ONE OF the few things that Presidents George W. Bush and Barack Obama have in common is that they reversed their long-standing approaches to counterterrorism during their very last years in office. They initially held diametrically opposed military policies, with Bush choosing invasion and occupation and Obama preferring disengagement and drone strikes. But by the end of their second terms they had both ended up in roughly the same place, with a central focus on indirect action—enabling local forces to achieve U.S. counterterrorism objectives.

Through long periods of trial and error, constrained by a common reluctance to change course, but in the end having their hands forced by growing terrorist threats and events spiraling out of control, both presidents finally came to adopt the only set of counterterrorism policies that have been shown to succeed over the long run. It is important that President Donald Trump avoid repeating this painful and time-consuming learning curve.

Doing so will require him to accept lessons from his predecessors’ experiences. President Bush’s central mistakes are relatively easy to avoid. Simply follow the advice offered repeatedly by strategists from Gen. Douglas MacArthur to Secretary of Defense Robert Gates and, if at all possible, avoid having American forces fight large conventional land wars in Asia. Eventually, President Bush largely extricated the United States from his self-dug hole through the combination of shifting to counterinsurgency operations, cultivating the Sons of Iraq, building the Iraqi Counter Terrorism Service and supporting the revolution in U.S. special-operations targeting led by Gen. Stanley McChrystal. But the success of “the surge” came only after the deaths of hundreds of thousands and the explosion of Salafi jihadist terrorism. It is difficult to argue that Al Qaeda in Iraq and the Islamic State would have emerged in the absence of the initial U.S. invasion and occupation of Iraq.

From the outset of his administration, President Obama clearly learned this lesson and was appropriately wary of any policy proposal that risked starting down the proverbial “slippery slope” to a large conventional land war. Even as the frequency of drone strikes against terrorists in multiple countries greatly expanded, he consistently stressed the need to marry these discrete direct actions to wider campaigns focused on indirect action. Obama seemed to understand that while U.S. direct actions can disrupt and even degrade foreign terrorist or insurgent groups, they rarely defeat and almost never destroy them. In military terms, direct action is a necessary line of operation, but indirect action is the decisive line of operation.

Some have caricatured this emphasis on indirect action as “leading from behind” and dismissed it as somehow being a sharp break from American tradition. This critique is itself ahistorical and more than a little ironic, since indirect action was a central element of the Reagan Doctrine that supported anti-Communist fighters in Afghanistan—which itself took a page from the Truman Doctrine that supported anti-Communist fighters in Greece. In both cases, the United States successfully enabled others to fight to meet its own national-security objectives.

But given the mixed record of the Obama administration—and the near certainty that future presidents will also want to combat terrorists indirectly rather than send hundreds of thousands of U.S. troops to occupy another country—it is critical that the Trump administration officials understand why indirect-action campaigns can fail, and what those same policymakers can do to maximize their likelihood of success.

SOME RECENT U.S. military campaigns focused on indirect action have been remarkably successful and thus attracted bipartisan support. Plan Colombia, which began under President Bill Clinton and expanded under President Bush, provided U.S. arms and advisers to fight the FARC, a group officially designated by Democratic and Republican administrations alike as both a terrorist organization and a leading drug cartel. This effort helped bring Colombia back from being a failed state in the 1990s and provided the necessary prerequisite to the recent agreement that promises an end to the Western Hemisphere’s longest insurgency. Less well known, but similarly successful, was the support provided by the Joint Special Operations Task Force in the Philippines under Presidents Bush and Obama. This task force, which was officially shut down in 2015 after over a decade of operation, sharply reduced the threat from terrorists allied with Al Qaeda and now the Islamic State, and improved the capacity of the Philippine armed forces to combat those who still remain in Mindanao.

Indirect action currently defines U.S. counterterrorism policy throughout Africa. Campaigns led by U.S. special-operations forces are successfully supporting counterterrorism operations by the African Union in Somalia, by the French and their partners across the Sahel, by the Ugandans against the Lord’s Resistance Army, and, most recently, by a multinational West African force in Cameroon against Boko Haram. These campaigns are far from complete, but have made real progress in degrading the capabilities of al-Shabaab, Al Qaeda in the Islamic Maghreb, and Joseph Kony and his followers.

But these accomplishments of the Obama administration are understandably outweighed in public perceptions by the high-profile failures in Yemen, Libya, Syria and Iraq. Washington promised to build a Libyan military that never materialized, announced an effort to train Syrian rebels that failed completely, supported a Yemeni government that crumbled before a Houthi insurgency, and spent billions training and equipping an Iraqi military that fled in the face of Islamic State. In each case, the terrorist organizations expanded their geographic areas of control, and the United States eventually found itself forced to engage in the types of direct military operations that it had previously precluded. Even in Afghanistan, the beneficiaries of over a decade of U.S. training and equipping have proven incapable of fighting the Taliban—or the relatively new local affiliate of the Islamic State—without continued U.S. support. And as a result, just as was the case with Iraq, President Obama eventually had to reverse his prior restrictions on U.S. military operations in Afghanistan.

Of course, when indirect-action campaigns fail, it is usually due to the weaknesses endemic to the political leaders and fighting forces that the United States is supporting, whether a national military, an irregular unit or a local authority. After all, it’s typically those with the deepest problems who most desperately need support. And even when American surrogates are militarily effective, their efforts may ultimately be for naught if wider political disputes prevent the conflict’s resolution. When compared with unilateral U.S. combat operations, indirect action works on a longer timescale and carries with it a higher risk of failure than many would prefer.

But it is inaccurate—and deeply unfair to those fighting and dying on the front lines against common enemies—only to blame others’ shortcomings for the inability of the United States to achieve its own military objectives through indirect action. Indeed, too often U.S. policymakers’ own decisions, made initially with the best of intentions, end up setting the stage for a subsequent failure. And these types of outcomes can be doubly tragic; when an indirect-action campaign fails, the remaining viable counterterrorism options can be particularly unappealing, as the Obama administration experienced firsthand.

After reviewing the recent record of successes and failures of the use of indirect action to combat terrorists, a few clear lessons emerge for U.S. policymakers. To maximize the probability of success, they must recognize the need to take action earlier than they might prefer. They must be willing to accept a greater and wider array of risks, both physical to military personnel in the field and political to leaders inside the Beltway. They must align themselves more fully with the interests of foreign partners rather than limiting actions to serve only U.S. interests. They must agree to delegate operational and tactical decisions to leaders in the field rather than micromanage them in Washington. They must integrate military actions with equally aggressive diplomatic strategy. And they must remain firmly and publicly committed to stay the course over an extended period of time. Easier said than done.

IT IS axiomatic that the best time to start an indirect-action campaign is before the threat has spiraled out of control. It takes considerable time to improve local forces’ military capabilities and even longer to build the necessary personal trust. Indeed, sometimes the window of opportunity for such an approach closes relatively early, especially if local actors become frustrated and increasingly distrustful after having multiple requests for U.S. assistance denied. It is better, therefore, to have a persistent but modest U.S. military, law-enforcement and intelligence presence operating largely out of the embassy, looking for emerging opportunities to partner with trusted locals committed to fighting terrorists. Because of the small size of the deployment and the unique nature of special-operations forces, the military component of this outfit can usually operate under something just short of a full status-of-forces agreement. Ideally, to ensure redundancy when setbacks emerge, multiple agencies would be working with various foreign partners in a given location, all coordinating under the U.S. ambassador and with the CIA’s station chief.

Similarly, it’s usually more effective to begin by building a small, local force than by attempting to train large foreign armies in scale. There are countless examples of U.S. special operators, law-enforcement agents (especially from the Drug Enforcement Administration) and other U.S. personnel successfully building elite, specialized foreign surrogates. But there are, arguably, no examples of the United States successfully taking the lead at building another country’s military in very large numbers from near scratch—as was the initial plan in Libya. American training can augment other countries’ concerted efforts to build their own forces in scale, but it cannot accomplish this for them. By focusing instead on building relatively small, but especially capable, surrogate counterterrorist units, the total number of U.S. personnel sent into the field can also be kept relatively small.

Surely, deploying “earlier and smaller” is far more preferable to “later and bigger.” Then why does the United States persistently prove reluctant to do so? Often, this failure comes down to a lack of strategic foresight. When dealing with countless local terrorists, narrow insurgencies or geographically limited violent criminal groups, the best option for the United States is usually to decline to involve itself militarily in matters without a clear link to vital U.S. national-security interests. But Salafi jihadist terrorists differ in at least one critical respect from other terrorists—even from other Muslim terrorist groups. Once they possess territory, sanctuary that allows them to act with impunity, they will eventually conduct external attacks. This evolution is inexorably driven by a dangerous combination of their own brand of religious ideology, the public narrative they are trying to project globally and the internal incentives that reward individual leaders who direct such attacks. In some places, it has taken longer than others for them to reach this point, but unless they are denied such sanctuary altogether, they will inevitably commit terrorist attacks abroad.

This was true when Al Qaeda gained sanctuary in Afghanistan and then Pakistan, when Al Qaeda in the Arabian Peninsula (AQAP) gained sanctuary in Yemen, and when the Islamic State gained sanctuary in Syria and Iraq. This was true whether in was in the context of Osama bin Laden’s prioritization of attacks against the “far enemy” before the “near enemy,” or in the context of Abu Bakr al-Baghdadi’s direction to attack the Shia, then the Saudi monarchy, and then the “crusaders and their bases.” In each case there were American analysts both inside and outside of government who initially argued that the terrorists’ goals were merely local; in each instance they were sadly proven wrong. But this dynamic does help explain why the United States has previously been slow to begin necessary indirect-action campaigns.

Furthermore, given perceived limits on U.S. resources it can be tempting for policymakers to focus their attention on the most immediately threatening terrorist groups and affiliates: those who have already successfully directed external attacks. But even a superficial review of budget levels over the past decade shows that the Congress has been remarkably forthcoming with military and other security budgets—indeed, perhaps too forthcoming—and thus the total amount of resources available is rarely the true limiting factor. More significant has been the executive branch’s failure to strategically prioritize within the counterterrorism assistance budgets provided, which are usually spread across a large number of differing bureaus and offices in multiple departments and agencies, each of which is institutionally reluctant to be told how to spend “its” money.

A more pernicious impediment is the concern among policymakers, sometimes deeply held but rarely spoken, that a campaign of indirect action will trigger a slippery slope that inevitably leads to a large conventional military deployment. This was a foundational lesson that many liberals took from the Vietnam War. Nevertheless, those who study indirect action argue the exact opposite—that a small deployment of military advisers is more often the critical element that prevents the later need for a large conventional military deployment.

BUT CONCERNS about a slippery slope shouldn’t be dismissed out of hand. Rather, U.S. policymakers should candidly recognize the risk and identify steps to mitigate it. Here America’s experience with Plan Colombia is instructive. When the Congress first approved the plan, during the final year of the Clinton administration and after much debate, it imposed an unsolicited legal limit on the number of U.S. military personnel and civilian contractors allowed in country: four hundred in each category. After a similarly robust debate, the Congress modestly increased the cap during the Bush administration, to eight hundred and six hundred, respectively. At the same time, aside from a prohibition on U.S. forces being involved in combat operations, the law did not micromanage the types of assistance that could be provided.

This combination of clear limits and broad authority incentivized creative military planning to adhere to the legal caps and accelerate the building of Colombian capabilities. The fact that these personnel ceilings were enshrined in specific law for Plan Colombia made the legislative and executive branches full partners in this indirect-action campaign and precluded the military from going back to the president for an increase in forces at the first sight of a roadblock. The number of U.S. personnel in Colombia, while still several orders of magnitude smaller than the prior conventional deployments in Iraq and Afghanistan, was the largest sustained indirect-action campaign until recent deployments to Iraq.

At this point in the policymaking process, concerns are also typically expressed about the nature of the force that the United States is to support. It is not unusual for allegations of corruption, criminal activity and gross human-rights abuses to surface, especially when considering irregular fighters or poorly trained regular military units. These concerns must always be taken seriously, both for moral reasons and because widespread corruption and criminality can undermine the military effectiveness of these units and cause them to lose the support of the wider local population. The Leahy Law provides a workable a framework for dealing with such allegations, should U.S. policymakers be willing to accept the domestic political risks inherent in these types of missions.

But a policy that requires perfection before partnering is usually a policy that doesn’t allow for indirect action at all—and thus increases the likelihood that the United States will eventually have to rely on its own conventional land forces to meet its counterterrorism objectives. This, of course, is exactly the outcome that many strategic planners are trying to forestall. Fortunately, the best way to mitigate human-rights violations over time is through thorough interagency vetting and active U.S. mentoring that accompany building relatively small, specialized forces. The less hands-on the U.S. support, the more likely it is that it will be misused. And the more American forces are trusted by the locals, the more they are likely to react thoughtfully and decisively to any accusations of human-rights impropriety.

Domestic political considerations have also hampered imperative indirect-action campaigns. After acceding to the Oval Office on an antiwar platform, President Obama was understandably reluctant to take steps that could misconstrued as new military fronts against terrorists. Asking a politician to risk the perception of flip-flopping—especially on the subject of war and peace—is never an easy sell. It took the Bush administration far too long to reverse its initial counterproductive prosecution of the Iraq War, and it took the Obama administration too long to recognize the need to send U.S. special-operations forces back into Iraq after they were withdrawn. Moreover, the American public does not easily distinguish between combat operations and advisory missions, especially if both can put U.S. military personnel at risk. Impassioned promises of “no boots on the ground” are as politically comforting as they are intellectually inchoate.

Once policymakers decide to send a relatively small contingent to support others fighting terrorists, they must also determine the range of support that American forces are allowed to provide. These can be roughly divided into equipping, training, advising, assisting and accompanying missions. At one end of this spectrum, the U.S. government simply sells or otherwise provides military or other specialized equipment and may or may not itself offer training on the proper use of that equipment. At the other end of this spectrum, U.S. forces help their counterparts plan operations against a specific terrorist target, provide a range of direct support for those operations and accompany them to the objective, allowing for continuous, real-time tactical advice—just short of joining them “on target” during the military operation itself. These are not combat operations but walk right up to that line, and in doing so, put U.S. forces in harm’s way. Indirect-action campaigns that allow the full spectrum of these activities typically maximize their overall effectiveness and accelerate the speed by which foreign forces become independently capable of successful counterterrorism operations. And this acceleration of local capability is the most effective way to limit the risk of the slippery slope that leads to a much larger conventional military deployment.

Building foreign surrogate units often requires a commitment to enable them when they get in the field, for operational as well as moral reasons. This means ensuring that they have everything U.S. forces would: command, control and communications; intelligence, surveillance and reconnaissance; medical-evacuation capabilities; and, when necessary, lift, fire and close air support.

Sometimes, for its own understandable reasons, the host government limits the range of support. But just as often, U.S. policymakers have imposed their own constraints—sometimes in direct opposition to partners’ requests. The further up the spectrum of support, the greater the “risk to force” even as “risk to mission” is mitigated. Since U.S. forces in small numbers are accompanying their foreign partners inside hostile territory, they must always be allowed to defend themselves if attacked. And since they must have rules of engagement defining the circumstances under which they can come to the defense of their surrogate unit, these indirect support and enabling missions always place American personnel in danger, even if they are not formally considered U.S. combat missions. And in a domestic media environment haunted by the specter of “Black Hawk Down” in Somalia and, more recently, the death of Amb. J. Christopher Stevens in Benghazi, it is not surprising that U.S. policymakers sometimes hesitate before accepting this “risk to force” and its wider political ramifications.

Nevertheless, while recognizing these considerations, U.S. policymakers should begin each indirect-action campaign with the assumption that the full range of enabling support will be offered to partners fighting Al Qaeda, the Islamic State and their affiliates. Should a convincing case be made to the president to limit U.S. support in specific circumstances—some will merit holding back—such a decision could be taken. But given the especially high opportunity cost of early inaction, the burden of proof should rest on those who would argue to do less. Too often, however, the reflexive inclination across multiple U.S. administrations has been to do the opposite. The implicit assumption is that indirect-action support should be started slowly, at the outset be limited only to the less risky (and thus less impactful) options, and expanded only incrementally and after lengthy review at the highest levels of the U.S. government. From the perspective of U.S. partners, this approach does not instill confidence in American commitment.

The trust and confidence of America’s partners is, of course, the center of gravity for a successful indirect-action campaign. And that is why “partnering” must always be a two-way street. A decision to rely on indirect action instead of direct U.S. military action means that Washington no longer has the luxury of viewing problems through the narrow prism of its own interests. If the U.S. military is going to be working by, with and through others, then its objectives must take their wider interests into account. This doesn’t mean that America needs to adopt all foreign interests as its own, but it does mean that the objectives of U.S. policies and the resulting scope of its assistance must be a product of negotiations with its partners, not a unilateral diktat to them.

For instance, a key element of Colombia’s successful military campaign against the FARC was the Clinton administration’s decision to approach the problem holistically. No longer would Washington pursue its counternarcotics objectives with tunnel vision; Colombia’s counterinsurgency and counterterrorism agenda were intertwined. In contrast, the United States publicly limited its military support for Yemen to focus on counterterrorism and not on Yemen’s wider counterinsurgency—at least until the Houthi insurgency took Sanaa and thus upended America’s counterterrorism campaign against AQAP. And since then, Washington has found itself supporting Riyadh’s air campaign against the Houthis to a greater degree than it previously supported the government of Yemen itself. A predictable rash of civilian casualties ensued. So once again, with a healthy dose of irony, a policy that was intended to limit U.S. involvement in a faraway war ultimately—as a by-product of the failure that resulted from those limits—expanded U.S. involvement far beyond the level initially feared.

A MILITARY strategy focused on indirect action requires more than just the military. It requires integrated actions across numerous U.S. government agencies. It requires a nimble flexibility in execution to manage relationships and adjust areas of emphasis on a tactical basis. It requires activist and creative U.S. diplomacy to manage coalitions of the willing and help local parties reach agreement on countless problems. This, in turn, requires empowered presidential emissaries to constantly be at the side of local actors—figuratively and literally—encouraging, cajoling and pressuring when necessary.

When America sends large numbers of conventional forces to war, the U.S. military commander can lead such an effort. But when the president has chosen to rely primarily on indirect action, the only person in the government properly positioned to play this role is the U.S. ambassador. Ideally, the ambassador in question wields an uncommon combination of skills that extend beyond the diplomatic prowess of the typical State Department officer to also include the ability to provide detailed, day-to-day oversight of all indirect-action operations, holding accountable the Chief of Station, law-enforcement attachés and military commanders in country. Fortunately, the Foreign Service has thus far been able to produce enough of these rare individuals. And where the U.S. military and other operational agencies have had success in indirect-action campaigns, it is typically due to the strong leadership they received from the ambassador.

In some cases, Washington-based policymakers eager to micromanage from afar risk undercutting the effectiveness of the ambassador and the operational teams he or she leads. Instead, once the president has established the policy direction, and his principals have defined the operational latitude and acceptable risks, the vast majority of operational and tactical decisions should then be delegated to the ambassador. Regular review of indirect-action campaigns should, of course, still take place inside the Beltway, and the president should hold the ambassador personally accountable if a pattern of failures emerges.

As has been the case during other successful indirect-action campaigns, U.S. military commanders in the field will come to regard their ambassador as an approval authority rather than merely one of many voices with whom they need to consult before taking action. The higher-level combatant commands will need to put aside their natural reluctance to allow civilian influence inside the chain of command. Chiefs of station will play important roles as civilian advisers to the ambassador for all indirect-action operations. Special-operations forces have historically been more comfortable with these kinds of relationships than have conventional military forces. Processes to raise fundamental disagreements up the formal chains of command must remain, but the benefit of the doubt should go to the U.S. ambassador in the field.

The most important role for senior policymakers in an indirect-action campaign, after they provide strategic policy guidance, is to project confidence and consistency in the strategy both at home and abroad. As much as possible, this support should be bipartisan and continue across multiple administrations, which requires thorough and transparent consultations with congressional leaders. Successful indirect-action campaigns take many years and are likely to encounter failures along the way—some will be public, and some may become the subject of partisan rancor back home. It is precisely at these moments that public support from the White House to stay the course is most critical. Similarly, when terrorist threats to a U.S. embassy emerge, policymakers should resist the urge to reduce presence—this is exactly what the terrorists desire. Past experience shows that once personnel have been withdrawn, it can take many years to restore them to their posts. Those absent years can setback an indirect-action campaign disproportionately.

A counterterrorism policy dependent on indirect action requires policymakers to fundamentally increase their willingness to accept measured risk, even the political variety. America’s partners will make mistakes. Its alliances will be uncertain. Tactical failures are inevitable. Money will be spent on programs that, in the end, don’t come to fruition. These risks need to be accepted up front.

And, most importantly, despite the best of efforts, some Americans conducting these indirect action campaigns near the front lines will likely die. This risk in the short term is much lower when indirect action is sharply constrained and U.S. counterterrorism operations are limited to merely shooting Hellfire missiles from remotely piloted aircraft. But targeted killing alone will not secure our enduring counterterrorism objectives. And then, when terrorist groups have expanded their reach and are planning external attacks—and the only remaining policy option is to start sliding down the slippery slope to a conventional war—the risk of U.S. casualties will increase exponentially. It would be far preferable to learn the lesson of the past two administrations and focus on indirect action.

William F. Wechsler, a senior fellow at the Center for American Progress, was deputy assistant secretary of defense for Special Operations and Combatting Terrorism from 2012 to 2015, and deputy assistant secretary of defense for Counternarcotics and Global Threats from 2009 to 2012.

Image: A U.S. Army soldier fires a simulated M136 AT4 anti-armor weapon. Flickr/U.S. Army