This is a follow up to my article about the attack on the MSF hospital in Kunduz that I just published on Stats.org.
To support the analysis I went through every US attack on civilian facilities listed in this article and scored them for how strong I judge the evidence to be that they were intentional attacks.
There is huge scope for debate here. The only certainty is that other people evaluating these incidents will form different judgements than I have. I would love to hear criticism from readers and will gladly modify my judgements in response to good arguments.
The first step in the direction of this discussion is to lay bare my judgements so I provide these below in the order the incidents appear in the article by Jon Schwartz.
Bear in mind that “intentional” means that the attackers knew they were hitting a civilian facility.
A score of “0” means I am certain an attack was unintentional. A score of 1 means I am certain it was intentional. (I never actually use 0 or 1 because I’m never that certain.)
Here they are:
Baby milk factory – 0.01
It looks quite clear to me that the attackers thought it was biological weapons facility. (This perception appears to be wrong.)
Air raid shelter – 0.05
The US insists it was a command centre. (This appears to be wrong.)
Sudan pharmaceutical factory – 0.01
The US thought it was an Al Qaeda chemical weapons facility. (This appears to be wrong.)
Train bombing in Serbia – 0.20
A pilot took out a bridge and didn’t seem to notice a train on the bridge. I think there is a chance that the pilot just wanted to complete his mission of taking out the bridge and decided to ignore the train.
In addition, the US released a video of the incident meant to show that although the pilot had the technology to divert the missile the situation developed too fast for him to react. But a later story came out saying that this film was sped up by 3 times.
Serb TV – 0.99
This one was clearly intentional. The US/NATO viewed this TV station as a legitimate target of war on the theory that it was a propaganda outlet. Bill Clinton confirmed this in a press conference.
Chinese Embassy – 0.3
A New York Times investigation concluded that this was down to ridiculous incompetence. The Guardian says they were told by several people in the know that it was deliberate because there was military activity inside the embassy. I’m thinking it was probably an accident but I’m not sure.
Red Cross Complex – 0.01, 0.01
This was bombed once, leading to a lengthy discussion of where all the Red Cross facilities were, and then the US bombed the same complex a second time. These are clearly accidents and also show just how incompetent such operations can be.
Al Jazeera Kabul – 0.10
Information is really sketchy but the fact that the BBC was also hit in the attack kind of suggests accident to me. Still, the Bush people really hated Al Jazeera although I’m not sure how strong this sentiment was so early in the Afghanistan war.
Al Jazeera Baghdad – 0.8
This one seems very likely to have been intentional, given the extent to which the Bush administration was up in arms over Al Jazeera.
Palestine Hotel – 0.03
Seems to have been unintentional. Journalists appear to generally accept this verdict.
Total 2.51 in 11 incidents for an average of 0.23
In my Stats.org article I ventured that this number overestimates the probability that the MSF attack was intentional since hospitals are special so we should go a bit lower than 0.23.
To be clear, I think that based just on the historical record the attack on the MSF hospital was probably unintentional but we should not rule out the possibility that it was intentional.
In addition, we can’t settle the question just from looking at the historical record. There needs to be a proper investigation.
I would add the following important point. Suppose the attack was intentional. In this case it is hard to have a lot of confidence that the US investigation will uncover this fact. Top US officials have already queued up to declare the attack a terrible mistake. In this environment both the investigators and the investigated will surely be loath to allow the investigation to find otherwise.
Finally, the above scoring is for the incidents I used to inform my Stats.org article. However, immediately after the article came out some people drew my attention to some other incidents. So I will soon write a follow up post incorporating these incidents into my analysis.