Thursday, 31 October 2013

Droning On: Amnesty and the efficacy of drone strikes

As the evidence I’ve previously presented shows: drones rarely kill civilians and they are quite effective at reducing terrorism. This shouldn’t be surprising: there is a general trend for Western military action to cause very little civilian loss and the latter proposition is only surprising to people who still believe in the idea of blowback. But two things have made me want to write this post: firstly, the report from Amnesty and secondly, laying out the evidence on effectiveness far more thoroughly than I previously had.

Amnesty International

The debate about drones is fundamentally about effectiveness and civilian causalities –and once those issues have been handled, the bulk of the debate is over – which is why I think the latest report by Amnesty doesn’t change much about that general debate. Glenn Greenwald stated that the report shows that the report shows ‘far more civilians than the USG claims’ have been killed. But thats been pretty much accepted by everyone outside of the US government. A meta-analysis showed the civilian casualty rate to be 8-17% (when you took out the lowest estimate). The main concern is civilian casualties not what the U.S says – and on that count, they are low. In fact, if you look at the appendix of the Amnesty report which lists drone strikes, only two (the ones documented below) are said to have killed civilians only (p.62).

The Amnesty report is not completely irrelevant as it can, however, raise specific incidents in attempt to spark a debate about those incidents. This is despite the general debate being settled. But it is my view that even in this narrower endeavour it fails. The report draws particular attention to two incidents: the drone strike on October 24 2012 which killed Mamana Bibi and the drone strike on 6 July 2012 which killed 18 individuals. Amnesty is, to its credit, aware of the limitations of their conclusions:

Because the US government refuses to provide even basic information on particular strikes, including the reasons for carrying them out, Amnesty International is unable to reach firm conclusions about the context in which the US drone attacks on Mamana Bibi and on the 18 laborers took place (p.8)

But, as we will see, this doesn’t go far enough. Mamana Bibi’s grandson told Amnesty that his grandmother was outside her home ‘gathering okra to cook that evening’ on October 24 2012. It was then that ‘before her family’s eyes, Mamana Bibi was blown into pieces by at least two Hellfire missiles’ (p.18-19). Amnesty states the following about the incident:

Witnesses and family members, interviewed separately and by different research teams at different times, all denied that any militants were anywhere near Mamana Bibi at the time of the attack. Amnesty International’s investigation found no evidence of military or armed group installations or fighters. (p.22)

It goes without saying that an innocent civilian’s death is regrettable. There does, however, seem to be more to the incident than Amnesty describe. The only thing that Amnesty cites that can be construed as a defence of the action is that a Pakistani source states there was signal intelligence which placed a Taliban fighter on a nearby road. Amnesty states that this is not enough to go on because (i) drones would have had time to see that Mamana Bibi was not a militant and (ii) the nearest roads were over 930ft away. I’m not sure how persuasive these responses really are (the first requires an assessment of the evidence and the second doesn’t seem relevant at all) – but that’s not the main thing thats wrong with their description.

After the drone strikes, source after source came out saying that there were militants not only in the area but were killed in the strike. As noted by the New America Foundation, different sources told CNN, AFP, Dawn, Associated Press and The News that militants had been killed. I stress that this (and everything that follows) is not conclusive. There are, however, two main reasons for siding with the New America Foundation over the Amnesty (and, unfortunately, the family of the victim). First, is the general trend of drone strikes and U.S military policy toward civilians. I have already written a full post on that so will only state the conclusion: U.S policy does not target civilians and does everything to minimise the loss of civilian life.

Second is a fact that Amnesty draws attention to. The reliability of witnesses is in doubt particularly on the question of whether militants are in the area (this response is, for reasons that will become apparent, extremely pertinent to the next incident). Amnesty gives the following example of a drone strike on 24 May 2012:

It is virtually impossible for residents to complain to the authorities about armed groups. For example, four foreign fighters and four local Taliban were killed instantly in a village in Esso Khel when a series of drone strikes hit the building they were resting in on the evening of 24 May 2012. While local residents confirmed details of the incident, most refused to confirm the presence of these fighters or whether they had any choice about them residing in their village (p.34)

There are no good reasons for thinking that the situation changes when Amnesty and their researchers enter the town, it remains ‘virtually impossible’ for people to share what really happened. Additionally, we know that militants were operating in Tappi over the long term, using the area as a shelter. Again, I stress this only gives good, not conclusive, reasons for discounting the Amnesty account.

Amnesty’s account of July 6 2012 incident has the same shortfalls. According to Amnesty a ‘group of laborers from Zowi Sidgi village had gathered at a tent after a long day of work in the summer heat’ (p.24). Then multiple drones carried out strikes against the individuals in this tent. After this first strike, individuals gathered to ‘search for survivors.’ A few minutes after this initial strike, more strikes were carried out and killed, according the Amnesty account, 18 people – going on to state:

All of the people who spoke to Amnesty International – each interviewed separately in detail and at different times and locations to corroborate testimony as accurately as possible – were adamant that all of those killed in the strikes were ordinary villagers, not fighters, and that none had engaged in attacks against US or Pakistani forces. Most of the victims worked as labourers (p.26)

In this incident the two reasons why we should discount the Amnesty account converge. According to the New York Times account of the incident:

At least 15 people suspected of being Taliban militants were killed by an American drone strike late Friday in northwestern Pakistan, according to a Pakistani intelligence official and local residents.

This goes to underscore the point about the reliability of the witnesses that Amnesty interviews. This is not just a general accusation about the reliability of witnesses but clear inconsistency. Aside from local residents, as the New America Foundation notes different sources – including ‘local tribesman’ - told AFP, The Guardian and Dawn that militants were killed in this attack. To really reinforce the point – and to give explanation for the second strike - compare these two statements from the Amnesty account and the New York Times:

Local residents said that soon after the strike Taliban militants cordoned off the area around the compound and searched for bodies in the debris (New York Times)

While residents said the area is not under the direct control of any armed group, it is not effectively governed by the Pakistani state either. Residents of Zowi Sidgi said some locals were sympathetic to the Taliban (Amnesty, p.26).

The last factual element of the report is about the effect of drone strikes on the Pakistani population. Amnesty quotes an individual stating the following:

“Local tribal people generally live in fear and stress and feel psychological pressure. They think they could be the target of a drone attack because wrong information might be given to drone operators” (p.31)

This is quite strange because I have only seen one poll asking about the accuracy of drone strikes and the results do not confirm this at all. A poll carried out by the Aryana Institute for Regional Research and Advocacy carried out a poll of those that lived under drone strikes and found the following:

– Do you see drone attacks bringing about fear and terror in the common people? (Yes 45%, No 55%)
– Do you think the drones are accurate in their strikes? (Yes 52%, No 48%)
– Do you think anti-American feelings in the area increased due to drone attacks recently? (Yes 42%, No 58%)

There is no doubt that if you poll across Pakistan, the last result is the opposite. But the quoted results do align with a general trend of other accounts and academic papers of people who actually live under the Taliban and other Al Qaeda affiliated groups (i.e., those who live under drone strikes).  The Economist recently went around the area and found that ‘surprising number of Pakistanis’ (‘many’) support drone strikes. Hussain Nadim found the similar results when asking individuals who actually suffer under the Taliban as he wrote in Foreign Policy:

Although no official polls have been conducted in Balochistan due to the lack of access in the area, I conducted an unofficial survey of 1,500 people from Balochistan, of which only 38% had a negative stance towards the United States.

While these reports should bring some comfort to people who support the use of drones, they should not be used as a determining factor in supporting strikes. If the question of efficacy (handled below) and civilian casualties (handled in these posts and above) is sorted to put it as diplomatically as I can: we must act in self-defence regardless and by ‘we’ I do not just mean the West.

Efficacy of drone strikes

There are two separate positions on drone strikes – the first is that they lead to radicalisation and thus further militancy and the second is that they reduce violence. For breadth, both claims will be handled quantitatively and qualitatively – and in proving the second, the first is further weakened.  

The academic literature and research on drones is fairly conclusive on the first question: drones do not lead to blowback. In a meta-analysis of the available research (specifically) on drones, Walsh (2013) concludes ‘from the existing research is that drone strikes that result in civilian deaths appear to have little relationship with subsequent insurgent violence’ (p.45). His own research is consistent with this view. By ‘plotting the number of civilians killed in drone strikes along with the number of terrorist attacks in Afghanistan and Pakistan,’ Walsh finds that ‘no clear patterns emerge[s]’ (p.32).

The qualitative element of this first point has been handled before and I will not repeat it (needless to say that the poll results quoted above are further evidence of the positions I have previously outlined: people do not suddenly become militant when they perceive themselves to be wronged). But both elements lead to the same conclusion stated by Aaron Y Zelin in Foreign Policy: ‘there is scant evidence that drones strikes have been mobilizing AQC to conduct attacks in response.’

On the question of whether the drone strikes reduce violence, Johnson and Sarbahi find that ‘drone strikes are associated with decreases in both the frequency and the lethality of militant attacks overall and in IED and suicide attacks specifically.’ Walsh own research finds that (i) pre-2011, the relationship between drones strikes and terrorist activity is the converse of what is claimed (thereby proving the first statement to be correct) and (ii) post-2011, drones strikes lead to a decline in terrorist activity:

[Until early 2011...] the pattern is one in which increases in terrorism are followed by more drone strikes. Something similar characterizes the data for Pakistan through 2010. After this date, though, a spike in drone strikes is closely associated with a decline in terrorist activity, suggesting that drones may have had their desired effect.

It’s worth stressing that drones were intensified in the periods where Walsh finds that they had their deterrent effect. Despite these two findings Walsh’s meta-analysis of available research (which includes the aforementioned empirical findings) states that ‘research efforts have not yet produced a consen­sus on how drones influence insurgent organizations’ and that ‘between drone strikes and terrorist attacks in Pakistan are quite variable’ (p.38, 46). This conclusion still supports the first proposition but not the second.

Why has Walsh concluded, contrary to his own findings and Johnson and Sarbahi, that it has little positive effect? Because of a paper by Jaeger and Siddique which finds that drone strikes have different effects on different organisations in Pakistan – including negative effects. My view is that less weight should be given to this paper for two reasons – firstly because as Walsh notes ‘Jaeger and Siddique’s findings on these relationships are not very robust’ (p.38) and secondly because it is inconsistent with the wider academic literature. Indeed, when we look at the academic literature outside of drone use, we see that wiping militants out works.

Instead of just linking to two old posts which show that this is the case, the study undertaken by Johnson which looks at 90 counter-insurgencies since 1970 finds that

when militant leaders are captured or killed militant attacks decrease, terrorist campaigns end sooner, and their outcomes tend to favor the government or third-party country, not the militants

Bearing in mind that ‘an estimated 3,300 al Qaeda, Taliban, and other jihadist operatives in Pakistan and Yemen’ have been killed, the second proposition seems to be a reasonable conclusion on the basis of all the quantitative evidence.[1] The qualitative evidence is just as clear and answers the questions as to why violence in FATA is still so frequent. The International Crisis Group in their report on drones makes the following conclusion:

The main causes for the spread of militancy in FATA are not drone strikes but domestic factors. These include the absence of the state and insecurity due to resulting og political, legal and economic vacuum; and the military's support of, provision of sanctuaries to, and peace deals with militant groups (p.24)

This really shouldn’t be controversial: not only is it the conclusion of our very own MI5, rather surprisingly it is the conclusion of The Guardian. In an editorial they state that

The public discourse in Pakistan suffers from a false binary that the TTP is a function of the drone strikes. The challenge it poses the state is more fundamental than that. Fundamentalism is a product of decades of official complicity, cowardice and appeasement. Sooner or later, Mr Sharif will be forced to realise that. Until then, he is merely kicking the can down the road.

It must have been Seamus Milne’s day off. Incidentally, this is exactly the same as the academic literature on the terrorist campaign in Yemen. Gregory Johnsen has repeatedly stated drones have increased Al Qaeda membership in Yemen on the basis of interviews he conducted. This is wrong not only because of the academic literature of drone strikes and leader incapacitation and the far larger samples of interviewers who found the opposite result but because it ignores the real causes of increased militancy.

Watts and Cillufo (2012) in a brilliant paper note that ‘several phenomena occurring outside Yemen’s borders have been the primary catalyst for AQAP’s emergence.’ First, after the Surge took effect in Iraq, foreign fighters returned home – and for many fighting in Iraq, home was Yemen. Second, Saudi military operations pushed AQ members from Saudi Arabia to Yemen. Third, the ‘intermittent military commitment’ of the Yemeni military has helped Al Qaeda. As Peter Bergen notes when the Yemeni military, in conjunction with the use of drones, conducts military operations Al Qaeda ‘lost all of these gains within about a year.’ Watts and Cillufo conclude the ‘logic behind this assertion [i.e., Johnsen’s] appears horribly backwards.’

Footnotes

[1] One further study which helps support this conclusion (beyond the vast academic literature in my previous posts) is this Journal of Conflict Resolution paper which finds that ‘(1) when the leader of a rebel group is captured or killed, wars are 398 percent more likely to end, (2) conflicts are less likely to end while rebel groups are being led by their founder.’ The reason that this is not quoted in the main body is that this study is specifically about leaders but the logic, it is submitted, is the same. Indeed, even if it weren’t, 51 militant leaders have been knocked out. 

No comments: