Category: 
Ten days ago, when I was too overwhelmed with work to comment, Brookings scholar Michael O’Hanlon and his researchers published an op-chart in the New York Times with statistics on how we’re doing in Afpakiraq. Many of the numbers seem to show improvement in our war operations, but under closer examination, they tend to reveal more about O’Hanlon’s credulity than about the headway we’re supposedly making in the War on Terror.
I want to focus on two pairs of stats that I found puzzling on first glance. Here’s the first pair:
Afghan Civilian Deaths From War: April 2008 … 136; April 2009 … 129; April 2010 … 150.
Civilian Deaths Caused Accidentally by NATO (percent of total, based on annual figures): April 2008 … 40; April 2009 … 30; April 2010 … 10–20.
As opposed to civilian deaths caused intentionally by NATO? I kid. Or do I? I asked the scholar himself for clarification, and here was his response:
The first is the % of all Afghan deaths from war that we cause, inadvertently (the resistance causing most of the rest).
So if Le Resistance kills most of the rest (accidentally?) not offed by NATO, there must be a third category of civilian war victim. Call it mystery meat. What is the source of mystery meat? We can only speculate. All we know is that NATO isn’t responsible.
Putting that worry aside, the third statistic in the second line bothers me more. How are we to understand the 10–20 percent figure from April 2010? It amounts to a range between 15 and 30 civilian corpses (i.e. 10–20 percent of 150) caused accidentally by NATO, with most of the 120+ other deaths caused by Le Resistance. But why are we offered a range, rather than a determinate percentage?
Imagine what that might mean in concrete terms: it’s as if there were 30 corpses of Afghan civilians lined up on the ground and NATO’s auditors couldn’t figure out from the evidence whether 15 of them were killed by NATO, Le Resistance, or some mysterious third cause. That doesn’t seem very plausible, does it?
No, it doesn’t. But that means that the categorization of civilian deaths isn’t based on actual investigations. If they were, how could NATO end up with such a wide variance? How could they investigate but not know how 15 Afghan civilians died (or think that NATO may have been involved but not be sure)? No, the more plausible explanation is that the percentages are estimates.
But if they’re mere estimates, why should we trust them? And why do they seem to know exactly how many civilians died (136, 129, and 150 don’t sound like guesses) but settle for rough estimates of how many were killed by NATO?
If you click through to the actual chart, you will see that the statistics are color coated from light, which represents more favorable conditions (for whom?), to dark, which represents the less favorable (for whom?). According to the colors, it counts as an improvement that NATO is causing a smaller percentage of civilian deaths than one year and two years ago in Afghanistan (and so too, presumably, does the corresponding fact that Le Resistance is killing a larger percentage—yes, that’s good news too). But if the accounting of civilian deaths is based on rough estimates and not actual investigations, why trust the numbers or the narrative that the scholar is trying to tell from them?
On to the second pair of troubling statistics:
Number of Aerial Drone Attacks by U.S. (monthly average): April 2008 … 3; April 2009 … 4; April 2010 … 8
Civilian Deaths From Drone Attacks (as percent of all casualties from drones): April 2008 … 30; April 2009 … 20; April 2010 … 5
I found the second line rather obscure, so I asked the scholar for clarification:
The second is the % of all deaths from drone strikes in Pakistan that are of innocent people rather than extremists.
Let me see if I understand this. There are two kinds of people in Pakistan dying from our drone missiles: innocents and extremists. (This is an exhaustive classification, the scholar seems to be saying.) And though we’re averaging more drone attacks now than in previous years, we’re killing more extremists and fewer innocents. Yes, that seems to be an improvement.
No, wait, I was being stupid. You see, the op-chart doesn’t say how many people our drones are killing in Pakistan. So the good news is actually the following: of the unknown number of people our drones are killing in Pakistan, a larger percentage of them are extremists and a smaller percentage of them are innocents than in previous years. Sure, our drones may be killing more innocents than before, but it’s all about the percentages. (Ask any sports fan.)
How do we know this? Presumably, as in Afghanistan, we conduct a thorough investigation of the deaths caused by our drone attacks. Special units are dispatched to the scene with Geiger counters, whose readings determine whether the corpses were previously animated by the souls of innocents or extremists. Oh, wait a minute. Sorry, we figured that we probably don’t investigate the causes of civilian deaths in Afghanistan; we estimate them. So perhaps these percentages from Pakistan are also based on estimates.
But look at the results. We’ve more than doubled the number of drone attacks in Pakistan over the past two years, but we’ve reduced the percentage of innocents killed from 30 to five. That’s really quite a coup for our joystick aces. How did they do it? Presumably they’ve picked up the game quickly and achieved higher and higher top scores. I bet two years ago they were spraying missiles all over God’s creation as they figured out which button-combos did what. However they managed to do it, there’s a story there. Why don’t we know more about how they improved their scores?
In all seriousness, what is the more likely cause of the dramatic reduction in the share of civilian deaths? Improved intelligence and skill? Or a change in the way the military estimates casualties?
It would be nice if we lived in a country in which the top scholars at our elite think tanks asked and answered such questions.
