WikiLeaks Data on Afghanistan Deaths Visualized

This post has been updated.

WikiLeaks unleashed some 76,000 military records this Sunday, but we have yet to see any independent visualizations of the data from outside the three newspapers that received the information ahead of time. We thought we'd get the ball rolling.

We created the following maps from the WikiLeaks war logs. You can see that this war is being conducted on the border with Pakistan, a country accused of being the true locus of the terrorism problem: from its porous border to the support its intelligence service allegedly gives to the Taliban.

Here are the details. Four columns in the giant WikiLeaks spreadsheet describe incidents in which different types of people were were killed in action: Friendly, Host Nation, Civilian, and Enemy.  We mapped the incidents against the geographic coordinates in the logs using GeoCommons, an excellent map masher-upper.

You can see screenshots of each map at the end of the post. Click any of those images and it will take you to the interactive GeoCommons map. Once there, you can click on any of the individual circles to see when and how many were killed in that spot. Larger circles indicate incidents with more deaths. (Here's links if you don't feel like scrolling: CivilianKIA, FriendlyKIA, HostNationKIA, EnemyKIA.)

It's important to note that these maps were generated using the raw logs that WikiLeaks obtained from the military. Some commentators, including the Guardian, have noted that it is difficult to ascertain how accurate some of the data is, and that the kill numbers specifically may be unreliable. Nonetheless, it seemed worthwhile to visualize what the dataset contained, even if we couldn't independently verify all these incidents.

But wait! We want to see more interpretations of the data. If you've created or seen any visualizations, let us know. The giant WikiLeaks spreadsheet came with 32 columns, which we've listed below. We also created a Google spreadsheet of the first 100 rows of data to give people a sense of what's in there.

[Editors' Note (4:15 pm): The current iteration of this visualization is not perfect. We have fixed one problem, which was that small circles -- indicating deaths -- were sometimes being displayed where there should have been none.

Here's what happened: We used the same spreadsheet for each of the four maps. Each map only displayed data from one of four fields: FriendlyKIA, EnemyKIA, HostNationKIA, or CivilianKIA. However, instances in which there were zero deaths were not ignored as we thought they would be, but rather showed up as small circles. Each map is now based on its own set of data, from which all rows in which there were zero deaths have been removed.

The other issue -- and this is a function of our development resources -- is that scale of the circles for each map is different. That makes them difficult to compare with one another. For clarity, @politicalmath, created a legend. The obvious takeaway is that enemy deaths look smaller while friendly and host-country deaths look larger:

Largest Civilian Circle: 67 casualties
Largest Enemies Circle: 181 casualties
Largest Friendly Circle: 16 casualties
Largest Host Circle: 27 casualties

A final thought: Data visualization is tough. It's intensely time consuming and very easy for things to go wrong . As we said, we hope that this provisional first effort stimulates others to come up with other ways of representing this information.]

Update (4:55pm): Commenter Pete Warden offers up a heat map of the data.

Civilian:

View full map

Enemy:

View full map

Friendly:

View full map

Host Nation:

View full map