February 26th, 2014

The Air Force’s Defenders Can’t Fly

I have mostly stayed out of the debate between Robert Farley and his air minded critics. But the recent War on the Rocks response to his piece makes me want to comment on how remarkably weak the responses to his book have been. Let’s be clear that I disagree with Farley’s thesis as well. But since this topic isn’t really that interesting to me, you’ll have to talk to me offline if you want to know why. 

Rather, what interests me is simply how little reflective thought has gone into the actual institutional and social scientific dilemmas this subject raises. 

There are two dominant problems that are clear to any causal observer of inter service architecture and the military institutions literature. 

(1) Institutions have biases, some of which may be general (e.g. Barry Posen’s argument that mil institutions will seek to maximize autonomy and offensive doctrines) and others specific to whatever culture or operational code structures institutional life. 

(2) What is optimal for one actor is not the basis by which that actor’s performance ought to be decided. Aggregate effectiveness, not institutional effectiveness, matters when the actor is part of an overall grouping that must cooperate to achieve some goal. 

This paragraph from the response flat-out denies the former issue and minimizes the second: 

An independent Air Force, drawing on decades of combat experiences in the air domain, is best suited to create air-centric doctrine. United States dominance in the five core Air Force missions would be diluted and dominance in the other domains would be at risk if the other services were forced to absorb the Air Force’s mission and responsibility for air domain doctrine. ….

Each branch of the U.S. military has unique, service-specific priorities based fundamentally on the domain in which it fights. ….By focusing on one domain, each service organizes itself to maximize effectiveness in that domain. Adding additional domain requirements would create an intrinsic conflict in organization.” 

This response is premised on the idea that discrete domains exist and each service — like a firm in economics — maximizes comparative advantage. But while domains are a nice way for DoD budgeteers to differentiate organizational roles and missions, Sam Liles convincingly argues that for the most part they are completely artificial: 

United States doctrine and force structure is built around the domains of air, sea, land, space and now cyber. Domains as defined create cylinders of capability that can be merged and fought within. The domain construct is as much a historical artifact as it is an efficient categorical system. The military force structure to fight within these domains is an air force, army, and navy. The Marine Corps is an expeditionary force between the sea and land (and other tasks as designated). This structure as defined has inherently created a strategic blindness to the capacities, capabilities, and risks of conflict where they meet. This is especially true when dealing with cyberspace.

Hold up your left hand and look at your fingers. Each finger denotes a domain that United States doctrine defines. The palm of your hand represents the joint functions of these domains. When formed into a fist this meshing of national power assets represents a significant amount of power that is bent toward national strategic objectives.

I encourage readers to click through and read his “alien invasion” hypothetical to see just how artificial the service-domain distinctions are. What this implies is that services don’t get a cookie for maximizing effectiveness in their particular area of specialization. If the price of one service achieving its maximal effectiveness in one area is an suboptimal overall military utility then optimal effectiveness isn’t the goal to be sought. 

To see just how ridiculous this idea is, imagine would happen if your car crashed and the manufacturer explained it away as “maximizing rear wheel domain effectiveness.” You want a functioning automobile, not something that accepts an overall task inefficiency because of the importance of having each discrete component optimized to the fullest extent possible. And even is this analogy is generous in that it requires accepting that organizational bias isn’t restraining the organization from maximizing its own discrete area of specialization. 

Here’s how to argue against Farley. 

(1) Either deal with the service culture/aggregate effectiveness arguments or argue that they’ve been misconceptualized. Don’t ignore them altogether. 

(2) Argue that the status quo, while inefficient in many respects, is superior in some way to Farley’s preferred alternative. Perhaps there may be a larger benefit to a 3-service architecture (such as, perhaps, using inter-service competition to reduce the military’s collective action power and thus optimize civilian control? Again, I’m not interested in in this issues so it’s up to you, dear readers). 

Right now, though, the responses to Farley’s book have looked as troubled as the F-35 production cycle. 

February 25th, 2014

A Computational Reading of Neo-Clausewitzian Strategic Theory

For most of my time in undergrad and my terminal MA, the dominant object of discussion revolved around several theme (which are still being debated today, I might add): 

  • Is there a higher “graduate” level of war? (and if so, is counterinsurgency that level?) 
  • Do domain-specific subsets of warfare require their own theory distinct from the general theory of war? And even if the “grammar” of warfare does not fundamentally change underlying logic, how much domain-specific explanation do we need to understand and practice different grammars? 
  • Are there discrete levels of war (tactics, operations, campaigns, strategy, etc)?
  • Does the nature of war change

The neo-Clausewitzian position, epitomized by the writings of Colin S. Gray, is that war has one nature but many characters. Understandings of a particular character that diverges in some fundamental way from underlying logic will only lead a theorist astray. 

That said, as I’ve often discussed with my friend Nick Prime, we often conflate theories of strategy with theories of war. A theory of strategy follows naturally from a theory of war, but concerns a subset of war — the organization of violence for political purpose. Much in the same way game theory can be used to explain the decisions of actors in the international system, but isn’t necessarily a theory of international politics in and of itself. Strategy in war is a heuristic hunch about how we can gain advantage with violence or the threat of violence. 

As a neo-Clausewitzian theorist, Gray has synthesized the writings of many different general and domain-specific military theorists into a unified whole, even if some of them (such as Wylie and Clausewitz) do not mesh together perfectly. Still, the fundamental contention of the neo-Clausewitzian school — which also to some extent approximates ideas of strategy from business theorists like Rumelt and Mintzberg — is that strategy itself does not fully specify its implementation. Strategy is a map that we use to get from here to there, but actually takes the form of tactics. 

This is in part why as a research matter and a theoretical matter strategy is pernicious. We can obviously see the tactics in any given operation, war, or campaign, but the underlying strategy that they served (or didn’t) is more elusive. Does the composition of tactics serve any coherent strategy? Or do we have “tactics without strategy?”

On the idea of operational art and levels of war — the idea of discrete levels of war, besides serving as a justification for political-military authority/budgets (“My bombers are *strategic,* your tanks are just *tactical*”) seems to fit with our uncertainty over whether or not everything we do in an armed contest can fit into a neat strategy/tactics distinction. After all, what we referred to as strategy in the 19th century seems fairly small-scale and evokes images of wars small enough that the fate of nations could be decided in large, decisive Napoleonic battles. Is there not a process of tactical composition that goes on, with differing tactical structures composed of substructure components? A theater of war vs. a particular campaign, a particular campaign vs. an operation, an operation vs. battle, etc etc. Indeed, that is what James J. Schneider argued in his military history of operational art and his metaphor of operational art as the “loose marble” of war. 

On that note, we can just say that all manifestations of strategy are tactics, but did not that Clausewitzian Aleksandr Svechin make some qualitative distinctions between layers?: 

Insofar as we try to achieve positive goals, an operation is an incomparably more economical way of expending military force than local battles. Soldiers are very capable of seeing the difference between operational rationalism and operational shoddiness and are much more eager to sacrifice themselves when they feel that they are on the way to achieving the ultimate goal of a war. Commanders who abuse local battles themselves give evidence of the poverty of their operational talents. What may be completely impossible on a local scale or will require incommensurate sacrifices may be achieved incidentally and much less expensively on an operational scale.

Second, the similarities between various domain-specific “grammars” of war — while obvious in theory — seems to be difficult to sort out in practice. Hence the perennial arguments about whether or not institutional knowledge about guerrilla warfare is perishable, whether or not we need new ideas of war because of cyber warfare, etc etc. On the latter subject, while I disagree with Rosa Brooks about Clausewitz and cyber-attacks, the more I learn about computers and computer science the less confident I am in Thomas Rid’s ideas about cyber attacks, Clausewitz, and warfare. Certainly I don’t think that cyber warfare invalidates the general neo-Clausewitzian idea that Gray expresses. But the specifics —as with debates about insurgency, terrorism, and Clausewitz — matter. 

These are all difficult questions. However, there is one way we can look at this that may resolve some of the tensions. First, I’m talking now of strategy — strategic interactions and our ways to model and explain them — not war itself. Let’s run through some assumptions of the neo-Clausewitzian corpus as a preliminary: 

  1. Strategy is a means of converting violence into political currency.
  2. Making strategy is a process of dialogue and negotiation.
  3. Strategy takes the form of tactics.
  4. War has one nature and many characters, and hence domain-specific theories of strategy can exist as long as they are congruent with a general idea of strategy. 
  5. Strategy is not a plan — it does not follow specify its implementation and evolves as a function of elements of war such as reason of state, enmity and passion, and the play of chance on the battlefield. 

I have already argued elsewhere that differing strategic problems produce variation in models of strategic reasoning. Here I will argue that we can utilize ideas from computation to further explore dimensions of strategy from a neo-Clausewitzian perspective. 

The art of computer programming rests on aspects of computation that are generalized and unchanging. Hence this is why a given algorithmic problem is often usually just a problem instance of a more general issue. Creativity in implementation is needed, but at the end of the day the core problems of computation are structured by formal properties of mathematics and physics. This explains why concepts from discrete and combinatorial mathematics can be literally represented as data structures or algorithms. The goal of computation is to take something as input and return a desired result, and the ingredients involved are structures and operations of various sorts. 

That said, when it comes to the art of crafting programs, we see a nesting effect. A programming language is characterized by primitive data types that are combined into more complex composites. Data structures can hold various forms of data. Operations on data produce programs, and various operations have differing scope, impact, and relationship to other operations. Finally, programs themselves are often the result of modular subprograms that are patched together to create a desired effect. As a machine, the problem of all of the interacting subparts can produce undesired effects (or “friction”). 

How programs are designed and composed is a product of rigorous debate. There are different styles of programming (functional, imperative, object-oriented), differing ideas of design (design patterns, etc), and differing debates about tools. There isn’t necessarily an objective answer to any of these questions, though practitioners can behave as such. 

The core takeaway, though, are the elements of nestedness of components. A program can be thought of as a structure composed of building blocks of various size — though the interaction of all of those building blocks does not flow solely in one direction. And while philosophies, methods, and design styles all differ the core dynamics have a foundation independent of particular style or technology. Most problems a software engineer faces are just instances of more general kinds of puzzles. 

The process of strategy has many similarities. Though at core, a strategy consists of combining tactical operations together to produce a desired result, tactics themselves are produced by combinations of smaller subcomponents. A strategy is modular at core, in that generalized tactics can be swapped and customized to fit particular needs. Land operations differ in application but are structurally similar — no matter the geographic location or political objective. There is not an discrete operational art, though the art of command from the perspective of Bagration and how many levels of composition were needed to produce it vs. Cannae is obviously distinct. 

One might also view the idea of historically recurring types of weapon systems and organizations (the argument that Archer Jones makes) as abstract data types vs. particular data structures. The recurring nature of different operations of war (siege, pursuits, intelligence, attack, defense, logistics, etc) throughout history is also an argument for the idea of an abstract and unchanging idea of strategy composed of abstract and unchanging tactics and tactical subcomponents. This is also something that Jones and Hans Delbruck both tried to emphasize in their particular work, although Jones was more explicit about the idea of strategy as composite.

The implication of this ability to see strategy as a process of combining nested abstract subcomponents is the realization that generalized/abstract strategic problems also re-occur throughout history. If the core problem is using violence to achieve the policy, then particular problems concerning the organization and use of violence will be transitive across history. Intelligence and strategic warning, for example, was just as much a problem in the Bible as it was in Pearl Harbor in 1941. Most problems that practitioners encounter are specific instances of more generalized problems, though practitioners will often be tempted to regard them as unique.  

I should hasten to add that war is neither an art nor a science, it is a practical activity. It also concerns human beings and social organizations, not computer programs. Knowledge of these particular abstractions will not, contra von Bulow or Jomini, provide a plug and play template for analyzing or conducting warfare. A particular context (Clausewitz’s “real” war not “ideal” war) and the will of the adversary will be judge of what is right to do and what is not. 

And while the actual structure and operations of strategy may fit the description I’ve listed above, the way decisions are made about particular compositions and organization is, as Gray and others argue, not just about structured problem-solving but also a function of politics, morality, culture, geography, resources, technology, and the enemy. Empirically, strategy as a whole shouldn’t be assumed to represent the optimal way of solving a political problem — only the way that political intercourse manifests itself in a particular organization of violence. 

That said, from the perspective of a descriptive theory of the instrumental aspect of strategy (e.g. structure and function), the neo-Clausewitzian idea that Gray describes can be further explained by the scheme I have outlined above. It also, while abstract, does not rest on ideas like “eternal human nature” or metaphysics. Rather, it argues that the process of organizing violence to accomplish a political goal and thwarting an adversary’s effort to do the same can be described and decomposed at a high level of formalization and abstraction, and contextual problems, structures, or operations thought to be unique can be shown to be part of a general class of things. Finally, it also leaves room for new structures and ideas (space power, cyberpower, etc) to be both added on as both new abstractions and nested under more general ones. 

Strategic theory is personality and domain-driven — theoretical progress will come when we, as Gray and Lawrence Freedman are currently doing, look at the practice independent of the domain or the personality. 

February 18th, 2014

The Imperial Camel Corps


The Imperial Camel Corps

From the National Army Museum

In 1884 a British Camel Corps was formed for the Gordon Relief Expedition to the Sudan. In 1915 the Camel Brigade (later the Imperial Camel Corps), made up of British, Indian, Australian and New Zealand troops, was raised for…

View Post

Camels on camels 

Reblogged from Think Defence
February 17th, 2014


A Question For Obama’s Syria Critics: What Are the Alternatives?

This weekend, on CNN’s State of the Union with Candy Crowley, John McCain reacted to the failure of the latest round of Syrian peace talks by declaring that the Obama administration’s “policy towards Syria has been an abysmal failure and a disgraceful one.”

It’s a common refrain for the Republican senator, one often accompanied by praise for the Gulf states’ comparatively greater and less cautious support of the Syrian rebels. “Thank God for the Saudis. Thank God for the Qataris,” he said at the Munich Security Conference this year. This time around, McCain said that there are viable options other than U.S. military intervention that Washington is not pursuing in Syria. But he failed to articulate them, with the exception of further boosting the Free Syrian Army (FSA). 

Notwithstanding the question of how the Saudis and Qataris feel about McCain thanking his God for their work, the senator is mistaken in thinking that the core interests of the Gulf states align with America’s. In Syria, as in Iraq, the Saudis see the conflict as a case in which fellow Sunnis have come under siege, which explains the kingdom’s support for hardcore Sunni Islamist fighters throughout the region. Saudi Arabia just announced that it will supply Syrian rebels with mobile anti-aircraft missiles, something that Joint Chiefs of Staff Chairman Martin Dempsey has strongly resisted.

Read more. [Image: Reuters/Goran Tomasevic]

A critic shouldn’t be obligated to propose alternatives. That said, any criticism of a policy also makes an implicit judgement about whether or not it is possible to move higher up on the solution landscape. If there’s no way to do it, then being stuck on a local maxima is still better than sitting at the function’s minima. 

If we’re looking at a problem like Syria, the constraints involved must be taken into account. And those constraints are obvious. There is simply no political appetite for intervention, and had the White House engaged in a discrete intervention its hawkish critics would have likely — as in Somalia — been the first to jump on the administration when things went haywire. The administration’s ability to influence events on the ground is hazy at best. 

The administration is doing what we would expect it to do given what we know about domestic politics — it is splitting the middle. Splitting the middle may or may not be worse than avoiding action, but given the necessity to square the circle between regional partners, hawkish domestic critics, and a public that does not want intervention, it is what we would expect to happen. 

Reblogged from The Atlantic
February 17th, 2014


Ottoman Kard Dagger

  • Dated: 18th-19th century
  • Measurements: overall length in sheath, 31 cm

The hilt of this dagger is made of translucent stone with geometrical motifs inlaid in gold while the ferrule is enameled. The straight, single-edged damascus blade features gold-inlaid decoration. The wooden scabbard comes with velvet covering and silver-inlaid iron mounts. 

Source: Copyright 2014 © Auction Flex

Reblogged from Art of Swords
February 17th, 2014

Anti-Access and Unmanned Deterrence?

This WOTR piece, while well-intentioned, fails to demonstrate any distinction between the supposedly new problem of unmanned systems and Cold War debates about centralized vs. decentralized control over attack authority. 

Indeed, the threat of C4ISR nodes being eliminated in a series of blinding first strikes is precisely why second-strike capability exists, “launch on warning” was such a big debate, and the Triad was built. If anything else one do plenty of things with code to guard against catastrophic events that they can’t do with human grey matter — there is no way to code (for example) exception handling into the human brain. There is no way to defensively program a human brain. 

Thus, this statement: 

The focus of unmanned investment shouldn’t be in capabilities needed to operate independently at the outset or initiation of conflict. Rather, it should be on deterrence, defense, and intelligence collection.

…..could be easily changed to “the focus of [human] investment shouldn’t be in capabilities needed to operate independently at the outset of conflict.” 

There is very little new about this problem. One can find a large number of period articles debating the AirLand Battle/Follow-On Forces Attack set of technologies that note that the very strength of these technologies constituted a threat to peace. Engaging Soviet forces in the “deep” area of operations as opposed to the close-in area required rapid standoff and fire and forget capabilities that critics also warned would make conflict too easy.

But at the end of the day, those critics lost out because of one key problem: political reasons dictated that American forces be committed to defend key allies in place, without relying on an elastic defense.  The Soviets had a local concentration of force, and one way to disrupt that advantage was to hit their command and control nodes and devastate their formations as they marshalled in their assembly areas. Does this sound vaguely familiar to some kind of relevant situation involving “access” and “denial?” 

Second, the author is a bit too sanguine about the idea of unmanned defense and deterrence. 

Conflict deterrence is a significant, if not predominant, consideration for the military, and unmanned systems have a significant role to play in this regard.  A robust unmanned combat network – supported from a stand off distance by manned weapons and ground stations – can convince a prospective adversary that his cost of entry to a conflict is high relative to that of U.S. forces.  Put another way, the fact that U.S. personnel losses are minimized reduces the adversary’s chances of forcing quick capitulation to his desired ends.  One could even presume that the adversary will still invoke a negative international response while simultaneously failing to impose his military and political will – certainly not a recipe for success in armed conflict.

First, the same arguments the author uses against robots as offensive power projection tools hold true for “defensive” and “deterrent” capabilities because — as Colin S. Gray argued — the distinction between offensive and defensive weapons is very hazy in practice. Park an AEGIS-equipped missile boat up along the Persian Gulf scanning Iranian military activity and suddenly things don’t look all that “defensive.” Second, the idea of a completely static defense neglects the need for offensive actions to defend one’s position — and this is how “defensive zones” have a nasty habit of expanding once units come under fire.

Regardless of what one believes about offensive and defensive weapons and the possibility of contextual distinction, the author’s argument against overly scripted response also holds true for deterrent capabilities. What if deterrent capabilities — to be credible — must be able to operate even after they have been cut off from command and control? Then we wind up in the same fix.  

Yes, the opponent’s costs may be raised by the existence of unmanned forces……which might simply motivate them to invest in capabilities capable of bypassing a robotic Maginot line. Such a capability already exists: nuclear weapons. And in terms of conventional counter-value attacks, one might replace the American troops along the DMZ with robots….but can you replace everyone in Seoul with robots? 

There are also counter-force options available to the opponent, such as further incentivizing him to strike your C4ISR nodes — and the prospect for an “algorithmic security dilemma” of sorts also exists. Build a strong defensive unmanned capability, the opponent then builds his own designed to cut through yours without expending his personnel so he can save them for the main effort.

There are particular computational implications for deterrence, strategy, and crisis stability that can be explored in later posts. But the lesson of this is that technology can neither be blamed for the essential problems of deterrence and compellence nor rescue us from having to grapple with them. 

February 10th, 2014

New Infinity Pub

I’m in Infinity Journal talking about strategic theory, ends-based rationality, and a more pluralistic approach to strategic reasoning. You can check out my article, “Beyond Strategy as a Means to an End,” here

February 10th, 2014

Generating Strategies

Re-upping this one for those interested in computational approaches to strategy. Helps make strategies for wargames, which helps with games without fixed rules typically seen in AI approaches to game strategy. 

February 10th, 2014


Japanese Sword Guards (Tsuba)

  1. Edo period, dated early 19th century
  2. Edo period, dated 18th century 
  3. Edo, dated early 19th century (Ito school, Musashi prov)
  4. Edo period, dated mid-19th century (Iwama Mitsuyoshi, Japanese, about 1875)

Source: Copyright 2014 © Bostom Museum of Fine Arts

Reblogged from Art of Swords
February 10th, 2014


Bronze Dagger

  • Dated: circa 1350-1000 B.C.E.
  • Culture: Iranian, Luristan
  • Medium: bronze
  • Measurements: overall length: 16 inches (40.4 cm). Hilt length: 5 inches (12.8 cm). Blade length: 10 3/4 inches (27.6 cm)

Source: Copyright 2014 © LACMA Museum

Reblogged from Art of Swords
Loading tweets...


A blog on states, communities, and organizations in conflict by Adam Elkus.

Portrait photo: Marshal Liu "One-Eyed Dragon" Bocheng