced, less coup-ridden, more modern political-military order in Pakistan need not be complete or confused with perfection for it to gradually pinch the space in which al Qaeda, the Taliban, and related groups now operate. Moreover, in South Asia, outsiders need not construct or impose this modernizing pathway as a neo-imperial project. The hope for durable change lies first of all in the potential for normalizing relations between Pakistan and India, a negotiation between elites in those two countries that is already well under way, without Western mediation, and is much more advanced than is typically appreciated. Its success is hardly assured, but because of the transformational effect such normalization would create, the effects of American policies in the region on its prospects should be carefully assessed.

Against this backdrop, a Taliban insurgency that increasingly destabilizes both Afghanistan and the border region with Pakistan would make such regional normalization very difficult, if not impossible, in the foreseeable future. Among other things, it would reinforce the sense of siege and encirclement that has shaped the Pakistan Army's self-defeating policies of support for Islamist militias that provide, along with a nuclear deterrent, asymmetrical balance against a (perceived) hegemonic India.

Conversely, a reasonably stable Afghan state supported by the international community, increasingly defended by its own Army, and no longer under threat of coercive revolution by the Taliban could create conditions for Pakistan's government to negotiate and participate in political arrangements in Afghanistan and the Central Asian region that would address Pakistan's legitimate security needs, break the Army's dominating mindset of encirclement, and advance the country's economic interests.

American and international success in Afghanistan could also enhance the space for civilians in Pakistan who seek to persuade the Pakistani Army to accommodate their views about national security; for the United States to insist that Pakistani interests be accommodated in a pluralistic, non-revolutionary Afghanistan; and for Pakistani elites, including the Army, to have adequate confidence to take on the risks associated with a negotiated peace or normalization with India. Conversely, yielding unnecessarily to an indefinite period of violence and chaos in Afghanistan, one in which the Taliban may seek to take power in Kabul while continuing to operate across the border in Pakistan, will all but guarantee failure along all of these strategic lines.

There are narrower objections that should be registered about the "counterterrorism-only" or "counterterrorism-mainly" argument. It is probably impractical over a long period of time to wage an intelligence-derived counterterrorism campaign along the Pakistan-Afghan border if a cooperating Afghan government does not have access to the local population; if American forces are not present; and if the Pakistani state has no incentive to cooperate. This is exactly the narrative that unfolded during the 1990s and led to failure on Sept. 11 for the United States. Recent improvements in targeting al Qaeda leaders in Pakistan seem to be a function, at least in part, of changing attitudes toward cooperation by the Pakistani civilian government and security services. These changes in turn are a function of the dynamic, complex internal Pakistani discourse sketched above. It is unlikely that an American willingness to allow Taliban hegemony in Afghanistan will result in greater cooperation from Pakistani intelligence; in fact, the opposite is more likely because, as in the past, some in the Pakistani security services seek such hegemony for ideological reasons, while others will likely see a need to protect their position with Islamist militias in order to defend against India in a volatile, heavily contested regional environment.

Also, if a problem in assuring Pakistan's stability lies in the country's anti-American attitudes (which may not be as important as Americans believe), then waging a prolonged war of assassination by flying robots within Pakistan's borders and without its government's participation, as some "counterterrorism only" advocates would prefer, does not seem a prescription for success. The goal of American policy in Pakistan should be to create conditions in which this unattractive manifestation of unilateral American aerial and technological power is no longer unilateral -- and control of such operations can be shifted to a responsible Pakistani government, without the fear that prevails currently in the U.S. government that Pakistani security officers will misuse targeting intelligence to protect Islamist allies.

Another objection to the U.S. investments in Afghan stability and population protection is that al Qaeda is not in Afghanistan at all, or at least not meaningfully. A related argument is that it is pointless to take risks and make new investments to prevent Afghanistan from becoming a prospective sanctuary because al Qaeda can easily find other sanctuaries, such as in Somalia and Yemen, where no American counterinsurgency or stabilization project is realistic. Osama bin Laden's presumed current base in the Federally Administered Tribal Areas (FATA) of Pakistan, near the Afghan border, according to Council on Foreign Relations analyst Stephen Biddle, has no "intrinsic importance ... no greater than many other potential havens -- and probably smaller than many." It is also argued by some that al Qaeda is best understood as an organization, network, or movement in which physical geography such as the FATA is not a defining feature -- in this view, hotel rooms in Hamburg, Germany, or rental houses near pilot training facilities in Florida are as fundamental to al Qaeda's operational footprint as its headquarters and training camps along the Pakistan-Afghanistan frontier.

These are credible, serious arguments that accurately describe some of al Qaeda's character as a stateless, millenarian terrorist group. But they misunderstand the history of al Qaeda's birth and growth alongside specific Pashtun Islamist militias on the Afghan-Pakistan border. It is simply not true that all potential al Qaeda sanctuaries are of the same importance, now or potentially. Bin Laden and his No. 2, Ayman Al-Zawahiri, have a 30-year, unique history of trust and collaboration with the Pashtun Islamist networks located in North Waziristan, Bajaur, and the Northwest Frontier Province of Pakistan. It is not surprising, given this distinctive history, that al Qaeda's presumed protectors -- perhaps the Haqqani network, which provided the territory in which al Qaeda constructed its first training camps in the summer of 1988 -- have never betrayed their Arab guests.

These networks have fought alongside al Qaeda since the mid-1980s and have raised vast sums of money in Saudi Arabia and other Gulf states through their connections. They possess infrastructure -- religious institutions, trucking firms, criminal networks, preaching networks, housing networks -- from Kandahar and Khost Province, and from Quetta to Karachi's exurban Pashtun neighborhoods, that is either impervious to penetration by the Pakistani state or has coopted those in the Pakistani security services who might prove disruptive. It is mistaken to assume that Bin Laden, Zawahiri, or other Arab leaders would enjoy similar sanctuary anywhere else. In Somalia they would almost certainly be betrayed for money; in Yemen, they would be much more susceptible to detection by the country's police network. The United States should welcome the migration of al Qaeda's leadership to such countries.

Because there is no nexus on Earth more favorable to al Qaeda's current leaders than the radicalized Pashtun militias in the Pakistan-Afghanistan border region, American policy in the region must take special account of this specific, daunting political-military geography. As counterinsurgency doctrine correctly argues, the only way to penetrate such territory and disrupt or defeat insurgents, including outside terrorists like al Qaeda's leaders, is to do so in partnership with indigenous forces that are motivated to carry out such a campaign because they see it as in their own interests. No such campaign is plausible if the Taliban rule Afghanistan. And no such campaign is plausible if Pakistan does not continue to receive the economic and political support from the international community that may lead its own elites to decide that they will be better off without the Haqqanis and other uncompromising Islamists than with them.

It is true, in a sense, that not all Afghan stability projects are created equal, from the perspective of an American-led campaign against al Qaeda. Afghanistan's mountainous, Shiite-influenced central Bamiyan province, to choose an exaggerated example, may always be of marginal importance to al Qaeda, just as it has long been less than decisive to successive Kabul governments. But to extrapolate such observations to argue that Afghanistan's national stability is only tenuously connected to Pakistan's stability defies history, demography, and observable current trends. More Pashtuns live in Pakistan than in Afghanistan. Their travel and connections to international finance, proselytizing, criminal, and diaspora networks overlap. If the Taliban captured Afghanistan, this would certainly destabilize Pakistan by strengthening Islamist networks there.

It would also be mistaken to believe, as some in the Obama administration have apparently argued, that a future revolutionary Taliban government in Kabul, having seized power by force, might decide on its own or could be persuaded to forswear connections with al Qaeda. Although the Taliban are an amalgamation of diverse groupings, some of which have little or no connection to al Qaeda, the historical record of collaboration between the Haqqani network and al Qaeda, to choose one example, is all but certain to continue and probably would deepen during any future era of Taliban rule in Afghanistan. The benefits of a Taliban state to al Qaeda are obvious: After the Sept. 11 attacks, the United States gathered evidence that al Qaeda used Afghan government institutions as cover for import of dual-use items useful for its military projects. Reporters with the McClatchy newspaper group's Washington bureau recently quoted a senior U.S. intelligence official on this subject: "It is our belief that the primary focus of the Taliban is regional, that is Afghanistan and Pakistan. At the same time, there is no reason to believe that the Taliban are abandoning their connections to al Qaeda ... The two groups ... maintain the kind of close relationship that -- if the Taliban were able to take effective control over parts of Afghanistan -- would probably give al Qaeda expanded room to operate." This assessment is consistent with recent history.

The United States and its allies can stabilize Afghanistan. They should try; but they may fail. To avoid failure, it will be important to account for some risks that are often underestimated in the current policy debate.

These risks arise from a tendency in Washington to underestimate the importance of Afghan politics to the outcome of any course of action selected by the Obama administration. Because President Hamid Karzai has disappointed international governments; because the recent presidential election was marred by fraud allegations; because politics in Kabul appears to be difficult and fractious; and because it is not an arena in which American leverage can be easily brought to bear, there is a tendency in Washington to whistle past Afghan political issues, or to give up on the subject altogether, and to focus on other policy corridors -- counterinsurgency doctrine, military deployments, civilian efforts to build schools or highways or to provide agriculture training, counternarcotics strategy, local governance. It sometimes seems that American strategy is being designed so that it can involve itself in everything but the problems of Afghan politics, national integration, and reconciliation. But Afghan history argues that this would be an almost certain pathway to failure.

One example of this risk is embedded in the project of building a larger and more capable Afghan National Army and police force, for which there is currently much enthusiasm in Washington. The political-military history of Afghanistan since 1970 is one in which outside powers have repeatedly sought to do with Afghan security forces what the U.S. proposes to do now. It is also a history in which those projects have repeatedly failed because the security forces have been infected with political, tribal, and other divisions emanating from unresolved factionalism and rivalry in Kabul. Armies -- especially poor, multiethnic armies, such as the one Afghanistan has -- can only hold together if they are serving a relatively stable and unified national government. This has generally not been available to the Afghan Army since 1970.

Arguably, there are at least three cases during the last four decades in which programs to strengthen Afghan security forces to either serve the interests of an outside power or suppress an insurgency or both failed because of factionalism and disunity in Kabul.

During the 1970s, the Soviet Union tried to build communist cells within the Army in order to gradually gain influence. The cells, unfortunately, split into two irreconcilable groups, and their squabbling became so disabling that the Soviets ultimately decided they had no choice but to invade, in 1979, to put things in order.

Then, during the late 1980s, faced with a dilemma similar to that facing the United States, the Soviets tried to "Afghanize" their occupation, much as the United States proposes to do now. The built up Afghan forces, put them in the lead in combat, supplied them with sophisticated weapons, and, ultimately, decided to withdraw. This strategy actually worked reasonably well for a while, although the government only controlled the major cities, never the countryside. But the factional and tribal splits within the Army persisted, defections were chronic, and a civil war among the insurgents also played out within the Army, ensuring that when the Soviet Union fell apart, and supplies halted, the Army too would crack up and dissolve en masse. (I happened to be in Kabul when this happened, in 1992. On a single day, thousands and thousands of soldiers and policemen took off their uniforms, put on civilian clothes, and went home.)

Finally, during the mid-1990s, a fragmented and internally feuding Kabul government, in which Karzai was a participant for a time, tried to build up national forces to hold off the Taliban, but splits within the Kabul coalitions caused important militias and sections of the security forces to defect to the Taliban. The Taliban took Kabul in 1996 as much by exploiting Kabul's political disarray as by military conquest. The history of the Afghan Army since 1970 is one in which the Army has never actually been defeated in the field, but has literally dissolved for lack of political glue on several occasions.

None of these examples offers a perfect analogy for the present, but the current situation in Kabul does contain echoes of this inglorious history. Karzai's opportunistic and unscrupulous campaign for re-election contains two overlapping patterns of political disunity that could undermine the effort to rapidly build up and deploy the Afghan Army during the next few years. The president assembled a coalition of warlords and war criminals in his campaign coalition. Some of these warlords, such as Abdul Rashid Dostum, an ethnic Uzbek, are the very same characters whose vicious infighting caused the Afghan Army to dissolve in the face of Taliban pressure during the nineties.

Also, the currently unresolved split between Karzai and Abdullah Abdullah, the opposition leader, could become a proxy for the national division between southern Pashtuns, from whom the Taliban draw their strength, and northern Panjshiri Tajiks, with whom Abdullah has long been affiliated (although one of Abdullah's parents is a Pashtun). If Karzai and Abdullah become virulently or violently at odds, it is easy to imagine a Kabul government divided from within by its warlords and undermined from without by the Taliban on one side and disaffected northern groups on the other. This is poor ground on which to build an army of illiterate volunteers while in a hurry.

To improve its chances for success, the United States and the international community must bring all of their leverage to bear to ensure the formation of a coalition government in Kabul that incorporates all of the meaningful sources of non-Taliban opposition and sets Afghan political and tribal leaders on a sustained, Afghan-led program of political, constitutional, and electoral reform.

Some analysts have suggested invoking the Afghan institution of a loya jirga to host some or all of this continuous reform process. Whether that specific institution is selected or not, the spirit of this suggestion is critical -- Afghans have many difficult but important political and constitutional issues to negotiate, and political business-as-usual will not carry these negotiations forward adequately at a time when the United States is risking blood and treasure in support of Afghan stability. Issues that require discussion and negotiation among Afghan leaders, both formal and informal, include the future of the electoral system, to ensure fraud on the scale alleged in the most recent election cannot recur; political party formation and activity; constitutional issues such as the election of governors and the role of parliament; and issues of national integrity such as the access of different ethnic, tribal and identity groups to government employment and opportunity in the expanding security services.

Political reform and Afghan-led negotiations of this type must be seen as fundamental to American policy in Afghanistan no matter what choices are made about troop levels and deployments. Such a process would be part and parcel, too, of national program of reconciliation and reintegration designed to provide ways for Taliban foot soldiers to find jobs and for their leaders to forswear violence and enter politics.

This emphasis on political stability through continuous Afghan-led negotiation and national reintegration, as opposed to grandiose state-building or policies premised on the pursuit of military victory by external forces, should not be seen as an adjunct wing of U.S. policy in Afghanistan, but as fundamental. It is clear that no realistic level of American and Afghan forces deployable in the foreseeable future can provide security to the population in every village of Afghanistan. Accepting this reality and developing a political-military strategy that best accounts for it will lead, inevitably, to support for Afghan-led political approaches at the national, provincial, district and sub-district level. This is how the late Gorbachev-backed government in Kabul achieved a modicum of stability in far less favorable circumstances.

America's record of policy failure in Afghanistan and Pakistan during the last 30 years should humble all of us. It should bring humility to the way we define our goals and realism about the means required to achieve them. It should lead us to choose political approaches over kinetic military ones, urban population security over provocative rural patrolling, and Afghan and Pakistani solutions over American blueprints. But it should not lead us to defeatism or to acquiescence in a violent or forcible Taliban takeover of either country. We have the means to prevent that, and it is in our interest to do so.

Chris Hondros/Getty Images