Yesterday, Wikileaks posted more than 76,000 records of military actions in Afghanistan covering the years 2004 through the end of 2009. At this point, just about everyone in the world has heard about the data set and discussion as to potential damage to military strategy and policy in Afghanistan has already begun. My opinion is, that no matter how damning the records may be, these are records of historical incidents and if we be damned, then we be damned. Whoever leaked this data should get a medal. This is what we as tax-payers in a free democracy demand. Remaining above board and transparent in our military efforts will go a long way to keeping the world straight.
However, news coverage of the Afghanistan conflict vastly overlooks the incredible human cost that has been incurred in the nearly 10 years that the United States military has been there. How can it not? Media coverage of foreign wars is like watching a kung fu movie without subtitles. The average American feels no common ground with the persons depicted, and a war torn, impoverished country is about as far away from typical American life as the moon is. This data set is important because it carefully details the wounded and the killed among the civilian population in frighteningly precise numbers. Our military has kept a careful and meticulous count of a populace who mostly seem to be caught in the crossfire from elements who have little interest in Afghanistan’s future (ourselves included).
Last post, I presented two maps, one representing civilian wounded and the other civilians killed. To create context for what follows, I present them again:
The center of Afghanistan is largely mountainous and uninhabited. From these maps, it’s clear that the Afghan conflict has touched nearly every community across the entire country. Given the extended conflict with the Soviet Union in the 1980’s, the brutally medieval Taliban regime of the 90’s and the ubiuitous nature of the present conflict, it’s hard to believe that the average Afghani knows anything but war and violence at this point.
District Level Civilian Deaths
Afghanistan is divided into 388 administrative districts much like the United States is divided into 50 states. Boiling the point data down to the district level, we can sum the number of civilians killed, divide that number by the size of the surrounding population and generate a map of the percentage of the population that has been killed in a district. Raw counts of deaths are interesting, but we also would like to know how that raw number stands in relation to the size of the surrounding population. We can extend this pseudo-percentage to represent a probability of being killed as a civilian in these district. Most of the killing occurs in the southern part of the country along the Pakistani border as could be seen in the point based map above. In this map, however, we can spot a couple of hot spots in the north, that have experience a disproportionate number of civilian deaths relative to their population size. The populations that reside in the darker areas experience the greatest risk of being killed in a conflict event.
Civilian Casualties in Events Instigated by Friends and Enemies
The data includes not only the numbers killed, but also the instigator of the event, be it “Friend” (the US and other Coalition forces) or “Enemy” (presumably, everyone else). There were 3255 events that involved wounded civilians and 1718 which included civilian deaths. Apparently, the “Enemy” is much better at killing and wounding civilians than the United States, assuming that the party that fired first was responsible for the brunt of the civilian deaths. In overall numbers, enemy led events resulted in more than 7 times the number of wounded as coalition led events, and more than 20 times the number killed. Enemy instigated events resulted in a 1 to 2 ratio of deaths to wounded, where coalition led events were 1 to 5. Comparing the data to the number of operations, I found that that enemy instigated events involving 1 or more civilian casualties resulted in twice as many deaths as US instigated events and three times as many wounded. Of course, the designation of the instigating party lies in the American military’s hand, and could very well be up to interpretation. However, given the incredible amount of data, we can at least assume that it’s correct at least some percentage of the time.
Civilian Wounded and Killed over Time
The United States entered Afghanistan shortly after Sept 11, 2001. The data begins at January 1st, 2004, leaving a three year blank. However, it is clear from a time-series plot of the data that the conflict has only increased since 2004. Below is a time series decomposition of the data from Jan 1st, 2004 to Dec 31st, 2009. The first of the four plots is the raw time series of civilians killed daily by all instigators. All time series are made up of three components: trend, seasonality and noise. Trend is what you would think; it is an increasing or decreasing pattern of change over time. Seasonality is like the weather seasons; it is a predictable pattern of events over a cycle. Noise is just what is says, random events that happen without any discernible pattern. The trend component is the most striking. Not only are civilians at risk for death from armed conflicts, but this pattern of death has only increased over the past 6 years. If you look at the seasonal plot, you can spot 6 fairly distinct humps; deaths increase in the summer months and wane in the winter.
Combatants Wounded and Killed Over Time
That civilians are meaninglessly killed and wounded is unforgivable. However, the costs in human life and welfare do not stop at the civilian body count. Combatants on both sides (all sides?) of the conflict has exhibited massive casualties over the course of the conflict and this will be of particular interest to policy makers on this end of the globe. The benefits and costs of continuing the war will likely hinge on the American body count, which is inherently related to the American electorate.
The spatial distribution of US and coalition forces killed indicates that the highest numbers of casualties are southwest region of the country. Specifically, the brunt of the killing occurs in the Hilmand, Kadahar, Farah and Uruzgam provinces.
Below, I present a time series decomposition of the number of Coalition combatants killed over time, i.e. US and other NATO forces. Trend is evident. The number of coalition combatants being killed has been increasing over the course of the war and shows no evidence of stopping. Seasonal patterns are frighteningly clear. More coalition forces are killed in summer than winter, and the pattern is easily predictable.
A plot of enemy killed vs. coalition members killed indicates that the number of number of enemy combatants killed increases slightly as the body count of coalition forces increases, but that this relationship may be a 1/x sort of relationship (although I couldn’t get the line to plot. later!). Basically, overall, enemy casualties increase with larger numbers of dead coalition combatants, but generally, when the number of enemy combatants killed is very high, coalition members don’t die very much, but when the number of dead coalition members is very high, there are low numbers of combatant casualties. So we play this game, and somehow we meet in the middle and neither gets anywhere.
What’s the conclusion here? War is bad. Period. People die and suffer and the wounded often remain that way for life. From the data here, we can learn at least that. We can also assume that non US and Coalition groups are responsible for not only larger numbers of casualties, but are also better at getting a good amount of civilian blood for their buck. While much of the point of releasing the data is to demonstrate that the US military is evil and hell bent on killing people wherever it can for money and gold, an idea which I may or may not always agree with, I think that we can conclude that the data is even more damning for those who seem to want to take control of the country at the expense of it’s local population. While it is certainly true that the US forces there cause death and destruction merely by being there, I think that everyone can agree that the killing just needs to stop and the solution may not be to simply for the US to pack up, leave and let it be someone else’s problem. The troubles in Afghanistan go much deeper than simply flawed American policy. Staying is obviously not much of an option either. I am no expert on the Afghan war, nor on Afghan history and politics, nor even on military policy. Regardless, the facts stand that the great costs in human welfare and life are increasing and showing little sign of letting up. My conclusion? Stop the killing, please.
Wikileaks today released an incredible database of military reports of actions in Afghanistan from 2004 to the present. The database includes type of action, number of people killed an wounded and whether the attack was initiated by us or someone else. In addition, it includes geocodes (lat/long coordinates) for every action. I wasted absolutely NO TIME in downloading this 77M beast and IMMEDIATELY started to work on it (despite being on the way to bed!). Here is a map of all attacks in Afghanistan from 2004-2010.
You can see that the diary also includes operations outside of Afghanistan.
I’ve only started doing some rudimentary cleaning and data exploration, but was able to produce maps of civilian killed and wounded. It’s amazingly shocking the extent of the civilian carnage throughout Afghanistan. No region has been left without some measure of blood. Civilian wounded ranges from no deaths to 147 deaths from a particularly heinous IED bombing. Killed ranges from none to 67 in a natural disaster in Bamyan Province. Hardly war related, but the US is still involved militarily, apparently. Maps displaying the incredible extent of loss of civilian life and welfare are below. It’s incredible to me that this war is hardly regional, but touches every habitable area of the country.
Total civilians wounded: 9044
Total civilians killed: 3994
I have yet to do some more interesting analysis on the data, but will be sure to devote time to it this week. This is a really exciting resource.
Previously, I had written on the incredibly tragic and needless death of Hiroyuki Muramoto. To continue that thread, I did some digging and found that theCommittee to Protect Journalists has been maintaining a running database of journalists killed since 1992 ostensibly to generate awareness for the importance of journalists and the need for a free and active press. Fortunately, the database is available for download from their website and they have online statistics available. Here is a monthly plot for the entire database:
The bloodiest month in the past 18 years was November of 2009. During that month a record 32 journalists were killed. 30 of these journalists were killed in the Phillipines during the almost not even heard about Maguindanao massacre in which armed men attacked a convoy carrying the 30 men and women and killed them all over a pointless political feud.
Iraq holds the record for country which has seen the most journalistic death, and Iraqi journalists themselves face the highest numbers of deaths. However, our good friends and trading partners Russia, Mexico, India and Pakistan all do an inordinate amount of killing in hopes of stifling free speech and information. Some is inspired by the state, some by private citizens and criminals. 763 men have been killed and 57 women for a total of 820 people in the past 18 years. It may be thought that many journalists are killed from being in the middle of a war zone, but the data indicates that more than 72% of journalists are in fact murdered. A good two thirds of these deaths go unpunished (524 out of 820).
I attempted some fancy time series analysis of the data, but could find nothing of interest, unfortunately. It appears that journalistic deaths, evaluated on their own, are a rather random event and follow no specific patterns.Thus, barring any more interesting analyses, I am presenting these plots of journalist deaths in Iraq, Russia, Mexico and India.
The trouble with getting old is watching heroes and friends slowly die off. This year has seen the loss of two icons. Daniel Schorr was one of them. If we can all grow old like Dan Schorr, then we’ve all won.
I haven’t been able to blog about politics, Malawi or do any pointless data manipulations due to intense house building. Please do not tell my graduate advisor. One day we’ll actually live in the place. For those who are not familiar, back in 2005 we decided to build a house situated in Bridgewater, MI. Now, of course, it is 2010 and we are almost done with it. Here it is from the outside:
and from the inside:
Upon completion, I’ll write a complete history.
A few months ago, I did a short post using data from the FBI demonstrating that recent cries of “spiraling crime” to justify Arizona’s gestapo like immigration law were based on, well, nothing at all. It turns out that others have also gotten on the evidence train and written about the incredible divide between reality and fantasy in US cultural politics.
Center for American Progress has done a quick fact check on Jan Brewer’s outlandish (and false) claims. I can assume that a) Gov. Jan Brewer is a bigot of the first class variety, b) Gov. Brewer is dirtily playing election year politics or c) she’s a total moron. I’m going to vote for all three, but it’s clear that she obviously wants to repel as many new residents and tourists from Arizona as possible.
NPR did an interesting segment on how people tend to believe what they want to believe despite evidence to the contrary, particularly when it comes to political issues. This doesn’t surprise me at all but it’s a good read. As a data nerd, I find it discouraging, however, knowing that presenting evidence, particularly statistical evidence, is largely fruitless and often ignored. If people want to hate, they will.