One of my earliest posts argued against those who claimed that the U.S was offered Bin Laden on a plate in September 2001. This idea wasn’t particularly popular - it was restricted to elements of the “anti-imperialist left”. The argument handled below is somewhat more pervasive – you may even hear it in a pub.  It is the idea that the U.S (and more broadly, Western) military does not care about civilians when operating in war zones. This, it is claimed amounts to what is in effect a policy of targeting civilians either through wilful actions or gross negligence. 
To deal with some admin: I want to make clear that I am really only talking about U.S actions in the last decade or so. This is for brevity not necessarily because I want to avoid the issue of pre-9/11 actions. The end-notes are elaborations, the sources are contained with the text itself.
Kahl (2007) in International Security gives us a framework for evaluating whether the norm of not killing civilians is being violated:
[The] three types of measures... used here to assess the degree of U.S. military compliance with the norm of non-combatant immunity [are] (1) levels of civilian casualties (an indirect measure); (2) conduct during military operations; and (3) responses to instances of noncompliance (p.10)
Numbers
The first is a relatively easy argument to make and I have made it several times: in Afghanistan, coalition forces are responsible for less than 14% of civilians deaths so far in 2013 (a consistent trend). In Iraq, Kahl estimates that coalition forces were responsible for roughly 10% for the period 2003-2006 (p.11-2). A more recent study by King’s College London for the period 2003-2008 found coalition forces responsible for 12% of civilian casualties [1]. In relation to drone strikes, a meta-analysis of several estimates found that, if you take out the lowest estimate, the civilian toll is between 8% and 17%.  If the world’s most powerful militaries had a wilful policy of killing civilians, they are failing miserably. But the argument that the policy exists by way of negligence may still be correct which is where we use conduct and responses to instances of non-compliance.
Note that the mere killing of civilians is not sufficient to warrant moral condemnation. As will be obvious from the second and third sections below, while the loss of life is regrettable, it often comes about because of militant activities (operating in civilian areas, not wearing uniforms) and the fog of war which can lead to errors. Even these errors should not be morally blameworthy – soldiers must act according to the best evidence and it is simply a fact of war that acting reasonably and making reasonable assessments can lead to the death of civilians (see particularly section three, sub-section one). In essence, these are R v Pagget situations.
Why are the casualties so low? Partly because of improvements in technology which mean that we can avoid civilian casualties. Our weapons are becoming more and more sophisticated which allows for precision. The overwhelming majority of munitions used in Iraq and Afghanistan are precision guided (p.21). But this is not the main reason. There has been a radical change in the internalisation of the rules of engagement:
Military culture is institutionalized, routinized, and reproduced in several ways, including education and training, career incentives, doctrine and war plans, budgetary priorities, procurement programs, and even force structures (p.38)
Military conduct
One of the ways we see this military culture of protecting civilians is through their conduct (other examples of post-operation matters which indicate that this culture exists will be handled in the third section). In the run up to the Iraq War, Kahl points out that
every potential target was vetted by judge advocates for compliance with the Law of War before it got on the list, and then vetted again after the list was complete. Certain operations directed against Saddam Hussein’s regime were deemed off limits if they targeted civilians or risked producing disproportionate damage to civilians and civilian infrastructure. Early in the planning process, the Pentagon drew up “no-strike” lists that included schools, mosques, sensitive cultural sites, hospitals, water treatment facilities, power plants, and other elements of the civilian infrastructure (p.16)
These efforts have even been noted by Human Rights Watch. In their report on the invasion stage of the Iraq War concluded that ‘U.S.-led Coalition forces took precautions to spare civilians and, for the most part, made efforts to uphold their legal obligations’ (p.5) The HRW report goes on to list worries in the targeting of dual-use buildings, particularly media buildings but that remains their main conclusion (p.54).
Even after the invasion stage, U.S military policy attempts to mitigate against civilian casualties. Kahl notes that before targeting, there is a standard ‘collateral damage estimation method’ (CDEM) which involves assessing the target’s military use, alternatives in terms of weaponry and attack, the number of civilians in the area that are likely to be killed and then authorisation from senior personnel. During air phase of combat in Iraq, each target ‘had been vetted by dedicated intelligence officers and reviewed three or four times by judge advocates for potential Law of War violations’ (p.18).
Authorisation is required for operations which lead to ‘high collateral damage.’ Where this was the case, the military took steps to avoid civilian casualties – for example through carrying out operations at times civilian numbers were low. And, unsurprisingly, it worked: both a ‘study by Human Rights Watch and a RAND study commissioned by the U.S. Air Force suggest that there were not significant numbers of civilian casualties from preplanned strikes’ (p.18). This same policy is reflected in U.S rules of engagement and unplanned operations: everything from surveillance drones going ahead to identify civilians to the policy of warning shots. General McChrystal enhanced these rules in Afghanistan. As the Los Angeles Times reported, 'commanders could not fire on buildings or other sites where they had reason to think civilians might be present unless their own forces were in imminent danger of being overrun.’
Further indirect evidence is provided by the fact that the military does not use artillery systems in urban centres. As ‘artillery systems have a large radius of destruction’ firing them would lead to a higher loss of civilian life - which is why it is avoided (p.20). Kahl goes through countless ‘mitigation techniques’ – this covers minor things like avoiding operations during the day, different attack angles to avoid civilian areas, giving pre-warning to civilians etc. etc.
It might be said that all of this is propaganda, some chump academics got fooled into printing the faux procedures and studies into compliance. This would be an error as it’s simply too farfetched to suggest so: we see these procedures play out not just in the numbers, correspondent accounts, rules of engagement, weaponry use and peer reviewed studies – but from accounts where they are simply mentioned as an after-thought.  Here are a few of my favourite examples:
1.     President Obama was presented with several plans for knocking Bin Laden off when his hideout was discovered. One of the plans included using ‘a pair of B-2 bombers to drop “a few dozen 2,000-pound bombs” on the compound.’ As Spencer Ackerman (who has recently deservedly moved to The Guardian) noted ‘the plan was called off, for [inter alia] fear of civilian casualties’ 2.      In 2007, the U.S thought it had located Bin Laden. The New York Times reported that ‘The military set into motion one of the largest strike missions of its kind, with long-range bombers, attack helicopters, artillery and commandos.’ The strike was called off, not only because of doubts about intelligence but because of ‘concerns about civilian casualties from the bombs.’ 3.   This one isn’t from the U.S military but it’s still a favourite of mine because it shows that Seamus Milne is brazen with his sources. Milne, in an attempt to malign the British military linked to a database of UK drone usage in order to show how they “deliver death and destruction in Afghanistan”. That database actually showed constant suspension of military operations where even a single civilian life was in danger.
Given the data, procedures, studies, weaponry use etc., is it likely that these accounts are false? Before moving on to the final section, I want to discount one prominent example. ‘Collateral Murder’ was perhaps one Wikileaks’ first big exposes – and it is claimed to support the meme that I am arguing against - except it supports what I’m saying. Here is what Julian Assange said when confronted with the fact that there were men with AK47s and RPGs in the group on the Colbert Report:
Colbert: What were these men doing in the streets carrying rifles and rocket propelled grenades? Assange: ... The permission to engage was given before the word RPG was ever used.
This is misleading for several reasons. First, in the first few minutes of the video, several men are identified as having AK47s and “weapons” (at 00:27). It was then that permission to engage was sought – and only then. Second, there are good reasons for thinking that the reference to “weapons” was a reference to RPGs. U.S military personnel ask over the radio who requested permission to engage and Hotel Crazyhorse One Eight responded:
I just also wanted to make sure you knew that we had a guy with an RPG cropping round the corner getting ready to fire on your location. That's why we ah, requested permission to engage (15:28).
There had been a fire-fight in the area (something Assange tries to downplay by calling it a “small-arms skirmish”) – why wouldn’t you request permission to engage with these combatants? Civilians were killed but as collateral damage, they were not targeted and it does not show – in any sense – “murder”. It is a video of military personnel following procedures and engaging combatants. How wide-spread is this commitment to not killing civilians? The evidence above indicates that it’s extremely widespread but there is one further which, when taken with all of the information so far given, confirms the argument I’m making:
4 percent of soldiers and 7 percent of Marines reported unnecessarily hitting or kicking a non-combatant, and 5 percent of soldiers and 7 percent of Marines reported a willingness to ignore ROE to accomplish a mission (Kahl, p.33)
These figures are too high but they at least go to show that the procedures outlined above are internalised by more than 93% of military personnel. And sure, it’s a survey of the soldiers - that comes with all the disadvantages of polls you learned in your sociology class. But again, given all this information, isn’t it likely that the gist of what it conveys is true?
Responses to norm violations
This is a good point to point out what I am categorically not saying: I am not saying that the U.S always adheres to the procedures given above. The U.S military falls short of them – while the US military is reluctant to use artillery in residential areas, it has used them (they are still directed at military targets but just placed in residential areas). Human Rights Watch makes an arguable – but still convincing - case that the use of other weapons would be more proportionate (see p.91-2 for a criticism of British military). Members of the military have engaged in interpreting merely suspicious activity as hostile – Kahl notes that the use of a phone has led to engagement (p.25).
But two points should be emphasised. Firstly, not meeting the standards admits that that the standard, the policy is not wanton murder. Indeed, many of these cases can be justified on the basis of the ‘fog of war.’ Second, the response of authorities showed is vital. The responses – listed by Kahl - are varied: that particular problem with artillery was rectified through technological advancements; after a HRW report into the use of checkpoints, procedures were quickly changes (p.27); the restrictions of the rules of engagement (listed above).
(i) Investigations; where there is a civilian death or other ‘escalation of force’ (I think in a non-planned attack) there will usually be an AR 15-6 Investigation. The American Civil Liberties Union has obtained many of these AR 15-6 investigations through freedom of information requests. There are far too many to go through each one of them – but the overwhelming majority show proper practices were employed: many involve civilians ignoring warnings; activity that a reasonable person would think makes them a combatant. Don’t take my word for it – read through a random sample of the summaries given by the ACLU.
The reason I’m not going to go through them in detail (aside from the sheer volume which only covers 2004-6) is because the simple fact these investigations exist is sufficiently indicative of the aforementioned military culture. To repeat: given the low numbers, all the military conduct listed above, the attitudes of military personnel – doesn’t the fact these investigations exist say something? It is too time-consuming and inconsistent with all of the evidence to suggest this is merely lip service.
(ii) Prosecutions: This has proven to be an area the U.S is lacking and the one that poses a challenge to my view. There have, of course, been several prominent examples of prosecutions – but as the Washington Post notes
[In the period between 2003-6 in Iraq] only 39 service members were formally accused in connection with the deaths of 20 Iraqis from 2003 to early this year. Twenty-six of the 39 troops were initially charged with murder, negligent homicide or manslaughter; 12 of them ultimately served prison time for any offense (noted in Kahl, p.35).
Some of this can be partly explained by the simple standards we expect in prosecutions: beyond reasonable doubt and regular problems with prosecuting individuals. A good comparison to make that would mitigate the shock of these figures would be rape statistics in Western countries. But then there are specific problems in the military context. As the New York Times notes, ‘collecting physical evidence and finding witnesses can be difficult because the killings often occur in unstable and dangerous areas, and the cases often come to light only after time has passed.’ In Iraq and Afghanistan, the practices of quick burial mean that autopsies cannot be carried out which places further burdens on prosecuting individuals. Witnesses will often refuse to give testimonies if it involves travelling (despite financial assistance with travelling).

But this cannot be a full explanation. The aforementioned Washington Post article contains a quote from Major serving in the military which isn’t picked up by Kahl: “I think there were many other engagements that should have been [criminally] investigated, definitely.” It should be emphasised that this is problem is not limited to killing civilians but goes to the core of any prosecution in the military [2]. Both Kahl’s paper and the WaPo’s report were published pre-2007, it is not clear how much the situation has changed since then. For the UK’s position see [3].
Conclusion
The idea that the U.S military is killing civilians left, right and centre is wrong. The idea that the U.S. military does not have sufficient regard to the protection of civilians is wrong. The only conclusion that comes from the fact that the numbers are low, the studies, the whole host of measures taken, that the rules of engagement are drummed in, that they spend an inordinate amount of time investigating these events is that the U.S does have sufficient regard to civilians. There are improvements that should be made – not least with prosecutions – but even without them, the argument still stands. Anti-imperialists will try to point to individual examples whilst ignoring the trend outlined above – and it is that trend which tells what U.S policy really is.  
[1] The War Logs released by Wikileaks which revealed civilian casualties (Iraq Body Count estimates that there were 12,438 new civilian deaths revealed) does not change this figure in any real sense because as the New York Times notes ‘most civilians, by far, were killed by other Iraqis.’ (I have addressed the perverse argument that the West is responsible for the actions of these militants elsewhere).
[2] Aside from the rape example in domestic context, the following example should leave no doubt that it's more to do with norms surrounding prosecuting individuals rather than not caring about civilians. The problem with military prosecutions extends to incidents involving U.S military personnel. As the New York Times reported in relation to the awful record of the U.S military in prosecuting servicemen and women for acts of sexual assault:
Though the Defense Department estimates that there were 26,000 sexual assaults in the military last year, fewer than 1 percent resulted in a court-martial conviction. Why? There is a deep institutional bias in the military’s justice system; senior officers can — and often do — intervene to prevent cases from being investigated and prosecuted.
Perhaps somewhat perversely after President Obama tried to – rightfully speak out against this – the situation didn’t improve, in fact his words were used to exonerate individuals. Again, from the New York Times:
... lawyers in dozens of assault cases have argued that Mr. Obama’s words as commander in chief amounted to “unlawful command influence,” tainting trials and creating unfair circumstances for clients as a result. Their motions have had some success.
At Shaw Air Force Base in South Carolina in June, a judge dismissed charges of sexual assault against an Army officer, noting the command influence issue. In Hawaii, a Navy judge ruled last month that two defendants in sexual assault cases, if found guilty, could not be punitively discharged because of Mr. Obama’s remarks
[3] Unfortunately, there isn’t much evidence on the trends of prosecution. The Guardian reported in 2011 that ‘the Royal Military police (RMP) had launched 99 investigations into "incidents in which Afghan civilians have allegedly been killed or wounded by British military personnel in Afghanistan" [for the period 2005-March 2010].’ That there were 99 investigations into British forces despite their lower numbers in Afghanistan is an indictment of the American military prosecution record which managed 39 in Iraq despite their far larger numbers.
In 2009, the Service Prosecuting Authority took over the role of UK military prosecutions but they have ‘refused to say how many prosecutions have been mounted against troops alleged to have killed or wounded civilians.’ Incidentally, the reason they gave for not providing the figures to The Guardian is one I have sympathy for:
When asked by the Guardian to provide this information, the authority's deputy director, Brigadier Philip McEvoy, said: "I am afraid that our dealings with your newspaper do not fill us with the confidence that our response will be fairly represented.
The reason for my sympathy is because I have shown The Guardian misrepresenting documents on more than one occasion (see here, here, here and here). It does not excuse the SPA entirely – I can see no valid reason why these figures should not be in the public domain. In 2012, The Guardian reported that there had been 126 criminal investigations but also reported that
documents obtained by Wikileaks revealed US army reports of 21 occasions in which British troops are said to have shot or bombed Afghan civilians. At least 46 Afghans were killed or wounded in those incidents. Only of one those incidents, in November 2007, led to an investigation by the Royal Military Police
It may be the case that there is insufficient evidence to prosecute or the actions taken were within the rules of engagement but sadly there is not enough information in the public domain to state whether the British do or do not have a problem with military prosecutions.