Monday, July 13, 2009

Drones and sovereignty

Today the Wall Street Journal editorial page continues the fightback against those who call for a cessation of drone strikes in northwest Pakistan.

Several Taliban training camps in the Pakistan hinterland were hit last week by missiles fired from American unmanned aerial vehicles (UAV), or drones, reportedly killing some 20 terrorists. Remarkably, some people think these strikes are a bad idea.

To get a sense of what U.S. drone strikes have accomplished in the past two years, recall the political furor that followed a July 2007 National Intelligence Estimate, which found that al Qaeda had "protected or regenerated key elements of its Homeland [i.e., U.S.] attack capability, including: a safehaven in the Pakistan Federally Administered Tribal Areas (FATA), operational lieutenants, and its top leadership. . . . As a result, we judge that the United States currently is in a heightened threat environment." The media declared we were losing the war.

Less than a year later, then-CIA director Michael Hayden offered a far more upbeat assessment to the Washington Post.

What changed? At least part of the answer is that the U.S. went from carrying out only a handful of drone attacks in 2007 to more than 30 in 2008.

I don't have any problem with the contention that the near-term campaign against Al Qaeda is aided by drone strikes; this is almost certainly true. On the broader question of strikes as a beneficial part of the overall counterinsurgency campaign, I've gone from being anti- to roughly agnostic.

Having said that, this particular editorial sort of misrepresents the case of those who oppose drone strikes on the Pakistani side of the border.

Lord Bingham, until recently Britain's senior law lord, has recently said UAV strikes may be "beyond the pale" and potentially on a par with cluster bombs and landmines. Australian counterinsurgency expert David Kilcullen says "the Predator [drone] strikes have an entirely negative effect on Pakistani stability." He adds, "We should be cutting strikes back pretty substantially."

In both cases, the argument against drones rests on the belief that the attacks cause wide-scale casualties among noncombatants, thereby embittering local populations and losing hearts and minds. If you glean your information from wire reports -- which depend on stringers who are rarely eyewitnesses -- the argument seems almost plausible.

As Andrew Exum and others (notably his co-authors on the CNAS report Triage) have remarked, U.S. military figures on civilian deaths as a result of Predator attacks are almost irrelevant; as much as we talk about "being first with the truth," stories in the Pakistani (and global) media, often fueled by enemy propaganda -- those very "wire reports" of which the Journal is so disdainful -- are generally considered more credible in the region.

Which brings me around to my point, which is this: more than just creating additional enemies through the death of innocents, one of the major down-sides to a continued drone campaign is that it underlines the Pakistani government's inability to exercise sovereignty and control over its own territory. If we recognize that the very most fundamental imperative of our engagement in South Asia is the prevention of regime collapse and security of nuclear weapons in Pakistan, then this concern is magnified.

Proponents of an assassination-from-the-air campaign typically try to gloss over more sophisticated objections by arguing that the deaths of some small number of civilians (certainly smaller than what's reported in the news, I tell you!) are justified by the significance of the targets we've eliminated. But sometimes you lose even when you win.


  1. Well, I'll say again the obvious:

    "While the USA has used a variety of means in its surge strategy, military force has remained central, with the predictable outcome of new civilian lives lost. Airstrikes—the most frequent mode of US military attack involving civilian victims—have continued with regularity throughout the surge, killing 252 civilians in 2006 [and] then—in the surge years—943 in 2007 and 365 in 2008 [by the end of November]."

    That was from Iraq Body Count, an open sourced acknowledgement of our kills, good or bad, and it likely tracks pretty well with our SIGACT reporting.

    So, perhaps the lesson learned from the Surge is that more CAS that kills more innocent people nevertheless somehow isn't the loser everyone thinks it is. And I can assure you that the fish bowl Arabic-language media most certainly had no problem portraying those strikes (does Iraq have an air force?) as solely the responsibility of craven, blood-sucking American neo-con vampires and their drones.

    So, apparently CAS that kills more innocents than ever before is OK in Iraq's "Surge," but not when done in Pakistan because...

    Let's be blunt: It's because some people want doctrine to drive strategy, people who have apriori notions about what should or should not work but no interest in data that show one or the other.

    If you're going to make the case that we should quit the drone strikes, then convince me that the strategic loss from civilian deaths is WORSE than the gains made by destroying or spooking enemy leaders.

    I say this because, frankly, the metrics from the "Surge" would suggest that CAS is fine.


  2. This comment has been removed by the author.

  3. I can only assume that you read the linked story, but not my post. The point that I've made is that there are reasons that drone strikes may be counterproductive beyond an increase in civilian casualties. This strikes me as particularly important in a campaign where one of the primary objectives (even if unstated) is (or ought to be) the bolstering of the government of a neighboring country.

    To try to draw lessons about the utility of drone strikes in ungoverned Pakistan, free of U.S. troops, from the unchallenged utility of CAS in surge-era Iraq, swimming with American forces, seems to me to be a losing effort. Your point here seems to be that killing bad guys is an essential component of effective counterinsurgency, and on this we don't disagree. Necessary, yes, but not sufficient. So shouldn't we be looking at big-picture objectives?

    If U.S. intervention on Pakistani soil inflames nationalistic, tribal, or pseudo-religious passions to the point that anti-Americanism becomes anti-Pakistani-regime-ism, and mobilizes collective action against the Pakistani army and government, and imperils that state, isn't that a bigger problem for us than an al-Qaeda dude in a cave staying alive?

    Maybe it's my conviction that we can never kill enough terrorists or deny enough safe havens to keep SOMEone SOMEwhere from being able to strike catastrophically at American interests, but for me, existential instability in Pakistan is a much, much, much bigger deal than anti- or counter-terrorism in Afghanistan.

    So why do you conflate CAS and drone strikes? They're not the same thing, and you know they're not. Drone strikes are CT, while CAS is (definitionally) supporting fires for troops in contact. How can they be considered even similar, when American troops are unable (thankfully) to engage on Pakistani territory? How can airpower in support of a counterinsurgency be compared even in vague terms to surgical strikes on a country in which U.S. forces are not deployed? Aren't you being slightly disingenuous in your dogmatic effort to re-establish the primacy of violence in counterinsurgency (which no one here really disputes)?

  4. Gulliver, that's a good point about the comparisons of CAS to drone strikes. My opinion on this matter isn't to do it or not, but let's do it better. There were likely too many civilian casualties from air strikes (I'll use the general term) than anyone would like during the Iraq Surge. However, they were fairly minimal considering how much ordinance was dropped. As far as I can tell, this was due to better strike approval procedures.

    If even our lower estimates of drone strike casualties are correct, that is way, way too many for what is supposed to be a precise strike capability. I won't go into the decision making cycle in Iraq due to OPSEC, but suffice it to say that it was rigorous to say the least. The question that needs an answer is not should we strike or not in Pakistan, but: How do we apply Iraq-level rigor there?

  5. CAS isn't merely called during TICs. An air strike famously blew al-Zarqawi, several of his henchmen and a large Iraqi family to chunks in 2006.

    "Support" that's from the air and that's close (which is to say, observed, not always by humans watching on the ground but often with the aid of drones) was delivered in Iraq by both jets AND drones, and it killed MORE civilians in 2007 and 2008 than it did in 2006 and 2005.

    While drone strikes might increase Pakistani anger over a range of issues, does that ire become MORE IMPORTANT to the US than the elimination of potential terror suspects?

    I'm not sure that the Pakistani government would disagree, considering that their leaders apparently have been part of the target selection process, which is to say that we've likely used to missiles to rub out some of their enemies, too.

    Drones are used in Pakistan as a sop to the state of Pakistan, which has an air force and would be nonplussed by watching our ply our trade there.

    There's a wink-wink aspect to all of this: Obviously, the drones not only are tolerated but are quietly applauded by the government of Pakistan. The problem becomes one of plausible deniability -- Pakistan must pretend that they are not quietly applauding them.

    I don't have either the time or the patience for this kabuki theater. I simply request to know whether we gain more strategically by killer or cowing enemy leaders than we lose from pissing off Pakistanis.

    The literature on this actually isn't all that vast. We have an op-ed from Exum and Kilcullen that not only overstates dramatically the numbers of civilian dead culled from the Pakistani press, but contains the outright lie that American drone strikes were a prime cause of the success of Islamist forces in Somalia.

    No one who knows a trifle about SOmalia's endemic civil war would suggest this absurdity, but it's been allowed to stand, I guess because Kilcullen thinks it is so.

    It's not.


  6. "How do we apply Iraq-level rigor there?"

    Very well, Slinger. If we apply "Surge"-like rigor, then we shall kill four times as many civilians in AfPak next year.

    Again, where is that lesson we should've learned in Iraq?

  7. Very well, Slinger. If we apply "Surge"-like rigor, then we shall kill four times as many civilians in AfPak next year.

    Again, where is that lesson we should've learned in Iraq?

    I'm pretty sure I read today that there were zero air strikes in Iraq last month. So maybe the lesson is that there was a period where operational tempo and intensity necessarily increased, with the attendant rise in civilian casualties, in order to come to a place where such violence would no longer be necessary. That's speculation, but it seems consistent with the historical record and the casualty numbers.

    I don't want to be flip about the loss of innocent life, but I think most parties would agree that if a 4x uptick in civilian casualties from air strikes over a one-year period means that U.S. engagement in the region is curtailed on a shorter timeline, they'd be satisfied.

  8. I may be presenting my case based on my brigade and division experience during the Surge in which our air strikes resulted in minimal civilian casualties. I can't speak for MND-B or -N, but considering the many hundreds of thousands of pounds of ordinance we dropped, our reported/known civilian casualties were minuscule. By that I mean somewhere in the order of magnitude of 10 for my brigade after 13 or so months of operations and lots and lots of bombs/artillery shells/mortars/rockets.

    Where does the multiplier of four come from?

  9. In the pre-"Surge" Iraq of 2006, 252 civilians died due to CAS, which included drone strikes. During the first "Surge" year of 2007, the rate of civilian deaths increased 3.7 times, then began to drop off.

    In 2008 after two years of "Surge" operations, about 400 civilians died during CAS attacks, a rate 63 percent above the 2006 toll.

    With all these dead Iraqi citizens and the resulting bad publicity in all the Arabic-language news outlets in Iraq and across the region, how ever did we manage to pacificy the population?

    Perhaps, one might suggest, it was because of all the dead insurgents and their leaders who expired alongside what likely amounted to 2,000 or so civilians between the beginning of the "Surge" and today.

    The reason why I say this is because over the same span of time, we lost an equal number of US troops, about 900 per year, and there was talk by Iraqi leaders about wanting to stop the air strikes, etc, etc, etc.

    So, why was this lesson "learned" in Somalia and apparently AfPak, but no Iraq?

    Let me tell you why: Because Kilcullen's notions about Somalia are invented nonsense, and in AfPak we have robots exclusively doing the killing (well, drones directed by USAF officers at Nellis), which creeps everyone out but which limits the likelihood of anyone in Pakistan protesting too much because, well, it's not like US people are in their territory, only our machines.

    The larger isssue is the one this blog is avoiding: If we agree that the civilian population in Pakistan is the "center of gravity" in the entire debacle, then there's not much we really can do in Afghanistan that's going to woo them or coerce them to do what we want.

    For political reasons, we can strike with drones but not people. We can't invade and occupy NW Pakistan, so that will remain for a very long time a safe harbor for al Qaeda and the Taliban, a border that's really a colonial fiction separating us, but not our enemy, from the "center of gravity."

    In the list of "less bad" decisions we can make, I'm not exactly sure that drone blastings of al Qaeda and Taliban leaders are the worst things we could do.

    One might suggest that it's the only means we have of extending our pace of operations onto the enemy in Pakistan. Controlling the tempo of the war should be the counter-revolutionary's primary goal, and that would include letting our drones chase al Qaeda and Taliban jackasses all over the Hindu Kush.

  10. Sorry, managed to miss the numbers in your first post. I stand corrected with regard to theater standards. However, how much of this was in Baghdad vs the countryside? I'm just guessing here that Afghanistan will resemble the fight in the Belts, not the cities. I would be interested in seeing those numbers.


  11. There was a great deal of sporadic CAS in Basra and East Baghdad in 2008 during the anti-Sadrist campaign, in Mosul against the remnants of AQI and allied organic militias, and throughout the "belts" in 2007.

    If we accept the Exum/Kilcullen approach that it doesn't matter how many civilians are killed but, rather, how many are said to have died at the hands of the evil Americans, then we're left with a paradox: The population in Iraq seems to have died at a drastically higher rate due to CAS in 2007 and 2008 than during the previous years of the occupation, and yet their "anger" seems to have been unimportant enough for us to keep doing it until we didn't need to do so much of it.

    Now, one reason a COINdinista might propose for the uptick in CAS was because the intel gained from "living amongst the people" perhaps gave us more targets. But I'm not buying that because I suspect much of the CAS came from TICs or info gleaned from the same intercepts and tracking we used with direct action operators in 2005 and 2006.

    We had more troops in Iraq in more concentrated areas AND we dropped more air ordnance. We also fired more indirect weapons (even for terrain denial in Diyala) AND killed more civilians, not fewer, even if we agree that the ROE was more restrictive.

    So, why is the lesson learned that we shouldn't fire as much indirect weaponry or CAS?

    It's because we're trying to sell something, a tactic Gallieni would approve.

    We can't just come out and admit that this pop-centric strategy is a load of steamy horseshit. We can't honestly assess just how coercive Kenya and Malaya and Phoenix/CORDs and all the other supposed population-centric "non-kinetic" campaigns have been throughout the 20th century.

    No, we must hold tightly to our aphoristic suggestion that the American way of "COIN" posits a happier, friendlier, less coercive sense of the operations, just as Lyauty or Gallieni's PR gurus highlighted the spreading ink spot instead of all the spreading blood of revolutionaries.

  12. So SNLII, if pop-centric COIN is all "a load of steamy horseshit" and all it really takes is the effective slaughter of the insurgents, then why did the Rhodesian counterinsurgency - with a kill ratio of 10:1 in favour of the counterinsurgents, plenty of strikes (including insertion of ground troops) on sanctuaries in Mozambique and Zambia, and extensive military and financial backing from South Africa - fail so miserably?

    Just so we're clear, I'm not arguing against killing insurgents - I'm arguing it's not the whole story.

    On the subject of drone strikes, I tend to think they may have more utility than not, given recent developments. The only good opinion polling on attitudes towards strikes in the Tribal Agencies from a Pakistani source (a research institute based in Peshawar) show roughly 50/50 division about accuracy, effectiveness, etc.

    More importantly, given that opinion in Pakistan proper has swung pretty hard against the TTP, I'm not sure the political consequences that opponents cite are as acute as they were several months ago.

  13. That's a strawman, MK. No one is suggesting that the most effective counter-revolutionary campaign is the "slaughter of the insurgents."

    Besides, you're burying your lede. Why did the highly coercive, anti-enemy operations in Kitson's Kenya work so well, and those in nearby Rhodesia fail to do so?

    No one is suggesting that it's an either/or syllogism. All I'm saying is that the platitude that suggests drone/CAS strikes that kill increasing numbers of civilians, or perceptions about the same, seems to be disproven by Iraq.

    The twin examples presented by Killcullen and Exum in their policy prescription were Somalia and Pakistan. The Somalia case study was little more than fiction and can't be considered seriously.

    The Pakistan example should be measured against what transpired in Iraq. If Iraq showed that CAS/drone strikes that kill increasing numbers of civilians -- and, along the way, far more insurgents -- don't affect the overall strategy, then why should we assume the opposite in Pakistan?

    Well, I gave you a clue: The big difference between the two is that US-led forces had complete freedom of movement in Iraq, despite al-Maliki's incompetent and corrupt government and their protestations against CAS/drone strikes. With the population in Pakistan's NW Frontier, dubbed the "center of gravity" in the AfPak campaign, we are blocked politically from entering the redoubt of the Taliban.

    So we can't exercise our CAS/drone strikes alongside other facets of our counter-revolutionary program to defeat the insurgency.

    Which should begin to beg some policy questions: If we can't do anything to affect the course of the war beyond very limited but expensive and probably self-defeating options in Afghanistan, what is the point of doing them?

    Well, there really isn't. And quietly the Obama administration has set a two year time period to escalate in Afghanistan and, if that doesn't work, to ditch Kabul.

    I question the arbitrariness of two years. If we know already that there's very little in America's strategic interests to be accomplished through a pop-centric, expensive and lethal campaign in Afghanistan, and indeed the distinct possibility that the US-led occupation (and NOT the drone strikes) is what's spurring the ongoing uprising in Pakistan, should we not consider other ways of managing the problem?

    Unlike in Iraq, I doubt we shall be so fortunate to have Mahdists running around doing our ethnic cleansing, nor a government so ruthlessly vindictive and efficient as Baghdad's Ministry of the Interior around to torture and disappear the bad guys for us. Unlike in the Gulf, the refugees from Afghanistan displaced by our operations or ethnic cleansing tend not to go to largely neutral and thereby confined places (Iran, surely, but not always), but instead east to Pakistan, into the clutches of the Taliban.

    What's truly the "load of steamy horseshit" isn't pop-centric COIN as theory, but rather how it's sold. In reality, the most successful campaigns have been far more kinetic than they're billed, almost invariably very coercive, and largely are recalled by historians as gross and persistent programs designed to deny even the most basic of human rights to the populations the counter-insurgent has chosen to ply his trade.

    In other words, there's almost the assumption that in the triad of terrain, enemy and population, the most successful counter-revolutionaries (Malaya, Kenya, PI) have looked at the population as both terrain and the enemy in order to succeed.

    That's the true story, even if others want to try to convince you otherwise. This is why, despite perhaps the most restrictive ROE in the free world, the US managed to kill increasingly larger numbers of civilians during the first two years of the so-called "Surge" than we did in 2005 and 2006, with the rates dropping only after the pacification set in.

    The numbers don't lie, which is why I put them prominently up top.


  14. Two things, first, I recently came across your blog from reccommendations by both Mike Innes and Tim Stevens, and I think you are off to a wonderful start. I look forward to reading your future posts!

    Second, after reading this post and the ensuing disucssion in the comments, you may be interested in an article in today FP about how the IDF justifies targetted killings: