General Stanley McChrystal's request to send more troops to Afghanistan has induced sticker shock for many Americans--including, apparently, President Obama. The integrated counterinsurgency, or COIN, strategy that McChrystal wants to pursue has many components: protecting Afghan civilians, rapidly expanding the Afghan army and police, reforming government, providing economic development assistance, weaning Taliban fighters and leaders away from Mullah Omar and Osama bin Laden, reconciling them into the new government, and targeting those who refuse. This makes it a demanding strategy that McChrystal reportedly believes will require providing at least an additional 10,000 to 40,000 U.S. troops and more than doubling existing Afghan forces to a total of 400,000 indigenous soldiers and police. (Full disclosure: I served as a member of General McChrystal's assessment team in June and July 2009, but I do not speak for his command, and the views expressed here are strictly my own.) This price tag has further galvanized opposition to a war whose support was already fading fast.
Few, however, actually want to leave Afghanistan outright. Instead, most pair their opposition to reinforcement with support for a middle way--a more limited presence intended to secure U.S. interests without the cost and risk of escalation. Opponents have proposed at least a half-dozen such "middle ways," ranging from greater reliance on drone-based counterterrorism strikes to early pursuit of a negotiated settlement to end the war. The specifics are often fuzzy; none has been articulated with the detail of McChrystal's proposal, particularly regarding troop requirements. But most are tantamount to splitting off a piece of McChrystal-style integrated COIN and executing it alone. Some critics propose pursuing pieces in combination, but none attempts the totality, and, especially, none includes McChrystal's large U.S. ground combat presence for protecting Afghan civilians. For all, the underlying idea is to reduce the cost of the war without abandoning the U.S. interest in denying Al Qaeda a base for attacking the West or destabilizing neighboring Pakistan.
It is easy to see why such middle ways are so popular. They could lighten the burden on the federal deficit. They could put fewer Americans in harm's way. They would seem to better fit the U.S. interests at stake, which are real but limited and indirect. They appeal to the centrism of many American voters. The problem is that they probably won't work.
The reasons vary from proposal to proposal, but the basic problem is that the pieces of COIN are interconnected and mutually reinforcing. The whole is greater than the sum of the parts; implementing just one or two pieces alone undermines their effectiveness. It might make sense to do less and accept a greater risk of failure, depending on one's tolerance for risk and cost. But there is no magic middle way between the McChrystal recommendation and total withdrawal that offers comparable odds at lower cost. In counterinsurgency, less is not more.
Train, Don't Fight
One of the most popular middle ways is to shift the U.S. role from combat to training and advising Afghan forces. This would put an Afghan face on the war effort, it would place the burden of the fighting on those with the most at stake, and it would ostensibly reduce U.S. exposure to casualties. Building the indigenous military is in fact a major element of General McChrystal's proposal. The real difference between his vision and that of his opponents is the latter's desire to accomplish this while reducing our combat presence. But training and combat are not exclusive of one another. The former requires the latter, and to field a large Afghan force faster will require more, not fewer, U.S. combat troops.
To build an indigenous security force in the middle of a war is not like teaching math to high school students--it cannot be done successfully by a handful of teachers in classrooms with chalk and blackboards. To field Afghan troops quickly without breaking them in the process requires close partnership on the battlefield, with experienced Western combat units that provide on-the-job training, mentoring, confidence-building, fire support, and stiffening in actual combat. And this requires Western troops, in large numbers, living and fighting together with Afghan forces at all levels of command. The faster the Afghans are to be fielded, the more Western combat forces are needed. If a large Afghan military is to be raised, then many tens of thousands of Americans will be needed, and those Americans will be exposed to combat, and to casualties.
And the process takes time even so. In the meantime, someone must protect not just key population centers but also the recruitment centers, supply depots, bases, and transportation connections needed to create the new Afghan formations in the first place. Close partnership with expanded indigenous forces is indeed the best way to pursue counterinsurgency in Afghanistan. But this is not a plausible route to reducing U.S. combat activity or troop strength there any time soon.
Another popular middle way is to rely on drone attacks, of the kind now ongoing in northwest Pakistan, to suppress Al Qaeda without a major ground commitment in Afghanistan. By killing key leaders and limiting the others' freedom of action, it is argued, the drone strikes make large-scale terrorism much harder. Drone-based counterterrorism cannot destroy Al Qaeda outright, but it might be able to constrain it far more cheaply than a major counterinsurgency campaign could. Like training, however, leadership targeting is a normal part of integrated COIN, where it is used to disrupt enemy command and control, keep the opposition off balance, and encourage insurgents to negotiate lest they be killed. McChrystal himself led the use of such strikes during the "surge" in Iraq and will surely employ them in Afghanistan, too. But, as an alternative to COIN, counterterrorism has important shortcomings.
The biggest challenge to relying on drones is the need for intelligence. Drones are not wonder weapons; in particular, they require information on targets' whereabouts that is normally provided by other assets--and especially by host government cooperation on the ground. It was Pakistani government penetration of the Tehrik-e-Taliban, for example, that reportedly enabled a U.S. Predator drone to kill terrorist leader Baitullah Meshud in August 2009. In general, such spies, informants, and other tipsters are key intelligence sources for drone attacks on secretive terrorist groups. This "human intelligence," however, is very hard to get if the government on the ground decides to deny it to the United States. Penetrating wary, secretive, ruthless terrorist organizations with human informers is hard enough anywhere, but to do so on the soil of an actively hostile state means facing multiple layers of state security and counterintelligence efforts before getting anywhere near the terrorists. For a sustained drone campaign to keep a resilient group like Al Qaeda on the ropes requires systematic intelligence of the kind that is very hard to get without the access enabled by a tolerant government.
Security trends in Afghanistan are dangerously adverse; a major reduction in the U.S. combat presence could easily lead to the collapse of the Karzai government. If the Taliban were restored to power in Kabul because U.S. counterinsurgency failed, we would obviously lose the access we now have in Afghanistan, significantly hampering our ability to pursue Al Qaeda there. Perhaps even worse, the Taliban could threaten the stability of the government in Islamabad. In 2000, the Taliban ruled Afghanistan and Pakistan was their ally, but, in 2001, the Bush administration coerced Pakistani President Pervez Musharraf into turning on his erstwhile allies and aligning with the United States. That switch in allegiances was never as complete as Washington wanted, but it was enough to persuade the Taliban and a variety of other Islamist terrorist groups within Pakistan that the secular government in Islamabad had become a tool of the hated Americans. The result has been a growing insurgency within Pakistan fueled by groups once aligned with (and, indeed, created by) the government they now attack.
These groups are diverse, and the connections between them are complex--they are not a monolithic, unified front. But they are interconnected, and they share great distrust of both the United States and a secular Pakistani government. Collectively, they have mounted an insurgency that is dangerous enough as it is; given the advantage of a state-scale haven across the Durand Line, their virulence could increase dramatically. If the failure of counterinsurgency in Afghanistan enabled insurgents to succeed in Pakistan, the effectiveness of U.S. counterterrorism efforts would plummet even as the need for them skyrocketed. Collapse of the Pakistani state would create an urgent need to prevent its nuclear weapons from falling into the hands of Al Qaeda or other radicals, yet the loss of government intelligence would hamstring our ability to do so.
To be sure, the connection between Afghan and Pakistani stability is incomplete. Karzai could fall, but, if the Pakistanis put their house in order and defeated their own insurgency, then the net consequences would be moderate; conversely, Pakistan could fail even if Afghanistan stabilizes. The problem is that the odds of success in Afghan COIN and Pakistani counterterrorism move in the same direction. COIN failure makes counterterrorism failure more likely, and vice versa. And this makes counterterrorism a poor substitute for counterinsurgency.
Reconcile with the Taliban
Another common proposal is to negotiate a power-sharing deal with some or all of the Taliban as a means of ending the war without the escalation embodied in the McChrystal recommendations. America's real interests are quite limited, it is often argued, so why not pursue a settlement to bring the Taliban into a coalition government on the proviso that they keep Al Qaeda out and deny the use of Afghan territory for destabilizing Pakistan? Others recommend lower-level reconciliation with individual Taliban foot soldiers. Many of these soldiers fight more for the financial incentive of pay to feed their families than for any ideological commitment to Mullah Omar; if so, then why not outbid the Taliban for their services and win them over to the government's side without killing or capturing them all?
Here, too, the only real difference with McChrystal's proposal is the assumption that this course of action is an alternative to reinforcement and escalation. After all, the McChrystal report itself explicitly calls for low-level reconciliation with Taliban fighters, and many successful counterinsurgencies end by negotiation rather than outright annihilation of the last insurgent. The problem is the idea that this can be done on the cheap and still produce terms the West can accept. It cannot.
Today, the Taliban have little incentive to accept any deal that the West could live with. Taliban leaders appear to believe that they are winning the war and will soon drive out the foreigners and topple the Karzai government. This is hardly unreasonable: The McChrystal report itself suggests that the security trends are now dangerously negative for the government. Why should Taliban leaders compromise for half a loaf when the whole bakery is available? Karzai has reportedly been reaching out to the Quetta Shura and Hekmatyar factions of the Taliban via Saudi intermediaries for some time now; the talks have never made real progress because the Taliban insist on a total withdrawal of foreign forces as a precondition for negotiation. This is an obvious nonstarter, and the Taliban surely understand this--the indications are that they are unwilling to accept any meaningful compromise because they think total victory is within reach.
A similar problem confronts related proposals for lower-level reconciliation with individual Taliban foot soldiers. In principle, the government could try to outbid the Taliban for their services. The Taliban, however, are not easy employers to leave. Foot soldiers who turn on them are subject to brutal retaliation against themselves and their families. If the government cannot protect them from the inevitable retribution, it would be suicidal for individuals or small groups of fighters to change sides. Nor is mere acceptance of government subsidies real evidence of realignment: The worst outcome would be paying fighters who join government militias but answer to Taliban commanders because of threats to their families.
This is not to say that negotiation is unwise, or that low-level reconciliation is impossible. If an improved Western military effort persuaded key factions that the payoff to fighting is lower and the cost higher than they think now, then deals might be possible. And a credible promise of real security under government pay could wean away an important fraction of today's Taliban foot soldiers. Such deals could substantially shorten the war and reduce its cost to all sides. But, without a major change in the tide of battle and the associated expectations, it is hard to see where the common ground lies. The heart of the McChrystal approach is to produce exactly that change in the tide of battle, but it will be hard to get the result without the price.
Buy Off Warlords
It is sometimes argued that the West should stabilize Afghanistan and control Al Qaeda by paying warlords, tribal leaders, or other local power brokers to police their own turf, rather than relying on the national government in Kabul to control the entire country. Afghanistan has never had a strong central government, and order in the provinces has often been maintained by local authorities, legal and otherwise. The British, it is said, found direct control impossible but managed to wield influence by paying tribal or factional leaders. If the United States is willing to settle for government-by-warlord, then it could avoid the expense and risk of an orthodox counterinsurgency campaign while still denying militants access to Afghan havens.
Some increase in the role of tribes or other local authorities is very likely, but here, too, the issue is whether this can substitute for the cost and expense of counterinsurgency and enable a solution on the cheap. Unfortunately, it probably cannot. The central problem is very similar to that of negotiating with the Taliban: Parties' alignments turn powerfully on their expectations for the outcome of the war. It is dangerous to end up on the wrong side, and today, too many Afghans think the Taliban will win. If the United States pulls back and tries to compensate with payola, this will just reinforce the perception that a Taliban takeover is coming. And why would coldly self-interested warlords risk their lives by crossing the Taliban on behalf of an absentee benefactor who clearly lacks the motivation to defend its local allies? The odds are that such factions will take U.S. money, then side with the enemy once the latter shows up and threatens superior force. Payments, subsidies, and aid of various kinds can be a powerful adjunct to a broader strategy that includes the forces needed to defend allies and defeat the insurgency. But, if it looks to Afghans like the West is losing to the insurgents, it will be dangerous for them to buck the tide at risk of their lives, and money alone is unlikely to change this.
What's more, some of the more prominent sub-national authorities are even less popular than the central government. The traditional tribal leadership is one thing, but many of Afghanistan's former warlords and current narcotics kingpins are hated figures whose predatory rule is disliked even more than that of the Taliban. Indeed, Western and central government complicity with such figures has helped the Taliban penetrate Afghan society. Better pay for warlords is hardly the way to create a government that Afghans will prefer to the insurgency. As a component of a broader strategy that includes both sufficient forces for security and an insistence on governance reforms to halt the worst of today's predatory excesses, some form of decentralization and acceptance of sub-national autonomy and authority is probably necessary and appropriate. But simply paying off warlords to do America's dirty work for it in the absence of a much broader counterinsurgency strategy is unlikely to succeed.
Some argue that the chief obstacle in Afghanistan is nationalist resistance to foreign occupation. In a proud, xenophobic society, they say, foreign forces stimulate something akin to an "antibody reaction" in which, regardless of the foreigners' mission or intentions, Afghans will fight to expel any alien presence. The larger and more visible the foreign presence, the harsher the resistance. If so, reinforcement will only make things worse, and a better strategy would be to reduce outside forces to minimal strength, adequate only to advise Afghan leaders on defending themselves.
Afghans surely resent foreign occupation, as would anyone. But it is far from clear that this is the primary problem, or that a drawdown to a "light footprint" could defeat the Taliban. After all, we've tried it, and it hasn't worked.
In 2004, there were only 15,200 American troops and under 9,000 NATO-led troops in Afghanistan; as recently as March of 2006, there were only 20,000 American soldiers on the ground and about 12,000 NATO-led troops in a country of about 30 million people.* The thinness of these deployments was defended by then–Defense Secretary Donald Rumsfeld precisely in terms of a perceived need to avoid nationalist resistance to a foreign occupation. If a light footprint could avert insurgency, then there should be no war in Afghanistan today. To put it mildly, it has not worked out that way: The Rumsfeld light-footprint policy gave us the mess we have now. It yielded too little security to protect the population from the Taliban, too few trainers and advisers to create an indigenous military, but enough of a foreign presence to alienate the public all the same.
By contrast, there is evidence to suggest that civilians will accept foreign forces when these are strong enough to bring security in exchange. Americans, for example, were hated and resisted in Iraq's Anbar Province between 2004 and 2006, when there were enough of them to be irritating but too few to provide real security. By late 2007, when reinforcements (together with the Anbar Awakening) enabled the troops to offer security, their presence was tolerated to the point where I was able to walk through the central marketplace in Fallujah handing out candy to schoolchildren. Afghanistan, by contrast, is now the worst of both worlds: There are enough foreign forces to be off-putting, but not enough to secure the population. Reducing those numbers would cede the country to the insurgents; increasing them to enable real security might make the price worth paying for Afghans.
Send Aid, Not Troops
Another proposal would shift the international contribution from combat to development assistance. Prosperity and an economic stake in the government, it is argued, can wean the population from the Taliban more effectively than force, which inevitably causes collateral damage and kills innocent civilians. Aid, moreover, is an asymmetric advantage for the West, which can vastly outspend the Taliban, and would avert the "antibody reaction" spurred by foreign soldiers, instead inoculating the population against an insurgency that can offer little more than violence. Military force is also extremely expensive; for the cost of today's U.S. troop deployments, the total aid budget for Afghanistan could be increased many times over. Given this, some advocate replacing large military deployments with major increases in development aid instead.
Aid is in fact a major component of orthodox counterinsurgency strategy. But to rely on it as an alternative to the security provided by military forces is impractical in an active war zone. Aid is inherently political and is clearly understood to be so by the Taliban, who systematically target Western aid projects for attack. Without large security forces to defend them, aid projects cannot survive. In fact, development projects in Afghanistan are often destroyed even when they are defended, if those defenses are inadequate. No sensible Taliban would allow aid projects to undermine their control over the population when insurgents have the means at their disposal to destroy them or to intimidate their staff. Aid without security in Afghanistan would be fruitless.
None of the usual middle-way proposals are thus likely to be effective as alternatives to reinforcement. Many are potentially important components of an integrated, properly resourced COIN strategy. But to pull pieces out of this integrated context and undertake them as substitutes for major troop deployments is to deny them essential preconditions they need to function. The pieces of orthodox COIN strategy interact: security enables development and governance, development and governance enhance security, governance facilitates counterterrorism, counterterrorism improves security, security enables negotiation and reconciliation. Each is a valuable complement to the others; none is a viable substitute. Integrated COIN is itself no guarantee of success. Social scientists have estimated its success rate at somewhere between 25 and 70 percent at best. But middle ways are even less promising because they lack the key enablers of an integrated strategy and the synergies that result.
Of course, the United States could in principle pursue the current COIN mission in Afghanistan with only the forces already committed. (At present, there are almost 68,000 U.S. troops, 35,000 non-U.S. foreign soldiers, 90,000 Afghan troops, and 80,000 Afghan police.) But muddling through a COIN strategy with insufficient resources increases the risk of failure, while leaving an already large and burdensome U.S. presence in place. Balancing cost and risk is central to the whole issue of U.S. strategy. General McChrystal's own troop request is reportedly framed as a choice between a large option offering the lowest risk, and a much smaller reinforcement with greater risks. No reinforcement at all would increase those risks accordingly. But, whereas many of the other proposed "middle ways" would at least reduce the burden dramatically in exchange, muddling through with an under-resourced version of integrated COIN would increase the danger of ending up in the worst of all worlds, with enough force deployed to be a heavy burden on the military, the budget, and the patience of Afghans and Americans, but not enough force to succeed. In other words, we might fail expensively rather than cheaply.
In a world of probabilities rather than guarantees, no strategy can ensure success. But integrated COIN offers a higher probability of success than any of the proposed middle ways; middle ways are cheaper, but also likelier to fail. The result is not a single, analytically derivable right and true answer for U.S. strategy--instead, we face a hard value judgment in choosing between better odds at a higher price or worse odds at a lower cost. What analysis does show, however, is that there is no middle way that offers COIN's odds without its sacrifices. Years of neglect and error have produced a situation in South Asia where none of the available options offers an easy or palatable way out of a difficult set of dilemmas. It is understandable that Americans would like a cheaper way to secure U.S. interests in Afghanistan than reinforcement and COIN; it is far less clear that a middle way exists that can accomplish this.
Stephen Biddle is the Roger Hertog Senior Fellow for Defense Policy at the Council on Foreign Relations.
*Correction: The article previously misstated that there were only 15,200 foreign troops in Afghanistan in 2004, and only 20,400 foreign troops in 2006. We regret the error.