Children in an IDP camp in Dand district, Kandahar province. Thousands of people fled their homes after Taleban offensives in Helmand and Kandahar provinces in October and November 2020, and ANSF counter-offensives. Photo: Javed Tanveer/AFP, 7 January 2021.
A year on from the US-Taleban agreement signed in Doha on 29 February 2020, and after several months of Taleban-government talks, Afghans can only look back at what has been a very violent winter. Civilian casualties in the first quarter of 2021, as tracked by UNAMA, are back up to 2019 levels. That comparison is important because civilian casualties dipped in the first quarter of 2020 due to several factors stemming from the agreement (more on which below).
In comparison with 2020, there have been marked changes in who is responsible for civilian casualties. What UNAMA calls ‘international military forces’, in reality only American since 2015, are now playing so minor a direct role in combat that they no longer appear in UNAMA’s breakdown. The Islamic State Khorasan Province, ISKP (also known as Daesh), is a much weaker group in 2021 and has inflicted far fewer civilian casualties – 91 in 2021, compared to 173 casualties in the first quarter of 2020. Instead, compared to the same period a year ago, there has been a 35 per cent surge in civilian casualties caused by both the Afghan National Security Forces (ANSF), which killed and injured 447 civilians, and the Taleban, who killed and injured 775 civilians. There was also a fifty-fold increase in the number of civilian casualties at the hands of what UNAMA calls “undetermined anti-government elements,” from 14 casualties in the first quarter of 2020 to 225 in 2021. It can be assumed these casualties were largely from unclaimed Taleban attacks, given the group’s domination of the insurgency.
In UNAMA’s breakdown of the ‘incident types’ which caused civilian casualties in the first quarter of 2021, suicide attacks do not feature. The Taleban have largely desisted from carrying these out from late 2019 in the run-up to its agreement with the US, and it seems the weakened ISKP has not been able to mount them as often as it used to. UNAMA also does not list search operations as a cause of civilian casualties. In recent years, Afghan forces associated with the CIA – the Khost Protection Force, Shahin Forces and NDS special forces – have been responsible for so many civilian deaths and injuries during search operations that they featured as a separate category in UNAMA’s breakdown of incident types. These CIA proxies stopped operations after the US-Taleban agreement, although reports that the Khost Protection Force had begun operations again surfaced in December 2020 (see also UNAMA’s annual report and AAN’s analysis of it). Airstrikes, strongly associated not only with the US military but also with the Afghan air force, were also greatly reduced as a cause of civilian casualties in 2021 – from 13 per cent of the total in both 2020 and 2019 to five per cent. Within that overall figure, however, the number of civilian casualties caused by the Afghan air force in the first quarter of 2021 rose by 31 per cent compared to the first three months of 2020.
In the first quarter of 2021, the types of violence that caused the most civilian casualties were ground engagements (up by 56 per cent from 2020), IEDs (up by 117 per cent) and targeted killings. The number of incidents of targeted killings rose by 40 per cent, although the number of casualties was about the same. This would reflect a shift towards shooting victims, which typically result in relatively less harm to passers-by, as is also evident from the fact that the number of those killed was higher than the number of those injured.
A look back at the last year
The patterns of violence have changed over the 15 months, mainly due to the US-Taleban agreement. The charts in this section, which map the conflict in 2020 and 2021, show some of these trends, quarter by quarter: civilian casualties by party (chart 1), civilian casualties by incident type (chart 2) and the number of conflict-displaced persons (chart 3).
Civilian casualties were lower than normal in the first quarter of last year mainly as a result of the Doha agreement: the eight-day ‘reduction in violence’ which ushered in the agreement (intended to be a confidence-building measure and proof that the Taleban leadership could control their forces); a halt to offensive action by the Afghan National Security Forces (ANSF) which followed it and lasted until the summer; the cessation of Taleban mass-casualty suicide and complex attacks in Kabul and other cities, which began in the last months of 2019 in the context of US-Taleban talks and which still largely holds and; an end to offensive airstrikes by the US. Added to these factors, a weakened Daesh also proved to be less deadly to Afghan civilians than in earlier years.
From March 2020 onwards, the Taleban slowly but steadily ratcheted up their violence as they pushed the boundaries of what they could do without attracting US airstrikes. The US always insisted the Taleban had broken a verbal agreement to reduce violence, made at the same time as the written one. The Taleban always denied this, but even so, progressed carefully, not knowing what might attract US reprisals. By contrast, after the Doha agreement, the ANSF shifted from an offensive to an ‘active defensive’ stance, defined as defensive actions which encompass pre-emptive strikes (see AAN reporting). The restraint was an attempt at confidence-building with the Taleban. One former senior MoD official told AAN it had been pushed by the commander of US and NATO forces in Afghanistan, General ‘Scott’ Austin Miller.
In the summer of 2020, these decisions by the warring parties were already having an impact. AAN research on the conflict carried out in June to August 2020 in three provinces found government forces demoralised and frustrated by what had happened since the Doha agreement – the first tranche of US troops withdrawing and the government order, which they felt was forcing them to be passive in the face of the insurgency. In contrast, the research found an emboldened Taleban increasing their pressure on members of the ANSF to defect. It also found that foreign fighters had become more visible.
The ANSF stance was re-set, however, as the former MoD official related: “When the Taleban didn’t change, we changed. We started targeting them then, and expanded [operations].” The Taleban were still driving the violence, but the tempo of ANSF operations also increased. This led to an intensification of the conflict in autumn and early winter, when fighting in Afghanistan usually dies down. UNAMA figures for civilian casualties show a sharp peak in those caused by insurgents in October and November, with casualties caused by pro-government forces showing a gentler rise, peaking in September and October.
One element to note here is the high number of civilians hurt by the ANSF in reckless responses to Taleban fire, when the ANSF or Afghan air force target Taleban who are firing at them from among the civilian population (noted in our reporting in March 2020, October 2020 and UNAMA’s 2020 annual report on civilian casualties; see also AAN’s analysis here).
The Taleban’s calibrated increase in deploying violence during 2020 culminated in major offensives in the south in the autumn that did attract US airstrikes. These strikes were partly responsible for stopping Taleban advances and, in some places, allowing Afghan ground forces to push the insurgents back, The Washington Post reported. The number of conflict-displaced Afghans, which peaked in autumn, was one consequence of those offensives (see chart3).
With the US and (earlier in the year) the ANSF not attacking Taleban positions, people in Taleban-controlled areas have reported that their lives have become more peaceful (see AAN reports in April and October. Those in contested areas and people living close to ANSF bases and posts have often reported increasing violence since the agreement, with Taleban carrying out attacks and the ANSF firing back, including into populated areas. People living in cities have seen a respite from mass casualty ‘spectacular’ attacks, only to see violence return to their streets as the year wore on, albeit in a different form.
The Taleban’s desire not to provoke the US appeared to lie behind the decision to launch a wave of unclaimed assassinations in Kabul and other cities over the winter. The author assumes the bulk of these attacks were carried out by the Taleban, as a sudden surge in criminally, personally or politically motivated attacks seems highly unlikely, and the only other insurgent group, ISKP, has never shown any hesitancy in claiming attacks. While members of the ANSF made up the majority of victims, there were also civilian government employees, judges and lawyers and – new targets for the insurgents – journalists and human rights defenders. Lists of media workers and activists supposedly singled out by the Taleban for assassination magnified the anxiety. These targeted killings were not the mass-casualty, urban attacks which the Taleban have said they agreed not to carry out, but had a similar aim, to terrorise the urban population in general and specific groups in particular. They demonstrated to the population that the government was unable to protect them and were also a sort of pre-emptive strike against independent-minded activists and journalists. They were a signal that such individuals would have no place in any re-constituted Emirate, even if they were critical of both the Taleban and the government.
In the last quarter of 2020, instead of the conflict abating as the weather cooled, as it has always done previously, it worsened. More civilians were killed and injured from October to December last year than in any other quarter of 2020 and more than double the number killed in the same period in 2019. October was the bloodiest month of the year, and November the worst November since UNAMA began its systematic recording in 2009. This aberrant last quarter of 2020 was a sign that the nature of the Afghan conflict had changed in the wake of the US-Taleban agreement. Significantly, it happened while the Taleban and government were conducting talks in Doha.
An analysis of trends in the conflict illustrates how the Doha agreement has already brought significant changes to how the war is being waged. President Biden’s decision, announced on 14 April 2021, to withdraw US troops fully by 11 September, will have even further-reaching ramifications. The date was set to suit a narrow US domestic political agenda, regardless of the consequences of the short timeline for Afghanistan. It is especially problematic for certain key parts of the ANSF which are still reliant on international military or contractor support (more on which below). Biden also decided to make the withdrawal unconditional, thereby discarding any residual leverage US military power might have over Taleban actions. The nations contributing to NATO and its non-combat mission have, not necessarily happily, followed suit.
The repercussions of the US decision to withdraw
Ahead of Biden’s announcement on troop withdrawal, as the American president weighed his options, there appeared to be abatement in some Taleban violence. For example, numbers of targeted killings in Kabul fell away (as seen in AAN’s own tracking of security incidents in the capital). The Taleban, however, were not mollified by the announcement of total and unconditional withdrawal but expressed outrage at the delay, given that the Doha agreement had specified 30 April as the date when international troops would leave. A statement published on 15 April hinted at threats, saying the “breach” of the Doha agreement “opens the way for the Mujahideen of Islamic Emirate to take every necessary countermeasure, hence the American side will be held responsible for all future consequences, and not the Islamic Emirate.” The Taleban may be contemplating attacking foreign military targets again, or as the mujahedin did when Soviet forces left in 1989, think it more in their self-interest to let the foreign troops leave peacefully. More importantly for Afghans, the decision by the US to withdraw all troops unconditionally will likely influence Taleban strategy towards their Afghan enemies this year, with inevitable consequences for the nation’s civilians. The Taleban’s future actions will also be affected by their assessment of whether the US still poses any residual threat. Biden’s words indicate that it will not.
President Biden said the US would continue to monitor and address ‘terrorism threats’, but his language lacked detail. It is clear that the focus will be on defending the US from the threat of international jihadists, and possibly their Taleban hosts, not on protecting Afghans from the Taleban:
We’ll reorganize our counterterrorism capabilities and the substantial assets in the region to prevent reemergence of terrorists — of the threat to our homeland from over the horizon. We’ll hold the Taliban accountable for its commitment not to allow any terrorists to threaten the United States or its allies from Afghan soil. The Afghan government has made that commitment to us as well. And we’ll focus our full attention on the threat we face today.
It is not obvious how the US expects to respond to perceived threats once its forces withdraw. In 2017, AAN considered the possibility of the US moving to a ‘drone-mainly’ mission if it stopped being “comprehensibly embroiled” in Afghanistan but still wanted to retain a “narrow counter-terrorism mission.” At that time, we imagined a mission with far fewer US troops, rather than none, but the possibilities and problems of conducting ‘over-the horizon’ operations discussed in that piece, still hold, only more so if there are no US military personnel left. Afghanistan is landlocked, meaning air or missile strikes would need to be launched from a yet-to-exist base in a neighbouring country (there is no obvious candidate – see discussion of this and more from Asfandyar Mir and Colin P Clarke) or from aircraft carriers faraway in the Gulf. The latter is not impossible but complicated and costly in terms of fuel and time. Targeting would also rely on Afghan government partners for intelligence, complicating issues of accountability if strikes go wrong – as they inevitably would – and civilians are killed.
More importantly than the practicalities, though, is the US intention. Biden’s stance looks like a return to something between Obama’s posture after the withdrawal of ISAF in 2015 when the US did not target the Taleban but did target ISKP and al-Qaeda, and the United State’s pre-2001 position on Afghanistan, when it was disengaged from the civil war, but concerned about the Taleban’s foreign jihadist ‘guests’. President Biden’s message suggests that what will motivate US engagement after the withdrawal will be the actions of al-Qaeda and similar groups in Afghanistan, and then only in so far as they threaten the security of the US and its (non-Afghan) allies, rather than what is happening to Afghanistan or Afghans.
Possible Taleban and ANSF responses to the US withdrawal
Now that the US has set its course on unconditional withdrawal, the Taleban have no need to hold back for fear this might influence US decision-making. Their reticence over deploying certain types of violence could evaporate altogether. AAN’s own monitoring of violence in the capital has already noted an increase in targeted killings since Biden’s announcement, although not yet up to the levels seen over winter. Suicide attacks have been a relatively rare event in the nation’s capital over the past 18 months. There were, however, concerns that the suicide attack on the ANSF convoy in Kabul on 20 April 2021, which wounded seven people, a mixture of civilians and security personnel, and then a huge car bomb targeting a guest house in Pul-e Alam in Logar province as people broke their fast on 30 April, which killed and injured dozens of people and damaged homes and a hospital, might be a herald of the shape of things to come.
As US troops withdraw over the next few months, the ANSF can expect decreasing air support from its powerful ally and finally little or nothing at all. The Taleban’s current position of harassing ANSF, consolidating control of territory and focussing on extorting money from travellers and other citizens could morph into them massing forces and launching offensives on provincial centres. Would a withdrawing America deploy its air force in 2021 as it did to defend Kandahar and Lashkargah in the autumn of 2020 if the Taleban start attacking Afghanistan’s cities in the next few months?
At the same time as emboldening the Taleban, the US and NATO withdrawal leaves the ANSF without international partners on the ground. Since the end of 2014 and the termination of the ISAF mission, NATO’s non-combat follow-on mission, Resolute Support, has been training, assisting and advising the ANSF, while the US has run its separate ‘can-be-combat’ mission, Operation Freedom’s Sentinel. Diminishing troop numbers and Covid-19 have increasingly restricted International forces’ contact on the ground. Even after this last year of much-reduced air support, the US was still offering it in extremis. While it is not a complete deterrent, the threat of US airstrikes has restricted the Taleban from massing to launch major attacks on cities because of the prospect of severe losses in its ranks. The Afghan air force has been conducting aerial strikes, but if its deterrent power proves to be less potent, major Taleban offensives will become more likely, with grave losses of life to both combatants and civilians.
It is noteworthy that the current Afghan air force is the first since 1978 not to deliberately target civilians seen as allied or linked to the opposition as a matter of policy or practice. Attacks on civilians have taken place. A memorable example would be the firing of rockets and heavy machine-gun fire by Afghan air force helicopters into an open-air madrassa graduation ceremony in Kunduz province, where hundreds of men and boys had gathered, on 2 April 2018. The government said the pilots were targeting Taleban, but the children in the crowds would have been clearly visible to them, and at least 81 children, along with civilian adults, were killed or injured in the attack. Even so, the incidence of such attacks by the current Afghan air force have been relatively few, paling in comparison to the deliberate attacks on civilians launched by all governments and armed groups who have had access to air power in the past – PDPA, mujahedin and Taleban. If Taleban attacks become very bloody, current Afghan air force practice may change. An inkling of what might happen already comes in the recklessness of ANSF and Afghan air force responses to Taleban attacks launched from populated areas in ground operations.
One key element helping to keep parts of the ANSF running are contractors. Globally, the US military has come to rely on contractors for even basic operations – logistics, procurement, catering, construction, intelligence gathering and targeting – and makes use of them also to support the ANSF. There are thousands of such contractors of various nationalities paid for by the US currently in Afghanistan. In testimony to a House Armed Services Committee hearing on 20 April, US Central Command commander General Kenneth McKenzie was unequivocal: “Everyone will leave. All U.S. defense contractors will leave as part of the withdrawal” (quoted in the Special Inspector General for Afghanistan Reconstruction (SIGAR) 2021 first-quarter report (p55). Khalilzad seemed less sure when asked about the contractors by the Congressional Foreign Relations Committee on 27 April, although his imprecise speech was confusing (see the transcript and video of his presentation, including questions and answers). After responding to a question as to whether the administration would be requesting substantial financial or material assistance to Afghan forces in the affirmative, he was asked if those funds could be used to pay for American or foreign contractors and replied:
The issue of US contractors staying, that is not part of the agreement that contractors could stay. So the contractors are also leaving, but the Afghans with our help are looking for others to be able to provide that service to them. And we are obviously very sympathetic to them to find alternatives for the needs they have in terms of maintenance and other needs being addressed. 
The continuing need for contractors was already evident from the Special Inspector General for Afghanistan Reconstruction (SIGAR) 2020 fourth-quarter report to Congress that there will be a continuing need for the US to pay for contractors to support the ANSF. The Department of Defence “has taken steps to develop sustainable Afghan air forces,” SIGAR said, but there would be a “need to provide continued contractor logistics support for years.” It also described ongoing work by contractors to provide maintenance services to ANSF ground vehicles and training for ANSF technicians. The aim, SIGAR says, is for the technicians to provide 90 per cent of Afghan National Army (ANP) maintenance and 65 per cent of Afghan National Police (ANP) maintenance by 2023, implying a future role for contractors on the ground even after that date. SIGAR’s latest report, published on 30 April ,was categorical: “Senior military leaders consider [US defense contractors] vital to maintaining Afghan military equipment such as aircraft and vehicles.”
US military veteran and writer, William M Arkin, has provided evidence that the US would continue to fund contractors after September. A day after Biden’s announcement, he listed almost 60 defence and intelligence contractors who continue to advertise positions in Afghanistan, “many in targeting and other direct counter-terrorism task.” The impression given is that the US will be funding military contractors, but whether just for ‘over-the horizon’ support to the US military or assistance also to the ANSF is not clear.
President Biden has said that the US will “continue to support the government of Afghanistan” and “keep providing assistance to the Afghan National Defenses [sic] and Security Forces.” While it seems likely funding will diminish once troops have left and international attention focuses elsewhere, maintaining adequate support for the Afghan state and security forces will be crucial. The Najibullah government survived the withdrawal of Soviet troops in 1987, only to fall three years later when funding from Moscow stopped.
The consequences of the US peace process for the conflict
Significant to the way the US is ending its twenty-year military deployment in Afghanistan is that its troops are leaving after Washington tried and failed to generate an intra-Afghan peace process. Negotiations between the Taleban and the government always seemed an unlikely project, given that neither side was truly interested in participating. US Special Representative for Afghanistan Reconciliation Zalmay Khalilzad, appointed by Trump and kept on under Biden, gave the Taleban much and asked little in return to get the group to sign the 29 February agreement (read the text here). In exchange for agreeing to sit down in negotiations with other “Afghan sides” – the agreement did not even commit the Taleban to speak to the Afghan government – and only required it to make some vague commitments on al-Qaeda. The Taleban, however, won a measure of international legitimacy, the withdrawal of their most dangerous enemy, the United States, from the battlefield, the release of up to 12,000 prisoners held by the government (5,000 were released, 7,000 were due to be) and the lifting of sanctions (not implemented). By contrast, Khalilzad had to force President Ashraf Ghani to make concessions, most notably the freeing of thousands of Taleban prisoners over the summer of 2020.
Khalilzad’s last-ditch attempt, as Reuters described it, to “fast-track an agreement between Taliban insurgents and the Afghan government” by bringing both parties and other key international and regional players together at a conference in Istanbul, has foundered. The Taleban, as they said of themselves, “showed no interest” in the meeting and, after the Biden announcement on troops, said they would not attend. Calling the conference an attempt “to push the Taliban, willingly or unwillingly, to a rushed decision which was needed by America,” the Taleban said it had aimed to “complete a for-show road map before the withdrawal of foreign forces” (reporting by AP here). Reports (see, for example, Tolonews) that the Taleban might have decided to attend the talks in Istanbul if Khalilzad had agreed to the release of 7,000 more prisoners and an end to Taleban blacklisting (both as per the Doha agreement) suggest that they hoped to get more concessions, just for agreeing to speak to their fellow Afghans.
The US, Turkey, Qatar and Pakistan were all reported as trying to convince the Taleban to attend the conference, which had been scheduled for 24 April to 4 May (see here) but failed. It has officially been delayed rather than cancelled, but whether it happens at all is surely in question. Certainly, efforts to generate enthusiasm for the meeting, which is now due to take place after Ramadan, look like flogging a dead horse.
In the wake of the Taleban decision to stay away, the foreign ministers of Turkey, Pakistan and Afghanistan issued a joint statement urging the Taleban to attend the delayed Istanbul conference and pointing to “the urgent need for an immediate ceasefire” to end the violence and “provide a conducive atmosphere” for peace talks. The United Nations, which accepted Khalilzad’s poisoned chalice to take a prominent role in the ‘peace process’ (Secretary-General António Guterres appointed former head of UNAMA, Jean Arnault, as his Personal Envoy on Afghanistan and Regional Issues, on 18 March) and was to have co-convened the Istanbul conference, along with Turkey and Qatar, has also spoken out. The current head of UNAMA, Deborah Lyons, vowed, in a series of tweets that the UN would “ramp up its engagement in support of an Afghan peace & end of the #Afghanistan war” and there would be “no pause in work to support Afghan peace negotiations. The UN is continuing its engagement with both Islamic Republic of #Afghanistan & Taliban representatives, to maintain a focus on peace efforts and the path to a just and durable peace settlement.” Yet even if the Taleban do attend, and the conference finally does go ahead, it is difficult to see how it would make any material difference to the conflict. Why should Istanbul succeed where Doha has not?
The seeds of failure appeared to be sown in Khalilzad’s strategy from the beginning because of the basic unconditionality of his approach. There was no Plan B if the Taleban did not seriously engage in a peace process because the US wanted to withdraw its troops anyway, so why would the Taleban concede anything substantial? This meant that instead of ‘talking and fighting’, a stance well-known during peace negotiations, the Taleban have been able to ‘talk to fight’, that is, use negotiations to secure better conditions for their forces on the battlefield. The Khalilzad project has left the Taleban stronger, with greater international legitimacy and has enhanced the morale of those within the movement who believe they can secure a military victory. Other Talebs who, like many other Afghans in February of last year, hoped the US-Taleban agreement might herald genuine negotiations and an end to the fighting (see our reporting), have been disappointed and sidelined.
On the Kabul side, even after the US had clearly signalled its decision to withdraw troops, the political class did not pull together, either to seriously engage with the Taleban in negotiations or to decide on and pursue an alternative strategy. 20 years of unconditional international support to Afghanistan, with the US and other international players backing Kabul regardless of what Presidents Karzai and Ghani said or did and irrespective of how
corrupt the Afghan state became, has left the Republic badly positioned to act decisively in the face of a crisis.
Since 2001, rewards have fallen into the laps of those in power, not because of any effort or outcome, but solely because of their access to ‘rent’ – unearned foreign income, in the form of aid or military spending. Scholars have described how breaking the relationship between work and reward encourages a ‘rentier mentality’ when those in power feel entitled to rewards without effort. In Afghanistan, one can trace how rent has helped shape a political class unused to making hard decisions (because the income always flows anyway), financially insulated from the population and organised to jockey for power and gain access to rent, rather than to do anything (all of this, of course, with honourable exceptions). The question now is whether the Republic’s political class is capable of the change needed, as international troops withdraw and international support diminishes.
The US withdrawal has begun, General Miller told journalists on 25 April. As this author has recently reported, when the US began their intervention and toppled the Taleban in 2001, there was a real opportunity for lasting peace. Feckless US decisions and oppressive behaviour, both by their military and the CIA as well as their Afghan allies, in the first years of the intervention eventually sparked rebellion. Even then, after the Taleban insurgency gained momentum, there were still opportunities for talks and reconciliation. Khalilzad’s last-ditch attempts to stitch up an end to the conflict with the proposed establishment of a transitional government have only left the Taleban stronger and the government weaker.
Finally, the US is leaving Afghanistan after 20 years, with a negotiated end to the war looking the least likely scenario. More probable is that the Taleban will, sooner or later, move to try to capture territory and attack Afghanistan’s provincial capitals. Yet, the likelihood of the movement taking power through military victory also seems remote. The Taleban show every sign of having underestimated the ANSF. There is every prospect that any Taleban drive to intensify the violence will be resisted, with immense suffering to countless Afghans.
Edited by Aunohita Mojumdar and Roxanna Shapour. Charts by Roger Helms.
 UNAMA gives only a percentage for those killed by undetermined anti-government elements; this works out at just over 14 people.
 According to a senior former Afghan security official “deeply familiar with the country’s counterterrorism programme” who spoke to Associated Press’ Kathy Gannon, the CIA is in the midst of transferring these proxies to the NDS. Two units based in Kunar and Paktia provinces have already been transferred to Afghan control,” Gannon reported him as saying that the plan is to “gradually” hand over the remaining four, which are in Kandahar, Kabul, Khost and Nangrahar.
 Media organisations and their employees had been targeted before, but rarely individual journalists.
 This was not mentioned in the agreement, but either appeared in a verbal agreement with the US and/or, as the Taleban say, was something they voluntarily agreed not to do. For discussion of this, see AAN reporting a month after the agreement was signed.
 Troops may be out even sooner, by 4 July, American Independence Day, the Associated Press reported, on 21 April 2021: “’The Resolute Support headquarters in Kabul are currently considering whether to shorten the withdrawal period,’ said German Defense Ministry spokesman David Helmbold. ‘The 4th of July is now being considered as a pullout date. The ministry informed the (German parliament’s) defense committee about this today.’”
 President Obama pulled back to a reduced mission, in terms both of US soldiers and aims, after the end of NATO’s ISAF and the US Enduring Freedom missions in Afghanistan, after security had officially been handed over to the ANSF at the end of 2014. In 2014, Obama had said: “America’s combat mission will be over” and from 1 January 2015, it would be “training Afghan forces and supporting counterterrorism operations against the remnants of al Qaeda.” The Afghan-US Bilateral Security Agreement (BSA), signed in 2014, was a ‘can-be-combat’ agreement, in that it said “US military operations to defeat al-Qaida and its affiliates may be appropriate” (see discussion and text of the BSA here). In 2015, the US began attacking foreign targets but allowing the Taleban to mass and take territory from government forces. However, the US air force and special forces were increasingly drawn back into the fighting to shore up the ANSF, especially at critical moments, such as the Taleban’s capture of Kunduz in September 2015; the US bombing of the MSF hospital in Kunduz at that time partly stemmed from a confusion between orders and the desire of the US military to protect Afghan comrades (see discussion here). In the aftermath of that disaster, which left more than 40 patients and medical staff dead, Obama made changes to what US forces could do, allowing offensive strikes on the Taleban. President Trump changed the brief further. Secretary of State for Defence Jim Mattis described in testimony to the US Congress on 4 October 2017 how much easier it had become to order airstrikes:
“At one time, sir, we could not help Afghan forces unless they were in extremis” — that is, under direct, urgent threat, Mattis said. “And then eventually that was rescinded, but they still had to be in proximity. They had to be in contact [ie coming under fire from insurgents]. Today, wherever we find them, the terrorists — anyone trying to throw the NATO plan off, trying to attack the Afghan people and the Afghan government — then we can go after them.”
The numbers of sorties flown and weapons released by the US air force reflect these different orders: they dipped in 2015, but then rose each successive year: 2,365 weapons dropped (2014); 947 (2015); 1,337 (2016), 4,361 (2017), 7,362 (2018), 7,423 (2019). Information was not released thereafter.
 At the time of the February 2020 US-Taleban agreement, we wondered if the clauses specifying that the US would defend itself against threats to itself and its allies would cover the ANSF, but it rarely did.
 As the author wrote in 2017: “War crimes reporting from before 2002 shows the sustained use of the Afghan Air Force to target civilians when it was controlled by the PDPA (a junior partner to the Soviet air force which carried out widespread bombing of rural areas in the 1980s) and later by Shura-ye Nazar and Jombesh, who got hold of the remnants of the air force after the fall of the PDPA regime in 1992 and used them to bomb civilian areas during the civil war in Kabul (all factions also used artillery indiscriminately in the capital). The Taleban, who seized some of those even fewer remaining air assets also used them to bomb civilians, including the hospital and the Oxfam office in Yakawlang, in Bamyan province in 2001 and opposition-held parts of Dara-ye Suf district, in Samangan province in 2000, even bombing people who fled the town and tried to seek shelter in the mountains.”
 Numbers are difficult to pin down. On 30 January 2021, SIGAR reported that the Department of Defence had 6,346 US contractors in the last quarter of 2020 (p51) and also that “almost one-fifth of the roughly 27,000 contractors [roughly 5,400 individuals] in Afghanistan were performing security functions, including some 3,000 armed [private security companies] (p161). During Khalilzad’s testimony to Congress on 27 April, US senator Brian Schatz said there were “a total of 17,000 US-government employed contractors” in Afghanistan (see footnote 10).
 Other questions and answers to Khalilzad about contractors were:
01.53 Senator Brian Schatz: We’ve got 2500 troops, about 450 US government staff – we kind of know what will happen with respect to that. I want to talk about the footprint of United States citizens. We’ve got a total of 17,000 US employed contractors. Can you talk to me about what we need to do to protect everyone else? The 2,500 withdrawn is the headline maker, but what is the presence going to be like?
Khalilzad: I believe there would be a corresponding reduction in the number of contractors, as well.
Schatz: Who drives that decision and can you give us a better fidelity on what that is going to look like?
Khalilzad: The withdrawal of contractors who supported the afghan security forces as part of the agreement we have with the Taleban, so that reduction has been there since we signed the agreement last year.
Schatz: So we’ve got 6,150 US Nationals and 16,000-17000 employed by the US. Should we expect a drawdown, and if so, what period of time?
Khalilzad: As far as diplomats are concerned, people working for the embassy, including protecting the embassy, that is one category. We will maintain a strong embassy presence. If your question is with regard to contractors that service the Afghan security forces as part of the agreement, they will draw down as the military forces do.
Schatz: At the same pace?
Khalilzad: Their numbers are larger, but yes, in terms of the contractors. They will be out as security forces are out. Those who provide services for security forces, the 2,500 or so that remain.
 For an in-depth look at the impact of international support on Afghanistan’s political economy, see AAN’s “The Cost of Support to Afghanistan: Considering inequality, poverty and lack of democracy through the ‘rentier state’ lens”.