Just before midnight in Kabul on Monday, Major General Chris Donahue walked onto a C-17 transport aircraft.
He was the last American soldier to leave Afghanistan.
The aircraft took off from Kabul airport, and America’s longest war was over.
There was no peace treaty. There was no victory parade. The government America spent twenty years supporting had already disappeared, the Taliban controlled Kabul, and the final days of the evacuation had been marked by an ISIS-K bombing that killed 13 American service members and scores of Afghans.
Then the runway was empty.
The Taliban celebrated.
For twenty years, American presidents had tried to answer the question of how the war in Afghanistan should end. George W. Bush started it. Barack Obama escalated it and then began reducing American forces. Donald Trump negotiated an agreement with the Taliban setting the withdrawal in motion. Joe Biden made the final decision to leave.
Four presidents. Roughly twenty years. Trillions of dollars spent. Thousands of American and allied troops killed, and vastly more Afghans.
And on Monday night, Afghanistan was again controlled by the Taliban.
The easiest conclusion is also the most emotionally powerful: it was all for nothing.
I don’t think that is quite true.
But saying it wasn’t all for nothing should not become a way of avoiding the scale of what failed.
The difficult task now is separating the two.
The United States invaded Afghanistan in October 2001 for a much narrower reason than most of the objectives it eventually pursued there. Al-Qaeda had carried out the September 11 attacks while operating from Afghanistan under the protection of the Taliban government. The initial American objective was therefore fundamentally about terrorism: destroy al-Qaeda’s ability to use Afghanistan as a sanctuary and remove the Taliban government protecting it.
That happened remarkably quickly. Taliban rule collapsed within months. Al-Qaeda’s infrastructure was severely disrupted. Osama bin Laden eventually fled Afghanistan and was killed by American forces in Pakistan in 2011.
If that had remained the definition of the mission, America might be able to describe Afghanistan very differently today.
But the mission expanded. The United States and its allies did not simply want an Afghanistan without al-Qaeda. They wanted an Afghan state capable of ensuring al-Qaeda could not return. That meant constructing security forces, supporting a new political system, holding elections, building government institutions and increasingly defending rights that had been brutally restricted under Taliban rule.
One objective created another.
Destroying a terrorist sanctuary required removing the government protecting it. Removing the government created a political vacuum. Preventing that vacuum from producing another terrorist sanctuary required constructing a new state. Protecting that state required an army. Protecting the army required foreign assistance. And suddenly a counterterrorism operation had become a twenty-year state-building project.
This is perhaps the first lesson I would take from Afghanistan: objectives expand much more easily than they contract.
Each additional objective can sound individually reasonable. Why not educate girls? Why not build democratic institutions? Why not create a professional Afghan army? Why not improve healthcare? Why not protect human rights?
The problem appears when those objectives become conditions for leaving.
If America could leave only once Afghanistan was democratic, peaceful, secure, economically sustainable and capable of defeating the Taliban independently, then withdrawal depended on achieving something vastly more ambitious than the reason America entered the country in the first place.
There was never an obvious moment when somebody could announce: Afghanistan is finished.
That made ending the war politically harder every year it continued.
Yet we should also be careful about looking at the Taliban in Kabul today and concluding that everything between 2001 and 2021 therefore disappears.
Millions of Afghan girls attended school during those years. Women entered universities, workplaces, journalism and politics. Afghanistan developed a much more open media environment. Cities changed. Healthcare and education expanded. A generation grew up with expectations that would have been almost impossible under the Taliban government of the late 1990s.
Those twenty years happened.
History does not retroactively delete somebody’s education because the government later collapses.
This distinction matters because political analysis becomes distorted when we judge every policy entirely by its final scene.
Imagine a programme improves millions of lives for twenty years and then collapses. The collapse is a failure. The previous benefits remain real.
But the opposite mistake would be equally serious.
Pointing towards schools, roads or elections cannot obscure the central failure of the American project: the political order built during those twenty years proved incapable of surviving America’s departure.
That is not a small qualification.
It is the difference between development and sustainability.
America demonstrated that enormous external resources could transform parts of Afghan society while those resources remained present. It did not successfully create institutions capable of preserving that transformation independently.
The collapse of the Afghan military makes this particularly obvious. For years, Washington could report how many Afghan soldiers had been trained, how much equipment had been supplied and how many operations Afghan forces conducted. Those figures created the appearance of measurable progress.
Then the test changed.
Could those forces operate when American contractors, intelligence, logistics and air support disappeared?
Too often, the answer was no.
This makes me increasingly sceptical of the way governments measure success. Governments prefer indicators they can count: money spent, troops trained, schools constructed, elections held.
Those are inputs and outputs.
They are not necessarily outcomes.
Training 100,000 soldiers is an output. Creating a military capable of defending the state is an outcome.
Holding an election is an output. Creating a political system that citizens consider legitimate is an outcome.
Building a school is an output. Creating an education system that continues functioning when foreign funding declines is an outcome.
Afghanistan appears to have accumulated enormous numbers of outputs without enough durable outcomes.
And because outputs are easier to count, perhaps policymakers convinced themselves that they were measuring the second when they were actually measuring the first.
There is also a question of opportunity cost that becomes unavoidable after twenty years.
Every dollar spent in Afghanistan could not simultaneously be spent somewhere else. Every diplomat concentrating on Afghanistan had limited time for another problem. Every military deployment consumed resources and political attention.
That does not mean the war was wrong simply because it was expensive. National security costs money.
But the relevant question is not whether Afghanistan produced benefits. It clearly did.
The question is whether those benefits justified the resources required to produce them compared with the alternative uses of those resources.
That is much harder to answer.
The United States did not make its Afghanistan decision in 2001 knowing exactly what the next twenty years would look like. Policymakers operated under uncertainty. September 11 had just happened. Another attack appeared possible. The costs of allowing al-Qaeda to continue operating freely looked enormous.
It would therefore be unfair to judge the original intervention entirely through information available in 2021.
But the same defence becomes weaker as the war continues.
Governments receive new information.
Strategies fail.
Assumptions are tested.
At some point, continuing a policy becomes a new decision rather than simply the consequence of the original one.
That means responsibility for Afghanistan cannot be assigned only to Bush because he invaded, or Biden because he withdrew.
Each administration inherited the war.
Each also made choices about what to do with it.
This is why the political argument in Washington over who “lost Afghanistan” feels increasingly inadequate.
A twenty-year institutional failure involving four presidents, Afghan governments, NATO allies, Pakistan, the Taliban and countless military and political decisions cannot honestly be compressed into a campaign slogan.
Biden deserves scrutiny for the evacuation. The speed of Kabul’s collapse clearly surprised his administration, and vulnerable Afghans should have been evacuated much earlier. The chaos at the airport was not an unavoidable consequence of every possible withdrawal.
Trump’s administration deserves scrutiny for negotiating directly with the Taliban and agreeing to withdraw American forces while reducing the Afghan government’s bargaining position.
Obama deserves scrutiny for escalating the war while simultaneously signalling that the escalation would be temporary.
Bush deserves scrutiny for allowing the original counterterrorism mission to expand into a much broader political project without a clearly achievable endpoint.
Afghan leaders deserve scrutiny for corruption, factionalism and dependence on foreign support.
Pakistan deserves scrutiny for the complicated relationship between elements of its security establishment and the Taliban.
The Taliban is responsible for choosing twenty years of violent insurgency.
These responsibilities are not equal.
But they coexist.
That is probably the second lesson Afghanistan leaves me with: a complex failure does not become more understandable because we identify one person to blame for it.
The more useful question is which mechanisms failed and why.
There is also a much larger question about American power.
The United States possesses military capabilities unmatched by any other country. It can destroy governments with extraordinary speed. It can strike targets thousands of kilometres away. It can deploy forces almost anywhere.
Afghanistan demonstrates the limit of what that power can accomplish politically.
Military force can remove a government.
It cannot automatically create a legitimate replacement.
It can capture territory.
It cannot force people to identify with a state.
It can train soldiers.
It cannot guarantee they believe the government is worth dying for.
It can construct institutions.
It cannot manufacture the informal relationships and political trust that allow institutions to survive.
This does not mean military power is useless.
It means military power is extremely effective at solving some problems and surprisingly poor at solving others.
Perhaps America’s mistake was not believing it was powerful.
America is extraordinarily powerful.
The mistake may have been assuming that one form of power could be converted easily into another.
Military dominance does not automatically become political legitimacy.
Money does not automatically become institutional capacity.
Elections do not automatically become democracy.
And removing a dictator or extremist government does not automatically create a stable replacement.
These distinctions seem obvious after the fact.
They were much less obvious when American forces entered Kabul twenty years ago.
Now Afghanistan belongs to the Taliban again, and the Taliban faces its own version of the same problem. It has proved that it possesses enough military and organisational power to capture the country. It has not yet demonstrated that it can govern it.
The Taliban inherits an economy dependent on foreign assistance, a population much younger and more urban than the Afghanistan it ruled in the 1990s, millions of women with expectations created during the last twenty years, an ISIS-K insurgency and a world deeply suspicious of its intentions.
Winning was one problem.
Government will be another.
Foreign governments now have to decide how to respond. Refusing to recognise the Taliban may preserve diplomatic pressure but cannot make the organisation disappear. Sanctions may constrain Taliban leaders while simultaneously worsening conditions for ordinary Afghans. Humanitarian assistance will be necessary, but delivering it without strengthening the new government will be difficult.
America may have left Afghanistan.
Afghanistan has not left American foreign policy.
That may be the final illusion destroyed by Monday’s departure: withdrawal is an event, but disengagement is a process.
There will still be terrorism concerns, refugees, humanitarian aid, diplomacy and arguments over recognition. There will still be Afghans who worked with foreign governments and were unable to leave. There will still be questions about what happens to women and girls.
The last soldier leaving does not settle any of them.
So what did twenty years achieve?
Enough that saying “nothing” is wrong.
Not enough that calling the intervention a success is remotely convincing.
America succeeded in destroying the al-Qaeda sanctuary that existed in 2001 and prevented another major attack originating from that same structure. It helped create twenty years in which millions of Afghans experienced opportunities unavailable under the previous Taliban government.
But the larger state-building project failed its most important test.
It did not survive without the state-builder.
Perhaps that should change how future interventions are judged from the beginning. Before asking whether an outside power can construct an institution, policymakers should ask whether the society and government receiving that institution will be capable of maintaining it once outside support disappears.
Before counting how many soldiers have been trained, ask whether the military can operate independently.
Before celebrating an election, ask whether the losing side trusts the institutions enough to accept defeat.
Before building a government, ask where its legitimacy will come from.
And before beginning a war, decide what conditions would allow it to end.
On Monday night, the final American aircraft disappeared above Kabul and the Taliban watched from below.
It is tempting to treat that image as the ending.
It isn’t.
It is simply the point at which the United States stops being able to postpone the judgement.
Twenty years is enough time to distinguish what was built from what merely remained standing while America was there to hold it up.
