An Analysis of Collateral Murder

It is tempting to watch this recording once and decide. Almost everyone does. But the footage is an odd kind of evidence: it shows a great deal and proves surprisingly little on its own, and the parts people argue hardest about are usually the parts the camera handles worst. This page goes through it slowly.

What kind of picture this is

An Apache targeting sight is not a camera in the ordinary sense. It is a narrow, high-magnification sensor built to find vehicles and hot engines at long range, and it renders the world in grey. Shapes are blocky. Edges smear. Depth is hard to judge. There is no colour to tell you that a jacket is a press vest.

The aircraft was also flying, so the frame drifts and swings. Anyone who has tried to identify an object in a moving, zoomed, monochrome image knows the feeling: you see what you are already expecting to see. That is not an insult to the crew. It is how vision works, and it is the single most useful thing to hold in mind while watching.

One thing to hold in mind while watching

You see what you are already expecting to see. That is not a criticism of the crew, it is how vision works under pressure, and it is the single most useful idea to carry into the next few minutes of footage.

The first burst: the strongest case for the crew

A group of eight to twelve men are standing at a street corner. Some are carrying things. At least one man is carrying what looks very much like a rocket-propelled grenade launcher, and at one point a man appears to crouch at the corner of a building and point something around it.

In 2007, in that district, a group of men in the street with at least one launcher among them was a target a crew could reasonably act on. Ground troops nearby had been taking fire. The crew reported what they saw, asked for clearance and were given it.

Here is the difficulty. Namir Noor-Eldeen was carrying a camera with a long telephoto lens. Hold one of those on your shoulder and step around a corner to get a shot, and the silhouette you make is genuinely close to a man with a launcher. The crew did not know Reuters staff were there, and there is no evidence anyone was hunting journalists. The argument is about whether the identification met the standard, not about whether it was malicious.

The radio traffic, which is its own problem

A lot of the shock this video produced did not come from the picture. It came from the sound. The crew talk the way people talk when they have been doing something frightening for a long time and have built a shell around it. There is eagerness before firing and satisfaction afterward. At one point, watching a badly wounded man drag himself along the kerb, they say they want him to pick up a weapon so that they are allowed to shoot again.

That line is not a crime. It is, strictly, a man describing the rules he is working under. But it is hard to listen to, and it is the moment most viewers stop being able to give the crew the benefit of the doubt. Whether that reaction is fair is a real question, and it is not the same question as whether the shooting was lawful.

The van: where the defence gets thin

Minutes later a plain van stops beside the wounded man. Two or three men get out and begin lifting him. No weapon is visible in the footage at any point. The crew report that people are collecting bodies and possibly weapons, ask for clearance, and fire into the vehicle.

Three things make this phase different from the first:

1. There is no visible weapon. The first burst at least has a plausible object in frame; this one does not.
2. The people were plainly doing something that reads, on any ordinary view, as rescue.
3. Protecting the wounded, and generally those who collect them, is one of the oldest rules in the law of armed conflict.

The crew could not see that two children were in the front seat, and there is no suggestion they knew. That is a real mitigation of intent and not at all a mitigation of outcome. The full page on this phase takes the legal argument apart properly.

The missiles, and the part nobody examines

About half an hour later the same crew fire three Hellfire missiles into a partly built residential block, reporting armed men going inside. A civilian can be seen walking past moments before. What was actually in that building has never been established in public, and the death toll from it is the main reason no one can give you a firm total for the day.

It is striking how little attention this phase gets compared with the first two. It is also the phase where the rules on proportionality bite hardest, because a missile fired into a structure in a populated street is a different kind of decision from a burst at people in the open.

What the footage cannot settle by itself

The recording cannot tell you what the crew genuinely believed, what the ground unit was reporting at that second, or what was inside the building. For those you need the Army's own investigation and the rest of the written record, which is set out on the sources page.

Reading Collateral Murder honestly is harder than reacting to it

The fair summary is uncomfortable for everyone. The first burst is defensible and may well have been a terrible mistake made in good faith. The van is very hard to defend and has never been properly answered. The building has never really been examined at all. And the voices on the radio, which are what most people actually respond to, are evidence of a culture rather than evidence of a crime.

If you only take one habit from this page, make it this: say which phase you are talking about. Most arguments about this video are two people discussing two different minutes and assuming they disagree. The myths page collects the claims that fall apart once you do that.