Should the Taliban ever take power in Kabul, the key to preventing them from destabilising the region will lie not in Washington but in Beijing. This is above all because such a Taliban regime would depend heavily on Pakistani backing, and Pakistan depends critically on Chinese backing. China, while it does not necessarily oppose the Taliban as such, has a vital interest both in the survival of Pakistan and in regional stability more widely. This interest is inherent to China’s ethno-religious makeup and geographical position, and will last long after the United States has gone home.

As to the idea of U.S. forces in Afghanistan as a useful tool against Iran, Russia and China, this is almost the opposite of the truth. In these terms, they are not U.S. assets but liabilities—even hostages. Iranian contacts with and limited help to the Taliban are not due to any affection for them (Iran’s natural allies in Afghanistan are the Shia Hazara, and Tehran almost went to war with the Taliban in 1998 in their defense) but simply because Iran sees the Taliban as perhaps the best way to hit back at the United States if Iran is itself attacked by the United States and Israel. Moscow too may be interested in this possibility.

THE GREATEST threat to the present Afghan state comes not from the Taliban but the country’s own internal divisions. This state was cobbled together by the United States in 2001–02 on the basis of elements which had previously formed part of the mujahideen groups that fought the Soviets and their Afghan allies in the 1980s. For several years, former Communist officers and officials were ignored and rejected by the United States, considerably delaying the creation of an effective Afghan army. Even more disastrously, Taliban commanders who wished to make a deal with the new regime were often not just rejected but targeted by U.S. forces on the advice of local rivals.

None of this would have mattered so much if the former mujahideen forces on which the United States did rely had been united. But as the dreadful example of their rule in the mid-1990s demonstrated, this was very far from being the case. The present Afghan state is split internally along multiple lines, but the most important are ethnic. In the wars of 1979 to 1992, and again after 2001, ethnicities that had previously played a subordinate role in Afghanistan—notably the Tajiks, Hazaras and Uzbeks—were able greatly to improve their position at the expense of the Pashtuns, the ethnicity that had always seen itself as the people of state. “Afghan” is indeed the traditional Farsi word for Pashtun. In a curious twist, however, because of the historic prestige of Persian culture, the government language and lingua-franca of Afghanistan was always Farsi, not Pashto, even though the dynasty and senior officials of the army and state were Pashtuns; and the mass of the bitterly poor Pashtun rural population never derived any real benefit from their “privileged” position.

The rise of the Tajiks and even more the previously downtrodden Hazara (despised by the Pashtuns as Shia heretics) in turn caused deep anxiety and resentment among the Pashtuns, which contributed to mass support for the Taliban among Pashtuns both before 2001 and today. In particular, the role played by the Tajik-dominated “Northern Alliance” forces in overthrowing the Taliban state in 2001 allowed them to gain a predominant role in the armed forces. Visiting Afghanistan in recent years, I have been depressed by the echoes of ethnic tensions that I experienced before the collapse of Yugoslavia and of Soviet rule in the Caucasus: the way in which every disputed issue and state appointment is portrayed in ethnic terms; the constant falsification of ethnic proportions of populations and government jobs; the constant appeals to mythical histories, and endless battles over symbolic issues. Thus, in Afghanistan, the state has attempted to limit controversy over the new national identity card by describing everyone as “Afghan.” This then led to protests from Tajiks and others who declared that “Afghan” is simply another word for “Pashtun.”

A very large part of U.S. political activity in Afghanistan since 2001 has been devoted to managing these ethnic tensions. In the wake of the Taliban’s overthrow, the Tajiks and others were able to block a return of the former (Pashtun) king, Mohammed Zahir Shah. But with the greatest Tajik leader, Ahmed Shah Masoud, assassinated, the other ethnicities had to accept the U.S. choice of a Pashtun from the old royal elites, Hamid Karzai, as president. Washington turned a blind eye to the corruption and heroin-dealing of the Karzai regime, and the rigging of presidential elections by Karzai in 2009 which contributed to the defeat of the Tajik-backed candidate, former foreign minister Abdullah Abdullah. In the presidential elections of 2014, the United States acquiesced yet again in what was almost certainly rigging in favor of the Pashtun candidate, Ashraf Ghani, resulting in another defeat for Abdullah.

What will happen in the 2019 presidential elections (if they happen), God alone knows. Following the 2014 elections, after long and bitter negotiations, then Secretary of State John Kerry persuaded Abdullah’s supporters (which included some Pashtuns, just as Ghani’s supporters were by no means solely Pashtun) to accept the result by initiating what was in effect an informal, unstated kind of ethnic power-sharing, in which a Pashtun took the presidency but Abdullah was recognised as a kind of prime minister, and government jobs were to be shared equally between them (something for which there is no provision at all in Afghanistan’s “democratic” constitution).

In practice, however, Ghani and his supporters have dominated the government, leaving Abdullah’s former supporters seriously aggrieved. It is likely to be even more difficult to broker such an agreement after the next elections, even supposing that the Trump administration has the knowledge, commitment and stamina to do so. Meanwhile, Kabul is full of unemployed individuals, products of the boom in “higher education” over the past seventeen years, convinced that their “degrees” entitle them to government jobs—perfect fodder for ethnic protests. To hold the Afghan state together, therefore, it is not enough for the United States to give military support and aid. It will also have to go on managing the internal political problems of the state and providing the financial patronage that keeps the elites more-or-less loyal.

DESPITE THIS miserable picture, it is still on balance worth the United States hanging on in Afghanistan for a variety of reasons. The first is what in the U.S. foreign and security establishment is called “credibility,” and which used to be called by the more straightforward word “prestige.” A senior U.S. general put it very well to me back in 2008. When I asked him to define victory in Afghanistan, he said that he couldn’t, and nor in his view could anyone else in Washington. But, he added, they could all define defeat. Defeat is a repeat of the fall of Saigon in 1975, with panic-stricken Afghan refugees clinging to helicopters on the roof of the U.S. embassy in Kabul as the Taliban storm the city: “Do not underestimate the determination of the U.S. armed forces to fight on for a very long time indeed to prevent images like that appearing again,” he told me.

Since then, the U.S. military has fought on for another decade. It was on the basis of the argument that U.S. prestige would suffer terribly from a Taliban victory that the Pentagon persuaded first Obama and then Trump to reverse their previous positions and continue the U.S. military presence in Afghanistan. It is worth remembering in this context that the Communist victory followed a peace agreement with North Vietnam and U.S. military withdrawal, which in turn provided the cue for a parsimonious Congress, disgusted with South Vietnamese waste and corruption, to radically reduce military and financial aid to the South. This, like the end of Soviet aid to the Afghan state in 1991–92, then ushered in a speedy victory by the other side.

There is a very cogent response to the “credibility” argument and the Vietnam analogy. The loss of U.S. prestige as a result of the fall of South Vietnam was only temporary and was followed not too long afterwards by the Soviet debacle in Afghanistan and then the collapse of the ussr itself. The real damage done by the Vietnam War was not by the Communist victory, but by the disintegration of the domestic U.S. consensus and the growing unwillingness of ordinary Americans to fight endless wars for unclear ends in far-off countries. Moreover, far from leading to the famed “domino effect” and the fall of more U.S. allies, ancient national hostility to China meant that within four years of the fall of Saigon, China and Vietnam were at war. Today, Communist Vietnam is moving towards becoming a U.S. ally. Can the same thing not happen in Afghanistan?

The answer is most probably no. The Vietnamese Communist party, the Vietcong and the North Vietnamese Army were (and are) formidably disciplined and united institutions with an iron determination to build a modern, united Vietnamese state. Ethnic Chinese were driven out for ethnic chauvinist reasons and because as the local commercial class they had taken sides against the Communists—producing the tragic spectacle of the “boat people.” The same happened to America’s allies, the ethnic Hmong. With these exceptions however, the Vietnamese state after the Communist victory was able successfully to impose enduring national unity and stability, albeit by ruthless means.